Course Hero - We put you ahead of the curve!

You have requested the below document.

Coursehero >> Texas >> Texas >> BENJAMINSM 042

by Copyright Maureen Reindl Benjamins 2004 The Dissertation Committee for Maureen Reindl Benjamins certifies that this is the approved version of the following dissertation: Religion and Preventive Health Care Use in Older Adults Committee: ___________________________ Robert A. Hummer, Supervisor ____________________________ Christopher G. Ellison ____________________________ Paula M. Lantz ____________________________ John Mirowsky ____________________________ Marc A. Musick Religion and Preventive Health Care Use in Older Adults by Maureen Reindl Benjamins, B.A.; M.A. Dissertation Presented to the Faculty of the Graduate School of the University of Texas at Austin in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy The University of Texas at Austin May 2004 Religion and Preventive Health Care Use in Older Adults Publication No. _________ Maureen Reindl Benjamins, Ph.D. The University of Texas at Austin, 2004 Supervisor: Robert A. Hummer Despite the many benefits of preventive services, they are often underutilized, particularly by specific subgroups such as the poor, uninsured, and racial and ethnic minorities. Social factors can figure prominently in these discrepancies by either creating barriers or facilitating use. Because religious beliefs and activities can affect individual lifestyles, worldviews, and motivations, it is possible that religion has an effect on behaviors involving health promotion and disease prevention. However, few studies have directly addressed this relationship. Using data from the Health and Retirement Survey (HRS), the current study examines the relationship between religious attendance, salience, and denomination and six different types of preventive services among adults aged 51-61 in the United States. The results indicate that individuals who attend religious services more frequently use more preventive services. For example, persons who attend once a week exhibit 1.49 iv times the odds of having a cholesterol screening in the past two years compared to persons who never attend services. Denominational differences also exist. In general, Mainline Protestants use more services than Evangelical Protestants, while Catholics, non-affiliated individuals, and members of Other religions are more likely to use fewer services. Using data from the 1998 General Social Survey (GSS), the role of attitudes toward health care as a possible mediator for this relationship is examined. Findings suggest that various aspects of religion are significantly associated with three sets of attitudes: personal trust in one s physician, public confidence in physicians, and attitudes toward the health care system in general. Individuals with higher levels of religious service attendance are generally associated with greater trust and more positive attitudes. Similarly, membership in a Mainline Protestant, Catholic, or Jewish denomination is also associated with higher trust and more positive attitudes compared to Evangelical Protestants. Finally, strength of affiliation is also positively associated with trust in one s physician. This study adds important information to the public health literature concerning factors that influence preventive service use. Furthermore, the findings add to the growing literature on religion and health by illuminating a possible mechanism that may help to explain the association between religion and physical health outcomes. v Table of Contents Chapter One: Introduction ...1 Chapter Two: Literature Review .6 Chapter Three: Theoretical Framework .23 Chapter Four: Religion and Preventive Service Use .53 Chapter Five: Religion and Attitudes Toward Health Care .. . 92 Chapter Six: Discussion and Conclusion . 118 Tables and Figures ... 142 References 176 Vita. ..193 vi Chapter One: Introduction Preventive health services have been shown to reduce premature morbidity, increase longevity, maintain physical functioning, enhance overall quality of life, and contain health care costs (American Cancer Society, 1998; Centers for Disease Control and Prevention, 2002a; Janes et al., 1999; Nichol, Margolis, Wuorenma, and Von Sternberg, 1994; US Department of Health and Human Services, 2000). Furthermore, underutilization of preventive health services and other behavioral factors are associated with approximately one-half of all deaths in recent years in the United States (Institute of Medicine, 2001). Despite this, recommended preventive services such as screenings and immunizations are underutilized, particularly by specific subgroups such as the poor, uninsured, and racial and ethnic minorities (Breen and Kessler, 1994; Breen, Wagener, Brown, Davies, and Ballard-Barbash, 2001; CDC, 1999; Coffield, et al., 2001; Drociuk, 1999; Hayward, Shapiro, Freeman, and Corey, 1988; Janes et al., 1999). With mounting evidence supporting the effectiveness of these services, preventive care has become an increasingly important part of the strategy to improve overall health within our society (USDHHS, 2000). Correspondingly, it is critical that health researchers continue to examine factors that affect preventive care utilization. As noted above, utilization levels vary widely within the U.S. population. In addition to the poor, uninsured, and minorities, individuals who are older, less educated, not married, and without a usual source of care often use fewer preventive services as well. These disparities may reflect the reasons often given for low levels of use, including issues of limited access, information, and motivation (Amonkar, Madhavan, 1 Rosenbluth, and Simon, 1999; Drociuk, 1999). Social factors can figure prominently in these discrepancies by either creating barriers or facilitating use. Understanding the low levels of use, and the potential role of social factors, could be especially important for older adults because the benefits of preventive care may be greater for this age group. This is due to the fact that older adults have a higher incidence of disease and, with increasing life expectancies, now have more time to benefit from the services (Coffield et al., 2001). Detecting existing conditions, rather than preventing new ones, is especially critical for individuals in the later stages of life (Goldberg and Chavin, 1997). Religion is one social factor that may affect preventive service utilization. Because religious beliefs and activities can influence individual lifestyles, worldviews, and motivations, it is possible that religion has an effect on behaviors involving health promotion and disease prevention. In fact, religion has been shown to be associated with other health behaviors such as smoking, drinking, drug use, and diet, as well as with general health care use (for reviews, see Koenig, McCullough, and Larson, 2001; Schiller and Levin, 1988). Thus, it is reasonable to assume that religion may also be associated with preventive service use. Few studies directly address this relationship, but those that do lend support to its existence. Most recently, religious salience and denomination were found to predict the use of six different preventive services in a nationally representative sample of elderly adults (Benjamins and Brown, 2003). Another recent study focused on a wider variety of religion variables, including church attendance and self-rated religiosity; however, these variables exhibited no significant effects on breast cancer screening. (Fox, Pitkin, Paul, 2 Carson, and Duan, 1998). Beyond these two studies, the majority of the prior research in this area focuses on denominational differences in female preventive service utilization. Most findings show that breast and cervical cancer screening utilization rates differ by religious affiliation (Miller, Norcross, and Bass, 1980; Miller and Champion, 1993; Murray and McMillan, 1993; Naguib, Geiser, and Comstock, 1968; Yi, 1994; Yi, 1998). This evidence is suggestive, but the contradictions in findings and lack of consistent measures indicate that other studies are clearly needed. In addition, most of the studies have methodological problems that limit their impact. For example, with the exception of the study by Benjamins and Brown (2003), all of the above-cited studies use cross-sectional data and non-representative samples with limited preventive service outcomes for females only. Furthermore, many test only bivariate relationships. Finally, while the Benjamins and Brown (2003) study made many improvements on past research, the sample was limited to adults over 70 years of age and one of the most important measures of religious involvement, service attendance, was not included. The present study attempts to fill these important gaps in the literature by considering the effects of religious attendance, salience, and denomination on six different types of preventive services within a nationally representative sample of preretirement aged adults. Services aimed at both disease prevention (flu shots) and disease detection (cholesterol screening, mammograms, breast exams, Pap smears, and prostate exams) are examined. This study adds important information to the public health literature concerning factors that influence preventive service use. This type of knowledge could help lead to more effective interventions, reduced sociodemographic 3 disparities, and an overall increase in preventive care utilization. The research goals of this dissertation are to examine the relationship between religion and preventive service use among older adults and to explore the role of several possible mediators. In particular, the role of attitudes toward health care providers and the health care system are considered. These attitudes are conceptualized as a potentially important mechanism because they have been shown to influence health care use and may differ by religious beliefs and level of involvement. To develop these ideas more clearly, I examine a broad range of religious predictors and preventive service outcomes. The study includes both genders as well as three race/ethnic groups to facilitate group comparisons and test for possible interactions. Differences by level of education and income are also examined. In order to accomplish these goals, the following analyses use two data sets that are unique in their ability to address different aspects of the research goals. To investigate the relationship between religion and preventive service use, data from the Health and Retirement Survey (HRS) are used. This data set enables me to look at three different measures of religion, three race/ethnic groups, and six preventive services within a nationally representative, longitudinal sample of older adults. To look at the relationship between religion and attitudes toward the health care system, I use the General Social Survey (GSS). This cross-sectional data set includes both religion and health attitude variables that allow me to examine the possible mediating role of these beliefs. 4 Organization of the Dissertation This dissertation is organized into the following chapters. Chapter Two summarizes the literature concerning religion and health care use. Studies focusing on religion and preventive health care use are reviewed in greater depth. Following this, other predictors of preventive health care use are summarized. The theoretical background and conceptual diagram of the study are included in Chapter Three. Rationale for the control variables and explanations for proposed mediators and moderators are provided in this chapter. Chapter Four provides detailed information on the data and methods used in the analyses of religion and preventive service use, and provides results for that portion of the work. Chapter Five presents the analyses concerning the potential mediating role of attitudes toward the health care system. Finally, Chapter 6 summarizes the findings from both of the analysis chapters and discusses how they relate to each other. The importance of the findings within the literature and the public health field is reviewed. Finally, study limitations and future work still needed in this area are discussed. 5 Chapter Two: Literature Review This chapter provides an overview of studies that are relevant to the current research questions. To begin, general background information on the relationship between religion and health is presented. Then, previous findings relating religion and health care use are summarized. Special attention is paid to studies that investigate the relationship between religion and preventive service use. Finally, information on other predictors of preventive service use is summarized. Religion and Health Individuals in the public health and medical communities have been noticeably reluctant to consider a connection between religion and health. Nevertheless, over a thousand studies, from a wide variety of disciplines, have examined this relationship and found fairly consistent results, with much of the empirical work published in the last two decades (Koenig, McCullough, and Larson, 2001). Most studies find a positive relationship in which higher levels of religious involvement are associated with better mental and physical health outcomes. Many studies report sizable effects of religion; for example, individuals who attend church more than once a week live an average of seven years longer than those who never attend (Hummer, Rogers, Nam, and Ellison, 1999). Many explanations have been posited to account for these relationships, including selectivity, social support, psychological resources, and health behaviors (for a review, see Ellison and Levin, 1998). Within the realm of health behavior, one area that has not received much attention is differences in levels of health care use. More specifically, religious beliefs and activities may influence how individuals use health care services 6 and this, in turn, may have a positive effect on a variety of health outcomes. Religion and General Health Care Use Within the religion and health literature, religious influences on overall health care have received more attention than those related to preventive care, most likely because only a small proportion of the national health expenditure is spent on these types of services. More specifically, only three to six percent of the total national expenditure on health care (which was $1,299.5 billion in 2000) is spent on prevention, while the rest is spent on hospital care, physician services, nursing homes, and other services (CDC, 1992; Levin, et al., 2002). Correspondingly, an extensive range of factors that may influence general health care utilization have been examined, including religion. Numerous studies have investigated the link between religion and a wide range of health care services, including mental health services, primary care, maternal and child care, dental care, and hospitalization. Two reviews of the literature have concluded that most of these studies find significant relationships (Koenig, McCullough, and Larson, 2001; Schiller and Levin, 1988). One of the most consistent findings is that Jewish individuals have significantly higher rates of utilization compared to members of other denominations and the non-affiliated, though this difference is often small (Linn, 1967; Mechanic, 1963; Scheff, 1966; Segal, Weiss, and Sokol; 1965; Solon, 1966; Wan and Soifer, 1974; Wan and Yates, 1975). Another fairly consistent finding is that individuals who have high levels of religious involvement use fewer services (Frankel and Hewitt, 1994; Linn, 1967). Self-rated religiosity has not received as much attention as other aspects of religion, but has also been found to be significantly correlated with a higher 7 number of physician visits in a study of Mexican Americans (Levin and Markides, 1985). Despite the breadth of this literature, few consistent patterns emerge in the findings and many methodological limitations are apparent. For example, the studies most commonly operationalize religion as religious affiliation (generally limited to Catholic, Protestant, or Jewish). It also must be noted that most of the studies did not control for potential selected mediating variables. Furthermore, the samples used often limit the representativeness of the results; many of the studies are also quite dated. Perhaps most importantly, no study has focused explicitly on the relationship between religion and general health care use (Koenig, McCullough, and Larson, 2001). Previous Studies As previously noted, the relationship between religion and preventive service use has not received much attention from either the medical community or social scientists. The findings of the few studies found are summarized in more depth here. It must be noted that only two studies have focused specifically on religion and preventive service use (Benjamins and Brown, 2003; Fox et al., 1998). These studies were also the only ones to offer any explanations for why this relationship should (or does) exist. The remainder of the studies simply included religion as a control variable or as one of the many possible predictors of utilization. One of the most recent studies examined the influence of religious salience and denomination on six different types of preventive services among adults aged 70 years and older (Benjamins and Brown, 2003). The results showed that individuals who report high levels of religious salience are more likely to use preventive services compared to 8 those with lower levels of salience. Similarly, compared to non-affiliated individuals, those claiming membership in some religious organization are more likely to report the use of preventive services. Of the denominations included in the study, Judaism was most significantly associated with increased service utilization. Although these findings are revealing, one of the most commonly used measures of religion, religious service attendance, was not included and the sample was limited to elderly adults. Fortunately, the possible influence of religious attendance (and other measures of religion) on preventive service utilization has been studied elsewhere. Fox et al. (1998) used a church-based sample of Los Angeles women to analyze the effects of the level of involvement in church activities, church attendance, partner s attendance, self-rated religiosity, presence of a church health committee, and importance of breast cancer to that committee on breast cancer screening rates. The findings show that none of these diverse religious characteristics were significant predictors of this type of screening; however, this study is based on a very limited sample. An older study also examined the effect of religious attendance on preventive service use (Naguib, Geiser, and Comstock, 1968). This study found that women who attended church once a week or more were more likely to use self-use cervical cancer screening pipettes than women who never attended religious services. Several prior studies also provide evidence that preventive service utilization varies by religious denomination. Using a convenience sample of women from four churches in a Midwestern city, Miller and Champion (1993) found that Catholic women were nearly six times as likely to have a mammogram in the past three years as 9 Protestant women. In addition, Yi (1994) found that, of a sample of Vietnamese women in Western Massachusetts, those who were Catholic or Protestant reported more cervical cancer screening than those who were Buddhist (using zero-order correlations). In a similar study using Vietnamese college students, Yi (1998) confirmed the earlier findings that Catholics report higher levels of Pap smear utilization than do Buddhists. She also found that non-affiliated students have relatively higher levels of use, while Protestant students report the lowest levels of use (Yi, 1998). While Catholics are often shown to use more services, Murray and McMillan (1993) found that women in Northern Ireland were more likely to report breast self-examinations if they were members of the Church of Ireland compared to Roman Catholics. Similarly, Miller, Norcross, and Bass (1980) found that Protestants were more likely to report breast self-examinations than Catholics. Although not supported by the Yi study (1998), it is possible that being involved with a religious organization, regardless of the denomination, also predicts preventive service use. For example, Fox et al. (1998) found that breast cancer screening rates from a church-based sample were significantly higher than those from a community sample in the same county. Similarly, the findings of Naguib, Geiser, and Comstock (1968) show that women who were members of a religious or social organization were more likely to utilize self-use pipettes for cervical cancer screening, while atheists or women not affiliated with a religious organization were less likely to use the pipettes. Limitations of Past Studies Prior studies have shown that use of preventive services differs by religious denomination, salience, and possibly by level of religious attendance, but the samples 10 used were limited to certain denominational, gender, and age groups. Moreover, the majority of these studies examined a narrow range of religion measures and preventive service outcomes. For example, no studies have investigated detailed denominational differences in preventive service utilization. Furthermore, few studies have attempted to discover the factors that may mediate or moderate this overall relationship. While these previous studies, and those on religion and general health care use, have provided a foundation for future studies, many of these questions have not yet been subject to careful scientific scrutiny (Koenig, McCullough, and Larson, 2001: 397). Specifically, only one study has used a nationally representative data set to examine the effect of multiple facets of religion on a range of preventive services, but this study was limited to adults over 70 years of age (Benjamins and Brown, 2003). Other Predictors of Preventive Health Care Use Despite this lack of attention to religion and preventive service use, other factors that influence this type of health care service have been well-studied. Typically, these studies focus on one type of preventive service, on a group of services aimed at preventing a particular disease, or on services targeted to specific groups, such as women. Findings show that the most common predictors of utilization include demographic factors (such as age, gender, and race/ethnicity), socioeconomic characteristics (such as education, income, and insurance coverage), and other health related factors (such as functional health status and the presence of a usual source of care). Several psychological and social characteristics are also important predictors. Although research on these characteristics is less common, factors such as social support, acculturation, 11 and depression have been found to be associated with levels of preventive service use. The effects of many of these predictors vary by type of preventive service, therefore, each of the six different preventive services will be discussed below according to the condition they are intended to prevent or detect. Influenza The first preventive service included in the current study is designed to prevent influenza. Millions of Americans suffer from influenza (the flu) each year, approximately 20,000 die annually from it, and nearly five times that many need to be admitted to the hospital (Centers for Disease Control and Prevention, 2002b). Being vaccinated from this contagious disease is critical, especially for older adults, those with other health problems, and young children. Even among healthy adults, flu vaccinations provide many benefits, including a 25 percent decrease in the number of upper respiratory illnesses and an average savings of almost 50 dollars per person vaccinated (Nichol, et al., 1995). The importance of being vaccinated regularly, especially for older adults, has been well publicized through recent public health campaigns. For example, flu shots were named one of the top ten leading health indicators for females in the U.S. Government s Healthy People 2010 Report (Maiese, 2002). Flu shots have become the center of attention for several reasons: they are relatively inexpensive and easy to obtain, numerous studies have documented their effectiveness, and current utilization rates are well below target (Nichol, et al., 1994). Flu Shots. Flu shots (which consist of injecting an inactivated virus into the blood stream) are the only type of preventive service included in this study intended to 12 prevent disease rather than to detect it. Groups that have been shown to have low utilization rates include racial and ethnic minorities. For example, in 1999, 69 percent of white older adults received a flu shot, compared to 58.6 percent of Hispanics, and 48.1 percent of African Americans (CDC, 2001). Studies have also found that whites are significantly more likely to have annual flu shots than African Americans, even after adjusting for several demographic and socioeconomic characteristics (Schneider, Cleary, Zaslavsky, and Epstein, 2001). Interestingly, various factors related to access to care have not been found to be significant predictors (CDC, 2001; Schneider, 2001). Age also predicts use, which can be seen by comparing the rates for older adults with those for individuals under 65 years of age. Rates of utilization are currently 20 percent for individuals 18-64 years of age and 65 percent for those over age 65 (Maiese, 2002). These rates are significantly lower than the target rates that the U.S. government has set for 2010 (90 percent utilization rates for all adults). Finally, health beliefs that affect flu shot use include perceived effectiveness of the vaccine and likelihood of side effects (Chapman and Coups, 1999). Due to the increased risks of influenza faced by older adults, the U.S. Preventive Service Task Force (USPSTF) has recommended that all persons over 65 years of age and those in selected high-risk groups receive a flu shot every year (USPSTF, 1996). Heart Disease Cardiovascular disease is the leading cause of death in the United States among adults over 35 years of age (CDC, 2002c). Diseases of the heart kill approximately 725,000 individuals each year in the U.S., which is more than 30 percent of all deaths 13 (Anderson, 2001). Furthermore, cardiovascular disease is the costliest disease; over $300 billion dollars is spent each year on health care expenditures and lost productivity resulting from diseases of the heart (CDC, 2002c). One of the primary risk factors for heart disease, a high cholesterol level, can be easily detected with screening. Cholesterol Screening. Tests that detect levels of cholesterol in the blood can provide individuals with important health information. This information can motivate individuals to alter diet and exercise patterns, to see health care providers regularly, or to begin a treatment regime of medications, if needed. Studies have shown that predictors of greater cholesterol screening use include being female or over 40 years of age, and having a higher education and income (Nelson, Norris, and Mangione, 2002). In addition, having a heart condition or diabetes and having health insurance all predict higher levels of screening (Nelson, Norris, and Mangione, 2002). Recommendations for cholesterol screening vary by source. The USPSTF strongly recommends that men over 35 years and women over 45 years be routinely screened for lipid disorders. Younger adults who have other risk factors for coronary heart disease should also receive annual tests. The recommendations further specify that the lipid tests should include measures of both total cholesterol and high-density lipoprotein cholesterol (USPSTF, 2002). The American College of Physicians, another group that regularly publishes screening recommendations, also provides guidelines for cholesterol screening. They recommend that this type of screening is appropriate, but not mandatory for men aged 35 to 65 years and women aged 45 to 65 years, unless other risk factors are present (Garber, Browner, and Hulley, 1996). 14 Breast Cancer The breast is the top site for new cancers and incidence rates for breast cancer have been steadily increasing over the past two decades (Jacobellis and Cutter, 2002; Makuc, Breen, and Freid, 1999). Correspondingly, breast cancer is the second leading cause of cancer deaths among women in the U.S. and the leading cause of cancer deaths for women ages 40 to 55 (American Cancer Society, 1998; Erwin, Spatz, Stotts, and Hollenberg, 1999; Wingo, et al., 1996). Preventive screening is especially critical because early detection and follow-up could prevent 15 to 30 percent of breast cancer deaths in women over 40 (CDC, 2002a). In addition, if the disease is diagnosed before the cancer spreads, the costs can be reduced by over 30 percent (CDC, 2002a). The American Cancer Society states that old age is the greatest risk factor for breast cancer. For example, the risk at age 40 is 1 in 67, while the risk at age 70 is 1 in 25 and more than one-half of the new cases of breast cancer occur in women over age 65 (ACS, 1998). Other demographic characteristics such as income, race, ethnicity, and education are also significantly related to the risk of developing breast cancer. Furthermore, having a family history of the disease, health insurance, and a usual source of care also predict levels of risk (Breen, Wagener, Brown, Davis, and Ballard-Barbash, 2001; Hayward, Shapiro, Freeman, and Corey, 1988; Kirkman-Liff and Kronenfeld, 1992; NCI Consortium, 1995). There are two primary ways to detect breast cancer: mammograms and breast exams. While breast cancer deaths fell in the last decade, incidence rates have increased and utilization rates for screening are still far below target (CDC, 2002c). Information 15 regarding rates and recommendations for each specific type of screening are described below. Mammograms. Mammograms are a particular type of X-ray that can detect tumors in the breast, even those too small to be found with manual examinations. Only one third of women over 40 years of age have had a mammogram in the past year (Barr et al., 2001). Like most preventive services, utilization varies by age, marital status, education, and other socioeconomic factors (Barr et al., 2001; Fox et al., 1998). In contrast to the other services, most studies find that blacks use mammograms significantly more than whites (NCI consortium, 1995), or at least at the same rate (Breen and Kessler, 1994; Breen et al., 2001). Acculturation to the U.S., health insurance coverage, having a usual source of care, doctors recommendations, and higher fertility have all been found to be predictors of mammography use as well (Katapodi, et al., 2002; Kirkman-Liff, 1992; Klassen, et al., 2002; O Malley, Kerner, Johnson, and Mandelblatt, 1999). Several psychological variables, such as depression, fatalism, anxiety, fear, and denial, are also related to mammogram usage (Fox et al., 1998; Michielutte, Dignan, and Smith, 1999). Specific health beliefs, such as concern about side effects and the belief that cancer is not a major risk, also provide barriers to the use of mammograms (Barr et al., 2001; Breen and Kessler, 1994). Despite current debates over the effectiveness of mammography (discussed below), the National Cancer Institute continues to recommend screening for women over 50 years of age every one to two years (NCI, 2002). Similarly, the USPSTF states that having a mammogram every 12-33 months significantly reduces the risk of breast 16 cancer mortality for women over 40 years of age (USPSTF, 2002). The benefits of mammography seem to be highest for women 50 to 69 years of age. Unlike other preventive services, the USPSTF maintains that the evidence concerning the benefits of mammograms is generalizable to women over 70, if they do not have other conditions that may limit their life expectancy (Randolph, Goodwin, Mahnken, and Freeman, 2002; USPSTF, 2002). Breast Exams. Breast exams, which can be conducted either by a health care provider or by the individual, involve examining the breasts for lumps or other unusual changes. The predictors of having regular breast exams are nearly identical to those for mammography use (Kirkman-Liff, 1992; O Malley, et al., 1999). For example, the probability of having a breast exam is significantly less likely for women over age 50 compared to those 20-49 years of age (Hayward, et al., 1988). Similarly, race differences are found, often with black women reporting more breast exams than white or Hispanic women (Breen and Kessler, 1994). Social support was found to be a significant predictor of both clinical and self breast exams (Katapodi, et al., 2002). Individuals with less social support are less likely to adhere to screening guidelines. The tendency to combine all types of breast cancer screening in studies has led to a relative lack of studies focused specifically on clinical or self breast exams. At least partially due to this, the USPSTF has concluded that there is insufficient evidence to recommend routine clinical breast exams or self-exams (USPSTF, 2002). Cervical Cancer The incidence and mortality rates of cervical cancer are both decreasing each 17 year (NCI , 2002). Pap smears have been recognized as the chief reason why cervical cancer deaths have decreased by 20-60 percent over the past several decades (USPSTF, 2002). The major risk factors for this type of cancer include age, sexual history, HIV or HPV infection, and family history (NCI, 2002). Most cases of cervical cancer occur to women between the ages of 20 and 30 and the chances of dying from this cancer increase with age (NCI , 2002). Pap Smears. Papanicolaou testing (a Pap smear) is the primary method for detecting cervical cancer. The test usually takes place in a doctor s office and involves taking a sample of cells from the cervix and upper vagina to analyze for abnormalities (NCI, 2002). Screening for this type of cancer in women is extremely important because early detection and follow-up could prevent nearly all deaths in women over 40 years of age (CDC, 2002). Predictors of Pap smear use include age, income, insurance, education, marital status, number of children, acculturation, and race (Hewitt, Devesa, and Breen, 2002; Kirkman-Liff, 1992; Yi, 1994; Yi, 1998). It is estimated that nearly two-thirds of women over age 18 have regular Pap smears (Hewitt, 2002). The USPSTF recommends that women should begin receiving Pap smears at age 18 or with the onset of sexual activity and that these tests should be repeated at least once every three years (USPSTF, 1996). A debate exists over whether or not these tests can be safely discontinued after age 65. However, the USPSTF maintains that there is insufficient evidence to put an upper age limit on the recommendations (USPSTF, 1996). 18 Prostate Cancer Prostate cancer is the most prevalent cancer for males in the U.S. and is the second leading cause of male cancer mortality (NCI, 2002). Prostate cancer survival rates are largely dependent on the progress of the disease at the time of diagnosis, so screening is critical (ACS, 1998). If prostate cancer is diagnosed while it is still localized, the individual has a 98 percent probability of surviving the next five years. However, if it is found after it has spread to other parts of the body, the five-year survival rate falls to 29 percent (ACS, 1998). Major risk factors include having a family history of the disease, age, race, and dietary factors. Prostate cancer rarely occurs in men under 50 years of age (NCI, 2002). Like other types of cancer, prostate cancer is more common in black males than white males. Mortality rates from this disease are also higher for blacks (ACS, 2002; NCI , 2002). However, it is possible that socioeconomic status accounts for a large portion of these racial disparities. Prostate Cancer Screening. There are several types of tests that screen for prostate cancer, including the digital rectal exam (DRE), the transrectal ultrasonography, and the prostate-specific antigen test (PSA) done with a blood sample. Screening rates vary by demographic and socioeconomic factors. White men are significantly more likely to receive regular prostate screening. This racial difference is present even after the role of socioeconomic factors was at least partially accounted for by including income controls and by offering free screening (Tingen, Weinrich, Boyd, and Weinrich, 1997). Looking at predictors of free screening is important because cost is the primary barrier to prostate screening utilization (Tingen, Weinrich, Boyd, and Weinrich, 1997). Age, 19 marital status, income, and health insurance are also important predictors of utilization (Merrill, 2001; Roetzheim, Pal, Tennant, Voti, Ayanian, Schwabe, and Krisher, 1999; Tingen, Weinrich, Boyd, and Weinrich, 1997). Of the health beliefs examined, the perceived benefit of screening and the perceived risk of prostate cancer were found to be significant predictors, while fatalism was not (McDavid, Melnik, and Derderian, 2000; Tingen, Weinrich, Boyd, and Weinrich, 1997). Arguing that proof of the effectiveness of any of the prostate screening tests is weak, the USPSTF does not recommend routine screening (USPSTF, 1996). In contrast, the American Cancer Society, the American College of Radiology, and the American Urological Association all recommend annual exams (PSA or DRE) for men over 50 years of age and for black men over 45 (ACS, 2002; USPSTF, 1996). Summary of Predictors of Preventive Service Use As noted in the above reviews of the literature, there are several demographic and socioeconomic factors that consistently affect the use of preventive services. Most notably, age, race, ethnicity, acculturation, marital status, education, income, and health insurance all predict use of preventive screening services. More specifically, older adults tend to use fewer preventive services, with the exception of flu shots. White individuals use more services than blacks or Hispanics, with the exception of female services. Generally, more acculturated individuals use more services as well. Finally, greater social and economic resources, such as marriage, education, income, and health insurance, are associated with higher levels of utilization of these services. The potential 20 roles of psychological and social factors, such as social support, mental health, and health beliefs, have also received limited support in the literature. Current Issues with Preventive Services Although the preventive services described above are generally considered to be valuable services, debates over the effectiveness of these procedures are present in the medical and epidemiological literatures. Not all experts agree on the utilization recommendations or even on the benefits received from different types of screenings. Concerns include the problem of false positives, the identification of diseases that may never present clinically, and the harm caused by unnecessary surgeries, as well as by the actual screening procedures, in some cases. A prime example of this type of debate involves breast cancer screening. Recently, a panel of cancer experts (the Physician Data Query Screening and Prevention Editorial Board) reported that there is insufficient evidence to show that mammograms prevent breast cancer mortality (National Cancer Institute, 2004). Without a definitive benefit to the tests, subjecting women to the potential harm caused by false positives, false negatives, and unnecessary biopsies seems imprudent to many health care professionals and researchers. And these risks may be larger than commonly realized; recent studies found that mammograms give false positive readings up to 15.9 percent of the time (Elmore, et al., 2002) and false negatives in 10-15 percent of women who have cancer that will manifest within one year (Wright and Mueller, 1995). However, a concerted effort by many of the most influential medical organizations in the U.S., including the American Medical Association and the National Cancer Institute, 21 contested the panel s conclusions and directed attention back to other prior studies that support the use of mammography (Charatan, 2002). Prostate screenings are another good example of the discord present among both researchers and practitioners. Although elevated levels of PSA and an enlarged, benignfeeling prostate are both indicators of prostate cancer, studies estimate that 80 percent or more of men with these symptoms do not have cancer (Catalona, 1995; Woolf, 1995). Furthermore, even if cancer is present, it is often not fatal. The risk of being diagnosed with prostate cancer over a lifetime is 12.3 percent, but the lifetime risk of prostate cancer mortality is only 3.8 percent (Jewett, et al., 2003). New screening procedures have caused incidence levels to increase dramatically, but mortality rates have remained stable (Chabner, Haluska, and Talcott, 1997). In addition, the most common surgery to treat prostate cancer may also be more harmful than the disease itself. Incontinence and impotence are both more frequent in men who have had this type of surgery (Holmberg, et al., 2003). Despite these debates over the benefits of preventive services, there is extensive research on the predictors of preventive service use. However, the role of religious factors in influencing utilization is largely absent in the literature. With growing evidence connecting religion to various health outcomes, determining the effect of religion on preventive service use is particularly important. Thus, the current study is valuable because it identifies religion as a potential predictor of preventive health care use and it adds an important dimension to the burgeoning religion and health literature. 22 Chapter Three: Theoretical Overview Chapter 3 provides the theoretical background for this study. First, previous frameworks are discussed. Then, a new conceptual diagram is developed by modifying previous models to better fit the research questions of the current study. Each intervening component of the model is discussed as it relates to the relationship between religion and preventive health service utilization. Particular attention is paid to the different facets of religion that may influence preventive service use. Finally, factors that may moderate the relationship between religion and preventive service use are discussed within the context of two competing hypotheses. Andersen Model of Health Service Use There are a number of theoretical frameworks that can be used to understand the relationship between religion and preventive service use. Many of these frameworks come from the medical sociology or public health fields and were originally designed to predict general health care use. One of the most enduring models of health care use is the one proposed by Andersen (1968) and further developed by Andersen and Newman (1973). In it, demographic, social, and health-related determinants of health care utilization are categorized as predisposing, enabling, or need characteristics. Most subsequent models of health care utilization still include these classifications, though possibly under different names, such as motivation, access, and need (Koenig, McCullough, and Larson, 2001). Predisposing characteristics include certain demographic and socioeconomic factors that affect an individual s motivation to use health care services. This category 23 is comprised of demographic variables, such as age and gender, that represent biological imperatives (Andersen, 1995), as well as social characteristics such as education and ethnicity. Health beliefs are also included here. Enabling resources are those factors that enable or impede use (Andersen, 1995). This category includes essential characteristics (of the individual or the family) such as income, health insurance, and access to a regular source of care. It also takes into account contextual factors such as the availability and cost of health care in the area, as well as region and urban-rural status. Finally, the illness level, or need, characteristics are included with (perceived and evaluated) measures of symptoms, diagnoses, and general health status (Andersen and Newman, 1973). Although numerous studies have used the Andersen model, or variations of it, to test the predictors of a wide variety of health (and social) services, few studies have used it to examine preventive services, especially among older adults. One study focusing on mammography use found only modest support for using the Andersen model to test predictors (Miller and Champion, 1996). Specifically, only predisposing and enabling characteristics were included, and only predisposing traits had any predictive ability (Miller and Champion, 1996). Another study also used this model, but focused primarily on the role of enabling factors, such as supplemental insurance coverage (Saag, et al., 1998). They found that enabling traits, such as type of insurance coverage and income, were important predictors of utilization for certain preventive services. 24 Religion and Preventive Health Care Use Only one study has used the Andersen model to test the role of religion in predicting preventive health service use (Miller and Champion, 1993). As discussed in the previous chapter, this study examined a convenience sample of 161 women selected from Catholic and Protestant churches in a metropolitan area. Affiliation (Catholic or Protestant) was included as a predisposing trait. The findings showed that Catholic women were almost six times more likely than Protestant women to use mammography screening. Overall, they found that the predisposing traits (religion, education, physician recommendations, and personal history) all predicted use, as did the enabling traits (income and having a regular source of care). In this study, need was operationalized as whether or not a physician recommended the mammogram. Although it was significant, the validity of measuring need in this way is questionable and no justification for the decision was offered (Miller and Champion, 1993). New Conceptual Framework Although empirical tests of the appropriateness of using general models of health care use to predict the relationship between religion and preventive service use are limited, concepts from this often used framework are still helpful. However, a new conceptual framework, one that borrows components from models specifically designed to examine the relationship between religion and health as well as from the models of general health care use determinants, may better illustrate this relationship and the potential variables that mediate it. The new framework is presented in Figure 1. The three categories of individual variables proposed by Andersen are included as the most 25 proximate predictors of service use; however, the model also considers the effect of background and religious characteristics, as well as other social and health-related mediators, on preventive care use. As noted by Andersen, predictors of preventive service use differ from those related to other types of health care (Andersen and Newman, 1973). One important change is that within the illness level category, actual need can be largely ignored since the goal of prevention is to stop or detect diseases before they result in declining health status (and, thus, increased need). Correspondingly, in the new conceptual framework, the role of need is represented by perceived need only. This is a measure that reflects the intentions of the Andersen model, but makes it more appropriate to preventive health service use. The original classification of perceived need included constructs such as the individual s awareness of symptoms, functional impairments, and overall health status. The current category of perceived need represents concepts similar to this, such as personal risk factors for diseases, as well as knowledge about disease prevalence and information on the availability and effectiveness of related services. Another factor that has often been overlooked in health care use models and that may be an especially important mediator of the effect of religion on health service utilization is attitudes toward health care professionals and the health care system in general. Although health beliefs are included in the Andersen model as one type of predisposing trait, specific attitudes, such as trust, are rarely tested in studies. While no studies were found that examined the influence of trust on preventive service utilization, trust has been found to be significantly associated with related outcomes, such as self26 reported adherence to medication (Thom, et al., 1999) and intention to follow the doctor s advice (Thom, et al., 2002). The addition of attitudes toward providers in the model clarifies another possible pathway in which religion may influence preventive service use, through its effects on motivation. As with much social science research, there are more conceptualized determinants than those that can be studied here. When health care use is the outcome of interest, factors such as presence of health insurance, education, and current health status may be some of the most important predictors. The influence of these variables is recognized here and, as much as possible, they are included in the models. Nevertheless, religious factors also have been shown to be associated with health outcomes, utilization of health care services, and other health behaviors net of these other powerful influences. Correspondingly, religion is central in this framework not because it is a primary predictor of health care use, but because the purpose of this study is to examine its relationship with preventive service use. Finally, the ordering of the conceptual model represents the expected causal relationships between groups of variables. The model is intended to characterize broad relationships that have been supported by previous studies and which will be tested here. Recognizing the complexity of causality, general hypotheses about the relationship between religion and preventive service use, rather than predictions about specific relationships between sets of variables, are represented in the model. 27 Background Demographic and Social Factors From the previous distillation of the literature, certain demographic and socioeconomic characteristics can be included in the models as probable influences on preventive service use. Many of these same characteristics also are associated with religious involvement (i.e. gender, age, and race/ethnicity). For example, nearly twothirds of women report that religion is very important in their life, compared to less than one-half of men (The Gallup Report, 1987). More importantly, studies that control for a broad range of covariates also find gender disparities. Using four national surveys, one study found that women displayed higher levels of religiosity in 13 of the 21 measures analyzed (Levin, Taylor, and Chatters, 1994). In addition to religion, gender may also influence each of the potential mediators. For example, women generally have higher levels of social support (Antonucci and Akiyama, 1987); however, they also have higher levels of mental health problems such as depression (Blazer, Hughes, and George, 1987). Age also affects religious involvement. Weekly church attendance rates increase at every age group, with those 65 and older reporting the highest level of attendance (The Gallup Report, 1987; Princeton Religion Research Center, 1994). Similarly, the percentage of individuals who report that religion is very important in their lives increases with age (Markides, 1983). This measure of religiousness peaks at the oldest age group with over three-fourths of elderly respondents stating that religion is very important (The Gallup Report, 1987; Princeton Religion Research Center, 1994). Age is also associated with many of the proposed social and health mediators. For example, 28 rates of functional limitations and chronic conditions both increase with age and this could limit the use of preventive services (Bould, Sanborn, and Reif, 1989). As with women and older adults, African Americans are often found to be more involved in religious activities (Ellison, 1995; Ferraro & Koch 1994; Levin, Taylor & Chatters 1994) and report higher levels of religious identity and use of religious consolation (Ferraro and Koch, 1994). Furthermore, approximately two-thirds of blacks report that religion is very important in their own life, compared to just over half of whites and Hispanics (The Gallup Report, 1987). Similarly, older Mexican Americans have been shown to have higher levels of church attendance, self-rated religiosity, and private prayer than whites (Markides, 1983). Race may also affect the potential mediators, such as health beliefs. One study found that black patients were more likely to mistrust the health care system than white patients (LaVeist, Nickerson, and Bowie, 2000). This mistrust may lead to lower levels of use of preventive services. Resources An individual s resources encompass various measures of socioeconomic status, as well as other types of resources relevant to health care use. Specifically, educational attainment, employment status, family income, total net worth, and the presence of health insurance are included in the models. These measures can be affected by each of the demographic and social factors discussed above. For example, race/ethnicity affects educational attainment and income among older adults, with minorities being less advantaged than non-Hispanic whites (Crystal and Shea, 1990; Federal Interagency Forum, 2000). In turn, these resources may influence various aspects of religion, as 29 well as the other social and health mediators. Individuals with lower incomes and education are more likely to attend church weekly or watch religious programming (The Gallup Report, 1987). While Andersen proposed that these types of resources are examples of enabling characteristics, I suggest that access to services incorporates a broader range of factors (discussed below). From this viewpoint, resources are just one of the many factors that contribute to an individual s access to preventive services, through a variety of pathways. Religion Three different facets of religion will be tested: organizational involvement, salience, and religious affiliation. Each of these measures is conceptually distinct and may affect preventive service use, and the mediators, in different ways. For example, it is possible that being involved in religious activities gives an individual access to greater amounts of social support, while having a higher level of salience has no effect on one s level of social support. Regardless of the pathways followed, it is expected that higher levels of organization or personal involvement with religion will be associated with higher levels of preventive service use. Organizational Involvement One important aspect of religion entails involvement with a religious organization. This involvement may affect preventive service use in several ways. For example, churches and synagogues frequently offer activities or information about health promotion topics that may lead (directly or indirectly) to greater use of services. It has been estimated that over 7,000 congregations in the U.S. use the services of a trained 30 health care professional (Carter, 1996). These health care professionals are often part of a parish nurse program. These programs recruit members of the church in health professions to fill the following roles: integrator of faith and health, health educator, health counselor, referral advisor, health advocate, developer of support groups, and volunteer coordinator (Parish Nursing website, 2003). Other types of related programs include health education campaigns and transportation services to health care providers. Numerous studies provide evidence supporting the effectiveness of these types of programs in promoting healthy behaviors and lifestyles (Davis, et al., 1994; Erwin, Spatz, Stotts, and Hollenberg, 1999; Fox, Stein, Gonzalez, Farrenkopf, and Dellinger, 1998; Lasater, Wells, Carleton, and Elder, 1986; Levin, 1984; Voorhees, et al., 1996). Part of this advantage may come from the fact that individuals with higher levels of church attendance have significantly more knowledge about health maintenance activities (Apel, 1986). Another example comes from the Thai Ministry of Public Health. This study found that an anti-smoking campaign by a Buddhist abbot resulted in significantly higher proportions of individuals who had stopped smoking compared to those in areas not affected by the campaign (Swaddiwudhipong, et al., 1993). Most importantly, of the reasons cited for quitting, the abbot was reported to be more influential (80.3%) than health care workers (72.1%) or family members (62.3%) (Swaddiwudhipong, et al., 1993). It is likely that individuals who are involved in religious organizations, through service attendance or other activities, will benefit from programs such as these. The effectiveness of these types of programs may be due to a variety of reasons. For one, time and transportation obstacles can be reduced because health programs can 31 be scheduled in conjunction with religious services or other congregational activities (Hale and Bennett, 2000). Furthermore, these religious services and activities can be used to disseminate information or to promote the health programs. Churches also have the benefit of being able to present the information at various times and in different formats to people of all ages (Hale and Bennett, 2000). Beyond these practical concerns, many researchers suggest that interventions done in churches are often successful because churches provide familiar, emotionally safe, and private surroundings (Schommer, et al., 2002). This may be in direct contrast to the unknown and chaotic environments present in many health centers, clinics, and hospitals. In addition to being an alternative to the impersonal atmosphere of the health care industry, church-based programs may also be successful because the faith community represents a powerful means of motivating and reinforcing positive behavior changes, stated David Satcher, chief of the CDC (quoted in Nation s Health, 1994). This influence is another way in which participation in religious organizations may be associated with utilization of preventive services. It is important to note that many individuals welcome this type of motivation from the church. Specifically, seventy percent of individuals in one study reported that the church should play a role in helping church members to meet their health needs (Swinney, Anson-Wonkka, Maki, and Corneau, 2001). The role of religious organizations in promoting (or even providing) preventive services may be increasingly important as responsibilities of health care delivery are beginning to be transferred from the formal health care system to communities (Swinney, et al., 2001). 32 Denomination It is also expected that denominational differences will be found. These differences may arise from theological distinctions, different social norms, or institutional discrepancies in programs or policies (including those discussed above). It is important to note that it is not exclusively theological differences that drive variations between denominations, but a combination of theology and group norms brought about by individual interactions (White, 1968), as well as organizational differences between denominations (and congregations). Theological differences may arise directly from the scriptures of a particular denomination or indirectly through interpretations. Group norms that exist within cultural subgroups (such as religious denominations) may also potentially influence how members understand health, disease etiology, and treatments (Jacobs and Giarelli, 2001; Turner, 1996). Finally, organizational differences may arise if the leaders (or members) of a certain denomination or church initiate programs or policies related to the health of their members. Within the early Christian traditions, belief in God s will guided ideas and behaviors related to health and illness. However, while early Christians were taught to accept illness as a punishment from God or a result of demonic possession (Richardson, 1991), the present-day interest in the physical and mental well-being of members of religious communities focuses more on health promotion and prevention than on acceptance and penitence. For example, members of Christian churches are often appealed to during weekly services to pray for the recovery of sick congregants. More institutional influences on health are also seen. For example, approximately one-third 33 of all hospitals in the United States are owned by Catholic or Protestant groups (Carter, 1994). It is possible that this scriptural and community focus on supporting and caring for the sick will lead to a greater use of preventive health services by Christians through higher levels of social support and greater institutional encouragement of health promotion activities. Catholicism. Within Christianity, it is possible that certain denominations promote health more than others. For example, Catholicism has always maintained a close tie to health services. In fact, Catholic nuns and priests established the first hospitals in the western world (Richardson, 1991). There are now 614 Catholic hospitals in the U.S., which is the largest group of not-for-profit hospitals in the country (Catholic Health Association of the United States website, 2004). These hospitals served over 5.5 million patients in 2001 alone (Catholic Health Association of the United States, 2003). Beyond its traditional involvement in the provision of health care, Catholicism also promotes religious healing and has several saints and shrines that are recognized and revered for their healing abilities (Numbers and Sawyer, 1982). These traditions may highlight the historical and current connections between religion and health that exist for many Catholics. However, the evidence on whether Catholics use fewer or more services than members of other denominations is split. On one hand, Catholics have often been found to be more likely to use preventive services than other Christian denominations (Miller and Champion, 1993) and Buddhists (Yi, 1994; Yi, 1998). On the other hand, Catholics have been shown to use fewer services than members of the Church of Ireland (Murray and McMillan, 1993) and Protestants (Miller, Norcross, and Bass, 1980). 34 Protestantism. Other Christian traditions, such as Protestantism, have also been associated with high levels of preventive service use compared to those in other denominations and the non-affiliated (Miller, Norcross, and Bass, 1980). This may be due to traditional group values, as well as to more institutional factors. Denominations that fall under Protestantism are becoming increasingly diverse. Each denomination s position on the continuum between liberal and conservative may affect the strength and direction of its influence on health behaviors. More general distinctions within Protestantism are made below, with a focus on some of the more relevant divisions for health care use. Evangelical Protestantism. Evangelical Protestants are distinct from more moderate and liberal Protestant groups because of their focus on the Bible, personal relationships with Jesus Christ, born again conversions, and moral absolutes (Sweet, 1994). Evangelicals have been at the forefront of many health movements, both in their home congregations and in missionary work. One writer even credits the development of public health to the sanitary reform movement directed by American and British Evangelicals (Sweet, 1994). More recent examples of health-related movements also exist. For instance, members of the 1999 Southern Baptist Convention voted to begin a nation-wide campaign against alcohol and other behavior-altering drugs within their churches (Hastings, 1999). Through this campaign and others, Evangelicals have been important players in the battles against drinking, drug use, addictions, and premarital sexual activity (Sweet, 1994). However, it must be noted that these campaigns were motivated by moral issues, not health ones, and the health benefits of such programs are 35 a positive, yet inadvertent by-product. Thus, programs and policies of these denominations most likely will not influence members use of preventive services. Within Evangelical Protestantism, predominantly Black denominations may have special properties that make these congregations particularly conducive to the good health of their members. Many researchers in the field have expounded upon the unique characteristics of Black churches. In general, Black churches are distinct because of their interest and involvement in all areas of their member s lives, not just religious or spiritual ones. Predominantly Black churches often participate in social, political, and educational activities, as well as those related to health (Cone, 1985). Black churches may be particularly likely to impart information and guidance for improved health and may also play a unique role in influencing health-related beliefs, behaviors and outcomes (Abrums, 2000; Levin, 1984; Scandrett, 1996). In addition, for many Blacks, membership and participation in a church is an expectation (Ellison and Sherkat, 1995), and rates of formal participation may also be higher. Consequently, Black churches may provide more community resources and social support (Levin, Taylor, and Chatters, 1994), and these resources may lead to greater preventive service use among members of Black churches. Mainline Protestantism. There is also a longstanding concern for health among certain Mainline Protestant denominations. For example, the work of John Wesley, recognized as the founder of Methodism, often focused on the relationship between spiritual health and physical health. John Ott summarized three fundamental themes within Wesley s teachings, as follows (Ott, 1991). First, he believed that maintaining a 36 well-functioning body was important because it allowed individuals to live and act according to God s will. He also emphasized that spiritual health had a direct impact on physical health, as well as the reverse. Finally, he believed that a healthy lifestyle, including regular exercise, was necessary to achieve this holistic concept of health. This historical concern for health and the priority placed on health behaviors from leaders such as John Wesley may have helped to establish a denominational interest in health within Mainline Protestant congregations. Furthermore, the influence of Wesley s teaching is still evident within the Methodist church. For example, numerous events were held to celebrate the 300th anniversary of his birth in 2003 and many courses, lectures, and exhibitions on his life and teachings are currently being offered by Protestant organizations around the world (General Board of Global Ministries website, 2004; John Rylands Library website, 2004; Wesley 2003 website, 2003). This linkage between the theological teachings of the church and health maintenance could lead to higher levels of preventive health care utilization among these denominations. Judaism. Any expected differences within the Christian faith are largely speculative due to the scarcity of previous studies in this area. However, relatively more research on denominational differences in health behaviors and beliefs has examined the Jewish faith. As noted previously, Jewish individuals generally use more health services than members of other religious denominations (Linn, 1967; Mechanic, 1963; Scheff, 1966; Segal, Weiss, and Sokol, 1965; Solon, 1966; Wan and Soifer, 1974; Wan and Yates, 1975). To explain this pattern, most theories have focused on certain norms or values, such as a more intense parental interest in children s well-being and an overall 37 concern for health, that are often found to be associated with Jewish culture (for discussion, see Jacobs and Giarelli, 2001). Furthermore, within the Jewish tradition, good health is often considered to be a community concern, rather than an individual one. This focus on health can be seen in the many traditions of the Jewish faith that are healthrelated, such as frequent hand-washing and dietary prohibitions against pork and shellfish (Jacobs and Giarelli, 2001). Jewish health beliefs and traditions have been the focus of several books, including Jewish Medical Ethics (Jakobovits, 1975) and Modern Medicine and Jewish Law (Rosner, 1972). A consequence of the elevated levels of awareness and concern for health could be increased health promotion and disease prevention activities for members of this denomination in comparison to non-affiliated individuals and those belonging to other denominations. Other Denominations. Certain non-traditional Christian denominations have also incorporated health-specific teachings into their tradition. Specifically, Jehovah s Witnesses, Mormons, Christian Scientists, and Seventh-Day Adventists all are distinguished by their health-related beliefs and activities (Numbers and Sawyer, 1982). Membership in non-Judeo Christian religions may also influence health behaviors. For example, a study of Muslim women in Iran found that one-third of the women believed that clinical breast examinations by male physicians went against their religious beliefs (Montazeri, Haji-Mahmoodi, and Jarvandi, 2003). Salience Religious salience is included to capture the possible effects of personal beliefs, faith, and commitment on preventive service use. Although the studies that have 38 examined salience and preventive service use have found conflicting results (Benjamins and Brown, 2003; Fox, et al., 1998), several studies have shown that salience is related to other health-related behaviors, such as smoking, promiscuity, and other risk-taking behaviors (for review, see Koenig, McCullough, and Larson, 2001). Unfortunately, the mechanisms underlying these relationships have not been examined. It is possible that they are due to such factors as attitudes toward the health care system and providers and better mental and physical health. More directly, belief in a higher power may encourage positive health behaviors due to feelings of responsibility or optimism, for example. Religious individuals may also be more motivated to maintain their health in order to lead a life consistent with their beliefs. In other words, staying healthy is necessary for individuals to volunteer in their community, stay involved in their family life, and participate in other religious activities. It is also possible that religious salience may have a negative influence on health promotion and disease prevention. Previous researchers have theorized that believing in the afterlife may undermine the importance of preventive services. If individuals believe that life continues after death and, perhaps more significantly, that life in the next world is more important, activities designed to improve their health or decrease their mortality risk may be meaningless (Wynder and Sullivan, 1982). Furthermore, others claim that religion has often impeded medical progress, by linking dissection, certain treatments, and even vaccinations to sacrilege, atheism, and witchcraft (Numbers and Sawyer, 1982). However, there is little research to suggest that these types of beliefs have any effect on present-day preventive service use. 39 Social and Health Mediators Attitudes Toward Health Care Providers and The Health Care System Health-related beliefs and attitudes are important predictors of all types of health care use, including preventive services (Speedy and Hase, 2000). Although these viewpoints incorporate a wide variety of issues and ideas including personal responsibility, the effectiveness of treatments, trust in physicians, fatalism, and confidence in the Western medical model, among others, the current framework will focus on attitudes toward physicians and the health care system. More specifically, different aspects of patient trust (in doctors and the system) will be considered. A broad range of demographic, socioeconomic, and psychosocial factors may affect both the type, and intensity, of these attitudes. Some of these factors include age, gender, race, ethnicity, education, income, health knowledge, values, and prior experiences. It is possible that religious involvement and affiliation also influence health attitudes, though no research on these relationships was found. In turn, trust and other health-related beliefs may affect health care use through various pathways. For example, patients who report lower levels of trust are less likely to intend to follow their doctor s advice (Thom, et al., 2002). Furthermore, patients with low levels of trust report less satisfaction (Thom, et al., 2002), and women who are less satisfied with their physician report lower mammography use (Barr, et al., 2001). More generally, these health-related beliefs may influence preventive service utilization through two of the Andersen mediators: motivation and perceptions of need for preventive services. 40 Several theoretical frameworks have been devised to incorporate certain types of health beliefs. One of the most commonly cited models is the Health Belief Model, first proposed by Rosenstock (1966) and later outlined by Becker and Maiman (1975). This model assumes that health care use is influenced by the following factors: susceptibility to the relevant disease, seriousness of the disease, potential benefits of the service, barriers to utilization, health motivation, and confidence. Many studies have tested the merits of this model to predict preventive service utilization and have found modest support (Chapman and Coups, 1999; Speedy and Hase, 2000). The current conceptual model takes many of these factors into account. Susceptibility to a disease and disease seriousness both may affect an individual s perceived need for the relevant preventive service. Moreover, the potential benefits of the service, health motivations, and confidence are examples of factors that may motivate an individual to use services. Finally, barriers to utilization reflect the level of access an individual has to a particular service. Religious involvement and affiliation may affect all of these types of health-related beliefs, through each of the mechanisms outlined in the section above (i.e. theology, group norms, and church policies). There are also more general health-related beliefs that may affect preventive service use and may vary by religious involvement and affiliation, such as fatalism, faith in a higher power, or the use of prayer to supplement or replace medical services and treatments. These beliefs may negatively influence preventive service use, as may other more obvious factors such as religious beliefs that dictate the refusal of medical treatments. The religious groups who hold these beliefs, particularly the Christian 41 Scientists and Orthodox Dutch Reformed, have been publicly attacked for their stances on refusing treatments for sick children, organ transplants, and other medical interventions. Because most health care services, including vaccinations, are forbidden by these groups, the use of flu shots and other preventive services should be negatively affected by the health beliefs of these denominations. However, the proportion of Americans that belong to one of these groups, or holds these beliefs, is very small. Therefore, these beliefs are beyond the scope of the current study. Social Support Social support can take many forms, including emotional, instrumental, cognitive, and financial support. Religious involvement may influence levels of social support in numerous ways; for example, involvement in religious organizations provides individuals with opportunities for social contact with others who are likely to hold similar values. One study found that the majority of friendships of older adults began with individuals they met at church (Koenig, Moberg, and Kvale, 1988). Similarly, shared experiences and beliefs may also serve to unite individuals who belong to the same religious organization and strengthen their relationships (Koenig, McCullough, and Larson, 2001). Empirical evidence supports religious variation in levels of social support. In fact, in The Handbook of Religion and Health, twenty studies linking religion and social support were found and 19 of these reported significant relationships (Koenig, McCullough, and Larson, 2001). More specifically, individuals who attend religious services frequently have larger and denser social networks and receive more types of social support than those who never attend (Bradley, 1995; Ellison and George, 1994). Moreover, they are 42 more satisfied with the quality of their relationships than non-religious individuals (Bradley, 1995; Ellison and George, 1994). Social resources such as social support and integration have also been shown to be associated with use of health care services. Individuals with higher levels of social support may use more preventive health care because social relationships can serve as a conduit for information and an incentive for positive health behaviors. Individuals with more support are likely to receive more monitoring, more information, more help getting to the services, and perhaps even more help paying for preventive services. While empirical evidence on this question is limited, at least one previous study has shown that adults with more social support are more likely to report both preventive medical and dental check-ups (Berkman and Syme, 1979). In sum, participation with a religious organization provides individuals with greater amounts of social support and this social support is expected to lead to higher levels of utilization of preventive services. Thus, several measures of social support are included in the models as possible mediating variables. The first variable, marital status, is an important indicator of the amount of support an individual receives (House, Umberson, and Landis, 1988) and is associated with religious involvement. For example, religious individuals are less likely to divorce and more likely to have satisfying marriages (Kaslow and Robison, 1996; Mahoney, Pargament, Tarakeshwar, and Swank, 2001). More generally, family members are an important source of social support. In addition, the quantity and quality of friendships are further indicators of social support. 43 For these reasons, various aspects of social support are expected to mediate the relationship between religion and preventive service use. Mental Health Higher levels of religious involvement are associated with greater well-being, hope, optimism, purpose, self-esteem, and coping as well as less anxiety, stress, and depression (for review, see Koenig, McCullough, and Larson 2001). Over a hundred studies have analyzed the relationship between religion and depression and the vast majority have found that, the religiously active are less likely to become depressed and less likely to stay depressed (Koenig, McCullough, and Larson 2001: 216). In addition to depression, religion also is associated with levels of anxiety, though differences regarding the direction of the effect have been found in the literature (for a review, see Koenig, McCullough, and Larson 2001). Because depression, anxiety, and other aspects of mental health affect the use of health services (Koenig, Sehlp, Golid, Cohen, and Blazer, 1989; Simon, Ormel, VonKorff, and Barlow, 1995; Unutzer, et al., 1997), the mediating role that these factors may play between religion and preventive service use is examined in the current study. Physical Health As noted in the review of previous studies, the relationship between religion and physical health has garnered an increasing amount of attention from researchers in the social, behavioral, and health sciences as well as from the popular media. These studies have generally found that religion has a salutary, or protective, effect on a variety of outcomes, including functional limitations (Idler and Kasl, 1997; Levin and Markides, 44 1985), self-rated health (Levin and Markides, 1985; Musick, 1996) and mortality risk (Hummer et al., 1999; Oman and Reed, 1998; Strawbridge, et al., 1997). However, this growing field has also attracted several critics (e.g., Sloan, Bagiella, and Powell, 1999) due to methodological limitations common in the literature, the lack of consensus on the best way to measure and model these associations, and because of differing viewpoints about the place of religion in scientific research. Despite these limitations, the majority of studies support a positive relationship between religion and physical health. Moreover, it is possible that physical health status is associated with levels of preventive service use. This relationship can most likely be explained by differing levels of access. Individuals with functional limitations, or other physical problems caused by illness or disease, may be less able to access preventive services due to difficulties with getting out of bed, getting dressed, leaving the house, taking transportation to the services, and performing any other necessary actions. However, it must be noted that this relationship could also work in the opposite direction. For example, an individual who experiences a serious health problem may be more likely to seek out preventive care for several reasons. For one, the health problem may create awareness of these services through increased contact with the health care system. Furthermore, health problems may scare individuals into improving their health behaviors. While it is possible that poor health status may increase preventive service utilization for these reasons, the reduction in access caused by poor health is assumed to be more influential. Therefore, physical health status must be included in the models as one way in which religion may positively influence preventive service use. 45 Andersen Mediators It is critical to recognize the importance of the predictors of health care use proposed by Andersen (1968). The broad categories of need, predisposing, and enabling characteristics may explain a substantial portion of the variation within preventive service use. However, data limitations preclude explicitly measuring these sets of factors. Instead, the background, religious, social, and health constructs previously discussed are assumed to work through these factors. Perceived Need While the need for preventive services differs from the need for other types of medical procedures or treatments, perceptions of how important a particular service is may still be an important predictor of preventive health service utilization. Studies examining the reasons why individuals do not follow utilization recommendations find that, often, the perceived need for the service is low. For example, of the Medicare beneficiaries who did not receive flu shots in the past year, approximately 20 percent said their reason for not receiving the vaccination was that, I did not know the flu shot was needed (Drociuk, 1999). Motivation Compared to perceived need, motivation is a much broader category. Koenig, McCullough, and Larson (2001:430) explained the role of motivation in this way: Even if the person recognizes the need for health services and the services are readily available, he may not have the energy, interest, or desire to obtain the necessary medical care. Motivation can be influenced by religion through the attitudes toward health care, 46 levels of social support, and mental health status. For example, individuals who believe their physicians are incompetent will not be as motivated to seek out health services compared to those who trust their physicians judgments. Similarly, individuals with strong social networks may receive more encouragement to obtain preventive services. Finally, individuals who have a mental health problem, such as depression, often have low levels of motivation for various activities, not just preventive services. Through these factors, religion may influence preventive service use. Access In addition to the social and economic resources discussed above, access incorporates other factors that enable individuals to use services. More specifically, the availability of services, actual ability to get the services (accessibility), and the acceptability of the services provided are also essential components of access. Religious individuals may have greater access to preventive services through their religious organization, though evidence on this topic is scarce. Moderating Factors As noted above, many social and demographic characteristics influence religious involvement and other predictors of preventive health care use. Correspondingly, these factors may moderate, or alter, the relationship between religion and preventive service utilization. Gender, age, race/ethnicity, and education are all factors that may influence the way in which religion affects preventive service use. The interaction between religious denomination and attendance may also be important. 47 Amplification versus Compensation Several theories can be suggested to explain how different social and demographic characteristics combine with religion to affect preventive health care use. Two such theories, primarily coming from the sociology of religion literature, are (resource) amplification and (resource) compensation (Schieman, 2003). The amplification theory posits that the interaction between the moderating variable and the other variables strengthens the effect of existing resources. The compensation theory, on the other hand, suggests that advantages are provided specifically to those who are lacking other resources. The usefulness of these competing theories to describe the moderating role of gender, age, race/ethnicity, and education within the relationship of religion and preventive service use will be assessed. Amplification. In the case of preventive health care use, religion may increase the likelihood of using health care among those who are already more likely to follow utilization recommendations. Individuals with more social and personal resources are more likely to have greater access to services, to be knowledgeable about what services they need and where to find them, and may be more motivated to seek out this type of health care. In part due to these reasons, advantaged individuals may find it easier to capitalize on these opportunities offered through churches. Thus, the possible influence of religious beliefs on health beliefs may be more easily translated into service use. Looking at it from another viewpoint, individuals with fewer resources may be influenced by the same religious beliefs, support groups, and programs, but may be unable to turn these influences into greater preventive service use. 48 Compensation. Conversely, religion may increase the likelihood of using preventive services among those who have fewer resources by compensating for these disadvantages. In this case, religion is a counterbalancing force (Schieman, 2003) working against the lower levels of perceived need, motivation, and access faced by certain groups. For the disadvantaged, religion may be the singular force encouraging individuals to pursue good health. While more advantaged individuals may be bombarded with advice, information, and opportunities for achieving better health, this influence may be stronger for those for whom other resources are lacking. In other words, for individuals who are educated, employed, wealthy, and have high levels of social support, health care programs provided by religious institutions may be unnecessary due to the abundance of other options. Therefore, it is possible that the effect of religion on preventive service use will be stronger for those who have fewer social and personal resources. Gender Previous studies have found that the effect of religion on various health status outcomes, including mental health and mortality, is stronger for women than for men (Idler, 1987; Mirola, 1999; Strawbridge, Cohen, Shema, and Kaplan, 1997) and this may be true for health services utilization as well. According to the amplification theory, women, who already use more preventive services, will benefit more from their religious involvement. If this difference is found, the gender gap in preventive service use will be increased. 49 Age At least one study has suggested that religion may have relatively larger consequences on various aspects of mental health for adults in their 70 s and 80 s compared to those in their 60 s (Blazer and Palmore, 1976). It is possible that age may also affect the relationship between religion and preventive care utilization. However, it is likely that the limited age range of this sample will preclude finding any significant age differences. Race/ethnicity Racial and ethnic differences may also influence the relationship between religion and preventive service use. It has been hypothesized that not only do levels of religious involvement and types of beliefs differ between racial groups, but that religious activities and values actually hold different meanings and consequences for black individuals compared to whites (Levin, Taylor, and Chatters, 1994). Furthermore, Black churches may play an especially important role in influencing the health-related decisions and behaviors of members. It is possible that the potential differences in the role of religion both in individuals lives and in the community, may lead to a stronger relationship between religion and preventive care use. This would be an example of resource compensation. Socioeconomic Status Following the premise of the compensation theory, individuals with lower levels of education and income are hypothesized to benefit more from religious influences than those with higher levels of these resources. In contrast, the amplification argument 50 supports a greater effect of religion on preventive service use for those with higher socioeconomic status. However, little work has been done on socioeconomic differences in religiosity and no support for either theory is available on which to base expectations. Thus, directional hypotheses will not be proposed for these potential moderators. Religious Denomination As mentioned previously, the influence of religion may stem from a combination of religious affiliation and participation in religious activities (White, 1968). At least one study has found that the effect of religious activities differs by denomination. Specifically, Tix and Frazier (1997) found that religious coping significantly improved psychological adjustment after an organ transplant for Protestants, but not for Catholics. While little work has been done in this area, especially in studies that distinguish between the many Protestant denominations, it is possible that the effects of religious attendance and salience may differ by affiliation. For example, the effect of attendance may be stronger for individuals belonging to affiliations that stress weekly attendance, such as Catholicism, compared to affiliations that do not traditionally place such emphasis on service attendance (i.e. Mainline Protestantism). General Expectations After reviewing relevant previous studies and theoretical frameworks proposed in the literature, several expectations can be postulated. More specific hypotheses are incorporated in each of the two analyses sections. The general expectations are as follows: 51 Expectation 1: More religious individuals (identified by higher self-rated religiosity or higher levels of involvement in religious activities) use more preventive services. Expectation 2: Use of preventive services varies by religious denomination. Expectation 3: Attitudes toward health care mediate the relationship between religion and preventive care. Expectation 4: The effect of religion on preventive care use differs by subgroup (gender, age, race/ethnicity, socioeconomic status, and denomination). 52 Chapter Four: Religion and Preventive Service Use Chapter 4 focuses on the relationship between religion and preventive health care service use and includes an examination of the possible mediators and moderators of this relationship. The chapter begins with a brief review of the theoretical framework, which leads into the specific hypotheses to be tested. Following this, the data, measures, and methods for the analyses are described. Finally, the results of the analyses are presented and discussed. Introduction As noted in previous chapters, prior studies have shown that women who belong to a religious organization use more preventive services than those who do not and that levels of preventive service use vary by religious affiliation and involvement (Benjamins and Brown, 2003; Fox, et al., 1998; Miller and Champion, 1993; Murray and McMillan, 1993; Naguib, Geiser, and Comstock, 1968; Yi, 1994; Yi, 1998). These relationships may work through various mechanisms, including mental and physical health, social support, and health attitudes. Because religion may either compensate for a lack of other social and personal resources or amplify existing resources, the relationship between religion and preventive service use will be analyzed separately for certain demographic subgroups of the population. Hypotheses Hypothesis 1: Higher levels of attendance will be associated with more preventive service use. Hypothesis 2: Levels of use will differ by denomination, with Jewish individuals reporting the highest levels of utilization and non-affiliated individuals the lowest. 53 Hypothesis 3: Higher salience will be associated with more preventive service use. Hypothesis 4: Relationships between the religion variables and preventive service use will be moderated by certain demographic and social characteristics.1 Hypothesis 4a: These relationships will be stronger for women. Hypothesis 4b: These relationships will be stronger for older adults. Hypothesis 4c: These relationships will be stronger for minority group men and women. Data and Methods Data Data for this analysis comes from the Health and Retirement Survey (HRS). The HRS is a nationally representative survey of non-institutionalized pre-retirement aged adults in the United States (HRS website, 2003). The purpose of this panel study is to better understand a wide range of issues that are critical to our aging society. It focuses on demographic, socioeconomic, and health characteristics such as family composition, retirement behavior, and physical and mental health status. The HRS has a multi-stage probability sampling design that includes oversamples for African Americans, Hispanics, and residents of Florida (Heeringa & Connor, 1995). The religion and control variables are measured in Wave 1 (1992), while the preventive service outcomes are measured in Wave 3 (1996). Preventive services were first measured in Wave 3. While it would be ideal to measure the possible predictors of these services in the previous wave (Wave 2, 1994), information on the religion variables Other potential moderators were also tested. These include education, income, and religious denomination. However, no previous work relating to the effect of these variables on the relationship between religion and preventive service utilization was available and, thus, no directional hypotheses were made. Furthermore, no significant moderating effects were found for these three variables (analyses not shown). 1 54 in this wave was only collected from individuals new to the sample. Therefore, the religion variables in Wave 2 were identical to those in Wave 1 for everyone except for the 102 new respondents. Correspondingly, Wave 1 was used instead. The only exception to these guidelines is the variable representing intrinsic religiosity (i.e. salience), which was not asked in Waves 1 or 2 and, therefore, is measured in Wave 3 along with the outcome variables. While this presents a slight problem with the temporal ordering of the variables, the analyses take place under the assumption that an individual s religiosity will not change substantially between waves. I discuss this limitation further in the concluding section of the dissertation. More recent waves of this data set were not used because after 1996 (Wave 3) additional samples were added to the HRS (the AHEAD, CODA, and War Babies samples). The benefits of a larger sample were outweighed by the lack of religious attendance and preventive service utilization information for the new respondents. Specifically, approximately 5,000 respondents out of a sample of 21,000 were asked the religion and preventive service use questions in 1998. The full sample was asked the preventive health questions in 2000; however, information on the religion variables was not available for all of these respondents. Variables and cases available in each wave are shown in Table 1. The HRS was designed to examine pre-retirement age adults (aged 51-61 in 1992) and their spouses. In total, 12,652 interviews were successfully collected from 7,705 households during Wave 1 (1992) of the HRS (Heeringa & Connor, 1995). Due to the absence of individual-level weights for individuals born before 1931 or after 1941, only 55 those individuals born between 1931 and 1941 with a valid weight (i.e. non-zero) are included (n=9,507). Additionally, the analyses are limited to non-Hispanic whites, nonHispanic blacks, and Hispanics because of the small number of individuals in the other racial and ethnic categories. This resulted in the exclusion of 208 individuals. Next, 19 cases were excluded due to missing religious affiliation and attendance data. Of those still in the sample, 7,886 individuals (85 percent) remained in the sample through Wave 3 (1996). An additional 11 cases were lost due to missing religious salience information in Wave 3. The final sample sizes range from 7,857 to 7,867 for the services applicable to both genders (flu shots and cholesterol screening, respectively) and from 3,607 to 4,256 for the gender-specific services (breast exams, mammograms, Pap smears, and prostate exams). Slight variation in these numbers reflects the number of cases with missing information for the dependent variables. Measures Preventive Services. There are six dependent variables of interest: flu shot, cholesterol exam, mammogram, Pap smear, breast exam, and prostate exam. Inquiries about each preventive service begin with the following question, Since we talked to you last in (previous wave interview month and year), have you had any of the following medical tests or procedures? Specifically, the questions are phrased in this manner: A flu shot? , A blood test for cholesterol? , Did you have a mammogram or x-ray of the breasts? , A Pap smear? , Do you check your breasts for lumps monthly? , and An examination of your prostate to screen for cancer? The final four questions are only asked of the appropriate gender. All preventive behavior variables are dichotomous 56 with 1 representing utilization and 0 representing non-utilization in the past two years (or monthly, for breast exams). In addition, two gender-specific summary variables are used to examine the total usage of preventive services. Female total usage was determined by summing cholesterol tests, flu shots, mammograms, pap smears, and breast exams. Scores for this variable range from 0 to 5. Male total usage sums cholesterol tests, flu shots, and prostate screenings. Correspondingly, this variable ranges from 0 to 3. Religion. The independent variables reflect three different facets of religion: religious denomination, attendance at organized religious services, and intrinsic religiosity. The first variable measures an individual s involvement by asking about his or her frequency of attendance at religious services. The respondents are asked how often they have attended religious services in the past year and then are prompted with the following probe: Would you say more than once a week, once a week, two or three times a month, one or more times a year, or not at all? Five dummy variables were constructed to represent each of these categories. Due to the non-linear relationship between attendance and preventive service use, the dummy variables were included in the models. Individuals who never attend were selected as the reference group. The second religion variable, religious denomination, indicates the group or belief system with which the individual is affiliated. It is measured with two questions. The first question asks, What is your religious preference: Protestant, Roman Catholic, Jewish, or something else? The respondent is then prompted to be more specific with the question, Which denomination is that? Seven denomination categories were 57 created as follows: Catholic, Evangelical Protestant, Mainline Protestant, Black Protestant, Jewish, other religion, and non-affiliated, as presented in Table 2. The assignment of specific denominations to each category was done in accordance to the classification scheme developed by Steensland and his colleagues, with a few exceptions (Steensland, Park, Regnerus, Robinson, Wilcox, and Woodberry, 2000). However, the category of Black Protestant was highly correlated with the race variable, non-Hispanic Black (r=.83, p<.001), and, therefore, it was excluded from the models. All of the specific denominations formerly categorized as Black Protestant were then categorized as Evangelical Protestant. This category, Evangelical Protestant, is the reference group. The final religion variable measures an individual s intrinsic religiosity, or salience. This measure is intended to tap the salience of religion to that individual. The wording of the question is as follows: How important would you say religion is in your life: is it very important, somewhat important, or not too important? Like religious attendance, dummy variables were used to measure these three categories of salience. The lowest category of salience represents the reference category. Demographic and Social Factors. Measures of demographic characteristics include those frequently found to be associated with the use of preventive services, such as age, gender, race/ethnicity, and foreign-born status (Barr et al., 2001; Breen et al., 2001; Breen and Kessler, 1994; Fox et al., 1998; Hewitt, 2002; Kirkman-Liff, 1992; Maise, 2002; Nelson, Norris, and Mangione, 2002; O Malley et al., 1999; Schneider, 2001; Tingen, Weinrich, Boyd, and Weinrich, 1997). Age is measured as a continuous variable. Gender is a dichotomous variable with females representing the reference 58 group. Race/ethnicity indicates whether the individual is non-Hispanic White, nonHispanic Black, or Hispanic. Finally, because nativity may be a significant predictor of preventive service use, a variable is included to ascertain whether the individual was born in the U.S. or not. Resources. Measures of socioeconomic resources that are relevant to preventive service utilization include education, income, net worth, and health insurance (Hewitt, 2002; Katapoldi et al., 2002; Kirkman-Liff, 1992; Klassen et al., 2002; Nelson, Norris, and Mangione, 2002O Malley et al., 1999; Roetzheim et al., 1999; Schneider, 2001; Tingen, Weinrich, Boyd, and Weinrich, 1997). Education is measured with a continuous variable that represents the respondent s completed years of formal education. The highest value (17 years) represents all levels of education past college. Income is a measure of the family income of the respondent in the past year. This variable is measured as a continuous variable. Net worth is considered a more reliable measure of economic resources than income, especially for older adults (Hurd, 1989; Smith and Kington, 1997) and, therefore, it is included in the models as well. Net worth is measured with a dichotomous variable that designates individuals in the lowest quartile of the distribution as low net worth (worth less than $32,500). Finally, the presence of health insurance in the past year is measured with a dichotomous variable that indicates whether or not the respondent currently has any type of health insurance. Insurance from government sources, an employer, or other sources are all included. Social Support. As noted above, being married provides individuals with additional social support (House, Umberson, and Landis, 1988) and this may predict 59 preventive service utilization (Berkman and Syme, 1979). Therefore, a marital status variable is included to specify if the individual is currently married or living with a partner. The reference group is comprised of individuals who are divorced or separated, widowed, or never married. Support from other family members and friends may also be important. The quality of friendships is measured with an item that asks respondents for their overall level of satisfaction with their friendships. Support from family is measured with a similar item that assesses the respondents satisfaction with their family life. For these two variables, higher scores indicated higher levels of satisfaction. Approximately four percent (n=312) of the values for the two satisfaction variables were missing. Rather than exclude these cases, the values were imputed to the median category. Mental Health Status. Mental health measures cover subjective health and depression. Self-rated emotional health is measured with the following question: What about your emotional health--how good you feel or how stressed, anxious or depressed you feel? The response choices are excellent, very good, good, fair, and poor with higher scores indicating worse self-rated emotional health. The CES-D depression scale (for descriptions see Radloff, 1977; Ensel, 1986) measures depressive symptoms with a scale created with twelve individual items. These items include questions about feeling depressed, poor sleep quality, being lonely, having a poor appetite, and other symptoms of depression (HRS website, 2003). Again, higher scores indicate poorer mental health. Physical Health Status. Measures of physical health status encompass functional limitations, chronic conditions, and subjective health. The first measure of health status is a functional limitations scale, consisting of 16 questions regarding activities of daily 60 life (ADL). This variable represents the total number of reported activity limitations for each individual and higher scores represent more functional limitations (see Katz, Ford, Moskowits, Jackson, and Jaffe, 1963). Measures representing chronic conditions are also included. Two chronic conditions that are directly associated with the preventive screening measures (heart conditions and cancer) are included individually. The variable heart conditions takes into account heart attacks, coronary heart disease, angina, congestive heart failure, or other heart problems. The cancer variable represents a malignant tumor in any part of the body, excluding skin cancer. A variable representing the total number of other chronic conditions is also included. This count represents the presence of hypertension, diabetes, chronic lung disease, arthritis, and stroke. All chronic conditions represent conditions or episodes that have happened at any time prior to the interview. The final measure of physical health is self-rated health. Like the subjective emotional health measure, this is measured with a question asking respondents to evaluate their current health using a 5-item scale that ranges from excellent to poor. Higher values indicate worse health. Methods Descriptive statistics are presented as follows. Univariate analyses provide the range, mean, and standard deviation for each variable included in the regression models. Following this, the relationship between the religion variables is examined. Specifically, the differences between mean levels of religious attendance and salience by denominational affiliation are determined and the significance of these differences is 61 tested. Next, bivariate relationships for the mediating and control variables with each of the six outcome variables and the two total usage variables are displayed. Due to the dichotomous nature of the individual service use outcome variables, analyses for the six individual service outcomes are conducted using logistic regression models, while negative binomial models are used for the total usage variables. Logistic regression models are used for the individual services because the estimates produced by these models describe the odds of the event (here, whether the preventive service was used or not) occurring (Powers and Xie, 1999). For the second set of outcomes, models using negative binominal distributions are more appropriate than those using normal distributions because total usage is measured by a count variable that has a non-normal distribution (Long 1997). Regression models estimating the effect of the religion variables on the preventive service outcomes are tested first. These models do not include controls. Four models are run for each outcome, demonstrating the effect of each set of religion variables alone and combined with the other measures. Finally, multivariate models including the controls, mediators, and all of the religion variables are run. The method of progressive adjustment is used to help determine which variables are responsible for the relationship between religion and preventive services (Mirowsky, 1999). The first model includes demographic characteristics and then resources, social support, and health factors are added sequentially. Separate models with interaction terms are also evaluated. The models for the total usage variables use a different regression technique (negative binomial models 62 instead of logistic regression), but otherwise maintain the same analytical strategy. All significance levels come from two-tailed tests. For the control variables with minimal amounts of missing data (i.e. data are missing for less than five percent of the respondents), mean values are imputed. Values for those with missing data in the primary independent and dependent variables were not imputed, but are instead dropped from the respective analysis. One exception is that individuals with missing data for religious attendance were included in the never attend category (HRS, 2003). The reason for this is that individuals who did not report a religious affiliation were not asked the attendance question based on the assumption that non-affiliated individuals rarely or never attend religious services. Therefore, respondents with missing data in this category were assigned to the never attend category instead of being excluded or having values imputed to the mean. Individual-level weights provided by HRS are used in all analyses to account for sample selection probabilities, missing values, and attrition (Heeringa & Connor, 1995). Due to the complex sampling design of the HRS, the variances of the estimates in the models may be understated if a simple random sample is assumed. Since this assumption is made in most statistical software packages, another step must be taken to correct for this potential bias. Thus, adjustments for the sample design effect were made using the Taylor series linearization procedures in STATA (Stata Corporation, 2003). 63 Results Univariate Statistics Descriptive statistics are shown in Table 3. The religious attendance distribution indicates that attendance is split fairly evenly across the five categories. Fifteen percent of the sample attends services more than once a week, with approximately another quarter of the sample attending weekly. Nearly half of the respondents report attending services once or twice a year or more or never. Salience is more skewed toward the higher levels. Well over half of the sample report that religion is very important in their lives. More than a quarter of the respondents report that religion is somewhat important to them, while only ten percent report that religion is not important. The salience estimates correspond almost exactly with national estimates reported by the Princeton Religious Research Center (PRRC, 1994). Nearly two-thirds of the respondents belong to a Protestant denomination. Of these, slightly more individuals are affiliated with an evangelical denomination compared to a mainline one. Approximately one quarter of the sample is affiliated with the Catholic Church. Few respondents report a Jewish affiliation (2%) or belong to any other denomination (3%). Finally, five percent of the sample reports no religious affiliation. Descriptive statistics for the preventive service variables show that utilization levels are generally high, but vary by type of service. Mean levels of utilization range from 37 percent for flu shots to 71 percent for mammograms. Approximately 70 percent of the sample also reported cholesterol screenings and Pap smears, while over 60 percent reported breast and prostate exams. The total usage statistics indicate that women use 64 slightly more than three out of five services, while men use nearly two of three possible services, on average. These levels of use are similar to other national estimates. For example, for women 50-64 years of age, 66.5 percent reported a mammogram in the past year in 1994 and 73.7 percent reported a mammogram in 1998, according to data from the National Health Interview Survey (Pastor, Makuc, Reuben, and Xia, 2002). Likewise, of all U.S. adults over 20 years of age, 70.8 percent reported having their cholesterol screened in 1999 (CDC, 2000). While the ages and dates do not correspond exactly, the levels of utilization seen in other national samples such as these support the validity of the current data. The demographic and social characteristics of the sample are presented next. The average age of the respondents is nearly 56 years and approximately equal percentages of males and females are included in the sample. Three-fourths of the sample is nonHispanic white, with smaller numbers of non-Hispanic blacks (17%) and Hispanics (9%). Just less than ten percent of the sample was born outside of the United States. The average respondent has a high school education. One quarter of the sample has a net worth below $32,500 and the average household income is $50,000. Over 80 percent of the sample has health insurance. Social support variables show that over three-fourths of the respondents are married or living with a partner. Furthermore, high levels of satisfaction with friendships and family life are reported. The health of the sample is measured with a number of mental and physical health variables. The average self-rated mental health score indicates that most respondents rate their health as good (3) to very good (4). The mean depression score is approximately 65 27 out of 48, which indicates that the average respondent has a moderate level of depressive symptoms. Just over ten percent of the sample reports having a heart condition, with another five percent reporting cancer. Of the five other chronic conditions included, the average respondent reports having one. Finally, the subjective measure of physical health indicates that respondents rate their health as fair (2) to good (3).2 Relationships among Religion Variables Associations among the three religion variables are shown in Table 4. Specifically, differences between the mean levels of religious attendance and salience by denominational affiliation are presented. It is important to examine the relationships between these variables because they will all be included simultaneously in the final regression models. The tests show that significant denominational differences are present for all levels of attendance and salience. Most significant differences exist between members of specific denominations and those who are unaffiliated; however, differences between the specific denominations also exist. For example, Evangelical Protestants and members of Other denominations are more likely to attend religious services more than once a week, while Catholics are most likely to attend once a week. Those who are unaffiliated with any religious group are, not surprisingly, most likely to be in the never attend category (99 percent). Members of Jewish and Mainline Protestant denominations are also more likely to attend services less frequently, or never, when compared to Note that certain control and mediating variables that were specified in the conceptual framework have been excluded from the descriptive statistics and will not be included in the models. Specifically, employment status and the functional health measure have been left out of all the tables and models due to their lack of a significant relationship with any of the outcome variables. Furthermore, removing these variables from the models did not affect the relationships between the religion variables and the preventive service outcomes in any way. 2 66 members of other denominations. Within the salience categories, similar patterns can be seen. Evangelical Protestants and those in the Other category are most likely to report that religion is very important in their lives, while over half of the unaffiliated individuals report that religion is not important. Similarly, Jewish and Mainline Protestant individuals are significantly more likely than individuals of other denominations to report low levels of religious salience. To summarize Table 4, the three religion measures show several patterns of association with one another. When looking across the denomination categories, most similarities among levels of attendance are seen for the middle level (2-3 times a month), while most differences are seen at the high and low levels. At the more frequent levels of attendance, Catholics, Evangelical Protestants, and members of Other religions are more predominant, while Jewish, Mainline Protestant, and non-affiliated individuals are more prevalent in less frequent attendance categories. Similar patterns are seen within the religious salience categories. It is important to note that these three indicators of religion, while related, are not overly correlated with one another. For example, there is a substantial distribution in the attendance categories for all denominations, even Evangelical Protestants. Furthermore, a non-negligible to moderate percentage of respondents within each denomination report that religion is not important in their lives (ranging from 4 to 53 percent). 67 Regression Models Estimating the Effect of the Control and Mediating Variables on Preventive Service Use Zero-order logistic regression models are presented in Tables 5 and 6 to display the associations between the control and mediating variables and the preventive service use outcomes. Each model shows the effect of the particular control variable independent of the other variables. It is important to examine these relationships in order to better understand the place of the control and mediating variables within the conceptual framework. For instance, demonstrating an association between social support and preventive service use is an important first step in determining its potential mediating role in the relationship between religion and preventive service utilization. While the results presented below do not provide direct support for any of the hypotheses of this study, they do provide preliminary evidence for the appropriateness of the conceptual framework. Table 5 shows that older age is related to a greater likelihood of using flu shots, cholesterol screening, and prostate exams and a lower likelihood of using mammograms and Pap smears. Gender differences show that females are more likely than males to utilize flu shots. Membership in a minority racial or ethnic group, as well as non-native status, is associated with significantly lower usage for most services. The one exception is that non-Hispanic Blacks are more likely to report breast self exams. Higher education is related to greater use of all services except breast self exams. The same patterns are seen for financial resources, measured by net worth, income, and health insurance; that is, persons who are more financially advantaged are more apt to use all of the preventive services except breast self exams. 68 Marital status is strongly associated with utilization, with being married correlated with greater use of all services. The measures of social support are not consistently related to preventive service utilization, but some relationships do exist. For example, individuals with higher satisfaction with their friendships are less likely to report receiving flu shots. However, those with more satisfying friendships and family relationships are more likely to report having breast self exams. Finally, poor mental and physical health is generally significantly associated with greater use of services. Exceptions to this include a tendency for more depressed individuals to report less use of flu shots and individuals with poorer physical health to report fewer mammograms and Pap smears. Associations with the combined service use variables (Table 6) follow similar patterns as those discussed above. For example, Hispanics and foreign-born individuals tend to use fewer services, while those with more resources and more physical health problems use more. In addition, some interesting differences between men and women can be seen. For example, age is positively related to greater use for males, but not for females. Similarly, race, nativity, satisfaction with friendships, and self-rated mental health are all significantly associated with utilization for males, but not for females. The characteristics that are related to female utilization, but not male, include satisfaction with family life and depressive symptoms. Regression Models Estimating the Effect of Religion on Preventive Service Use The first set of regression models estimate the associations between religion and the preventive service use outcomes, without controls or mediating variables. These models are important to determine how each set of religion variables is associated with 69 the outcome variables, alone and with the other religion measures. Because the final regression models (discussed later) will include all of the religion measures together, these models are necessary to ascertain whether any of the religion variables overwhelm the influences of the others. As shown earlier (in Table 5), the religion measures are not entirely uncorrelated. Thus, it is possible that religious affiliation (for example) may be significantly associated the use of a particular service on its own, but when the other religion variables are included in the model, the effect disappears. Therefore, these models provide information that will be valuable in interpreting the results of the final models. The models are presented in Tables 7-10 and the results are discussed separately for each type of service and the total usage outcomes. Flu Shot The results presented in Model 1 of Table 7 show that moderate levels of attendance are associated with greater use of flu shots. Individuals who attend services two to three times a month or once a week are both at least 20 percent more likely to report a flu shot in the past two years compared to individuals who did not attend any religious services. The denomination estimates in Model 2 reveal that Catholic individuals are significantly less likely to use flu shots (O.R.=.88, p<.10), while Mainline Protestants are more likely to use flu shots (O.R.=1.15, p<.10), both compared to Evangelical Protestants. Model 3 shows that individuals who believe that religion is very important in their lives are 23 percent more likely to report a flu shot compared to individuals who say that religion is not important. When all of the religion variables are 70 included in the model together, the attendance and denomination effects remain, but the salience effect loses significance. Cholesterol Screening Model 1, on the right-hand side of Table 7, shows that individuals in all categories of religious attendance display higher odds of utilizing cholesterol screening when compared to those who never attend. The odds ratios indicate that individuals who attend religious services are 44 to 70 percent more likely to report having their cholesterol screened than those who do not attend. Three denominations- Catholic, Jewish, and Mainline Protestant report more cholesterol screening than Evangelicals (O.R.=1.15, p<.05; O.R.=1.62, p<.05; O.R.=1.23, p<.01, respectively), while individuals who are not affiliated with any denomination report usage levels that are 33 percent less than Evangelical Protestants (O.R.=.67, p<.01). Like flu shots, individuals who report the highest levels of salience are associated with greater cholesterol screening in the past year (O.R.=1.37, p<.001), compared to those reporting the lowest level of salience. In the final model, all levels of attendance remain significant, as do the denominational effects, with the exception of the non-affiliated category. The influence of salience is no longer significant. Mammogram Table 8 shows that religious service attendance is strongly related to use of mammograms at all levels. Most notably, women who attend services once a week are approximately twice as likely to report a mammogram in the last two years than women who never attend (O.R.=1.91, p<.001). Similarly, in Model 2, Jewish women are more 71 than twice as likely to report mammogram use (O.R.=2.33, p<.01) compared to Evangelical Protestant women. Mainline Protestant women are also more likely to report this type of service (O.R.=1.52, p<.001) compared to Evangelicals. Religious salience is not significantly associated with mammogram use, either in the simple or more complex model. Model 4 shows that all of the attendance and denomination variables retain their significance in the combined model. Pap Smear As seen in the right-hand side of Table 8, the results for Pap smears are very similar to those of mammograms. Specifically, individuals at all levels of religious attendance have a greater odds of using this service, compared to those who never attend. In Model 2, the estimates indicate that both Jewish and Mainline Protestant women are more likely to report having a Pap smear in the past year compared to Evangelical Protestant women. Here, Jewish women are over three times as likely to report this type of service use (O.R.=3.71, p<.01). Salience is not significantly associated with Pap smear utilization. In the combined model (Model 4), almost all of the attendance and denomination variables that were significant predictors remain so, with only the lowest level of attendance losing significance (though the estimate does not decrease in size). Additionally, in the fourth model, women who are not affiliated with any denomination are moderately more likely to report the use of Pap smears compared to Evangelical Protestants. 72 Breast Self Exams The left-hand side of Table 9 illustrates that religious attendance is only associated with the use of breast self exams for those women attending services two to three times a month. This group has 1.48 times the odds of reporting a breast self exam in the past year compared to women who never attend services. As seen in Model 2, Catholic women are moderately less likely to report this type of service than Evangelical Protestant women (O.R.=.82, p<.10), as are those who belong to a denomination in the Other category (O.R.=.70, p<.10). The religious salience model (Model 3) shows that women who consider religion to be somewhat or very important are both associated with odds of reporting breast self exams that are 40 percent higher, compared to those who report that religion is not important (O.R.=1.41, p<.05; O.R.=1.49, p<.05, respectively). These salience effects remain significant even after the addition of the other religion variables. Similarly, the attendance and denomination effects also remain significant in Model 4. Prostate Screening Model 1, on the right-hand side of Table 9, shows that religious service attendance is associated with the use of prostate screening in the same direction and with a similar magnitude as with the other outcomes. For example, men who attend services once a week are 1.48 times more likely to report having a prostate exam in the past two years compared to men who never attend. The only religious affiliation variable that is significantly related to reporting prostate screening is the lack of an affiliation. These men are 26 percent less likely to report having a prostate screening in the past year compared 73 to Evangelical Protestants. Model 3 shows that men with high levels of religious salience are associated with higher odds of reporting a prostate screening (O.R.=1.31, p<.01), compared to men with the lowest levels of salience. These findings change somewhat when all religion variables are specified in Model 4. For instance, the association between attendance and utilization is less significant at all levels and men with high levels of salience are no longer significantly more likely to report prostate screening than their counterparts who report that religion is not important in their lives. Conversely, men who are not affiliated with any religious denomination are no longer less likely to report a screening than Evangelical Protestants. Female Total Usage Table 10 shows that the female total usage variable, which consists of the five individual outcomes that are applicable to females, has a similar relationship to religion as the individual outcome results described above. That is, each of the service attendance variables are related to higher levels of female total usage compared to non-attenders, with the middle levels of attendance showing the strongest associations. These attendance findings remain significant in the combined model. Also in Model 4, Mainline Protestant use more services than Evangelicals, as do Jewish women (b=.09, p<.001; b=.14, p<.10, respectively). When salience is the only predictor in the model, the highest category is significantly associated with total usage, in comparison to the lowest category of salience (b=.08, p<.05). However, this significance disappears when the other religion variables are included in the model. 74 Male Total Usage The right-hand side of Table 10 presents the estimates for the association between religion and male total usage. Like the female total usage results, religious service attendance is strongly associated with male total usage. Compared to those who never attend, men who attend religious services one to two times a year or more use more preventive services. All levels of attendance remain significant after the other religion variables are included in Model 4. The only denominational effect for males is that nonaffiliated men are significantly less likely to use these services than Evangelical Protestant men (b=-.14, p<.05). However, this association loses significance when attendance and salience are included in the model. Finally, the highest category of salience is again significantly correlated with greater total usage in the simple model (Model 3), but not in conjunction with the other religion variables (Model 4). Summary of the Association between Religion and Preventive Service Use Despite the varied nature of the preventive services analyzed here, the relationships between the religion variables and these services are fairly consistent across outcomes. Several patterns can be discerned from the eight sets of models discussed above. The first, and most consistent, pattern is that religious attendance is strongly associated with the use of most types of preventive services. All categories of religious attendance predict the use of certain preventive services, while one category attending two to three times a month significantly predicts greater use of all services compared to the never attend category. The influence of religious attendance on preventive service use is not a dose-response effect. In other words, the association of attendance and 75 preventive health care use does not increase consistently with each increase in level of attendance. Most often, the middle two levels of attendance are associated with the most frequent utilization of preventive services. Furthermore, for most outcomes, the highest level of attendance is less strongly associated with utilization; for flu shots and breast exams, this relationship is not even significant. However, in general, the models show that, compared to individuals who never attend religious services, those who attend religious services have increased odds of preventive service use and these effects remain significant even after the other religion variables are added to the models. The magnitude and consistency of these attendance effects are not repeated by the other religion variables. Nonetheless, certain patterns are seen within the various religious denominations. Most notably, Mainline Protestants are significantly more likely to report the use of five of the six individual preventive services, as well as the female total usage, when compared to Evangelical Protestants. Even after the other religion variables are controlled for, this increase in utilization ranges from a 18 percent greater use of flu shots to a 62 percent greater use of mammograms. Jewish individuals also use more services, on average, than Evangelical Protestants. Specifically, they report greater utilization of cholesterol screening, mammograms, and Pap smears. On the other hand, Catholic individuals tend to report less preventive service utilization compared to Evangelical Protestants. For instance, Catholics report levels of flu shots and breast self exams that are 13-19 percent lower than Evangelical Protestants. Individuals who do not belong to a religious denomination also report lower levels of use than Evangelicals for many preventive services. However, these effects lose significance when other religion 76 variables are included in the models, meaning that the influence of religious affiliation is often captured by the other measures of religion. Finally, belonging to a denomination other than those specified in this study is not associated with preventive health care use compared to Evangelical Protestants. The third religion variable, salience, is most weakly related to preventive service use. While there is a positive graded relationship between salience and service use, the association is only significant at the highest level of salience compared to the lowest level. Moreover, this effect disappears when the other religion variables are added, indicating that the influence of religious salience seems to be overwhelmed by the effects of religious attendance and denomination. These patterns hold for the majority of the individual preventive services, as well as for both of the total usage variables. The only exception is for breast self exams. The following models will test whether these relationships are affected by the addition of demographic and social controls and possible mediators. Full Regression Models The full regression models, shown in Tables 11-19, will most rigorously test the hypotheses. As shown at the beginning of this chapter, these hypotheses propose that higher levels of attendance and salience will be associated with more preventive service use. Furthermore, it is expected that denominational differences in preventive service use will also be present. Specifically, Jewish individuals are presumed to use more services than Evangelical Protestants, while non-affiliated individuals are expected to use fewer services. The final hypothesis states that relationships between the religion variables and the preventive service outcomes will be moderated by certain social and demographic 77 characteristics. Each of the following tables represents a specific preventive service outcome. Sets of variables are progressively added to the models to examine the role of control and mediating variables. Each outcome is discussed separately and, following this, a summary of how the data support each of the hypotheses is provided. Flu Shot The relationships between the religion variables and the use of flu shots are presented in Table 11. The results show that individuals who attend services once a week or two to three times a month are more likely to report the use of a flu shot, in comparison to those who never attend. In fact, utilization is 25 percent higher for those who attend services two to three times a month compared to those who never attend, even after the demographic and social controls and mediators are included. Certain religious affiliations are also associated with the use of flu shots. Specifically, being Catholic is associated with twenty percent lower odds of reporting a flu shot, compared to Evangelical Protestants. Again, this decreased utilization is not influenced by the addition of the control or mediating variables. Finally, religious salience is not significantly associated with the use of flu shots in any of the four models. The control variables show similar patterns to those seen in the bivariate models with age, gender, race, and the socioeconomic resources being particularly strong predictors of service use. Most of the proposed mediators, especially social support and physical health status, are also significantly associated with the use of flu shots; however, they do not significantly alter the relationship between the religion variables and the flu shot outcome. 78 Cholesterol Screening Religious attendance is strongly associated with the use of cholesterol screenings, as seen in Table 12. All levels of attendance are associated with an increase in the odds of cholesterol screening utilization of at least 25 percent, net of other controls and in comparison to non-attenders. The addition of the resource variables (education, net worth, income, and health insurance) in Model 2 attenuates the benefits of religious attendance slightly. None of the proposed mediators significantly influence the relationship between attendance and cholesterol screening use. Religious affiliation is also associated with the use of this type of service. In particular, Mainline Protestants are more likely to report a cholesterol screening (O.R.=1.16, p<.05) in the full model, compared to Evangelical Protestants. Model 1 shows that Jewish individuals are also more likely to report this type of screening compared to Evangelicals; however, the association is accounted for by the addition of the resource variables in Model 2. Also, individuals who are unaffiliated with any denomination appear to be less likely than Evangelical Protestants to use this service, although the effect is not significant in all of the models. Salience does not have a statistically significant relationship with the use of cholesterol screening, although those with high levels of salience show some tendency to use this service more than individuals who report that religion is not important in their lives. While the bivariate models showed that minorities and those born outside of the U.S. were less likely than White and native-born individuals to report cholesterol screening, these models indicate that this disadvantage can be accounted for by controlling for socioeconomic resources. With these resource variables in the models, Hispanics 79 and foreign-born individuals are actually more likely to report this type of screening than non-Hispanics and native-born individuals in the final model (O.R.=1.31, p<.10; O.R.=1.41, p<.05, respectively). Of the possible mediators, physical health characteristics are particularly strong predictors of cholesterol screenings, although, as with previous models, the religion estimates do not change very much with the addition of these variables. Mammogram Table 13 shows that all levels of religious attendance are strongly associated with the use of mammograms, net of all controls and mediators. In the final model, the size of this effect ranges from a 34 percent increase (for those attending one to two times a year or more) in odds to almost double the likelihood of reporting a mammogram (for those attending service once a week), compared to those who never attend religious services. Like the previous outcomes, the relationship between attendance and mammogram use is weakened by the addition of the resource variables in Model 2. In all four models, Mainline Protestants are more likely than Evangelical Protestants to report having a mammogram. Similarly, Jewish women are over two and a half times more likely to report a mammogram in Model 1, compared to Evangelical women; however, this effect is also attenuated by the inclusion of the socioeconomic characteristics. Religious salience is not significant related to mammography utilization in any of the models. Within the demographic factors, younger age is moderately associated with greater usage of mammograms. Furthermore, similar to trends seen in previous models, being Hispanic is only associated with decreased utilization compared to Whites before 80 controlling for resources. In the final model, Blacks and foreign-born individuals are both more likely to report the use of mammograms, compared to Whites and native-born individuals. Socioeconomic resources and marital status are also important predictors of use, while the mental and physical health characteristics are generally unrelated to mammogram utilization. Pap Smear Other types of female preventive services, such as Pap smears, are also associated with levels of religious attendance, as shown in Table 14. Compared to women who do not attend religious services, those who attend two to three times a month are over 50 percent more likely to report a Pap smear, while those who attend once a week or more are over 30 more likely to report this type of preventive service, net of control and mediating variables. The effect of attendance on Pap smear use is attenuated by the addition of the resource variables in Model 2, but remains stable with the addition of the mediators in Models 3 and 4. Religious denomination is also associated with the use of Pap smears. In the final model, Jewish women are almost three times more likely to report Pap smears, compared to Evangelical Protestants (O.R.=2.88, p<.01). Mainline Protestants are also more likely than Evangelical Protestant women to report utilization of this service. Interestingly, women who are not affiliated with any denomination are also more likely than Evangelical women to report a Pap smear; however, this association loses significance with the addition of the resource variables. Again, religious salience is not significantly related to Pap smear utilization. 81 The estimates for the demographic control variables indicate that younger women are more likely to have a Pap smear than older women. Furthermore, once resources are taken into account, Blacks and foreign-born women are also more likely to report a Pap smear, compared to Whites and native-born women. The social and economic resource measures are again strong predictors of utilization, while the potential mediating variables have inconsistent effects. For example, social support through marriage is associated with greater use of Pap smears, but support from family and friends is not. Moreover, the addition of this set of social support variables does not alter the relationship between the religion measures and Pap smear utilization. The additions of the mental and physical health measures also do not change this relationship, although better subjective health and the total number of chronic conditions are associated with greater use of Pap smears. Breast Self Exams Religion appears to be related to the use of breast self exams differently than other preventive service outcomes. Table 15 shows that the higher categories of religious attendance are not associated with greater use of breast self exams compared to nonattenders, in contrast to previous outcomes. Specifically, only the two lowest levels of attendance are associated with the likelihood of reporting breast self exams in the final model (O.R.=1.46, p<.01; O.R.=1.24, p<.05, respectively), compared to women who never attend religious services. Furthermore, religious affiliation is not significantly related to breast exams, but religious salience is. Women who report that religion is somewhat or very important in their lives are 38 percent more likely than women who say 82 religion is not important to have a breast self exam, after controlling for demographic variables and possible mediators. Many of the demographic, social, and health measures are also significantly associated with the use of breast self exams. In particular, being Black is associated with a 67 percent increased likelihood of reporting a breast self exam compared to Whites. In contrast, Hispanic women are 29 percent less likely than non-Hispanic women to report a breast self exam. Both being married and having greater satisfaction with one s family also are related to greater use of this preventive service compared to individuals who are unmarried and have lower levels of satisfaction with their family. Finally, more depressive symptoms and poorer self-rated physical health also are associated with a greater likelihood of reporting a breast self exam. Prostate Exam Table 16 illustrates that the relationship between religion and prostate exams is relatively weak. After the inclusion of the demographic and social controls and mediating variables, only religious salience has a significant association with this preventive service. In the first model, three levels of religious attendance are significantly related to a greater use of prostate exams compared to non-attendance, but these effects lose significance with the addition of the social and economic control variables in Model 2. Within the religious affiliation variables, only a lack of affiliation is associated with prostate exam utilization when compared to Evangelical Protestants. However, this lower level of use does not reach significance in the final model. The only religious factor significantly related to prostate screening is religious salience. Higher categories of salience are associated with 83 greater use of prostate screenings. Though only marginally significant, the findings indicate that men who consider religion to be very important in their lives are 25 percent more likely to report a prostate screening, compared to men who consider religion to be unimportant. Other factors that are associated with the utilization of prostate exams include age, education, financial resources, marital status, and chronic conditions. Specifically, older men are more likely than younger men to use this service, as are the married and the more educated, compared to the unmarried and those with lower levels of education. Men with higher net worth and income, as well as health insurance, are also more likely to report a prostate screening than those with fewer economic resources. Finally, men with heart conditions and a greater number of other chronic conditions are more likely to have their prostate examined, compared to those with a healthy heart and fewer chronic conditions. The next two sets of models estimate the influence of religious attendance, affiliation, and salience on the total number of preventive services utilized. The estimates come from negative binomial regression models and the findings are presented as unstandardized regression estimates (b), in contrast to the odds ratios presented for the previous individual outcomes. The results for females are presented first (Table 17) and the models include the following preventive services: flu shot, cholesterol screening, mammogram, Pap smear and breast self exam. The models for males (Table 18) include flu shot, cholesterol screening, and prostate screening. 84 Total Female Usage As might be expected, the relationships between the religion variables and the total number of preventive services used by females follow the patterns established by the individual models. In general, attendance is the religion variable that is most strongly associated with overall levels of service utilization. The highest three levels of attendance all are associated with a greater number of services used, compared to women who never attend religious services. Attending services two to three times a month appears to be especially advantageous (b=.11, p<.001), compared to non-attending women. The findings for religious affiliation also resemble those seen earlier. For example, both Jewish and Mainline Protestant women report using more services than Evangelical Protestants. The effect of being Mainline Protestant remains significant throughout the addition of the control and mediating variables (b=.05, p<.05). Finally, religious salience is not significantly associated with the utilization of preventive services in the final model, though marginally significant effects are seen in Models 2 and 3. Findings from these two models indicate that higher levels of religious salience are associated with greater utilization of preventive services for females; however, the association is reduced to nonsignificance when the health status variables are included. Within the demographic controls, only nativity has a significant relationship with total female usage. Specifically, foreign-born women use more services than native-born women. All of the socioeconomic resources are important predictors of total service utilization. As in previous models, people with higher levels of education, income, and net worth use more services than those with fewer economic resources. Having health 85 insurance also predicts greater use when compared to women without insurance. Of the mediating variables, only marital status and the presence of cancer and other chronic conditions are associated with total female usage. Married women use more services than unmarried women, as do those with cancer and a greater number of other chronic conditions, compared to those without these conditions. Total Male Usage The association between religion and total preventive service utilization for males is shown in Table 18. As with the previous models, religious attendance has the most consistent, and significant, association with total male utilization. Here, attending religious services from one to two times a year to once a week is positively related to the use of a greater number of services for males, compared to those who never attend. This effect is strongest in Model 1 and is attenuated by the addition of the socioeconomic resource variables in Model 2. Religious affiliation is associated with total male usage only for certain denominations. In particular, men who are members of Other denominations use fewer services than Evangelical Protestant men (b=-.15, p<.10). Men who are not affiliated with any affiliation also use fewer services than Evangelicals, but this pattern does not remain significant throughout all of the models. Salience is not related to total utilization for males. The demographic and social measures show that older men use a greater number of services, as do those with more socioeconomic resources. In addition, many of the proposed mediators are also significantly associated with the use of preventive services for men. In terms of social support, those who are married use more services than the 86 unmarried, while those who are more satisfied with their family use fewer services than those who are less satisfied. The health measures reveal that men with poorer mental and physical health use more preventive services than those with better health status. Interactions Interaction models were run to determine if the relationships between the religion variables and the preventive service outcomes differ by age, gender, race/ethnicity, education, or income. Additionally, possible interactions between the religion variables were examined. Due to the extremely large number of interactions tested, some significant findings could be expected, based on chance alone. Specifically, 472 interaction terms were tested for their influence on the six individual preventive service outcomes (128 for attendance, 160 for denomination, 64 for salience, and 120 for attendance by denomination). At the .05 level of significance, one might expect approximately 24 of the 472 effects to appear significant, by chance alone. To reduce this level of randomness, significance tests were conducted at the .01 level instead of the .05 level. At this level, the increased accuracy results in an expectation of only 5 of the interaction terms appearing significant by chance. Although other methods of correction are available (e.g. Bon-Ferroni Correction Tests), this strategy is a simple way to make the tests more conservative. In contrast to the expectations proposed by the hypotheses, none of the moderators were found to be significant in any consistent manner (analyses not shown). Specifically, only fifteen of the interaction terms were significant at the .01 level. While the number of significant terms is more than could be expected if the variables were 87 completely unrelated, it does not indicate strong support for the role of the proposed moderating variables primarily because the significant interactions were spread out over the six outcome variables and no clear patterns emerged. While it still would be possible to speculate about the possible explanations (and implications) for some of the significant findings (e.g. the influence of education on breast exams is weaker for Mainline Protestants and members of Other religious denominations), the findings are too irregular and too weak to merit this type of discussion. Support for Hypotheses The findings for the six individual service outcomes and the two total usage outcomes are summarized below, in relation to the specific hypotheses. Hypothesis 1: Higher levels of attendance will be associated with more preventive service use. Reviewing the models shown in Tables 11-18, religious attendance is found to be significantly associated with utilization for almost all of the preventive services, as well as for total utilization by both genders. Although not attending any religious services is consistently associated with the lowest levels of preventive service utilization, the association between attendance and service use does not follow a dose-response pattern. In other words, the magnitude of the association between attendance and preventive service use does not increase as the frequency of attendance increases. Throughout the six individual services and the two total usage outcomes, the magnitude of the relationship between attendance and service utilization varies from 25 percent higher odds than those 88 who never attend to 86 percent higher odds, net of the control variables. These findings indicate that the data support the first hypothesis. Hypothesis 2: Levels of use will differ by denomination, with Jewish individuals reporting the highest levels of utilization and non-affiliated individuals the lowest. The second hypothesis received mixed support from the data. The first part, that denominational differences in levels of use exist, found some support. However, the predictive ability of the religious denomination variables was not consistent across the range of preventive service outcomes. The second part of the hypothesis, that Jewish individuals would have the highest levels of service use and non-affiliated individuals the lowest, was not entirely support by the data. Only for Pap smears did Jewish individuals report greater utilization than Evangelical Protestants in the final model. However, for several other outcomes, being Jewish was weakly, or inconsistently, associated with relatively higher levels of utilization. Specifically, for cholesterol screening, mammograms, and total female usage, Jewish individuals were more likely to report utilization than Evangelical Protestants. In a similar manner, individuals who are not affiliated with any denomination exhibited lower levels of utilization, though the relationships rarely remained significant in the final models. In particular, having no religious denominational affiliation was associated with a decreased use of cholesterol screening, prostate exams, and total male usage in comparison to Evangelical Protestants. Overall, the strongest finding is that Mainline Protestants are more likely to report cholesterol screening, mammograms, Pap smears, and total female services, compared to Evangelical Protestants. Other findings suggest that Catholics are less likely to report a 89 flu shot and individuals affiliated with one of the Other denominations are associated with a decrease in total male utilization, both compared to Evangelical Protestants. Therefore, preventive service utilization does differ by religious affiliation, as expected; however, it is Mainline Protestant, not Jewish, individuals who have consistently higher levels of use. Hypothesis 3: Higher salience will be associated with more preventive service use. Salience was only associated with two of the eight preventive service outcomes. Specifically, women who report that religion is a very or somewhat important factor in their lives were more likely to report breast self exams than women who report that religion is not important to them. Similarly, men who report that religion is very important to them are more likely to report the use of prostate exams, compared to those who reported that religion was not at all important in their lives. While salience was not significantly associated with the other outcomes, all of the relationships (with the exception of mammograms) were in the expected direction. Based on these findings, this hypothesis received only weak support from the data. Hypothesis 4: Relationships will be moderated by certain demographic and social characteristics. In general, the hypotheses regarding possible moderators were not supported by the data. According to the findings of the interaction tests, the relationships between the various religious variables and the preventive service outcomes are not significantly affected by gender, age, race, ethnicity, education, income, or religious denomination. In other words, religion is associated with preventive service utilization in a similar manner 90 for individuals in all of the demographic and social groups tested in this study. Summary Overall, religion appears to play a role in influencing individuals decisions to utilize preventive health services. Notably, however, the association between religion and preventive service utilization depends on both the measure of religion used and the type of preventive service. Most importantly, attending religious services is associated with increased preventive service utilization. Additionally, belonging to certain religious denominations is also related to the odds of an individual using particular preventive services. Even after controlling for numerous demographic and social characteristics, many associations between religious attendance, denomination, and preventive service use remain significant. Explanations for the relationships are quite difficult to discern from the data. Three sets of possible mediators were investigated and the findings reveal that although social support and mental and physical health status are significantly associated with preventive health care use, the relationships between the religion variables and the preventive service outcomes are not substantially affected by their inclusion in the models. Finally, while the mediators tested here did not play a significant role, attitudes toward health care may provide some explanation for the relationship between religion and preventive health care use and the following analyses will explore that possibility. 91 Chapter Five: Attitudes Toward Health Care Chapter 5 investigates the role of one potential mediator in the relationship between religion and preventive service use. Specifically, the association between religion and attitudes toward the health care system and providers is tested and discussed. To begin, the theoretical foundation for the chapter is summarized, followed by the specific hypotheses to be tested. Like the previous analytical chapter, detailed descriptions of the data, measures, and methods for the analyses are also given. Results of the are analyses then presented. Introduction The Health Belief Model proposes that ideas and attitudes, including those concerning the health care system, influence individual health care utilization levels. If religious beliefs affect how much individuals trust their physicians and the health care system, then these attitudes may mediate the relationship between religion and preventive service use. The following analyses test whether religious service attendance, affiliation, and strength of affiliation are associated with various types of health attitudes. Again, special attention is paid to subgroups for which religion may be particularly influential. Hypotheses Hypothesis 1: Health attitudes are associated with frequency of religious attendance. More positive attitudes toward the health care system and its providers are associated with more frequent attendance. Hypothesis 2: Attitudes toward the health care system and providers differ by religious denomination. 92 Hypothesis 3: Health care-related attitudes are associated with an individual s strength of religious affiliation. Individuals who report a stronger affiliation are expected to have more positive attitudes toward the health care system and its providers. Hypothesis 4: Relationships will be moderated by certain demographic and socioeconomic characteristics. Stronger relationships between the religion variables and the health care attitudes will be seen for women, older adults, and blacks. i Data and Methods Data Data for this section come from the General Social Survey (GSS). The GSS is a nationally representative survey of U.S. households fielded by the National Opinion Research Center (NORC). Over 38,000 respondents have been interviewed in the 22 surveys that have taken place since 1972 (NORC, 1999). This time span makes the GSS the longest running, current cross-sectional survey of public opinion in the U.S. The surveys cover a broad range of topics concerning demographic and socioeconomic characteristics, attitudes and beliefs, and other aspects of important social issues. The GSS was designed to foster social research through the widespread diffusion of its data. The questionnaires include several types of items: permanent questions, rotating questions, and a small number of occasional questions. The surveys also include various modules. In total, over 3,000 questions have been asked in the GSS. Extensive coverage, replication of questions, and high quality data are a few of the strengths of the GSS (GSS, 2002). For the current study, data come from the 1998 survey. In 1998, the response rate was 76.4 percent, which resulted in a total sample size of 2,832 (for more detailed 93 information on the sampling and methodology, see Davis, Smith, and Marsden, 1998). However, using a split ballot approach, only half of the respondents were asked the questions regarding health care beliefs. Respondents with missing information for the religion variables were also excluded (n=89). Of those with complete information for the independent and dependent variables (n=1,278), individuals who reported belonging to a racial group other than white or black were also excluded (n=81). This resulted in a final sample size of 1,197. For the control variables with minimal amounts of missing data (i.e. data are missing for less than five percent of the respondents), mean values are imputed. Measures Attitudes Toward Health Care. The dependent variables of interest involve respondents attitudes toward the health care system and providers. In 1998, the GSS included a section on health-related beliefs that was asked of approximately one-half of the respondents for that year. Twenty questions were asked to gauge attitudes toward health care providers and perceptions about the quality of care provided by the health care workers. The attitudes were assessed by asking individuals to respond to the provided statements in the following manner: strongly agree, agree, uncertain, disagree, or strongly disagree (NORC, 1999). Negative prompts were reverse coded so that more positive responses (agree and strongly agree) consistently reflect more positive attitudes toward the health care providers or the health care system. Those who responded, Don t know were excluded from the analyses, as were those who provided no answer. 94 An exploratory factor analysis was done to reduce the large number of questions into a smaller number of latent factors. The factor analysis was done using proc factor in SAS, with promax chosen to obtain an oblique rotation. The results showed four underlying constructs. Cronbach Alpha values were then calculated to further assess the four factors. In addition to these quantitative analyses, information on past use of these variables was gathered from the literature (Pescosolido, Tuch, and Martin, 2001). From this combination of sources, the following three indices were created: personal trust in one s physician, public confidence in physicians, and attitudes toward the health care system. Two questions ( Doctors always avoid unnecessary patient expenses and I hardly ever see the same doctor when I go for medical care ) were excluded because they exhibited a poor fit with the other questions. The individual questions that are included in each index are described in Table 19. Religion. Three aspects of religion are measured: attendance, denomination, and strength of affiliation. The attendance variable is measured with the following question: How often do you attend religious services? (NORC, 1999). Nine categories were given to prompt respondents. These categories were then collapsed into five broader groups. Respondents who reported that they attend services several times a week were placed in the highest attendance group. Respondents who reported that they attend services every week were placed in the high attendance group. Those who reported going to services 2-3 times a month or nearly every week were labeled medium attenders. Respondents who attend services less than once a year, once or twice a year, or about 95 once a month were placed in the low attendance category. Finally, those who reported never going to religious services were placed in the never group. The religious denomination variable is measured with a question that asks the respondent for their religious preference. This preference is then clarified with the phrase, Is it Protestant, Catholic, Jewish, some other religion, or no religion? (NORC, 1999). If the response is no religion, the interviewer moves on to the next question. If the religious preference given is Protestant or Jewish, the respondent is then asked to specify a denomination. Like the previous analyses, seven denomination categories were created (Catholic, Evangelical Protestant, Mainline Protestant, Black Protestant, Jewish, other religion, and non-affiliated), largely based on the Steensland et al. (2000) classification scheme. Strength of affiliation is included as a proxy for religious salience. It is measured by a question that asks, Would you call yourself a strong (preference named in previous question) or a not very strong (preference)? (NORC, 1999). Response choices were strong, not very strong, or no religion. In addition, significant numbers of voluntary responses were given for somewhat strong. These were recoded into a variable ranging from 1 (no religion) to 4 (strong). Demographic and Social Factors. The models include several demographic and socioeconomic control variables. Age is measured as a continuous variable for adults over 18 years of age. Gender is measured as a dichotomous variable. Race is included as a dichotomous measure of whether the individual is white or black. All other race and ethnic groups (including Hispanics) are combined into one category in the data and, 96 therefore, are excluded because of the difficulties this causes with interpretation. Nativity distinguishes between respondents who were born in this country and those who were not. Finally, marital status is categorized as currently married or not. Resources. The socioeconomic variables included in the analyses are education, income, and health insurance. Education is measured with a question that asks for the respondent s highest level of completed education and is coded as a continuous variable. Income is measured with the following question: In which of these groups did your total family income, from all sources, fall last year before taxes? (NORC, 1999). Respondents were given a card with the response category choices and were instructed to report only the letter that corresponds with the correct category. There are 23 categories ranging from Under $1,000 to $110,000 or over. Individuals who refused to answer or who reported that they did not know were imputed to the mean (approximately 10.9 percent of the sample). A missing income variable was created to determine if those with missing information were significantly different. In the full models, this variable was not significant and, thus, was excluded from the models. Individuals who have any type of health insurance are placed in the insured category. A measure of health was also included to control for the possible confounding effect of physical health status. Self-rated health was measured as a scale ranging from poor (1) to excellent (4). Methods Like the previous analyses, the univariate and bivariate statistics are first calculated, followed by the multivariate regression models. The regression models for 97 this section use ordinary least squares regression and the estimates are presented as unstandardized regression coefficients. This type of model is most appropriate because of the ordinal nature of the outcomes used here. Three sets of models are run, one for each of the health care attitude indices. Each set includes simple models to analyze the relationship between religion and health care attitudes, as well as more complex multivariate models that control for the demographic, social, resource, and health characteristics. Results Univariate Statistics Descriptive statistics are shown in Table 20. The religious attendance distribution shows that relatively few individuals in this sample (7 percent) attend services more than once a week. Nearly 40 percent attend either once a week or two to three times a month and almost another 40 percent attend one to two times a year or more. Nineteen percent report never attending. The affiliation distributions show that the majority of the sample is Catholic or Protestant. Most notably, slightly more than one-third of the sample is Evangelical Protestant. Small percentages report a Jewish or Other affiliation, while fifteen percent report no affiliation at all. The strength of affiliation mean indicates that the average respondent considers herself/himself a somewhat strong member of her/his particular denomination. The distribution of the attendance categories is significantly different from the HRS sample when looking at all ages. However, when the GSS sample is restricted to individuals between the ages of 51 and 61 (to match the HRS sample), the differences 98 are reduced. When comparing these two samples (not shown), only the lowest two levels of attendance are significantly different. This difference is most likely due to dissimilar question wording. More specifically, the GSS uses nine categories of attendance, compared to five in the HRS. In fact, the GSS includes four categories of very low attendance that are covered by just one category in the HRS. Allowing this level of specification for those at this end of the spectrum increases the number of individuals in the second lowest category and decreases the number of individuals in the never attend category. Except for this difference, the attendance distributions within that age range are similar. The three dependent variables are personal trust in physicians, public confidence in physicians, and attitudes toward the health care system. Because each variable was constructed by summing the responses for the relevant questions (which each have five response choices, ranging from strongly disagree to strongly agree), the means can be interpreted by determining the average answer for the questions in that category. More specifically, the mean response is determined by dividing the possible maximum value by five (to determine how many questions are included in each scale) and then dividing the mean value by this number. Thus, levels of personal trust in physicians and attitudes toward the health care system are slightly positive (3.59 and 3.40, respectively), with the average response of most respondents being between neutral and agree, while the level of public confidence in physicians (3.03) is neutral, on average. The demographic characteristics are presented next. The average age of the sample is 46 years and slightly over half of the respondents are female. The vast 99 majority of the sample is white (85 percent) and the other 15 percent are Black. Only five percent of the respondents were born outside of the U.S. In addition, approximately half of the sample is married. The socioeconomic variables indicate that the average education is just over a high school degree, the average household income is within the $30,000$34,999 category, and most respondents have health insurance. Finally, the average respondent reports good self-rated health. Relationships among Religion Variables Table 21 demonstrates the relationships between the three religion variables by examining differences in attendance and strength of affiliation by religious affiliation. The tests show that significant denominational differences are present for all levels of attendance and strength of affiliation. As expected, members of specific denominations and those who are unaffiliated show the most significant differences; however, there are also differences between the specific denominations. For example, over twice as many Evangelical Protestants attend religious services more than once a week compared to Mainline Protestants. In fact, Evangelical Protestants are significantly more likely to attend more than once a week than all of the other denominational groups, except Other. Similarly, Catholics are more likely to attend once a week than other groups, while significantly more Jewish, Other, and non-affiliated individuals fall into the never attend category. Strength of affiliation also differs by denomination. Evangelical Protestants and members of Other denominations rate themselves as more strongly affiliated than do members of the other denominations. 100 In general, there are several patterns of association among the three religion variables. When looking across the denomination categories, Catholics, Evangelical Protestants, and Mainline Protestants generally have higher levels of attendance, while Jewish, Other, and non-affiliated individuals are more prevalent in less frequent attendance categories. Similar patterns are seen for the strength of affiliation variable. While patterns do exist, it is important to note that there is considerable variation across the attendance categories for all denominations, as well as differences in strength of affiliation by denomination. These patterns are largely consistent with the relationships seen in the HRS data. Regression Models Estimating the Effect of Religion on Attitudes Toward Health Care The first set of regression models estimates the association between the independent and dependent variables before the inclusion of the confounding or mediating variables. These models are important to determine how each set of religion variables is associated with the outcome variables, alone and with the other religion measures. The full regression models (presented later) include all of the religion measures together, so these models are helpful to determine whether any of the religion variables overwhelm the influences of the others. The models are presented in Tables 4-6 and the results are discussed separately for each outcome. Personal Trust in One s Physician Table 22 displays the association of the three religion variables with personal trust in one s physician, as well as a more complex multivariate model that includes all three religion measures together. Model 1 shows that higher levels of attendance generally 101 are associated with a greater level of trust in one s physician. While attending more than once a week is not related to higher levels of trust, each of the other attendance categories is positively, and significantly, associated with trust when compared to individuals who never attend services. Religious affiliation is also associated with level of trust in one s physician. As seen in Model 2, Catholics, Jews, and Mainline Protestants have significantly higher levels of trust, while members of Other affiliations have a significantly lower level, compared to Evangelical Protestants. Finally, greater strength of affiliation is related to more trust in one s physician. When these three religion variables are included together (in Model 4), all of the findings remain significant. Higher levels of attendance are still generally associated with greater trust, though the magnitude of these associations are somewhat reduced. Affiliation effects are also still seen, though the influence of belonging to a Catholic or Other affiliation, in comparison to Evangelical Protestants, is decreased to some extent. In a similar manner, the association between strength of affiliation and personal trust in one s physician is also attenuated by the addition of the other religion variables, but remains significant. Public Confidence in Physicians The relationship between the religion variables and public confidence in physicians is shown in Table 23. All levels of religious attendance are strongly associated with increased public confidence in physicians compared to individuals who never attend services. The effect is not graded, but instead is strongest for those attending once a week. Within the denomination categories, Mainline Protestants tend to report higher levels of public confidence, while non-affiliated individuals report significantly lower levels (both 102 in comparison to Evangelical Protestants). Again, greater strength of affiliation is related to higher levels of confidence in physicians. In Model 4, the magnitude and significance of all of the religion variables is reduced. The attendance variables all remain significantly associated with public confidence, though the effect sizes are smaller. Mainline Protestants still report higher levels of public confidence in physicians than Evangelical Protestants, but the effect for non-affiliated individuals is no longer significant. Finally, greater strength of affiliation is now only modestly associated with higher levels of public confidence. Attitudes Toward the Health Care System Table 24 displays the associations between attendance, denomination, and strength of affiliation and the general attitudes toward the health care system. Like the previous models, individuals who attend religious services more frequently exhibit more positive attitudes toward the health care system compared to individuals who never attend. This association is seen for all but the lowest level of attendance. Jewish and Mainline Protestant individuals also have more positive attitudes toward the health care system than Evangelical Protestants. Not being affiliated with any denomination is associated with more negative views of the health care system, compared to Evangelical Protestants. Strength of affiliation is again associated with more positive attitudes, as seen in Model 3. The full model shows that the attendance and strength of affiliation effects are reduced when all religion variables are included in the same model, with strength of affiliation reduced to non-significance. In contrast, the coefficients for Jewish and Mainline 103 Protestant affiliations are larger in Model 4. This represents a suppression effect caused by a negative association between attendance and these affiliations. Full Regression Models Although the above models provide some degree of support for the original hypotheses, the full regression models, shown in Tables 25-27, offer more stringent tests of these expectations. As discussed at the beginning of the chapter, the hypotheses propose that higher levels of attendance and strength of affiliation will be associated with greater personal and public trust in physicians as well as more positive attitudes toward the health care system. In addition, it is expected that denominational differences in these attitudes will also be present. Finally, relationships between the religion variables and these attitudes will be moderated by certain social and demographic characteristics, including age, gender, race, education, and income. Tables 25-27 show results from the three different outcome variables. The method of progressive adjustment is used to determine the effects of the control and mediating variables. Each outcome is discussed separately and, following this, a summary of how the data support each of the hypotheses is provided. Interaction models were run to determine if the relationships between the religion variables and the health attitudes differed by age, gender, race, education, or income. Due to the large number of interactions tested, some significant findings could be expected, based on chance alone. Specifically, 183 interaction terms were tested for their influence on the three outcome variables (60 for attendance, 75 for denomination, 15 for strength of affiliation, 15 for attendance by denomination, 15 for denomination by strength of 104 affiliation, and 3 for attendance by strength of affiliation). To make the tests more conservative, .01 significance levels were used instead of .05 levels. At the .01 level, the increased precision results in an expectation of only two of the interaction terms appearing significant by chance. Since more than two significant interaction terms were found, and some patterns can be seen, the significant interaction effects are displayed in the following tables (Tables 25-27). Additionally, possible interactions between the religion variables were examined and are discussed separately (Table 28). Finally, additional analyses were done to further explore the relationship between religion and attitudes toward the health care system. A new outcome variable, measuring trust in the individuals running medical institutions, was tested in a similar manner to the previous models. Furthermore, tests were done to discern whether or not religion is associated with general trust in all people. Finally, a previously untested aspect of religion, biblical literalism, is included in the models. These supplemental analyses take advantage of additional information available in this data set to help to shed more light on the relationships discussed above. Personal Trust in One s Physician The association between religion and personal trust in one s physician is seen in Table 25. As in the previous sets of models that did not include demographic and social covariates, religious service attendance is related to levels of trust. Except for the highest level of attendance, more frequent attendance is associated with greater trust compared to individuals who never attend services, even after controlling for demographic factors (Model 1). This association remains significant with the addition of the socioeconomic 105 and health variables in Model 2, although the estimate sizes decrease slightly. Religious affiliation is also associated with amount of personal trust in physicians. Specifically, Catholic, Jewish, and Mainline Protestant individuals report greater trust in their physicians than do Evangelical Protestants. The estimate for Jewish individuals is especially strong and remains so even after the social resource variables are included in Model 2 ( =1.71, p<.01). In contrast, individuals who are affiliated with Other denominations have significantly lower levels of trust in their own physicians, compared to Evangelical Protestants ( =-2.65, p<.05). Again, these effects remain significant after controlling for possible confounding variables. Finally, strength of affiliation is also related to levels of personal trust in physicians. Though only marginally significant, individuals who report a stronger affiliation with their denomination have greater levels of trust compared to those with weaker affiliations ( =.36, p<.10). Other variables that are associated with personal trust in doctors include age, gender, race, income, health insurance, and self-rated health. Older individuals have greater levels of personal trust, as do males, Blacks, and those with lower incomes, health insurance, and better subjective health. It is interesting to note that nativity, marital status, and education are not associated with levels of trust in one s physician in this sample. The interaction terms shown in Models 3 and 4 indicate that the relationship between religion and personal trust in one s physician is moderated by education. Specifically, increases in education result in greater trust in physicians only for individuals who are in the highest two categories of attendance (more than once a week and once a 106 week). That is, individuals who are both highly educated and religiously involved report a high level of trust in their physicians. In contrast, increases in education lead to significantly lower levels of trust in physicians for individuals belonging to Other denominations. Interactions between the religion variables and age, gender, race, and income were not significant. Public Confidence in Physicians Table 26 shows the OLS estimates for the relationship between religion and public confidence in physicians. Religious service attendance has a strong and consistent association with public confidence in physicians. The effect is seen for all levels of attendance and remains throughout the addition of the demographic, social, and health variables. In general, individuals who attend services more frequently report greater public confidence in doctors than those who never attend services. However, the pattern is not fully graded. For example, attending services more than once a week is significant, but it does not display the largest coefficient. Religious affiliation also matters, though only for Mainline Protestants. Specifically, Mainline Protestants report greater public trust in doctors than do Evangelical Protestants ( =.69, p<.10). The final measure of religion, strength of affiliation, is not related to public confidence in physicians. The demographic, social, and health variables that are associated with public confidence in physicians include age, nativity, health insurance, and self-rated health. Again, older individuals report higher levels of confidence. Greater public confidence in physicians is also perceived by native-born respondents, those with health insurance, and those with better subjective health. There were no significant interaction terms, 107 indicating that the relationship between religion and public confidence in doctors remains constant throughout the social and demographic groups examined here. Attitudes Toward the Health Care System The relationships between aspects of religion and attitudes toward the health care system are displayed in Table 27. Unlike the previous outcomes, general attitudes toward the health care system are not related to religious attendance after the demographic, social, and health variables are included in the models (Model 2). However, similar to previous models, significant patterns were seen within religious affiliations. In particular, Jewish and Mainline Protestant individuals both have more positive attitudes toward the health care system than Evangelical Protestants ( =1.38, p<.05; =.55, p<.01, respectively). These effects decrease in size and significance when the social resource and health variables are added in Model 2, but are still marginally significant. Strength of affiliation was not related to attitudes toward the health care system. Other variables with significant relationships with attitudes toward the health care system include age, marital status (marginally), education, health insurance, and self-rated health. Being older, married, more highly educated, in better health, and possessing health insurance are all related to more positive attitudes toward the health care system. Interactions between religion and attitudes toward the health care system were also tested. Significant effects were seen for the interaction between religious denomination and education. In particular, the positive association of education with these attitudes is weaker for Catholic individuals. Again, no significant interactions are seen for age, gender, race, or income. 108 Interactions Among Religion Variables Interactions were tested among all three of the religion variables for each outcome. Personal trust in one s physician was found to be associated with multiplicative combinations of the religion variables. As seen in Table 28, the interactions between denomination and attendance and those between denomination and strength of affiliation have significant effects. More specifically, using the .01 level of significance, the interaction terms for Catholicism and attendance ( =1.19, p<.01), no affiliation and attendance ( =1.68, p<.01), and those for Other denomination and strength of affiliation ( =-3.65, p<.01) are significant. The first two interaction terms indicate that the positive association between attendance and personal trust in one s physician is stronger for Catholics and those not affiliated with any religious organization. In contrast, the third interaction term suggests that the positive relationship between strength of affiliation and personal trust is much weaker for those belonging to Other denominations. None of the other interactions between the religion variables were significant. Models with Individuals Between 51-61 Years of Age As discussed earlier, the previous analyses were conducted to discern the association between religion and health care-related attitudes in order to determine if these types of attitudes may help to explain the relationship between religion and preventive health care use. In order to more closely match the sample used in the HRS analyses, the full regression models seen in Tables 25-27 were run again, this time with the sample limited to adults ages 51-61. This age limitation reduces the sample to 177 respondents. In contrast to the findings using the whole GSS sample, no significant findings were 109 found using this subsample (analyses not shown). The results suggest that religion is not associated with personal trust in one s physician, public confidence in physicians, or attitudes toward health care for adults ages 51-61; however, the lack of significant findings is most likely because of the very modest sample size of adults in this age range of the 1998 GSS. Testing Other Outcome Variables There are two other variables in the 1998 GSS that may provide further insight into the relationship between religion and attitudes related to health care. One question concerns trust in the individuals running the medical system and the other is a measure of general trust in all people. The first question is included to further investigate how religion may be related to health-related attitudes and the second question may help determine if differing levels of trust in health care providers and the medical system by religious involvement or religious affiliation is a reflection of differences in levels of trust in people in general. Confidence in medicine is measured with a question that asks about the amount of trust the respondent has in the individuals running certain institutions, such as banks, major companies, education, and medicine. The exact wording and descriptive statistics are shown in Table 29. Multinomial logistic regression models examining the association between this variable and the religion variables are shown in Table 30. This type of model was chosen because it is the most appropriate choice for generating estimates with categorical outcome variables. The results, presented here in odds ratios, are similar in direction to those found for the previous outcome variables. For example, religious 110 attendance is strongly related to levels of confidence in medical institutions. Individuals who attend services more frequently have higher levels of confidence when compared to individuals who never attend services. This effect remains significant throughout the addition of the demographic, social, and health variables. In fact, in the final model, individuals who attend services more than once a week are approximately seven times more likely to have high levels of confidence in the people running medical institutions then those individuals who never attend services (O.R.=7.05, p<.01). In contrast, most religious affiliations are unrelated to confidence in the people running medical institutions. Only individuals who report belonging to Other denominations are significantly less likely to report a higher level of confidence compared to Evangelical Protestants. In fact, in the full models (Model 3), individuals belonging to Other denominations are over 80 percent less likely to have a medium or high level of confidence, compared to Evangelical Protestants. Finally, unlike in the models for the previous outcomes, strength of affiliation shows a modest negative association with confidence in medicine. In other words, the more strongly individuals are affiliated with their religion, the less confidence they have in the people running the medical system. However, this effect is only moderately significant for one of the models (Model 1, high). Other related variables include gender, race, education, health insurance, and self-rated health. Females, Blacks, and those with health insurance, better education, and better health report higher levels of confidence in the people running medical institutions. The second additional outcome tested here is general trust in people. This variable is measured with an item that asks, Generally speaking, would you say that 111 most people can be trusted or you can t be too careful in life? (GSS Codebook, 1999) and the distribution is described in Table 29. While the response choices were implied in the question, a certain number of respondents replied with some other answer (such as It depends ). These responses were excluded from the analyses, leaving a total sample size of 1,138. The results of the logistic regression models shown in Table 31 indicate that the relationship between religion and trust in health care providers and the health care system differs from the relationship between religion and general trust in people. In fact, religion is generally unrelated to general trust in people, after controlling for demographic, social, and health characteristics. As seen in the top section of each model, religious attendance is completely unrelated to general trust in people. Although religious affiliation is associated with general trust in people in Models 1 and 2, these effects are reduced (or eliminated) with the addition of the demographic, social, and health covariates. In the final model (Model 3), only Mainline Protestants are more likely than Evangelical Protestants to have high levels of trust in people and this result is only moderately significant (O.R.=1.41, p<.10). Other characteristics that are related to general trust include age, race, education, and subjective health. Individuals who are older, white, more educated, and in better health report higher levels of general trust in people than do those who are younger, non-White, less educated, and in poorer health. Additional Religion Measures There are many other measures of religion in this data set; however, the GSS uses a split sample design in which individuals within the sample receive different sets of 112 questions. Therefore, most other measures of religion could not be analyzed in conjunction with the health attitude questions. Of the other religion items, one that could be expected have an association with health attitudes, based on previous research or theoretical frameworks, is biblical literalism. This aspect of religion was not included in the previous analyses for two reasons. One, nearly 400 individuals (one-third of the sample) did not have information for this variable. Secondly, biblical literalism was not included in the HRS analyses and, therefore, the addition of this variable to the main analyses in this section would not address the theoretical links posed earlier in this dissertation. However, to take advantage of the data and to further explore the association between religion and health care-related attitudes, the relationship between biblical literalism and the outcome variables is discussed below. Biblical literalism, which is often used as a measure of religious conservatism, is measured as follows. Respondents were given three statements regarding the nature of the Bible and asked to select the one that most closely describes their feelings. The choices were as follows: The Bible is an ancient book of fables, legends, history, and moral precepts recorded by men, The Bible is the inspired word of God but not everything in it should be taken literally, word for word, and The Bible is the actual word of God and is to be taken literally, word for word. Correspondingly, the variable ranges from one to three. OLS models are used to estimate the effects of biblical literalism and the other religious variables and covariates on the three original health-related attitude outcome variables. 113 As seen in Table 32, biblical literalism is not strongly associated with any of the three health attitude outcomes. In fact, when the other religion measures and covariates are included in the models (Model 3 for each outcome), biblical literalism is only related to one of the three outcomes, personal trust in one s physician, and this association is only marginally significant ( =.59, p<.10). Individual interpretations of the Bible also are associated with public confidence in physicians; however, this association disappears when the other religion variables are included in the model. Biblical literalism is completely unrelated to general attitudes toward the health care system, as seen in the right-hand side of the table. Overall, biblical literalism is moderately associated with how much individual s trust their own physicians, but it appears to be unrelated to perceptions of public confidence in physicians and overall attitudes toward the health care system. Support for Hypotheses The findings for the three sets of models are summarized below, in relation to the specific hypotheses. Hypothesis 1: Health attitudes are associated with frequency of religious attendance. More positive attitudes (personal and public) toward health care providers are associated with more frequent attendance compared to those who never attend, as seen in Tables 25 and 26. Note, however, that this relationship is not graded. For example, the highest level of attendance has a smaller and weaker effect than the middle levels of attendance for both outcomes. In contrast to these positive relationships, religious attendance is unrelated to general attitudes toward the health care system, once 114 demographic, social, and health characteristics are taken into account (Table 27). In other words, there is strong support for the relationship between attendance and attitudes toward health care providers, but no support for the association between attendance and attitudes toward the health care system. Therefore, the data provide good, but not complete, support for the first hypothesis. Hypothesis 2: Attitudes toward the health care system and providers differ by religious denomination. Trust in the health care system and health care providers appears to differ by religious affiliation. The main finding, consistent for all three primary outcomes, is that Mainline Protestants have more trust in health care providers and the health care system than do Evangelical Protestants. Other affiliations, most notably Catholic and Jewish, are also associated with higher levels of trust than Evangelical Protestants for most of the outcomes. In contrast, members of Other affiliations tend to have lower levels of trust than Evangelical Protestants. Thus, this hypothesis is supported by the data. Hypothesis 3: Health care-related attitudes are associated with an individual s strength of religious affiliation. Only personal trust in one s physician is related to strength of affiliation. Specifically, individuals who are more strongly affiliated with their religion have greater levels of trust in their physicians. Public confidence in physicians and attitudes toward the health care system are unrelated to strength of affiliation. Thus, this hypothesis receives only weak support from the data. 115 Hypothesis 4: Relationships will be moderated by certain demographic and socioeconomic characteristics. Stronger relationships will be seen for women, older adults, and blacks. Several significant interactions were seen. Most notably, the association between education and personal trust in one s physician is stronger for individuals who attend services more frequently (compared to non-attenders) and the positive effect of education on personal trust in physicians and attitudes toward health care is weaker for certain denominations. Significant interaction terms were also seen within the religion variables. Specifically, the positive association between attendance and personal trust in one s physician is stronger for Catholics and the positive relationship between strength of affiliation and personal trust is weaker for those belonging to Other denominations. No differences in the relationships between the religion variables and the outcomes were seen for other subgroups, such as racial minorities or females. Finally, tests for age interactions also produced insignificant results. Thus, the primary relationships examined here remain stable over most demographic and social characteristics; however, education and other religion variables do moderate certain relationships. Summary The results from the current study show that religious attendance, denomination, and strength of affiliation are significantly associated with attitudes toward health care, even after controlling for numerous confounding demographic and social variables. Individuals who attend services, belong to a Mainline Protestant or Jewish denomination, and are more strongly affiliated with their denomination generally have more positive 116 attitudes toward health care providers and the health care system. These relationships remain constant over most demographic and social characteristics, except for education. Since general levels of trust in people do not vary in the same manner by religion, the more positive attitudes toward the health care system by individuals with greater religious involvement and within certain affiliations seem to reflect an interesting pathway by which religion may be related to more positive health outcomes in the United States. 117 Chapter Six: Discussion and Conclusion Preventive health services have many potential benefits, including improved physical health status, increased longevity, enhanced quality of life, and lowered health care costs (American Cancer Society, 1998; Centers for Disease Control and Prevention, 2002a; Janes et al., 1999; Nichol, Margolis, Wuorenma, and Von Sternberg, 1994; U.S. Department of Health and Human Services, 2000). For these reasons, public health programs are increasingly focused on the role of preventive services in improving overall levels of health in our country. For example, Healthy People 2010 has several objectives related to preventive service utilization, including increasing insurance coverage of these services, improving health promotion and disease prevention training for all health care providers, and providing more funds for population-based prevention research (USDHHS, 2000). However, utilization rates for these types of services are currently less than ideal and new strategies to increase levels of use are needed. Religious organizations, as well as other facets of religion, may provide an important means of reaching individuals and increasing utilization by expanding access, offering information, and increasing motivation. The current study illustrates the significant role that religion plays in influencing the use of six types of preventive health services and also examines the possible mediators and moderators of this relationship. These additional tests reveal that attitudes toward health care may help to explain the relationship between religion and preventive service use and that the relationships appear to remain stable across most social and demographic groups. 118 Summary of Findings: Religion and Preventive Service Use In the first half of the study, data from a nationally representative, longitudinal data set are used to examine the influence of religious attendance, denomination, and salience on six individual preventive services and levels of total utilization by gender. As expected, individuals who attend religious services are more likely to report the use of preventive services compared to individuals who do not attend services. This effect is strong (often associated with an increased odds of 25 percent or more) and consistent (significant for five of the six individuals service outcomes and both of the total utilization variables). Individuals from certain religious denominations are also more apt to use preventive services. In general, Mainline Protestants are more likely to use preventive health care, compared to Evangelical Protestants. Of the other denominations, Catholics, members of Other denominations, and non-affiliated individuals are less likely to use certain services than Evangelical Protestants, while Jewish individuals tend to use more services. The final aspect of religion tested in the current study, religious salience, has the weakest relationship with preventive service utilization. The results show that salience predicts the use of breast self exams and prostate screening, but is not associated with the other outcomes. To examine these relationships more closely, potential mediators were tested. While the possible mediating role of attitudes related to health care could not be tested with the HRS, several other potential mechanisms, such as social support, and mental and physical health, were investigated. The findings show that, although these factors are significant predictors of preventive service utilization, the influence of religion on 119 preventive service use is largely independent of them. Thus, this portion of the study sheds little light on the possible mechanisms through which religious involvement and affiliation affect the use of preventive services. Also of interest is whether or not the relationship between religion and preventive health care utilization is similar for all individuals. Interaction tests were run to determine the moderating role of several social and demographic characteristics that are known to influence religious involvement and other predictors of preventive health care use. These factors- gender, age, race, ethnicity, education, and income- may alter the relationship between religion and preventive service use in two ways: by amplifying existing resources or by compensating for an absence of other resources. Little prior research was available on which to make predictions about the direction of the effects; however, the findings show that, within this data set and with the available measures, the relationships between the religion and preventive service variables remained constant and, therefore, neither of the two perspectives are consistently supported by the data for any of the outcomes. Summary of Findings: Religion and Attitudes Toward Health Care The purpose of the second portion of the study is to examine the relationship between religion and different attitudes toward the health care system and health care providers. It was hypothesized that these health-related attitudes may partially explain the connection between religion and preventive health care utilization. Three different sets of attitudes were examined: personal trust in one s physician, public confidence in physicians, and attitudes toward the health care system. The findings reveal that 120 various aspects of religion are, in fact, associated with attitudes toward health care. Of the religion variables, attendance has the strongest relationship with attitudes toward health care. The results for each of the three outcome variables are discussed below. Personal trust in one s physician is associated with all three of the religion measures. Individuals who attend religious services more often are likely to have higher levels of trust in their doctors compared to those who never attend services. Among the religious denominations, greater trust is positively associated with membership in a Catholic, Jewish, or Mainline Protestant denomination (compared to Evangelical Protestants). In contrast, individuals who belong to an Other denomination are likely to have less trust in their physicians. Finally, greater strength of affiliation is associated with higher levels of personal trust. The relationships between religion and public confidence in physicians are similar to those discussed above. Namely, religious attendance has the strongest association with public confidence, in which higher levels of attendance are related to greater trust compared to individuals who never attend services. Mainline Protestants again have greater levels of confidence in doctors, compared to Evangelical Protestants. In contrast to the previous outcome, strength of affiliation is unrelated to public confidence in physicians. The final category of health-related attitudes measures general feelings about the health care system. Unlike the other measures, these general attitudes are unrelated to religious attendance. However, significant relationships with the denominational measures still exist. Specifically, Jewish and Mainline Protestant individuals have 121 more positive attitudes toward the health care system than do Evangelical Protestants. Strength of affiliation is not significantly associated with attitudes toward health care. Additional analyses of this data set revealed that religious attendance and affiliation were also associated with levels of confidence in the people running medical institutions. More specifically, individuals who attend religious services more often have higher levels of confidence, compared to those who never attend services. Within the denomination categories, individuals who belong to Other affiliations have less confidence in the people running the medical institutions than do Evangelical Protestants. While it is possible that this health-related attitude, along with the other three discussed above, may be due to more religiously involved individuals having higher levels of trust in all people, this idea is not supported by the supplemental analyses. Religious attendance was found to be unrelated to general trust in people. However, belonging to a Mainline Protestant affiliation was associated with greater levels of trust in people. Thus, it is possible that the denominational differences in attitudes toward the health care system, but not differences by level of attendance, may be at least partially due to differences in levels of general trust. Finally, biblical literalism was found to be moderately associated with personal trust in one s physician, but unrelated to the other outcomes. Placing Findings within the Theoretical Framework While identifying support (or lack of support) in the data for each of the individual hypotheses is important, it is also beneficial to place the results in a broader context. Thus, the original theoretical framework (presented in Figure 1 and discussed 122 in Chapter 3) is used to review the expected relationships between all of the variables included in the models. Demographic and Social Factors and Resources To begin, this framework proposes that the relationship between religion and preventive service utilization is affected by several groups of factors. Accordingly, social, demographic, and economic control variables are included in the models as possible influences on religion (as well as on other mediators and preventive service utilization). As seen in Tables 5 and 6, all of these control variables are important predictors of preventive service utilization. For example, older adults are more likely to use preventive services, with the exception of the female-specific ones. Similarly, women and married individuals generally report higher levels of utilization. Social and economic characteristics are especially strong predictors of preventive service use, with more socioeconomic resources predicting greater utilization for all preventive services but breast self exams. These relationships follow patterns seen in previous studies (Barr et al., 2001; Fox et al., 1998; Hewitt, 2002; Kirkman-Liff, 1992; Nelson, Norris, and Mangione, 2002; Yi, 1994; Yi, 1998). Religion While the above findings simply confirm relationships that have been welldocumented in the literature, the next section of the framework examines relationships that have received limited attention from previous researchers. The theoretical foundation underlying these key relationships presumes that various facets of religious involvement and beliefs influence an individual s use of preventive health care 123 services. As discussed above, and in accordance with previous studies (Benjamins and Brown, 2003; Miller and Champion, 1993; Miller, Norcross, and Bass, 1980; Murray and McMillan, 1993; Naguib, Geiser, and Comstock, 1968; Wan and Yates, 1975; Yi, 1994, 1998), the data confirm the connection between religion and preventive service utilization. These relationships are particularly robust when only the religion and outcome variables are included (Tables 7-10), but often retain their significance even after the addition of all of the control and mediating factors (Tables 11-18). While religious service attendance is the strongest predictor of utilization, both denomination and salience are also associated with different types of preventive services. All three facets of religion were hypothesized to have significant relationships with preventive service use, although the mechanisms responsible for these associations were expected to differ. Other researchers have noted these potential differences in the predictive strength of various aspects of religion. For example, Regnerus and Elder (2003) stated that, differential forms of religiosity lend themselves to prescribing or proscribing different types of behavior (p. 21). While they were not referring to health behaviors in particular, it is likely that attendance, denomination, and salience all influence preventive service use in different ways. Religious Service Attendance. Attendance at religious services predicts the utilization of flu shots, cholesterol screening, mammograms, Pap smears, and breast self exams, net of the demographic, social, and health covariates. In addition, attendance is also significantly associated with total utilization rates for both genders. For each of these outcomes, attendance at religious services is associated with higher odds of 124 utilization compared to individuals who never attend religious services. It is interesting to note that as the level of attendance increases, equivalent gains in service utilization are not seen. In fact, it is often the middle two categories of attendance that are associated with the highest odds of utilization. Moreover, the most frequent attendance category (more than once a week) is not a significant predictor of utilization for flu shots, breast self exams, or total male usage. Thus, it appears that attending religious services nearly every week is most beneficial to an individual s preventive service utilization rates and that increasing service attendance to more than once a week does not give the individual any additional advantages. There are several possible explanations for these findings. For example, involvement with a religious organization may be associated with increased utilization of preventive services for the reasons discussed above, but extreme levels of involvement may come at the expense of other forms of social interactions. In other words, individuals who attend religious services more than once a week may be isolated from other relationships and this may affect their use of preventive services by limiting their access to, and knowledge of, these services. It is also possible that individuals may significantly increase their attendance at religious services to cope with serious problems, and these problems may reduce their use of preventive services. Similarly, individuals who attend services more than once a week may be using the religious organization to escape from a bad environment at home. If this is the case, unmeasured variables concerning the individuals social and economic resources may temper the positive associations between attendance and preventive service use for the highest attenders. 125 The positive relationship between attendance and preventive health care use was expected to work through numerous pathways. At the most basic level, attending services implies involvement with a particular religious organization. The benefits of membership may include increased social support, both informal and formal, as well as access to various programs, some of which may be health-related. Service attendance may also lead to better physical and mental health, which could also increase utilization levels. However, the data did not support the mediating role of social support or health status, which leaves room for other possible explanations. For example, the relationship between attendance and utilization may be due to some underlying personality trait or lifestyle characteristics that influence both behaviors. It is possible that certain individuals are predisposed to play it safe or follow unwritten guidelines for living well. Attending religious services and participating in preventive health behaviors may both be part of this concept of conservative living. Along these same lines, a holistic view of health, in which health encompasses spiritual and emotional dimensions in addition to just physical ones, may also help to explain the connection between religious attendance and preventive health care utilization. Again, both religious attendance and preventive health care utilization would be important to individuals holding this view of health. Other related personality traits, such as general levels of discipline, may also differentiate those who regularly attend religious services and use preventive services and those who do not. A certain amount of order and control is necessary to set aside the time (weekly for attendance, yearly for certain preventive services) and to make the effort to physically get to religious services or health care 126 providers. Thus, having this level of discipline may influence both religious service attendance and preventive health care utilization. While all of these pathways could not be tested directly with the available data, the strong and consistent effect of attendance on all types of preventive services supports the assumption that attendance works through various mechanisms. Perhaps also due to this broad scope of influence, attendance has been found to be an important predictor for many other types of health behaviors as well (for review, see Koenig, McCullough, and Larson, 2001). Religious Affiliation. Affiliation, which represents a combination of the theological perspectives, social norms, and organizational structures of a particular denomination, may also predict health behaviors. Previous research on theological and organizational differences between religious affiliations indicates that members of Protestant, Catholic, or Jewish denominations may be more inclined to practice positive health behaviors (Benjamins and Brown, 2003; Mechanic, 1963; Miller and Champion, 1993; Murray and McMillan, 1993; Miller, Norcross, and Bass, 1980; Wan and Yates, 1975; Yi, 1994; Yi, 1998). Additionally, a recent study found that individuals belonging to any religious denomination are more likely to use preventive services than those who are unaffiliated, but this relationship is particularly strong for Jewish individuals (Benjamins and Brown, 2003). In fact, a large amount of information details the relationship between Judaism and health, much of it offering explanations for the relatively positive health behaviors and outcomes of Jewish individuals (Jacobs and Giarelli, 2001; Jakobovits, 1975; Linn, 1967; Mechanic, 1963; Rosner, 1972; Scheff, 127 1966; Segal, Weiss, and Sokol, 1965; Solon, 1966; Wan and Soifer, 1974; Wan and Yates, 1975). Information on the health beliefs and behaviors of Protestants is also available. However, no distinctions between Protestant denominations had been made in studies examining preventive service use prior to the current study. Thus, it is particularly interesting to note that the most consistent denominational differences in the current study exist between Mainline and Evangelical Protestants. Although Evangelical Protestant denominations incorporate many health-related beliefs and activities into their teachings and programs, individuals belonging to these organizations report less utilization of many preventive services. It is possible that certain beliefs linked to the more literal interpretation of the Bible favored by Evangelicals may account for some of these disparities. For example, members of conservative denominations may be more likely to disregard modern medical treatments and rely on their faith to keep them healthy. Biblical verses such as the following may provide believers with confidence in such a decision: He forgives all my sins and heals all my diseases, (Psalm 103:3) and Trust in the Lord with all your heart and lean not on your own understanding This will bring health to your body and nourishment to your bones (Proverbs 3:5, 8). Evidence of the widespread acceptance of ideas such as these within Evangelical Protestant denominations includes the tradition of many Pentecostal churches to lay hands on sick members of the congregation, in order to heal them. While rejecting medical services is not dictated by the church, a certain amount of faith in the healing power of God is expected. In fact, the Doctrine of the 128 United Pentecostal Church states that, God is the Great Physician. His knowledge of the human mind and body is complete. He can do more for the sick and the diseased than can all earthly doctors and surgeons combined. He created us; is it not reasonable, then, to believe that He can heal us when we are sick? (United Pentecostal Church International website, 2004). It is possible that having official teachings and activities that assert the healing powers of God over the healing powers of traditional health care may influence members of certain Evangelical denominations to limit their use of health care services, including preventive care. It is also important to note that the higher utilization rates of Mainline Protestants compared to Evangelical Protestants is not explained by the inclusion of the socioeconomic resources or the proposed mediators. However, it is still possible that some unmeasured aspects of social or personal resources may be influencing the higher utilization rates of Mainline Protestants. Despite, or perhaps because of, the lack of understanding regarding the cause of the denominational disparities, these results strongly support the separation of Protestant groups into these two categories (or even more, if possible) in future studies. Religious Salience. In contrast to these organizational and interpersonal influences of religion on preventive service utilization, religious salience was expected to affect levels of use through personal beliefs, faith, and commitment. Belief in a higher power may encourage positive health behaviors due to feelings of responsibility or optimism, for example. While making predictions for each religion variable with each preventive service outcome would have been implausible due to the lack of previous 129 evidence on the topic, particular patterns within the findings warrant a brief discussion. In this case, salience predicted the use of breast self exams and prostate screening, but not the other four individual outcomes or the two total usage by gender outcomes. It is possible that the mechanisms that link these two outcomes with religious salience are different, or merely stronger, than those mechanisms that link salience with the other outcomes. Interestingly, the findings for breast and prostate exams were not consistent with the general patterns of the other preventive services. Specifically, religious denomination and attendance are less important predictors of these types of utilization. Breast exams (as measured in the current study) are conducted by the women themselves, and this may differentiate this type of preventive behavior from the other female services. The other services may all be performed during a woman s visit to a physician (e.g. during annual check-ups), but breast self exams must be done periodically, at home. Although visiting a physician requires some degree of motivation, performing self-exams may require more effort to both remember and carry out without any external encouragement. It is possible that a sense of responsibility to a higher being may account for this higher likelihood of more religious women to complete breast self exams. It may be for these same reasons that attending religious services and being affiliated with a specific denomination does not have a strong influence on breast self exams. For example, religious organizations can not provide this type of screening, nor can many of the other means of encouraging utilization, such as providing transportation, facilitate use. Further, other potential mediators, such as physical health, may play less of a role in breast self exams. 130 The other outcome affected by religious salience, prostate screenings, also has several distinguishing characteristics that may account for its unusual relationships with the religion variables. For one, this type of screening is the only male-specific type of preventive service. While women often receive several services during a single visit to a health care provider, and have more motivation and encouragement to see a physician regularly, men receive fewer regular check-ups at which preventive services may be recommended or offered. For example, in 1996, 76.8 percent of female adults in the U.S. visited a physician, but only 59.6 percent of males did, according to data from the Medical Expenditure Panel Survey (Xu and Borders, 2003). Secondly, until recently, prostate cancer was considered a disease of old age and many middle age men were not aware of their risks for this illness. For example, the percentage of men age 50 years or older whose physicians told them that they should have a prostate screening increased from 58.2 percent in 1994 to 76.9 percent in 1997 (McDavid, Melnik, and Derderian, 2000). Finally, the effectiveness of prostate exams in lowering mortality risks is currently the subject of an intense debate (USPSTF, 2002). For these reasons, messages or programs promoted by religious organizations (or coreligionists) may be less prevalent and less effective and, thus, religious attendance and affiliated are less likely to be associated with the use of prostate screening. Explanations for the link between religious salience and prostate screening are not clear, but may be similar to those proposed for breast self exams. 131 Mediators Despite these differences between the various religion measures, the proposed model suggests that attitudes toward health care, social support, and mental and physical health are pathways through which all facets of religion influence preventive service use. For example, all three facets of religion have been linked to more positive mental and physical health outcomes (Koenig, McCullough, and Larson, 2001), and this may allow them better access to all types of health care services, including services aimed at prevention. However, the findings provide some support for the possible mediating role of attitudes toward health care, but suggest that the other proposed mediators do not fit this role in the current study. These proposed attitudinal, social, and health mediators were anticipated to work through even more proximate mechanisms. This set of proposed mediators includes those first conceptualized by Andersen need, motivation, and access (1968). Although data limitations precluded directly testing these factors, speculation on the presumed role of these mediators in the current study is possible. For example, attitudes toward health care are assumed to have an effect on preventive service use through the concepts of motivation and (less directly) perceived need and access. The findings showed that individuals who attend religious services more frequently have greater trust in physicians. This elevated trust may lead to a greater respect for the doctors recommendations and more frequent office visits and these differences, in turn, may bring about increases in preventive care utilization. 132 Overall, the findings from this study provide limited support for the proposed theoretical framework. While the current data suggest that several of the proposed mediators (social support, mental health, and physical health) can be excluded from the model, more research is needed before these theoretically important concepts are discounted. More research is also needed to confirm the role of attitudes toward health care as a mediator, since data limitations of this study allowed for only indirect tests of this link. Nonetheless, the data indicate that the proposed framework provides an important structure on which to base future studies. Relationship of Findings to Current Literature on Religion and Health As noted earlier, research on the connection between religion and health has been steadily increasing. Studies have investigated a vast range of physical, mental, and emotional health outcomes, with a wide variety of data sets. Furthermore, many of these studies have searched for group differences in the relationships, as well as possible explanations for the apparent connection. Commonly proposed mechanisms include social support, positive emotions, mental health, and health behaviors (for a review, see Ellison and Levin, 1998). Health behaviors may be the most direct and easiest to measure mechanisms and numerous studies have found significant differences by levels of religious involvement and religious beliefs. These health behaviors encompass a wide variety of activities, such as smoking, drinking, drug use, promiscuity, exercise, nutrition, and health care use. The current study provides additional evidence for a relationship between various aspects of religion and preventive health care use. As more complex models concerning the relationship between religion and health are developed, this 133 study provides valuable support for the necessity of addressing the role of preventive health care as a probable mediator for religion-health associations. While several previous studies have also examined how religion influences the use of preventive health care services, this literature is sporadic and undeveloped. The majority of studies come from the public health or medical literatures and merely include religion as one of many control variables. Furthermore, because researchers in the fields of religion and medical sociology have, thus far, shown little interest in studying this relationship, or have been unable to due to the lack of appropriate data sets, the theoretical link between religion and health care use has been virtually neglected. Thus, the current study also adds to this literature by creating a theoretical framework for the relationship and by testing several possible mediators. Relationship of Findings to Current Literature on Preventive Service Use Although the literature on religion and health care use is still in its infancy, the current study provides strong support for the role of religious attendance in influencing preventive service utilization. Religion has been a particularly neglected social factor in health research and findings such as those shown here may compel health care workers and health researchers to pay more attention to various aspects of religion as potentially significant determinants of health care utilization. While the use of general health care services is mainly determined by an individual s need for such services, utilization levels for preventive health care are more susceptible to other factors, particularly social and psychological ones. This essential difference, along with empirical evidence from studies 134 such as this, challenge researchers and practitioners in the health care field to further consider the effect of religion on the utilization of preventive services. Implications In addition to the impact this research has on the relevant literatures, other implications may stem from the present findings. For example, the results provide some support for the idea of integrating religious organizations and health care. Findings, such as these, may be useful to encourage religious organizations to initiate health programs as part of their social ministry. National organizations designed to promote the link between religious organizations and health care providers are growing in number and in visibility. For example, the Parish Nursing Movement was developed less than twenty years ago, but now has an international following of hundreds of churches and was recently designated as a specialty practice by the American Nursing Association (American Nursing Association website, 2003; International Parish Nurse Resource Center website, 2003). The groups exist within specific religious denominations and within types of health care providers and have a variety of objectives, such as providing information for potentially interested volunteers, offering strategies for implementing programs within an organization, and giving information to individuals who are looking for services. As more direct evidence exploring the usefulness of services provided through religious organizations becomes available, it is possible that these programs will continue to multiply, churches will increasingly consider adding a health component to their social services, and that more health care providers will volunteer through religious organizations. 135 Furthermore, in times of increasing costs and decreasing funds for health care, governments and insurance agencies may find religious organizations to be effective promoters of preventive health care and even providers for certain services. For example, the federal government is currently exploring the role of religious organizations in the provision of various social services. The extent of these faith-based initiatives may be influenced by the accumulation of research on topics such as the current one. While intervention studies and other, more policy-oriented, studies can provide more direct answers to these issues, empirical support for a positive relationship between attendance at religious services and the use of preventive health services may encourage government agencies to consider collaborating with religious organizations to better serve the health care needs of the population. Limitations While the results of this study may open up new areas of research and encourage the integration of health programs within religious organizations, several limitations of the data and the study must be taken into consideration. For one, the main data set used here is only representative of U.S. adults between the ages of 51 and 61. A data set more representative of the adult population would be more useful in making generalizations about the effect of religion on preventive service use. While no significant age interactions were found within this sample, it is possible that the relationship between religion and preventive service utilization changes with age. Thus, the results may not be generalizable to younger adults or the elderly, for example. Secondly, many measures, such as those representing social support in the first portion of the study, are not ideal. 136 While they are the best measures available in the data used here, more precise measures would improve the estimates of the related associations. Furthermore, not all of the concepts put forth in the theoretical model are included in the study. As noted earlier, the most proximate determinants (the Andersen mediators) were not available in the data set and were, thus, untestable in the current study. In addition, one of the primary predictor variables, religious salience, was measured concurrently with the outcomes in Wave 3 of the HRS. While a relationship was found between salience and two preventive services (breast self exams and prostate screening), the lack of temporal ordering precludes any causal interpretation of this relationship. However, previous studies have found that religiosity is relatively stable during adulthood (Courtenay, Poon, Martin, Clayton, and Johnson, 1992; Markides, Levin, and Ray, 1987). This earlier study looked specifically at changes in religious salience over two four-year periods and found that salience changed only slightly in the first period and not at all during the second (Markides, Levin, and Ray, 1987). Thus, most aspects of religion, including salience, remain fairly stable within adulthood. Because of this, and in accordance with the theoretical framework put forward above, salience is believed to influence the use of preventive services and not vice versa. Another significant limitation of this study is that the test of the mediating role of attitudes toward health care was incomplete. As discussed earlier, data limitations constrained the models to testing only the relationship between religion and the healthrelated attitudes. The influence of these attitudes on health care utilization had to be assumed based on previous research (Bogart, 2001; Bogart, Bird, Walt, Delahanty, and 137 Figler, 2004). While very plausible, this leap must be acknowledged. In addition, the primary models used to test these relationships were conducted using a sample of all U.S. adults over the age of 18. When the models were re-run using an age-restricted sample comparable to the HRS analyses, the association between the religion variables and the health-related attitudes were insignificant. A relatively small sample size may explain the lack of significance; however, for this age group, religion does not appear to influence attitudes toward physicians or the health care system. While this is important to recognize this when addressing the original hypotheses of this dissertation, the findings have a larger significance for the religion and health literature. Namely, attitudes toward health care appear to differ by level of religious attendance, denomination, and strength of affiliation. Future Research Despite these limitations, the findings of this study make a significant addition to the current literature. Within the field of religion and health research, preventive health care use has emerged as a viable mechanism linking religious involvement and beliefs to a wide variety of health outcomes. While still untested, the inclusion of preventive service utilization in studies investigating the influence of religion on various aspects of morbidity and mortality will be the next step in investigating the role of this potentially illuminating piece of the puzzle. With regards to the limitations mentioned above, future studies of religion and preventive service utilization should examine the relationship within a sample of adults of all ages. While the religion measures in this study were better than previous studies, 138 more aspects of religion also need to be considered. For example, measures of private religious practice, such as reading the Bible, may provide more insight into how religion affects both attitudes toward health care and preventive service use. More specific denominational information may also be useful. Especially considering the large distinctions between members of different Protestant denominations, more exact information on affiliation (and larger samples) may help to clarify the relationship between religious beliefs and preventive service use. Along these same lines, more specific information regarding the affiliations grouped together in the Other category in the present study would also be valuable. It is possible that the relationships between religion, health-related attitudes, and preventive service use work differently for individuals belonging to non-Western religions, for example. Little, if anything is currently known about these relationships. Although the percentage of individuals in the U.S. who are affiliated with these denominations is small, exploratory work could be conducted using the current data set. Finally, more information on beliefs stemming from theology, church teachings, or norms that may affect health knowledge and behaviors would be beneficial. The role of networks also deserves closer examination. Previous research has found that religious involvement increases the size and density of one s social network (Bradley, 1995; Ellison and George, 1994). Networks may influence health care utilization because they are conducive to the spread of information and the building of routines. While membership in a religious organization and participation in religious activities provide an individual with numerous opportunities to develop meaningful 139 relationships, this type of religious involvement also presents the individual with an array of weaker ties. According to Granovetter (1973), these weaker relationships may be particularly beneficial to the process of social integration. Correspondingly, certain benefits of integration, such as the sharing of information, may be imparted to those individuals who have developed a network of friends and acquaintances through their involvement in a religious organization. For each of these reasons, religious differences in the size and density of networks may be an important link between religion and preventive service use. However, more work in this area is needed. Final Summary Religious denomination and involvement influence whether or not middle-age adults in the U.S. use flu shots, cholesterol screening, mammograms, Pap smears, breast self exams, and prostate screening. Attending religious services, adhering to the beliefs of a particular denomination, and religious salience all are associated with the utilization of these preventive services. The influence of attendance on preventive service use is particularly strong, often increasing the odds of service use by 25 percent or more. Moreover, there is some connection between an individual s faith in religion and his or her faith in the health care system that may partially explain this effect. These findings add to the burgeoning literature on religion and health and the increasing amount of research investigating possible determinants of preventive service utilization. They may also provide valuable support for the integration of health programs into religious organizations and may influence policy decisions regarding this association. Finally, the 140 results and implications of the current study will hopefully motivate other health and social science researchers to pursue this worthwhile area of research. 141 Tables and Figures Figure 1. Conceptual Model of the Relationship Between Religion and Preventive Service Utilization. Attitudes Toward Health Care Demographic And Social Factors Perceived Need Resources Religion Attendance Salience Affiliation Social Support Motivation Mental Health Access Preventive Service Use Physical Health Background Religion Social and Health Mediators Andersen Mediators Outcomes Notes: Analyses do not test the role of the Andersen mediators. Model assumes direct effects may remain even after adjustment for mediating factors. 142 Table 1. HRS Variables Available by Wave (1992-2000) Wave 1 1992 Wave 2 1994 Wave 3 1996 Wave 4 a 1998 Wave 5 2000 Religion Denomination 12,652 102 196 5,093 b 262 Salience --10,964 21,384 19,580 Attendance 12,652 102 ---12,652 11,596 10,964 21,384 19,580 Total N 12,652 11,494 10,298 N (from original sample) Preventive Health Care Flu Shot --10,964 5,093 19,580 Cholesterol Screening --10,964 5,093 19,580 Breast Self Exam --10,964 2,838 19,580 Mammogram --10,964 2,838 19,580 Pap Smear --10,964 2,838 19,580 Prostate Exam --10,964 2,255 19,580 10,964 21,384 19,580 N Note: Each cell denotes number of respondents for a particular variable. a Waves 4 and 5 not used in current study b Denomination information for these 5,093 respondents comes from the preference question that has the following five categories (Protestant, Catholic, Jewish, None, or Other). Those who responded Protestant were asked for more detailed information and, of the 2,818 who provided it, four more categories were created. 143 Table 2. Coding Specifications for Religious Affiliation (HRS, 1992) DENOMINATION PROTESTANT: REFORMATION ERA Congregational Episcopalian, Anglican, Church of England Evangelical and Reformed Lutheran Presbyterian Reformed, Dutch Reformed, Christian Reformed United Church of Christ PROTESTANT: PIETISTIC African Methodist Episcopal; AME Zion Baptist Disciples of Christ Methodist United Brethren or Evangelical Brethren Mennonite; Amish Church of the Brethren Christian, no denomination given PROTESTANT: FUNDAMENTALIST Church of Christ Church of God, "Holiness", Church of Living God, Church of God in Prophecy Church of God in Christ Fundamentalist Baptist: include Primitive, Free Will, Missionary and Gospel Baptist Nazarene or Free Methodist Pentecostal or Assembly of God Plymouth Brethren Salvation Army Sanctified Seventh Day Adventist Southern Baptist United Missionary or Protestant Missionary; Christian and Missionary Alliance Missouri Synod Lutheran Other Fundamentalist; Apostolic; Charismatic; Bible Church, Word of Faith, Foursquare PROTESTANT: GENERAL Protestant, no denomination given Non-denominational Protestant church Community church--no denominational basis Born again Christian Evangelical Other Protestant-Berean, AZUA, United Church of Canada CATHOLIC; EASTERN ORTHODOX Roman Catholic; Catholic Greek Rite Catholic Orthodox; Eastern, Greek/Russian, other Orthodox NON-TRADITIONAL CHRISTIAN Christian Scientist Jehovah's Witnesses Latter Day Saints, Mormons Quakers Spiritualists Unitarian or Universalist Unity Other non-traditional Christian JEWISH Jewish NON-JUDEO-CHRISTIAN Nation of Islam; Islam; Muslim; Moslem; Mohammedan World Community of Islam in the West Buddhist Hindu Bahai Other non-Judeo-Christian religion NO RELIGION CODE MP a MP EP MP MP MP MP MP EP MP MP MP EP EP MP EP EP EP EP EP EP EP EP EP EP EP EP EP EP MP EP EP EP EP MP C C C O O O O O O O O J O O O O O O 144 None; No Preference Atheist; Agnostic Other (including Mason, New Age, RC and Jewish, RC in summer/Lutheran in winter, Mile-High Church of Rel. Science, Science of Mind, Yeshuan) Don t Know; don't remember Not Applicable Note: a MP=Mainline Protestant, EP=Evangelical Protestant, C=Catholic, O=Other Denomination, J=Jewish, N=Non-Affiliated N N O O N 145 Table 3. Sample Characteristics from the Health and Retirement Survey, 1992-1996 a b Range Mean S.D. Religion Service Attendance More than 1/Week 0-1 .35 .15 a Once/Week .23 0-1 .42 2-3 Times/Month .15 0-1 .36 1-2 Times/Year or More .22 0-1 .41 Never .25 0-1 .43 Salience Very Important .64 0-1 .48 Somewhat Important .26 0-1 .44 Not Important .10 0-1 .30 Affiliation Catholic .27 0-1 .44 Jewish .02 0-1 .13 Mainline Protestant .30 0-1 .46 Evangelical Protestant .34 0-1 .47 Other .03 0-1 .16 Not Affiliated .05 0-1 .21 Preventive Service Use Flu Shot 0-1 .37 .48 Cholesterol Screening 0-1 .70 .46 Mammogram 0-1 .71 .45 Pap Smear 0-1 .68 .47 Breast Self Exam 0-1 .63 .48 Prostate Exam 0-1 .64 .48 Total Female Usage 0-5 3.12 1.41 Total Male Usage 0-3 1.68 1.04 Demographic and Social Factors Age 51-61 55.7 3.08 Female 0-1 .54 .50 Race/ethnicity NH White 0-1 .75 .44 NH Black 0-1 .17 .37 Hispanic 0-1 .09 .28 Foreign Born Status 0-1 .09 .28 Resources Education (in years) 0-17 12.1 3.18 Net Worth Low ( $32,500) 0-1 .25 .43 Household Income (in 1000 s) $0- $598 50.2 46.0 Health Insurance 0-1 .79 .41 Social Support Married/Living with Partner 0-1 .77 .42 Satisfaction with Friendships 0-5 4.53 .76 Satisfaction with Family Life 0-5 4.58 .78 Mental Health Self-Rated Emotional Health 1-5 3.48 1.08 Depression Scale 0-48 26.9 5.16 Physical Health Chronic Conditions Heart Conditions 0-1 .12 .33 Cancer 0-1 .05 .22 Total Other Conditions 0-5 .98 .94 Self-Rated Physical Health 1-5 2.55 1.18 Notes: a Unweighted, N=7,875 b Proportions may not add to 1 due to rounding 146 Table 4. Differences Between Mean Levels of Religious Service Attendance and Salience by Affiliation (HRS, 1992-1996) a Religious Affiliation Mainline Evangelical Not Catholic Jewish Protestant Protestant Other Affiliated Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Religious Attendance More than 1/Week .02 (.12) .09 (.29) .25 (.43) .37 (.48) .00 (.00) .10 (.30) c Once/Week .03 (.17) .20 (.40) .20 (.40) .18 (.39) .00 (.00) .37 (.48) 2-3 Times/Month .12 (.32) .17 (.37) .16 (.37) .14 (.35) .00 (.00) .16 (.36) 1-2 Times/Year or More .64 (.48) .27 (.44) .17 (.38) .15 (.36) .01 (.10) .23 (.42) Never .19 (.40) .28 (.45) .22 (.41) .16 (.37) .99 (.10) .15 (.35) Salience Very Important .31 (.46) .56 (.50) .79 (.41) .78 (.41) .25 (.41) .62 (.48) Somewhat Important .40 (.49) .33 (.47) .17 (.37) .14 (.35) .25 (.43) .29 (.46) Not Important .29 (.45) .11 (.31) .04 (.20) .08 (.27) .53 (.50) .08 (.27) 2,130 129 2,356 2,669 215 376 N Notes: a Unweighted data, N=7,875 b Affiliation differences indicate significant differences (at p<.05) between the groups as follows: a. Catholic and Jewish b. Catholic and Mainline Protestant c. Catholic and Evangelical Protestant d. Catholic and Other e. Catholic and Not Affiliated f. Jewish and Mainline Protestant g. Jewish and Evangelical Protestant h. Jewish and Other i. Jewish and Not Affiliated j. Mainline Protestant and Evangelical Protestant k. Mainline Protestant and Other l. Mainline Protestant and Not Affiliated m. Evangelical Protestant and Other n. Evangelical Protestant and Not Affiliated o. Other and Not Affiliated c Proportions may not add to 1 due to rounding Differences by b Affiliation (p<.05) a,c,d,e,f,g,h,j,k,l,m,n,o a,b,c,d,e,f,g,h,l,n,o e,i,l,n,o a,b,c,d,e,f,g,h,i,j,k,l,n,o b,c,e,f,i,j,k,l,m,n,o a,b,c,d,e,f,g,h,j,k,l,n,o a,b,c,e,f,g,h,i,j,k,l,n,o a,b,c,e,f,g,h,i,j,k,l,n,o 147 Table 5. Zero-Order Logistic Regression Odd Ratios for Control and Mediating Variables with Individual Preventive Service Outcomes (1992-1996) ab Zero-Order Effects on Time 2 Individual Preventive Service Variables Flu Cholesterol Pap Breast Prostate Variable Shot Screening Mammogram Smear Self Exam Exam Demographic and Social Factors Age 1.06 *** 1.01 0.98 + 0.93 *** 1.00 1.04 ** Female 1.18 *** 1.12 ----Race/ethnicity NH Black 0.68 *** 0.87 1.05 0.97 1.67 *** 0.77 * Hispanic 0.75 *** 0.98 0.66 ** 0.71 + 0.72 * 0.54 *** Foreign Born Status 0.80 * 1.23 1.08 1.28 + 0.81 0.61 *** Resources Education (in years) 1.04 *** 1.07 *** 1.13 *** 1.11 *** 0.99 1.12 *** Net Worth Low (> $32,500) 0.78 *** 0.63 *** 0.50 *** 0.55 *** 1.04 0.47 *** Household Income (in 1000 s) 1.00 *** 1.01 *** 1.01 *** 1.01 *** 1.00 1.01 *** Health Insurance 1.25 ** 1.82 *** 2.30 *** 2.02 *** 1.04 2.31 *** Social Support Married/Living with Partner 1.20 *** 1.35 *** 1.69 *** 1.53 *** 1.26 *** 1.62 *** Satisfaction with Friendships 0.90 *** 0.94 0.97 0.99 1.15 ** 0.95 Satisfaction with Family Life 0.98 1.04 1.09 1.05 1.20 *** 1.00 Mental Health Self-Rated Health 0.96 1.01 1.11 ** 1.09 ** 1.07 * 1.12 *** Depression Scale 0.99 * 1.00 1.02 ** 1.03 *** 1.03 *** 1.01 Physical Health Chronic Conditions Heart Conditions 1.73 *** 2.02 *** 1.06 0.84 * 1.02 1.55 *** Cancer 1.52 *** 1.45 ** 1.13 1.17 1.20 1.72 * Total Other Conditions 1.27 *** 1.32 *** 0.98 0.93 * 1.04 1.40 *** Self-Rated Health 1.14 *** 1.07 ** 0.89 ** 0.85 *** 1.04 0.97 N 7,867 7,857 4,257 4,257 4,256 3,607 Note: a Weighted HRS data b Each estimate represents a separate model + p .10; * p .05; ** p .01; *** p .001 148 Table 6. Zero-Order Negative Binomial Estimates for Control and Mediating Variables with Time 2 Total Usage Preventive Service Outcomes by Gender (1992-1996) ab Zero-Order Assocations of Total Usage Preventive Service Variables Variable Female Male Demographic and Social Factors Age -.00 .02 *** Female --Race/ethnicity NH Black -.00 -.07 * Hispanic -.10 ** -.20 *** Foreign Born Status -.03 -.13 * Resources Education (in years) .02 *** .04 *** Net Worth Low (> $32,500) -.13 *** -.30 *** Household Income (in 1000 s) .00 *** .00 *** Health Insurance .18 *** .26 *** Social Support Married/Living with Partner .11 *** .16 *** Satisfaction with Friendships .00 -.04 * Satisfaction with Family Life .02 * -.00 Mental Health Self-Rated Health .01 .03 * Depression Scale .01 ** -.00 Physical Health Chronic Conditions Heart Conditions .06 * .24 *** Cancer .06 * .19 ** Total Other Conditions .03 *** .10 *** Self-Rated Health .00 .02 N 4,240 3,601 Note: a Weighted HRS data a Each estimate represents a separate model + p .10; * p .05; ** p .01; *** p .001 149 Table 7. Logistic Regression Models for the Effect of Religion Variables on Flu Shots and Cholesterol Screening (1992-1996) ab Flu Shot c Cholesterol Screening Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Religion Service Attendance (Never) d More than 1/Week 1.08 1.70 *** 1.04 Once/Week 1.20 ** 1.18 * 1.53 *** 2-3 Times/Month 1.24 * 1.20 + 1.49 *** 1-2 Times/Year or More 1.05 1.04 1.44 *** Affiliation (Evangelical Protestant) Catholic 0.88 + 0.87 + 1.15 * Jewish 0.87 0.94 1.62 * Mainline Protestant 1.15 + 1.18 * 1.23 *** Other 0.85 0.86 1.05 Not Affiliated 0.84 0.98 0.67 ** Salience (Not Important) Very Important 1.23 * 1.16 1.37 *** Somewhat Important 1.08 1.04 1.14 R-Square .00 .00 .00 .01 .01 .00 .00 N 7,867 7,857 Notes: a Weighted HRS data b Logistic regression odds ratios c The reference category is individuals who did not use the type of service specified by the model d Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) Model 4 1.64 *** 1.40 *** 1.38 *** 1.33 *** 1.17 * 1.78 ** 1.33 *** 1.01 0.94 1.12 1.00 .02 150 Table 8. Logistic Regression Models for the Effect of Religion Variables on Mammograms and Pap Smears (1992-1996) ab Mammogram c Pap Smear Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Religion Service Attendance (Never) d 1.91 *** 1.29 * More than 1/Week 1.66 *** 1.36 *** Once/Week 2.13 *** 1.91 *** 1.58 *** 2-3 Times/Month 1.82 *** 1.73 *** 1.47 *** 1.20 + 1-2 Times/Year or More 1.49 *** Affiliation (Evangelical Protestant) 0.99 1.04 Catholic 1.06 Jewish 2.33 *** 2.57 ** 3.71 ** 1.62 *** 1.34 ** Mainline Protestant 1.52 *** Other 1.01 0.97 1.18 1.19 1.19 Not Affiliated 0.79 Salience (Not Important) 1.03 Very Important 1.11 0.86 0.93 0.83 0.98 Somewhat Important R-Square .02 .01 .00 .03 .01 .01 .00 N 4,257 4,257 Notes: a Weighted HRS data b Logistic regression odds ratios c The reference category is individuals who did not use the type of service specified by the model d Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) Model 4 1.50 ** 1.54 *** 1.71 *** 1.20 1.02 4.04 ** 1.39 ** 1.16 1.57 + 0.94 0.98 .02 151 Table 9. Logistic Regression Models for the Effect of Religion Variables on Breast Exams and Prostate Screening (1992-1996) ab Breast Self Exam c Prostate Screening Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Religion Service Attendance (Never) d 0.90 1.33 * More than 1/Week 0.97 1.48 *** Once/Week 1.07 1.09 1.43 * 1.35 ** 2-3 Times/Month 1.48 ** 1.19 + 1.32 *** 1-2 Times/Year or More 1.16 Affiliation (Evangelical Protestant) 0.81 + 1.08 Catholic 0.82 + Jewish 0.69 0.68 1.01 0.86 + 1.11 Mainline Protestant 0.87 Other 0.70 + 0.72 0.80 1.17 0.74 * Not Affiliated 0.92 Salience (Not Important) Very Important 1.50 ** 1.49 * 1.31 ** 1.43 * 1.41 * 1.21 Somewhat Important R-Square .01 .00 .00 .01 .01 .00 .00 N 4,256 3,595 Notes: a Weighted HRS data b Logistic regression odds ratios c The reference category is individuals who did not use the type of service specified by the model d Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) Model 4 1.29 + 1.40 ** 1.28 + 1.26 * 1.05 1.05 1.16 0.78 0.95 1.12 1.09 .01 152 Table 10. Negative Binomial Regression Models for Total Utilization Levels by Gender (1992-1996) ab Female Total Usage Model 1 Model 2 Model 3 Model 4 Model 1 Religion Service Attendance (Never) c .14 ** More than 1/Week .07 * .08 ** .17 *** Once/Week .10 *** .09 *** .15 *** 2-3 Times/Month .12 *** .12 *** .06 * .12 ** 1-2 Times/Year or More .07 * Affiliation (Evangelical Protestant) Catholic -.01 -.01 Jewish .11 .14 + .09 *** Mainline Protestant .08 *** Other .01 .01 .06 Not Affiliated -.04 Salience (Not Important) Very Important .08 * .05 .05 .04 Somewhat Important -Square/Degrees of Freedom .63 .63 .63 .63 .62 N 4,240 3,588 Notes: a Weighted HRS data b Negative binomial estimates c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) Male Total Usage Model 2 Model 3 Model 4 .13 ** .16 *** .14 ** .10 ** .01 .02 .02 -.12 -.14 * .09 * .04 .62 -.00 .04 .04 -.13 -.03 01 -.01 .61 .61 153 Table 11. Estimated Net Effects of Religious Attendance, Affiliation, Salience, and Other Controls on the Use of Flu Shots (1992-1996) ab Flu Shot Model 1 Model 2 Model 3 Model 4 Religion Service Attendance (Never) c More than 1/Week 1.03 0.96 0.98 1.01 Once/Week 1.18 * 1.11 1.12 1.14 + 2-3 Times/Month 1.25 * 1.19 + 1.20 + 1.25 * 1-2 Times/Year or More 1.07 1.03 1.04 1.05 Affiliation (Evangelical Protestant) Catholic 0.82 * 0.78 ** 0.78 ** 0.80 ** Jewish 0.87 0.76 0.74 0.75 Mainline Protestant 1.07 1.01 1.01 1.02 Other 0.84 0.80 0.79 0.78 Not Affiliated 0.94 0.88 0.88 0.91 Salience (Not Important) Very Important 1.13 1.17 1.17 1.16 Somewhat Important 1.01 1.04 1.04 1.02 Demographic and Social Factors Age 1.06 *** 1.06 *** 1.06 *** 1.05 *** Female 1.14 ** 1.17 *** 1.20 *** 1.20 *** Race/ethnicity (NH White) NH Black 0.67 *** 0.63 *** 0.61 *** 0.66 *** Hispanic 0.82 + 0.97 0.97 1.01 Foreign Born Status 0.89 0.92 0.90 0.94 Resources Education (in years) 1.03 *** 1.03 ** 1.05 *** Net Worth Low (> $32,500) 0.93 0.94 0.86 + Household Income (in 1000 s) 1.001 * 1.001 + 1.001 * Health Insurance 1.03 1.03 1.12 Social Support Married/Living with Partner 1.16 ** 1.17 ** Satisfied with Friendships 0.87 *** 0.90 ** Satisfied with Family 0.98 1.02 Mental Health Self-Rated Health 1.04 Depression Scale 0.99 Physical Health Chronic Conditions Heart Conditions 1.46 *** Cancer 1.30 * Total Other Conditions 1.20 *** Self-Rated Health 1.12 *** Pseudo R-Square .02 .02 .03 .05 N 7,867 Notes: a Weighted HRS data b Logistic regression odds ratios c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 154 Table 12. Estimated Net Effects of Religious Attendance, Affiliation, Salience, and Other Controls on the Use of Cholesterol Screening (1992-1996) ab Cholesterol Screening Model 1 Model 2 Model 3 Model 4 Religion Service Attendance (Never) c More than 1/Week 1.62 *** 1.41 ** 1.43 ** 1.49 *** Once/Week 1.40 *** 1.21 + 1.22 + 1.25 * 2-3 Times/Month 1.39 *** 1.23 * 1.25 * 1.30 ** 1-2 Times/Year or More 1.33 *** 1.22 * 1.23 * 1.25 * Affiliation (Evangelical Protestant) Catholic 1.13 1.01 1.01 1.05 Jewish 1.70 * 1.35 1.32 1.36 Mainline Protestant 1.28 *** 1.13 + 1.13 + 1.16 * Other 0.99 0.94 0.93 0.93 Not Affiliated 0.92 0.79 + 0.79 + 0.83 Salience (Not Important) Very Important 1.13 1.19 1.19 + 1.18 Somewhat Important 1.01 1.04 1.04 1.02 Demographic and Social Factors Age 1.01 1.01 1.01 1.00 Female 1.04 1.12 + 1.14 * 1.13 * Race/ethnicity (NH White) NH Black 1.06 1.03 0.85 + 1.07 Hispanic 0.82 1.22 1.22 1.31 + Foreign Born Status 1.26 1.33 + 1.30 + 1.41 * Resources Education (in years) 1.03 * 1.03 * 1.05 *** Net Worth Low (> $32,500) 0.85 * 0.85 * 0.77 ** Household Income (in 1000 s) 1.001 *** 1.001 *** 1.001 *** Health Insurance 1.44 *** 1.44 *** 1.59 *** Social Support Married/Living with Partner 1.12 1.12 Satisfied with Friendships 0.89 ** 0.92 * Satisfied with Family 1.00 1.03 Mental Health Self-Rated Health 1.00 Depression Scale 1.00 Physical Health Chronic Conditions Heart Conditions 1.82 *** Cancer 1.37 * Total Other Conditions 1.37 *** Self-Rated Health 1.08 * Pseudo R-Square .01 .03 .04 .06 N 7,857 Notes: a Weighted HRS data b Logistic regression odds ratios c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 155 Table 13. Estimated Net Effects of Religious Attendance, Affiliation, Salience, and Other Controls on the Use of Mammograms (1992-1996) ab Mammogram Model 1 Model 2 Model 3 Model 4 Religion Service Attendance (Never) c More than 1/Week 1.95 *** 1.58 *** 1.61 *** 1.66 *** Once/Week 2.18 *** 1.79 *** 1.82 *** 1.86 *** 2-3 Times/Month 1.85 *** 1.60 *** 1.63 *** 1.66 *** 1-2 Times/Year or More 1.49 *** 1.31 * 1.35 ** 1.34 ** Affiliation (Evangelical Protestant) Catholic 1.04 0.89 0.90 0.91 Jewish 2.54 * 1.68 1.63 1.63 Mainline Protestant 1.62 *** 1.32 ** 1.32 ** 1.33 ** Other 0.96 0.89 0.90 0.89 Not Affiliated 1.20 1.00 0.99 1.00 Salience (Not Important) Very Important 0.87 0.96 0.95 0.94 Somewhat Important 0.84 0.89 0.89 0.88 Demographic and Social Factors Age 0.97 * 0.98 0.98 0.98 + Female ----Race/ethnicity (NH White) NH Black 1.44 * 1.02 1.41 * 1.47 ** Hispanic 0.56 *** 1.03 1.04 1.05 Foreign Born Status 1.29 1.44 * 1.43 * 1.45 * Resources Education (in years) 1.07 *** 1.08 *** 1.09 *** Net Worth Low (> $32,500) 0.77 ** 0.81 * 0.78 ** Household Income (in 1000 s) 1.01 *** 1.01 *** 1.01 *** Health Insurance 1.54 *** 1.52 *** 1.56 *** Social Support Married/Living with Partner 1.34 *** 1.34 *** Satisfied with Friendships 0.90 * 0.92 Satisfied with Family 1.02 1.05 Mental Health Self-Rated Health 0.96 Depression Scale 1.00 Physical Health Chronic Conditions Heart Conditions 1.18 Cancer 1.13 Total Other Conditions 1.12 * Self-Rated Health 1.01 Pseudo R-Square .03 .06 .07 .07 N 4,257 Notes: a Weighted HRS data b Logistic regression odds ratios c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 156 Table 14. Estimated Net Effects of Religious Attendance, Affiliation, Salience, and Other Controls on the Use of Pap Smears (1992-1996) ab Model 1 Pap Smear Model 2 Model 3 1.31 * 1.35 ** 1.53 *** 1.09 0.90 2.85 ** 1.18 + 1.06 1.36 1.06 1.07 0.93 *** -1.31 * 1.11 1.56 ** 1.05 *** 0.78 ** 1.001 *** 1.48 *** 1.18 0.96 1.00 Model 4 1.33 * 1.36 ** 1.54 *** 1.10 0.90 2.88 ** 1.18 + 1.07 1.37 1.08 1.07 0.93 *** -1.32 * 1.13 1.59 ** 1.05 *** 0.78 ** 1.001 *** 1.46 *** 1.19 * 0.96 1.01 0.91 * 1.01 0.99 1.28 1.11 * 0.91 * .06 Religion Service Attendance (Never) c More than 1/Week 1.56 *** 1.30 Once/Week 1.61 *** 1.34 ** 2-3 Times/Month 1.72 *** 1.52 *** 1-2 Times/Year or More 1.20 1.08 Affiliation (Evangelical Protestant) Catholic 1.02 0.90 Jewish 4.03 *** 2.90 ** Mainline Protestant 1.40 *** 1.18 + Other 1.12 1.05 Not Affiliated 1.57 * 1.36 Salience (Not Important) Very Important 0.99 1.07 Somewhat Important 1.01 1.07 Demographic and Social Factors Age 0.93 *** 0.93 *** Female --Race/ethnicity (NH White) NH Black 1.28 0.95 Hispanic 0.66 ** 1.10 Foreign Born Status 1.44 * 1.57 ** Resources Education (in years) 1.05 *** Net Worth Low (> $32,500) 0.76 ** Household Income (in 1000 s) 1.01 *** Health Insurance 1.49 *** Social Support Married/Living with Partner Satisfied with Friendships Satisfied with Family Mental Health Self-Rated Health Depression Scale Physical Health Chronic Conditions Heart Conditions Cancer Total Other Conditions Self-Rated Health Pseudo R-Square .03 .06 N 4,257 Notes: a Weighted HRS data b Logistic regression odds ratios c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) .06 157 Table 15. Estimated Net Effects of Religious Attendance, Affiliation, Salience, and Other Controls on the Use of Breast Exams (1992-1996) ab Model 1 Breast Self Exam Model 2 Model 3 0.88 1.06 1.45 ** 1.24 * 0.92 0.77 0.93 0.81 1.32 1.39 * 1.37 * 1.01 -1.71 *** 0.74 0.92 0.98 1.11 1.001 1.01 1.34 *** 1.10 * 1.15 ** Model 4 0.87 1.06 1.46 ** 1.24 * 0.93 0.79 0.91 0.80 1.31 1.38 * 1.38 * 1.00 -1.67 *** 0.71 * 0.92 0.98 1.13 1.001 1.01 1.33 *** 1.08 1.11 * 1.03 1.03 ** 1.00 1.19 1.04 1.09 * .03 Religion Service Attendance (Never) c More than 1/Week 0.89 0.90 Once/Week 1.05 1.06 2-3 Times/Month 1.41 ** 1.43 ** 1-2 Times/Year or More 1.20 1.21 Affiliation (Evangelical Protestant) Catholic 0.92 0.92 Jewish 0.74 0.75 Mainline Protestant 0.92 0.93 Other 0.75 0.76 Not Affiliated 1.24 1.25 Salience (Not Important) Very Important 1.45 * 1.42 * Somewhat Important 1.40 * 1.36 * Demographic and Social Factors Age 1.00 1.00 Female --Race/ethnicity (NH White) NH Black 1.55 *** 1.58 *** Hispanic 0.76 0.74 Foreign Born Status 0.93 0.91 Resources Education (in years) 0.98 Net Worth Low (> $32,500) 0.98 Household Income (in 1000 s) 1.001 Health Insurance 1.06 Social Support Married/Living with Partner Satisfied with Friendships Satisfied with Family Mental Health Self-Rated Health Depression Scale Physical Health Chronic Conditions Heart Conditions Cancer Total Other Conditions Self-Rated Health Pseudo R-Square .01 .01 N 4,256 Notes: a Weighted HRS data b Logistic regression odds ratios c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) .02 158 Table 16. Estimated Net Effects of Religious Attendance, Affiliation, Salience, and Other Controls on the Use of Prostate Screening (1992-1996) ab Model 1 Prostate Screening Model 2 Model 3 1.05 1.13 1.11 1.17 0.94 0.70 0.91 0.70 0.74 + 1.23 + 1.13 1.05 *** -1.04 1.01 0.75 * 1.08 *** 0.76 ** 1.01 *** 1.46 *** 1.25 * 0.92 0.91 Model 4 1.05 1.13 1.19 1.23 0.94 0.70 0.94 0.73 0.79 1.25 + 1.14 1.04 ** -1.01 1.12 0.83 1.09 *** 0.70 *** 1.01 *** 1.57 *** 1.26 * 0.95 0.93 1.09 0.99 1.38 ** 1.49 1.53 *** 1.02 .10 Religion Service Attendance (Never) c More than 1/Week 1.29 1.03 1.11 Once/Week 1.38 ** 2-3 Times/Month 1.34 * 1.10 1-2 Times/Year or More 1.31 ** 1.16 Affiliation (Evangelical Protestant) Catholic 1.12 0.94 Jewish 1.05 0.73 Mainline Protestant 1.09 0.90 Other 0.79 0.73 Not Affiliated 0.94 0.73 + Salience (Not Important) Very Important 1.13 1.22 Somewhat Important 1.07 1.12 Demographic and Social Factors Age 1.04 ** 1.05 *** Female --Race/ethnicity (NH White) NH Black 1.01 0.73 * Hispanic 0.56 *** 1.02 Foreign Born Status 0.71 * 0.75 Resources Education (in years) 1.08 *** Net Worth Low (> $32,500) 0.75 ** Household Income (in 1000 s) 1.01 *** Health Insurance 1.46 *** Social Support Married/Living with Partner Satisfied with Friendships Satisfied with Family Mental Health Self-Rated Health Depression Scale Physical Health Chronic Conditions Heart Conditions Cancer Total Other Conditions Self-Rated Health Pseudo R-Square .02 .07 N 3,607 Notes: a Weighted data b Logistic regression odds ratios c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) .07 159 Table 17. Estimated Net Effects of Religious Attendance, Affiliation, Salience, and Other Controls on Total Female Use (1992-1996) ab Total Female Use Model 1 Model 2 Model 3 Model 4 Religion Service Attendance (Never) c More than 1/Week .09 ** .05 .05 .07 * Once/Week .10 *** .06 * .07 * .07 * 2-3 Times/Month .12 *** .10 ** .10 ** .11 *** 1-2 Times/Year or More .07 * .04 .05 + .05 Affiliation (Evangelical Protestant) Catholic -.01 -.04 -.04 -.03 Jewish .13 + .06 .06 .06 Mainline Protestant .08 *** .05 * .05 * .05 * Other .01 -.01 .00 -.01 Not Affiliated .06 .02 .03 .03 Salience (Not Important) Very Important .06 .07 + .07 + .06 Somewhat Important .04 .05 .05 .05 Demographic and Social Factors -.00 -.00 Age -.00 -.00 Race/ethnicity (NH White) .04 .03 NH Black -.02 .03 Hispanic -.10 ** .01 .01 .01 Foreign Born Status .04 .06 + .06 .06 + Resources Education (in years) .01 ** .01 ** .02 *** Net Worth Low (> $32,500) -.05 * -.03 -.04 + .00 *** .00 * .00 ** Household Income (in 1000 s) .10 *** .10 *** .11 *** Health Insurance Social Support Married/Living with Partner .08 *** .08 *** Satisfied with Friendships .01 .02 Satisfied with Family .02 -.02 Mental Health Self-Rated Health -.01 Depression Scale .00 Physical Health Chronic Conditions Heart Conditions .05 Cancer .06 + Total Other Conditions .05 *** Self-Rated Health .01 Intercept 1.15 .86 .78 .72 Log Likelihood 1,847.4 1,891.4 1,896.7 1,924.0 N 4,240 Notes: a Weighted HRS data b Negative binomial estimates c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 160 Table 18. Estimated Net Effects of Religious Attendance, Affiliation, Salience, and Other Controls on Total Male Use (1992-1996) ab Total Male Use Model 1 Model 2 Model 3 Model 4 Religion Service Attendance (Never) c More than 1/Week .13 * .06 .06 .06 Once/Week .15 *** .09 * .10 * .10 * 2-3 Times/Month .15 *** .09 * .09 * .11 * 1-2 Times/Year or More .11 ** .08 * .08 * .10 * Affiliation (Evangelical Protestant) Catholic .01 -.04 -.04 -.05 Jewish .03 -.06 -.08 -.06 Mainline Protestant .02 -.03 -.03 -.03 Other -.13 -.15 + -.16 + -.15 + Not Affiliated -.04 -.11 + -.11 + -.08 Salience (Not Important) Very Important .02 .03 .04 .05 Somewhat Important -.01 -.00 -.00 .00 Demographic and Social Factors .02 *** .01 * Age .01 ** .02 *** Race/ethnicity (NH White) -.00 -.02 NH Black -.11 ** -.01 Hispanic -.17 ** .00 .00 .04 Foreign Born Status -.05 -.02 -.03 .01 Resources Education (in years) .02 *** .02 *** .02 *** Net Worth Low (> $32,500) -.14 *** -.13 *** -.16 *** .00 ** .00 ** .00 *** Household Income (in 1000 s) .13 *** .13 ** .16 *** Health Insurance Social Support Married/Living with Partner .08 * .08 + Satisfied with Friendships -.03 -.01 Satisfied with Family -.05 * -.04 + Mental Health Self-Rated Health .03 + Depression Scale -.00 Physical Health Chronic Conditions Heart Conditions .17 *** Cancer .15 * Total Other Conditions .10 *** Self-Rated Health .03 + Intercept -.23 -.74 -.49 -.58 Log Likelihood -2,862.9 -2,807.9 -2,801.0 -2,743.0 N 3,601 Notes: a Weighted HRS data b Negative binomial estimates c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 161 Table 19. Categories of Health Care Beliefs and Attitudes from the General Social Survey (1998). Personal Trust in One s Physician (alpha = .83) 1. My doctor is willing to refer me to a specialist when needed. 2. I doubt that my doctor really cares about me as a person. 3. I trust my doctor's judgments about my medical care. 4. I feel my doctor does not do everything s/he should for my medical care. 5. I trust my doctor to put my medical needs above all other considerations when treating my medical problems. 6. My doctor is a real expert in taking care of medical problems like mine. 7. I trust my doctor to tell me if a mistake was made about my treatment. Public Confidence in Physicians (alpha = .72) 1. Doctors aren't as thorough as they should be. 2. Doctors always do their best to keep their patient from worrying. 3. Sometimes doctors take unnecessary risks in treating their patients. 4. Doctors are very careful to check everything when examining their patients. 5. Doctors always treat their patients with respect. 6. Doctors cause people to worry a lot because they don't explain medical problems to patients. 7. The medical problems I've had in the past are ignored when I seek care for a new medical problem. 8. Doctors never recommend surgery (an operation) unless there is no other way to solve the problem. Attitudes Toward Health Care System (alpha = .79) 1. I worry that my doctor is being prevented from telling me the full range of options for my treatment. 2. I worry that I will be denied the treatment or services I need. 3. I worry that my doctor will put cost considerations above the care I need. 162 Table 20. Sample Characteristics from the General Social Survey, 1998 ab Range Mean S.D. Religion Service Attendance More than 1/Week 0-1 .07 .26 Once/Week 0-1 .19 .39 2-3 Times/Month 0-1 .17 .38 1-2 Times/Year or More 0-1 .38 .49 Never 0-1 .19 .39 Affiliation Catholic 0-1 .24 .43 Jewish 0-1 .02 .13 Mainline Protestant 0-1 .23 .42 Evangelical Protestant 0-1 .34 .47 Other 0-1 .02 .14 Not Affiliated 0-1 .15 .36 Strength of Affiliation 1-4 2.78 1.14 Attitudes Toward Health Care Personal Trust 0-35 25.15 4.83 Public Confidence 0-40 24.26 5.26 Health Care System 0-15 10.20 2.65 Demographic and Social Factors Age (in years) 18-89 46.02 17.03 Gender Female 0-1 .59 .49 Race White 0-1 .85 .36 Black 0-1 .15 .36 Nativity Foreign Born 0-1 .05 .22 Marital Status Married 0-1 .48 .50 Resources Education (in years) 0-20 13.30 2.90 Household Income 1-23 15.65 4.92 Health Insurance 0-1 .87 .33 Health Status Self-Rated Health 1-4 3.06 .80 Notes: a Unweighted, N=1,197 b Proportions may not add to 1 due to rounding 163 Table 21. Differences Between Mean Levels of Religious Service Attendance, and Strength of Affiliation by Affiliation (GSS, 1998) a Religious Affiliation Differences Mainline Evangelical Not by b Catholic Jewish Protestant Protestant Other Affiliated Affiliation Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) (p<.05) Religious Attendance More than 1/Week .09 (.29) .01 (.07) c, g, j, l, n .00 (.00) .06 (.23) .15 (.36) .02 (.15) c Once/Week d, e, l, n .19 (.40) .22 (.41) .09 (.29) .02 (.13) .14 (.36) .26 (.44) 2-3 Times/Month .02 (.15) c, e, l, n .19 (.40) .23 (.42) .17 (.39) .14 (.36) .16 (.37) 1-2 Times/Year or More .28 (.45) .39 (.50) .38 (.49) c, j, n .43 (.51) .44 (.50) .46 (.50) Never d, e, f, g, i, l, n, o .11 (.32) .26 (.45) .58 (.50) .29 (.46) .12 (.32) .09 (.29) 3.26 (.92) 1.12 (.54) c, e, i, j, l, n, o 3.14 (.96) 3.01 (.96) 3.22 (.92) Strength of Affiliation 2.94 (.92) 286 21 278 405 23 185 N Notes: a Unweighted data, N=1,197 b Affiliation differences indicate significant differences (at p<.05) between the groups as follows: a. Catholic and Jewish b. Catholic and Mainline Protestant c. Catholic and Evangelical Protestant d. Catholic and Other e. Catholic and Not Affiliated f. Jewish and Mainline Protestant g. Jewish and Evangelical Protestant h. Jewish and Other i. Jewish and Not Affiliated j. Mainline Protestant and Evangelical Protestant k. Mainline Protestant and Other l. Mainline Protestant and Not Affiliated m. Evangelical Protestant and Other n. Evangelical Protestant and Not Affiliated o. Other and Not Affiliated c Proportions may not add to 1 due to rounding 164 Table 22. OLS Regression Models for the Association between Religion Variables and Personal Trust in One s Physician (GSS, 1998) a Model 1 Model 2 Model 3 Model 4 Religion Service Attendance (Never) b More than 1/Week -.09 -.53 1.83 ** Once/Week 2.47 *** 2-3 Times/Month 1.59 ** 1.12 + 1-2 Times/Year or More 1.16 ** .90 * Affiliation (Evangelical Protestant) .73 + Catholic .83 * Jewish 2.43 * 2.50 * Mainline Protestant .88 * .87 * -2.65 * Other -2.82 * Not Affiliated -.50 .78 Strength of Affiliation .48 *** .40 * Intercept 23.09 23.95 22.94 21.82 Adjusted R-Square .02 .02 .01 .03 N 1,197 Notes: a Weighted data b Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 165 Table 23. OLS Regression Models for the Association between Religion Variables and Public Confidence in Physicians (GSS, 1998) a Model 1 Model 2 Model 3 Model 4 Religion Service Attendance (Never) b More than 1/Week 2.60 *** 1.64 * Once/Week 2.92 *** 1.96 *** 2-3 Times/Month 2.68 *** 1.78 *** 1-2 Times/Year or More 1.59 *** 1.06 ** Affiliation (Evangelical Protestant) Catholic -.12 .04 Jewish .74 1.16 Mainline Protestant .85 * 1.03 ** Other -.49 -.14 Not Affiliated -2.15 *** -.56 Strength of Affiliation .85 *** .32 + Intercept 23.35 25.31 22.80 22.89 Adjusted R-Square .04 .03 .04 .06 N 1,197 Notes: a Weighted data b Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 166 Table 24. OLS Regression Models for the Association between Religion Variables and Positive Attitudes Toward the Health Care System (GSS, 1998) a Model 1 Model 2 Model 3 Model 4 Religion Service Attendance (Never) b More than 1/Week .75 * .49 Once/Week .96 *** .62 * 2-3 Times/Month .90 *** .60 * 1-2 Times/Year or More .27 .07 Affiliation (Evangelical Protestant) Catholic .11 .22 Jewish 1.35 * 1.53 ** Mainline Protestant .53 ** .64 ** Other -.73 -.60 Not Affiliated -.57 * .05 Strength of Affiliation .31 *** .15 Intercept 9.71 10.12 9.32 9.29 Adjusted R-Square .02 .02 .02 .03 N 1,197 Notes: a Weighted data b Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 167 Table 25. Estimated Net Effects of Religious Attendance, Affiliation, Strength of Affiliation, and Other Controls on Personal Trust in One s Physician (GSS, 1998) ab Model 1 Model 2 Model 3 Model 4 Religion Service Attendance (Never) c More than 1/Week -.41 -.58 -8.94 ** -.77 Once/Week 1.83 ** 1.71 ** -4.08 + 1.57 ** 2-3 Times/Month 1.18 * 1.05 + .13 .91 1-2 Times/Year or More .99 * .88 + -.83 .78 + Affiliation (Evangelical Protestant) Catholic .88 * .85 * .83 + 1.25 Jewish 2.54 * 2.57 * 2.77 * 4.94 Mainline Protestant .91 * .85 * .85 * .21 Other -2.34 * -2.65 * -2.49 * 14.98 * Not Affiliated .96 .78 .86 5.80 * Strength of Affiliation .37 + .36 + .32 .38 + Demographic and Social Factors Age (in years) .02 ** .02 * .03 ** .03 ** Gender (Male) Female -.72 * -.78 * -.79 * -.78 * Race (White) Black .77 + .72 .71 .67 Nativity (US Born) Foreign Born -.18 -.10 .14 .05 Marital Status Married .05 .19 .19 -19 Resources Education (in years) .04 -.18 .13 Household Income -.11 ** -.11 ** -.12 ** Health Insurance 1.73 *** 1.76 *** 1.73 *** Health Status Self-Rated Health .54 ** .57 ** .57 ** Interaction Terms d Attendance * Education Highest * Education .65 ** High * Education .45 ** Medium * Education .09 Low * Education .14 Denomination * Education -.04 Catholic * Education Jewish * Education -.18 Mainline * Education .04 Other * Education -1.21 ** None * Education -.38 * Intercept 20.95 19.10 21.74 17.84 Adjusted R-Square .04 .07 .07 .07 N 1,197 Notes: a Weighted data b OLS regression estimates c Reference category in parentheses d Only significant interaction terms shown + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 168 Table 26. Estimated Net Effects of Religious Attendance, Affiliation, Strength of Affiliation, and Other Controls on Public Confidence in Physicians (GSS, 1998) abc Model 1 Model 2 Religion Service Attendance (Never) d More than 1/Week 1.59 * 1.39 * 1.88 *** 1.70 ** Once/Week 2-3 Times/Month 1.89 *** 1.69 ** 1-2 Times/Year or More 1.17 ** 1.04 * Affiliation (Evangelical Protestant) Catholic -.06 -.12 Jewish .77 .63 Mainline Protestant .80 * .69 + Other .22 -.05 Not Affiliated -.52 -.70 Strength of Affiliation .25 .25 Demographic and Social Factors Age (in years) .02 ** .02 ** Gender (Male) Female .02 .04 Race (White) Black -.61 -.53 Nativity (US Born) Foreign Born -1.35 * -1.29 * Marital Status Married .35 .27 Resources Education (in years) .04 Household Income -.04 Health Insurance 1.46 *** Health Status Self-Rated Health .47 * Intercept 22.04 19.51 Adjusted R-Square .06 .08 N 1,197 Notes: a Weighted data b OLS regression estimates c Reference category in parentheses d No significant interaction terms for the models shown here + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 169 Table 27. Estimated Net Effects of Religious Attendance, Affiliation, Strength of Affiliation, and Other Controls on Positive Attitudes Toward the Health Care System (GSS, 1998) ab Model 1 Model 2 Model 3 Religion Service Attendance (Never) c More than 1/Week .49 .26 .29 .54 + .32 .35 Once/Week 2-3 Times/Month .65 * .41 .45 1-2 Times/Year or More .16 .04 .05 Affiliation (Evangelical Protestant) Catholic .20 .11 2.73 ** Jewish 1.38 * 1.10 + 4.52 * Mainline Protestant .55 ** .39 + 1.24 Other -.48 -.74 -2.26 Not Affiliated .13 -.06 .98 Strength of Affiliation .10 .10 .07 Demographic and Social Factors Age (in years) .02 *** .02 *** .02 *** Gender (Male) Female -.12 -.10 -.09 Race (White) Black .04 .19 .23 Nativity (US Born) Foreign Born -.02 .02 -.04 Marital Status Married .29 + .09 .10 Resources Education (in years) .06 * .15 ** Household Income .01 .01 Health Insurance 1.01 *** 1.00 *** Health Status Self-Rated Health .36 *** .35 *** Interaction Terms d Denomination * Education Catholic * Education -.20 ** Jewish * Education -.24 -.07 Mainline * Education Other * Education .09 -.09 None * Education Intercept 8.50 5.65 4.62 Adjusted R-Square .04 .07 .08 N 1,197 Notes: a Weighted data b OLS regression estimates c Reference category in parentheses d Only significant interaction terms shown + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 170 Table 28. Interaction Effects between Religion Variables (Only Dependent Variables and Interaction Categories with Significant Terms Shown) a b Personal Trust Denomination*Attendance c Catholic*Attendance 1.19 ** Jewish *Attendance -2.82 * Mainline*Attendance .88 * Other*Attendance -.66 None*Attendance 1.68 ** Denomination*Strength of Affiliation Catholic*Strength .75 + Jewish*Strength -1.26 Mainline*Strength .54 Other*Strength -3.65 ** None*Strength 1.09 N 1,197 Notes: a Weighted data, OLS regression estimates b Models control for religion measures, and the demographic, social, and health variables c Attendance is measured as a continuous variables, ranging from 0-5. The dummy variables used in previous analyses were not used here due to small cell sizes. * p .05; ** p .01; *** p .001 (two-tailed test) 171 Table 29. Descriptions of the Alternative Outcomes: Confidence in Medicine and General Trust in Other People. Question Wording Range Mean (S.D.) N As far as the people running 1-3 2.35 (.63) 781 Confidence in Medicine these institutions are concerned, would you say you have a great deal of confidence, only some confidence, or hardly any confidence at all in them? Medicine? Generally speaking, would you 0-1 .42 (.49) 1,138 General Trust in People say that most people can be trusted or that you can t be too careful in life? 172 Table 30. Estimated Net Effects of Religious Attendance, Affiliation, Strength of Affiliation, and Other Controls on Confidence in Medicine (GSS, 1998) ab Model 1 Model 2 Model 3 Level of Confidence (Reference=Low) Medium High Medium High Medium High Religion Service Attendance (Never) c 8.19 ** 3.97 + 7.05 ** 7.54 ** 4.72 * 4.93 * More than 1/Week 2.36 + 2.19 2.59 + 1.73 2.14 2.29 + Once/Week 3.10 * 1.52 2.62 + 1.87 1.97 3.12 * 2-3 Times/Month 3.09 ** 2.83 ** 2.90 ** 3.21 ** 3.09 ** 3.19 ** 1-2 Times/Year or More Affiliation (Evangelical Protestant) 1.45 1.19 1.49 1.38 1.21 1.12 Catholic ------Jewish d 1.21 1.30 1.05 1.16 Mainline Protestant 1.13 1.15 .26 + .12 ** .19 * .35 .18 Other .19 * 1.38 1.23 1.13 Not Affiliated 1.44 1.33 1.45 * .72 + .86 .76 .84 .73 Strength of Affiliation .86 Demographic and Social Factors 1.00 1.00 .99 Age (in years) .99 Gender (Male) .98 .57 + .59 + Female .99 Race (White) 1.80 2.17 + 1.68 1.49 Black Nativity (US Born) 1.09 1.96 1.07 1.08 Foreign Born Marital Status 1.04 .87 1.00 1.20 Married Resources 1.14 * 1.08 Education (in years) 1.00 1.00 Household Income 3.08 ** 2.54 * Health Insurance Health Status 1.13 1.49 * Self-Rated Health Intercept 3.71 4.75 3.75 6.80 .24 .44 Likelihood Ratio 103 1328.6 1365.4 Notes: a Weighted data, N=781 b Multinomial regression odds ratios c Reference category in parentheses d Category too small to estimate + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 173 Table 31. Estimated Net Effects of Religious Attendance, Affiliation, Strength of Affiliation, and Other Controls on General Trust in People (GSS, 1998) ab Model 1 Model 2 Model 3 Religion Service Attendance (Never) c More than 1/Week 1.32 1.38 1.05 Once/Week 1.34 1.32 0.98 2-3 Times/Month 1.21 1.39 1.02 1-2 Times/Year or More 1.12 1.22 1.08 Affiliation (Evangelical Protestant) Catholic 1.50 * 1.19 1.04 Jewish 4.11 ** 2.97 * 1.87 Mainline Protestant 2.17 *** 1.77 *** 1.41 + Other 2.41 * 2.30 + 1.68 Not Affiliated 1.75 * 1.71 * 1.34 Strength of Affiliation 1.11 1.09 1.08 Demographic and Social Factors Age (in years) 1.01 ** 1.02 *** Gender (Male) Female 0.87 0.92 Race (White) Black 0.36 *** 0.42 *** Nativity (US Born) Foreign Born 0.98 0.97 Marital Status Married 1.33 * 1.21 Resources Education (in years) 1.19 *** Household Income 1.00 Health Insurance 1.38 Health Status Self-Rated Health 1.39 *** Intercept -1.15 -1.49 -5.19 -2Log Likelihood 1517.5 1468.2 1389.8 N 1,138 Notes: a Weighted data b Logistic regression odds ratios c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 174 Table 32. Estimated Net Effects of Religious Attendance, Affiliation, Strength of Affiliation, Biblical Literalism, and Other Controls on Attitudes Toward Health Care Providers and the Health Care System (GSS, 1998) ab Personal Trust Public Confidence Attitudes Toward System Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Religion -.02 .28 .03 -.14 .57 * .13 .59 + Biblical Literalism .61 * .54 + Service Attendance (Never) c .48 .65 1.68 * 1.94 * -.96 More than 1/Week -.88 .09 1.68 * .27 1.86 ** .95 Once/Week 1.02 .19 1.23 + .32 1.35 * .79 2-3 Times/Month .93 .01 .01 1.27 * 1.29 * .94 + 1-2 Times/Year or More .92 Affiliation (Evangelical Protestant) .08 .14 -.19 -.37 .87 .71 Catholic 1.31 + 1.67 * 2.73 * 2.01 + 2.84 * 2.85 * Jewish .34 .43 + .95 * .69 .59 .51 Mainline Protestant -.95 -1.11 -.24 -.03 -2.61 + -2.93 * Other -.01 -.11 -.73 -.77 1.18 .91 Not Affiliated .17 .21 + .24 .19 .56 * .61 * Strength of Affiliation Demographic and Social Factors Age (in years) .03 * .02 .02 *** Gender (Male) Female -.60 -.03 -.31 Race (White) Black .95 + -.55 .21 Nativity (US Born) Foreign Born -.59 -2.10 ** -.36 Marital Status Married .14 .49 .24 Resources Education (in years) .01 .05 .03 Household Income -.04 -.01 .01 Health Insurance 1.96 *** 1.18 * 1.08 *** Health Status Self-Rated Health .60 * .58 ** .39 ** Intercept 22.86 20.23 16.10 24.00 22.89 18.66 10.11 9.58 5.91 Adjusted R-Square .01 .03 .06 .01 .05 .08 .00 .02 .07 Notes: a Weighted data, N=774 b OLS regression c Reference category in parentheses + p .10; * p .05; ** p .01; *** p .001 (two-tailed test) 175 References Abrums, Mary. 2000. ""Jesus will fix it after awhile": meanings and health." Social Science and Medicine, 50: 89-105. American Cancer Society. 1998. Cancer Facts and Figures, 1997. New York: American Cancer Society. American Cancer Society. 2002. Can Prostate Cancer Be Found Early? Retrieved on October 2, 2002 from http://www.cancer.org/eprise/main/docroot/ CRI/content/CRI_2_4_3X_Can_prostate_cancer_be_found_early_36?sitearea=PED. American Nursing Association website, 2003. Retrieved on September 8, 2003 from Nursing World at http://nursingworld.org/snas/ia/parish.htm Amonkar, Mayur M., Suresh Madhavan, Sidney A. Rosenbluth, and Kenneth J. Simon. 1999. Barriers and Facilitators to Providing Common Preventive Screening Services in Managed Care Settings. Journal of Community Health, 24(3): 229-247. Andersen, Ronald M. 1968. A Behavioral Model of Families Use of Health Services. Chicago: Center for Health Administration Studies. Andersen, Ronald M. 1995. Revisiting the Behavioral Model and Access to Medical Care: Does It Matter? Journal of Health and Social Behavior, 36: 1-10. Andersen, Ronald M. and John F. Newman. 1973. Societal and Individual Determinants of Medical Care Utilization in the United States. The Milbank Memorial Fund Quarterly, 51: 95-121. Anderson, Robert N. 2001. Deaths: Leading Causes for 1999. National Vital Statistics Reports, 49(11): 1-88. Antonucci, Toni C. and Hiroko Akiyama. 1987. An Examination of Sex Differences in Social Support Among Older Men and Women. Sex Roles, 17: 737-749. Apel, Mary Dean. 1986. The Attitudes and Knowledge of Church Members and Pastors Related to Older Adults and Retirement. Journal of Religion and Aging, 293: 3143. Barr, Judith K., Susan Reisine, Yun Wang, Eric F. Holmboe, Karin L. Cohen, Thomas J. Van Hoof, and Thomas P. Meehan. 2001. Factors Influencing Mammography Use 176 Among Women in Medicare Managed Care. Health Care Financing Review, 22(4): 49-61. Becker, M.H. and L.A. Maimon. 1975. Sociobehavioral Determinants of Compliance with Health and Medical Care Recommendations. Medical Care, 13: 10-24. Benjamins, Maureen Reindl and Carolyn Brown. 2003. Religion and Preventative Health Care Utilization Among the Elderly. Social Science and Medicine, 58: 109118. Berkman, Lisa F. and S. Leonard Syme. 1979. "Social Networks, Host Resistance, and Mortality: A Nine-Year Follow-Up Study of Alameda County Residents." American Journal of Epidemiology, 109(2): 186-204. Blazer, Dan, Dana C. Hughes, and Linda K. George. 1987. "The Epidemiology of Depression in an Elderly Community Population." The Gerontologist, 27(3): 281287. Blazer, Dan G. and Erdman Palmore. 1976. Religion and Aging in a Longitudinal Panel. The Gerontologist, 16: 82-85. Bogart, Laura M. 2001. Relationship of Stereotypic Beliefs About Physicians to Health Care-Relevant Behaviors and Cognitions Among African American Women. Journal of Behavioral Medicine, 24(6): 573-586. Bogart, Laura M., Sheryl Thorburn Bird, Lisa C. Walt, Douglas L. Delahanty, and Jacqueline L. Figler. 2004. Association of Stereotypes About Physicians to Health Care Satisfaction, Help-Seeking Behavior, and Adherence to Treatment. Social Science and Medicine, 58: 1049-1058. Bosworth, Hayden B. and K. Warner Schaie. 1997. "The Relationship of Social Environment, Social Networks, and Health Outcomes in The Seattle Longitudinal Study: Two Analytical Approaches." Journal of Gerontology: 52B: 197-205. Bould, Sally, Beverly Sanborn, and Laura Reif. Eighty-Five Plus: The Oldest Old. Belmont, CA: Wadsworth Publishing Company, 1989. Bradley, D. E. 1995. Religious Involvement and Social Resources: Evidence from the Data Set Americans Changing Lives. Journal for the Scientific Study of Religion, 34:259-67. 177 Breen, Nancy and Larry Kessler. 1994. Changes in the Use of Screening Mammography: Evidence from the 1987 and 1990 National Health Interview Surveys. American Journal of Public Health, 84(1): 62-67. Breen, Nancy, Diane K. Wagener, Martin L. Brown, William W. Davis, and Rachel Ballard-Barbash. 2001. Progress in Cancer Screening Over a Decade: Results of Cancer Screening from the 1987, 1992, and 1998 National Health Interview Surveys. Journal of the National Cancer Institute, 93(22): 1704-1713. Carter, Jimmy. 1994. The Church s Preventive Medicine. Christianity Today, 38(7): 30-34. Carter, Jimmy. 1996. "Offering a Healing Hand: Religious Groups Can Bolster the Health of Their Surrounding Communities." Time, 148(19): 54-54. Catalona, William. 1995. New PSA Test Enhances Detection of Cancers, Cuts False Positives. Modern Medicine, 63(6): 22. Catholic Health Association of the United States website. 2004. Facts about the Catholic Health Association of the United States. Retrieved from: http://www.chausa.org/ABOUTCHA/CHAFACTS.ASP. Retrieved on: February 27, 2004. Catholic Health Association of the United States. 2003. Catholic Health Care in the United States. Ministry Engaged, September, 2003. Centers for Disease Control and Prevention. 1999. Surveillance for Use of Preventive Health-Care Services by Older Adults, 1995-1997. Morbidity and Mortality Weekly Reports, 48: 51-88. Centers for Disease Control and Prevention. 2000. Morbidity and Mortality Weekly Reports, 49(33): 750-755. Centers for Disease Control and Prevention. 2001. Influenza and Pneumococcal Vaccination Levels Among Persons Aged > 65 Years- United States, 1999. Morbidity and Mortality Weekly Reports, 50(29): 532.537. Centers for Disease Control and Prevention. 2002a. Reducing the Health and Economic Burden of Chronic Disease. Retrived on August 3, 2002 from www.cdc.gov/nccdphp/upo.intro.htm 178 Centers for Disease Control and Prevention. 2002b. Influenza: The Disease. Retrieved on September 15, 2002 from www.cdc.gov/ncidod/diseases/flu/fluinfo.htm. Centers for Disease Control and Prevention. 2002c. Chronic Disease Fact Sheets: Cardiovascular Disease. Retrieved on September 10, 2002 from www.cdc.gov/nccdphp/upo/factsheets.htm#cardiovascular. Chabner, Bruce A., Frank G. Haluska, James A. Talcott. 1997. "Screening Strategies for Cancer: Implications and Results." Journal for the American Medical Association 277(18): 1475-1476. Chapman, Gretchen B. and Elliot J. Coups. 1999. Predictors of Influenza Vaccine Acceptance among Healthy Adults. Preventive Medicine, 29: 249-262. Charatan, Fred. 2002. The Great American Mammography Debate. British Medical Journal, 324: 423. Coffield, Ashley B., Michael V. Maciosek, J. Michael McGinnis, Jeffrey R. Harris, M. Blake Caldwell, Steven M. Teutsch, David Atkins, Jordan H. Richland, and Anne Haddix. 2001. Priorities Among Recommended Clinical Preventive Services. American Journal of Preventive Medicine, 21(1): 1-9. Cone, James H. 1985. Black Theology in American Religion. Journal of the American Academy of Religion, 53: 755-771. Courtenay, Bradley C., Leonard W. Poon, Peter Martin, Gloria M. Clayton, and Mary Ann Johnson. 1992. Religiosity and Adaptation in the Oldest-Old. International Journal of Aging and Human Development, 34(1): 47-56. Crystal, Stephen and Dennis Shea. 1990. Cumulative Advantage, Cumulative Disadvantage, and Inequality Among Elderly People. The Gerontologist, 30(4); 437-443. Davis, Donna T., Ana Bustamante, Perry Brown, Girma Wolde-Tsadik, Edward W. Savage, Xiaoguang Cheng, and Letitia Howland. 1994. The Urban Church and Cancer Control: A Source of Social Influence. Public Health Reports, 109(4): 500506. Davis, James A., Tom W. Smith, and Peter V. Marsden. 1998. General Social Surveys: 1972-1998. Chicago: National Opinion Research Center. 179 Drociuk, D. 1999. Reasons Reported by Medicare Beneficiaries for Not Receiving Influenza and Pneumococcal Vaccinations United States, 1996. Morbidity and Mortality Weekly Reports, 48(39): 556. Ellison, Christopher G. 1995. Race, Religious Involvement, and Depressive Symptomatology in a Southeastern US Community. Social Science and Medicine, 40: 1561-1572. Ellison, Christopher G. and Linda K. George. 1994. Religious Involvement, Social Ties, and Social Support in a Southeastern Community. Journal for the Scientific Study of Religion, 33(1): 46-61. Ellison, Christopher G. and Jeffrey S. Levin. 1998. "The Religion-Health Connection: Evidence, Theory, and Future Directions." Health Education & Behavior, 25(6): 700-720. Ellison, Christopher G., and Darren E. Sherkat. 1995. "The `Semi-Involuntary Institution' Revisited: Regional Variations in Church Participation Among Black Americans." Social Forces, 73: 1415-1437. Elmore, Joann, Diana L. Miglioretti, Lisa M. Reisch, Mary B. Barton, William Kreuter, Cindy L. Christiansen, Suzanne W. Fletcher. 2002. Screening Mammograms by Community Radiologists: Variability in False-Positive Rates. Journal of the National Cancer Institute, 94(18): 1373-1381. Ensel, Walter M. 1986. "Measuring Depression: The CES-D Scale." In Social Support, Life Events, and Depression, Nan Lin, Alfred Dean, Walter M. Ensel (eds.). New York: Academic Press. Erwin, Deborah O., Thea S. Spatz, R. Craig Stotts, and Jan A. Hollenberg. 1999. Increasing Mammography Practice by African American Women. Cancer Practice, 7(2): 78-85. Federal Interagency Forum on Aging Related Statistics. 2000. Older Americans 2000: Key Indicators of Well-Being. Ferraro, Kenneth F. and Jerome R. Koch. 1994. Religion and Health Among Black and White Adults: Examining Social Support and Consolation. Journal for the Scientific Study of Religion, 33(4): 362-375. 180 Fox, Sarah A., Kathryn Pitkin, Christopher Paul, Sally Carson, and Naihua Duan. 1998. "Breast Cancer Screening Adherence: Does Church Attendance Matter?" Health Education and Behavior, 25(6): 742-758. Fox, Sarah A., Judith A. Stein, Rosa E. Gonzalez, Mary Farrrenkopf, and Ann Dellinger. 1998. A Trial to Increase Mammography Utilization Among Los Angeles Hispanic Women. Journal of Healthcare for the Poor and Underserved, 9(3): 309-321. Frankel, B. G. and W. E. Hewitt. 1994. Religion and Well-Being Among Canadian University Students. Journal for the Scientific Study of Religion, 33: 62-73. Garber, A.M., W.S. Browner, and S. B. Hulley. 1996. Cholesterol Screening in Asymptomatic Adults, Revisited. Annals of Internal Medicine, 124:518-531. General Board of Global Ministries website. 2004. John Wesley: Holiness of Heart and Life. http://gbgm-umc.org/umw/wesley/. Retrieved on April 5, 2004. General Social Survey. 2002. General Social Survey 1972-2000 Cumulative Codebook. www.icpsr.umich.edu/gss/. Goldberg, Todd H. and Stephen I. Chavin. 1997. Preventive Medicine and Screening in Older Adults. Journal of the American Geriatric Society, 45: 344-354. Granovetter, Mark S. 1973. The Strength of Weak Ties. American Journal of Sociology, 78(6): 1360-1380. Hale, W. Daniel and Richard G. Bennett. 2000. Building Healthy Communities Through Medical-Religious Partnerships. The Johns Hopkins University Press: Baltimore and London. Hastings, Dwayne. 1999. Southern Baptists Affirm Alcohol, Drug Abstinence. Retrieved on November 4, 2002 from the Southern Baptist Convention website: http://www.sbcannualmeeting.org/sbc99/news34.htm. Hayward, Rodney A., Martin F. Shapiro, Howard E. Freeman, and Christopher R. Corey. 1988. Who Gets Screened for Cervical and Breast Cancer? Archives of Internal Medicine, 149: 1177-1181. Health and Retirement Survey website. 2003. Retrieved on September 5, 2003 from http://hrsonline.isr.umich.edu/. 181 Heeringa and Connor. 1995. Technical Description of the Health and Retirement Study Sample Design. Online version; originally published as HRS/AHEAD Documentation Report DR-002. Retrieved on September 2, 2003 from http://hrsonline.isr.umich.edu/docs/userg/HRSSAMP.pdf Hewitt, Maria, Susan Devesa, and Nancy Breen. 2002. Papanicolaou Test Use Among Reproductive-Age Women at High Risk for Cervical Cancer: Analyses of the 1995 National Survey of Family Growth. American Journal of Public Health, 92(4): 666-669. Holmberg, L, A. Bill-Axelson, F. Helgesen, J. O. Salo, P. Folmerz, M. Haggman, et al. 2003. "Radical Prostatectomy or Watchful Waiting in Early Prostate Cancer?" Canadian Medical Association Journal, 168(1): 67-67. House, James S., Debra Umberson, and K.R. Landis. 1988. Structures and Processes of Social Support. Annual Review of Sociology, 14: 293-318. Hummer, Robert A., Richard G. Rogers, Charles B. Nam, and Christopher G. Ellison. 1999. "Religious Involvement and U.S. Adult Mortality." Demography, 36(2): 273285. Hurd, Michael. 1989. The Economic Status of the Elderly. Science, 244: 659-664. Idler, Ellen L. 1987. "Religious Involvement and the Health of the Elderly: Some Hypotheses and an Initial Test." Social Forces, 66(1): 227-238. Idler, Ellen and Stanislav Kasl. 1997. Religion Among Disabled and Nondisabled Persons II: Attendance at Religious Services as a Predictor of the Course of Disability. Journal of Gerontology, 52B: 306-316. Institute of Medicine. 2001. Promoting Health: Intervention Strategies from Social and Behavioral Research. Smedley, Brian D. and S. Leonard Syme (Eds.). Committee on Capitalizing on Social Science and Behavioral Research to Improve the Public's Health, Division of Health Promotion and Disease Prevention. International Parish Nurse Resource Center website. 2003. Retrieved on September 8, 2003 from http://ipnrc.parishnurses.org/aboutrev.phtml 182 Jacobellis, Jillian and Gary Cutter. 2002. Mammography Screening and Differences in Stage of Disease by Race/Ethnicity. American Journal of Public Health, 92(7): 1144-1150. Jacobs, Linda A. and Ellen Giarelli. 2001. Jewish Culture, Health Belief Systems, and Genetic Risk for Cancer. Nursing Forum, 36(2): 5-13. Jakobovits, Immanuel. 1975. Jewish Medical Ethics: A Comparative and Historical Study of the Jewish Religious Attitude to Medicine and Its Practice. 2nd Edition. New York: Bloch Publishing Co. Janes, Gail R., Donald K. Blackman, Julie C. Bolen, Laurie A. Kamimoto, Luann Rhodes, Lee S. Caplan, Marion R. Nadel, Scott L. Tomar, James F. Lando, Stacie M. Greby, James A. Singleton, Raymond A. Strikas, and Karen G. Wooten. 1999. Surveillance for Use of Preventive Health-Care Services by Older Adults, 19951997. Morbidity and Mortality Weekly Reports, 48(8): 51-88. Jewett, Michael A. S., Neil Fleshner, Lawrence H. Klotz, Robert K. Name and John Trachtenberg. 2003. "Radical Prostatectomy as Treatment for Prostate Cancer." Canadian Medical Association Journal, 168(1): 44-45. John Rylands Library website. 2004. John Wesley: An On-Line Exhibition. Retrieved from: http://rylibweb.man.ac.uk/data1/dg/methodist/jwol1.html. Retrieved on April 5, 2004. Kaslow, Florence and James A. Robison. 1996. Long-Term Satisfying Marriages: Perceptions of Contributing Factors. American Journal of Family Therapy, 24(2): 153.168. Katapoldi, M. C., N. C. Facione, C. Miaskowski, M. J. Dodd, and C. Waters. 2002. The Influence of Social Support on Breast Cancer Screening in a Multicultural Community Sample. Oncology Nursing Forum, 29(5): 845-852. Katz, Sidney, A.B. Ford, R. W. Moskowits, B.A. Jackson, and M.W. Jaffe. 1963. Studies of Illness in the Aged The Index of ADL: A Standardized Measure of Biological and Psychosocial Function. Journal of the American Medical Association, 914-919. 183 Kirkman-Liff, Bradford and Jennie Jacobs Kronenfeld. 1992. Access to Cancer Screening Services for Women. American Journal of Public Health, 82(5): 733735. Klassen, Ann C., Ann L. M. Smith, Helen I. Meissner, James Zabora, Barbara Curbow, and Jeanne Mandelblatt. 2002. If We Gave Away Mammograms, Who Would Get Them? A Neighborhood Evaluation of a No-Cost Breast Cancer Screening Program. Preventive Medicine, 34: 13-21. Koenig, Harold G., Michael McCullough, and David Larson. 2001. The Handbook of Religion and Health. Oxford University Press: New York. Koenig, Harold G., F. Shelp, V. Goli, Harvey J. Cohen, and Dan G. Blazer. 1989. Survival and Health-Care Utilization in Elderly Medical Inpatients with Major Depression. Journal of the American Geriatrics Society, 5(2): 123-131 Lasater, Thomas M., Barbara L. Wells, Richard A. Carleton, and John P. Elder. 1986. "The Role of Churches in Disease Prevention Research Studies. Public Health Reports, 101(2): 125-131. LaVeist, Thomas A., Kim J. Nickerson, and Janice V. Bowie. 2000. Attitudes about Racism, Medical Mistrust, and Satisfaction with Care among African American and White Cardiac Patients. Medical Care Research and Review, 57(Supp. 1): 146161. Levin, Jeffrey S. 1984. The Role of the Black Church in Community Medicine. Journal of the National Medical Association, 76: 477-483. Levin, Jeffrey S. and Kykiakos S. Markides. 1985. "Religion and Health in Mexican Americans." Journal of Religion and Health, 24(1): 60-69. Levin, Jeffrey S., Robert J. Taylor, and Linda M. Chatters. 1994. Race and Gender Differences in Religiosity Among Older Adults: Findings from Four National Surveys. Journal of Gerontology: SS, 49: S137-145. Levin, Katharine, Cynthia Smith, Cathy Cowan, Helen Lazenby, and Anne Martin. 2002. Inflation Spurs Health Spending in 2000. Health Affairs, 21(1): 172-181. 184 Linn, Lawrence S. 1967. Social Characteristics and Social Interaction in the Utilization of a Psychiatric Outpatient Clinic. Journal of Health and Social Behavior, 8(1): 314. Long, J. Scott. 1997. Regression Models for Categorical and Limited Dependent Variables. Sage Publications: Thousand Oaks, CA. Mahoney, Annette, Kenneth I. Pargament, Nalini Tarakeshwar, and Aaron B. Swank. 2001. Religion in the Home in the 1980s and 1990s: A Meta-Analytic Review and Conceptual Analysis of Links Between Religion, Marriage, and Parenting. Journal of Family Psychology, 15(4): 559-596. Maiese, Deborah R. 2002. Healthy People 2010 - Leading Health Indicators for Women. Women s Health Issues, 12(4): 155-164. Makuc, Diane M., Nancy Breen, and Virginia Freid. 1999. "Low Income, Race, and the Use of Mammography." Health Services Research, 34(1): 229-238. Markides, Kyriakos S. 1983. "Aging, Religiosity, and Adjustment: A Longitudinal Analysis." Journal of Gerontology, 38(5): 621-625. Markides, Kyriakos S., Jeffrey S. Levin, and Laura A. Ray. 1987. Religion, Aging, and Life Satisfaction: An Eight-Year, Three-Wave Longitudinal Study. The Gerontologist, 27(5): 660-665. McDavid, Kathless, Thomas A. Melnik, and Hrak Derderian. 2000. Prostate Cancer Screening Trends of New York State Men at Least 50 Years of Age, 1994 to 1997. Preventive Medicine, 31: 195-202. Mechanic, David. 1963. Religion, Religiosity, and Illness Behavior: The Special Case of the Jews. Human Organizations, 22:202-208. Merrill, Ray M. 2001. Demographic and Health-Related Factors of Men Receiving Prostate-Specific Antigen Screening in Utah. Preventive Medicine, 33: 646-652. Michielutte, Robert, Mark B. Dignan, and Bonnie Lanier Smith. 1999. Psychosocial Factors Associated With the Use of Breast Cancer Screening by Women Age 60 Years or Over. Health Education and Behavior, 26(5): 625-647. Miller, Anna M. and Victoria L. Champion. 1993. "Mammography in Women >50 Years of Age." Cancer Nursing, 16(4): 260-269. 185 Miller, Anna M. and Victoria L. Champion. 1996. "Mammography in Older Women: One-Time and Three-Year Adherence to Guidelines." Nursing Research, 45(4): 239-245. Miller, Steven L., William A. Norcross, and Robert A. Bass. 1980. Breast SelfExamination in the Primary Care Setting. The Journal of Family Practice, 10(5): 811-815. Mirola, W. A. 1999. A Refuge for Some: Gender Differences in the Relationship Between Religious Involvement and Depression. Sociology of Religion, 60: 419437. Mirowsky, John. 1999. "Analyzing Associations between Mental Health and Social Circumstances." Handbook of the Sociology of Mental Health Ed. Carol S. Aneshensel and Jo C. Phelan. New York: Kluwer Academic/ Plenum Publishers, pg. 105-123. Montazeri, Ali, Mehregan Hai-Mahmoodi, and Soghra Jarvandi. 2003. Breast SelfExamination: Do Religious Beliefs Matter? A Descriptive Study. Journal of Public Health Medicine, 25(2): 154-155. Murray, Michael and Carol McMillan. 1993. "Social and Behavioral Predictors of Women's Cancer Screening Practices in Northern Ireland." Journal of Public Health Medicine, 15(2): 147-153. Musick, Marc A. 1996. "Religion and Subjective Health Among Black and White Elders." Journal of Health and Social Behavior, 37(3): 221-237. Naguib, S. M., P. B. Geiser, and George W. Comstock. 1968. Responses to a Program of Screening for Cervical Cancer. Public Health Reports, 83: 990-998. National Cancer Institute Cancer Screening Consortium for Underserved Women. 1995. Breast and Cervical Cancer Screening Among Underserved Women. Archives of Family Medicine, 4: 617-624. National Cancer Institute. 2002. Cancer Screening Overview. Retrieved on September 10, 2003 from http://www.cancer.gov/cancerinfo/pdq/screening/overview. 186 National Cancer Institute. 2004. Screening For Breast Cancer: Summary of Evidence. Retrieved on April 10, 2004 from http://cancer.gov/cancerinfo/pdq/screening/breast/healthprofessional. National Opinion Research Center (NORC). 1999. General Social Surveys, 1972-1998 Cumulative Codebook. University of Chicago, IL. Nation's Health (editorial). 1994. "Churches and Other Partners Help Extend Public Health's Reach." Nation s Health, 24(11): 20-21. Nelson, Karin, Keith Norris, and Carol M. Mangione. 2002. Disparities in the Diagnosis and Pharmacologic Treatment of High Serum Cholesterol by Race and Ethnicity: Data from the Their National Health and Nutrition Examination Survey. Archives of Internal Medicine, 162: 929-935. Nichol, K. L., K. L. Margolis, J. Wuorenma, and T. Von Sternberg. 1994. The Efficacy and Cost Effectiveness of Vaccination against Influenza among Elderly Persons Living in the Community. New England Journal of Medicine, 331(12): 778-784. Nichol, K. L., A. Lind, K. L. Margolis, M. Murdoch, R. McFadden, M. Hague, S. Magnan, and M. Drake. 1995. The Effectiveness of Vaccination against Influenza in Healthy, Working Adults. New England Journal of Medicine, 333: 889-893. Numbers, Ronald L. and Ronald C. Sawyer. 1982. Medicine and Christianity in the Modern World. In Health/Medicine and the Faith Traditions. Editors Martin E. Marty and Kenneth L Vaux. Fortress Press: Philadelphia. O Malley, Ann S., Jon Kerner, Ayah E. Johnson, and Jeanne Mandelblatt. 1999. Acculturation and Breast Cancer Screening Among Hispanic Women in New York City. American Journal of Public Health, 89(2): 219-227. Oman, D. and D. Reed. 1998. Religion and Mortality Among the Community-Dwelling Elderly. American Journal of Public Health, 88: 469-1475. Ott, Philip. 1991. John Wesley on Health as Wholeness. Journal of Religion and Health, 30(1): 43-57. Parish Nursing website. 2003. Address: http://www.parishnurses.org/whatis.phtml. Retrieved on April 10, 2003. 187 Pastor P., D.M. Makuc, C. Reuben, H. Xia. 2002. Chartbook on Trends in the Health of Americans. Health, United States, 2002. Hyattsville, Maryland: National Center for Health Statistics. 2002. Pescosolido, Bernice A., Steven A. Tuch, and Jack K. Martin. 2001. The Profession of Medicine and the Public: Examining Americans Changing Confidence in Physicians from the Beginning of the Health Care Crisis to the Era of Health Care Reform. Journal of Health and Social Behavior, 42(1): 1-16. Powers, Daniel A. and Yu Xie. 1999. Statistical Methods for Categorical Data Analysis. Academic Press. Princeton Religious Research Center. 1994. Importance of Religion Climbing Again. PRRC Emerging Trends, 16(1): 1-4. Radloff, Lenore S. 1977. "The CES-D Scale: A Self-Report Depression Scale for Research in the General Population." Applied Psychological Measurement, 1(3): 385-401. Randolph, Whitney M., James S. Goodwin, Jonathan D. Mahnken, and Jean L. Freeman. 2002. "Regular Mammography Use is Associated with Elimination of Age-Related Disparities in Size and Stage of Breast Cancer." Annals of Internal Medicine, 137: 783-790. Regnerus, Mark D. and Glen H. Elder. 2003. Religion and Vulnerability Among LowRisk Adolescents. Social Science Research, In Press. Richardson, Andrea. 1991. Compassion and Cures: A Historical Look at Catholicism and Medicine. Journal of the American Medical Association, 226(21): 3063. Roetzheim, Richard G., Naazneen Pal, Colleen Tennant, Lydia Voti, Jaohn Z. Ayanian, Annette Schwabe, and Jeffrey P. Krischer. 1999. Effects of Health Insurance and Race on Early Detection of Cancer. Journal of the National Cancer Institute, 91(16): 1409-1415. Rosenstock, I. M. 1966. Why People Use Health Services. Milbank Memorial Fund Quarterly, 44: 94-121. Rosner, Fred. Modern Medicine and Jewish Law. 1972. New York: Yeshiva University Press. 188 Saag, K. G., B. N. Doebelling, J. E. Rohrer, S. Kolluri, R. Peterson, M. E. Hermann, and R. B. Wallace. 1998. Variation in tertiary Prevention and Health Service Utilization Among the Elderly: The Role of Urban-Rural Residence and Supplemental Insurance. Medical Care, 36(7): 965-976. Scandrett, Alfonso, Jr. 1996. "Health and the Black Church." Journal of Religion and Health, 35(3): 231-244. Scheff, Thomas J. 1966. Users and Non-Users of a Student Psychiatric Clinic. Journal of Health and Human Behavior, 7:114-121. Schieman, Scott, Kim Nguyen, and Diana Elliott. 2003. Religiosity, Socioeconomic Status, and the Sense of Mastery. Social Psychology Quarterly, 66(3): 202-221. Schiller, Preston L. and Jeffrey S. Levin. 1988. Is There a Religious Factor in Health Care Utilization? A Review. Social Science and Medicine, 27(12): 1369-1379. Schneider, Eric C., Paul D. Cleary, Alan M. Zaslavsky, and Arnold M. Epstein. 2001. Racial Disparity in Influenza Vaccination. Does Managed Care Narrow the Gap Between African Americans and Whites? Journal of the American Medical Association, 286(12): 1455-1460. Schommer, Jon C., Sandra R. Byers, Linda L. Pape, Gerald L. Cable, Marcia M. Worley, and Thomas Sherrin. 2002. "Interdisciplinary Medication Education in a Church Environment." American Journal of Health-System Pharmacists 59(1): 423-428. Segal, Bernard E, Robert J. Weiss, and Robert Sokol. 1965. Emotional Adjustment, Social Organization, and Psychiatric Treatment Rates. American Sociological Review, 30: 548-556. Shah, Babubhai V., Beth G. Barnwell, and Gayle S. Bieler. 1997. SUDAAN user s manual. (Release 7.5). Research Triangle Park, NC: Research Triangle Institute. Simon, G. E., J. Ormel, M. VonKorff, and W. Barlow. 1995. Health Care Costs Associated with Depressive and Anxiety Disorders in Primary Care. American Journal of Psychiatry, 152: 352-357. Sloan, Richard P., E. Bagiella, and T. Powell. 1999. Viewpoint: Religion, Spirituality, and Medicine. The Lancet, 353: 664-667. 189 Smith, James P. and Raynard Kington. 1997. "Demographic and Economic Correlates of Health in Old Age." Demography, 34: 159-170. Solon, J. A. 1966. Patterns of Medical Care: Sociological Variations Among a Hospital s Outpatients. American Journal of Public Health, 56: 884-894. Speedy, Sandra and Steward Hase. 2000. Health Beliefs and Perceptions of Women Presenting or Not Presenting for Mammographic Screening in a Rural Health Setting. Australian Journal of Rural Health, 8: 208-213. StataCorp. Stata Statistical Software: Release 8.0. 2003. College Station, TX: Stata Corporation. Steensland, Brian, Jerry Z. Park, Mark D. Regnerus, Lynn D. Robinson, W. Bradford Wilcox, and Robert D. Woodbury. 2000. "The Measure of American Religion: Toward Improving the State of the Art." Social Forces, 79(1): 291-318. Strawbridge, William, J., R.D. Cohen, S.J. Shema, and George J. Kaplan. 1997. Frequent Attendance at Religious Services and Mortality Over 28 Years. American Journal of Public Health, 87:957-961. Swaddiwudhipong, W., C Chaovakiratipong, P. Nguntra, P. Khumlam, and N. Silarug. 1993. "Influence of Religious Leaders on Smoking Cessation in a Rural Population." Morbidity and Mortality Weekly Report, 42(19): 367-369. Sweet, Leonard. 1994. Health and Medicine in the Evangelical Tradition. Trinity Press International: Valley Forge, Pennsylvania. Swinney, Jean, Cecilia Anson-Wonkka, Elizabeth Maki, and Jeannette Corneau. 2001. "Community Assessment: A Church Community and the Parish Nurse." Public Health Nursing 18(1): 40-44. The Gallup Report. 1987. Religion in America. The Gallup Report. 259. Thom, David H., Richard L. Kravitz, Robert A. Bell, Edward Krupat, and Rahman Azari. 2002. Patient Trust in the Physician: Relationship to Patient Requests. Family Practice, 19(5): 476-483. Thom, David H., Kurt Ribisi, Anita L. Stewart, Douglas A. Luke, and The Stanford Trust Study Physicians. 1999. Further Validation and Reliability Testing of the Trust in Physician Scale. Medical Care, 37(5): 510-517. 190 Tingen, Martha S., Sally Weinrich, Marlyn D. Boyd, and Martin C. Weinrich. 1997. Prostate Cancer Screening: Predictors of Participation. Journal of the American Academy of Nurse Practitioners, 9(12): 557-567. Tix, Andrew P. and Patricia A. Frazier. 1997. The Use of Religious Coping During Stressful Life Events: Main Effects, Moderation, and Medication. Journal of Consulting and Clinical Psychology, 66: 411-433. Turner, Dwayne C. 1996. The Role of Culture in Chronic Illness. The American Behavioral Scientist, 39(6): 717-728. United Pentecostal Church International Website. 2004. Doctrine of the United Pentecostal Church Divine Healing. Retrieved on January 12, 2004 from http://www.upci.org/doctrine/divine_healing.asp. United States Department of Health and Human Services (USDHHS). 2000. Healthy People 2010: Understanding and Improving Health. 2nd Edition. Washington, DC: US Government Printing Office. November 2000. United States Preventive Services Task Force. 1996. Guide to Clinical Preventive Services. 2nd Edition. Retrieved on August 5, 2002 from http://www.ahcpr.gov/clinic/cpsix.htm. United States Preventive Services Task Force. 2002. Guide to Clinical Preventive Services. 3rd Edition. Retrieved on August 5, 2002 from http://www.ahcpr.gov/clinic/cps3dix.htm. Unutzer, J., D. L. Patrick, G. Simon, D. Grembowsky, W. Walker, C. Rutter, and W. Katon. 1997. Depressive Symptoms and the Cost of Heatlh Serivces in HMO Patients Aged 65 and Older: 84 Year Prospective Study. Journal of the American Medical Association, 277: 1618-1623. Voorhees, Carolyn C., Frances A. Stillman, Robert T. Swank, Patrick J. Hagerty, David M. Levine, Diane M. Becker. 1996. "Heart, Body and Soul: Impact of ChurchBased Smoking Cessation Interventions on Readiness to Quit." Preventive Medicine, 25: 277-285. Wan, Thomas. T. and Scott. J. Soifer. 1974. Determinants of Physician Utilization: A Causal Analysis. Journal of Health and Social Behavior, 15: 100-108. 191 Wan, Thomas. T. and A. S. Yates. 1975. Prediction of Dental Services Utilization: A Multivariate Approach. Inquiry, 12:143-156. Wesley 2003 website. John Wesley 2003 Anniversary Events. Retrieved from: http://www.wesley2003.org.uk/events.htm. Retrieved on April 5, 2004. White, Richard H. 1968. Toward a Theory of Religious Influence. Pacific Sociological Review, Spring: 23-28. Wingo, P.A., S. Bolden, T. Tong, S. L. Parker, L. M. Marting, C. W. Heath Jr. 1996. Cancer Statistics for African Americans. Cancer Journal for Clinicians, 46:113125. Woolf, Steven. 1995. "Prostate Screening: The Pros and Cons." Modern Medicine, 63(2): 28-28. Wright, Charles J. and C. Barber Mueller. 1995. "Screening Mammography and Public Health Policy: The Need for Perspective." The Lancet, 346(8966): 29-32. Wynder, Ernst and Mary-Carroll Sullivan. 1982. Preventive Medicine and Religion: Opportunities and Obstacles. In Health/Medicine and the Faith Traditions. Editors Martin E. Marty and Kenneth L. Vaux. Fortress Press: Philadelphia. Xu, K. Tom and Tyrone F. Borders. 2003. Gender, Health, and Physician Visits Among Adults in the United States. American Journal of Public Health, 93(7): 1076-1081. Yi, Jenny Kisuk. 1994. Factors Association with Cervical Cancer Screening Behavior among Vietnamese Women. Journal of Community Health, 19(3): 189-200. Yi, Jenny Kisuk. 1998. Acculturation and Pap Smear Screening Practices among College-Aged Vietnamese Women in the United States. Cancer Nursing, 21(5): 335-341. 192 VITA Maureen Reindl Benjamins was born in Elmhurst, Illinois on June 13, 1976, the daughter of Sally Ann Reindl and Frank John Reindl, Jr. After completing her studies at Columbus East High School, Columbus, Indiana, in 1994, she entered Duke University in Durham, North Carolina. She received the degree of Bachelor of Arts from Duke in May 1998. During the following year, she was employed as a professional volleyball player for Montreux Volley Feminin in Montreux, Switzerland, and attended European University, an international business school. After returning to the United States in 1999, she entered the graduate program in sociology at the University of Texas at Austin. Permanent Address: 1640 Park Valley Dr., Columbus, IN 47203 This dissertation was typed by the author. 193

Find millions of documents here - Study Guides, Homework Solutions, Papers, Exam Answer Keys and more. Course Hero has millions of course related materials that will enable you to learn better, faster and get an A in all your courses.
Below is a small sample set of documents:

simpsonal13317.pdf
Path: Texas >> SIMPSONAL >> 13317 Fall, 2009

hamiltont84490.pdf
Path: Texas >> HAMILTONT >> 84490 Fall, 2009
kotrlaka518287.pdf
Path: Texas >> KOTRLAKA >> 518287 Fall, 2009
harrisont86130.pdf
Path: Texas >> HARRISONT >> 86130 Fall, 2009
brandonjc99738.pdf
Path: Texas >> BRANDONJC >> 99738 Fall, 2009
MATH107A46024536.doc
Path: MD University College >> ASIA >> 2092 Fall, 2009
crawforda65881.pdf
Path: Texas >> CRAWFORDA >> 65881 Fall, 2009
achacosom07761.pdf
Path: Texas >> ACHACOSOM >> 07761 Fall, 2009
jarroldwl86380.pdf
Path: Texas >> JARROLDWL >> 86380 Fall, 2009
sharyginany026.pdf
Path: Texas >> SHARYGINAN >> 026 Fall, 2009
goncalvesac026.pdf
Path: Texas >> GONCALVESA >> 026 Fall, 2009
zieglerkj47418.pdf
Path: Texas >> ZIEGLERKJ >> 47418 Fall, 2009
burtnerjc90760.pdf
Path: Texas >> BURTNERJC >> 90760 Fall, 2009
alvarezla07232.pdf
Path: Texas >> ALVAREZLA >> 07232 Fall, 2009
MATH012A46124534.doc
Path: MD University College >> ASIA >> 2092 Fall, 2009
bonningew86532.pdf
Path: Texas >> BONNINGEW >> 86532 Fall, 2009
CMIS141AA44024445.doc
Path: MD University College >> ASIA >> 2092 Fall, 2009
CMIS102AA42086692.doc
Path: MD University College >> ASIA >> 2088 Fall, 2009
STAT200A42186896.doc
Path: MD University College >> ASIA >> 2088 Fall, 2009
kulkarnis86095.pdf
Path: Texas >> KULKARNIS >> 86095 Fall, 2009
chapmanbg60287.pdf
Path: Texas >> CHAPMANBG >> 60287 Fall, 2009
slattonkc78713.pdf
Path: Texas >> SLATTONKC >> 78713 Fall, 2009
michalskylo026.pdf
Path: Texas >> MICHALSKYL >> 026 Fall, 2009
batemanmt33508.pdf
Path: Texas >> BATEMANMT >> 33508 Fall, 2009
lodowskid97061.pdf
Path: Texas >> LODOWSKID >> 97061 Fall, 2009
raichlend29983.pdf
Path: Texas >> RAICHLEND >> 29983 Fall, 2009
perkinsjd44616.pdf
Path: Texas >> PERKINSJD >> 44616 Fall, 2009
mehdiabadinj026.pdf
Path: Texas >> MEHDIABADI >> 026 Fall, 2009
borisovasa86653.pdf
Path: Texas >> BORISOVASA >> 86653 Fall, 2009
Abuhakema504399.pdf
Path: Texas >> ABUHAKEMA >> 504399 Fall, 2009
hw03_solution.doc
Path: Penn State >> ME >> 581 Fall, 2009
oestreichj19588.pdf
Path: Texas >> OESTREICHJ >> 19588 Fall, 2009
evstatieve01477.pdf
Path: Texas >> EVSTATIEVE >> 01477 Fall, 2009
paschvaldesg042.pdf
Path: Texas >> PASCHVALDE >> 042 Fall, 2009
alvaradocg86236.pdf
Path: Texas >> ALVARADOCG >> 86236 Fall, 2009
martinssonpj026.pdf
Path: Texas >> MARTINSSON >> 026 Fall, 2009
makowitza504694.pdf
Path: Texas >> MAKOWITZA >> 504694 Fall, 2009
andersonmw81540.pdf
Path: Texas >> ANDERSONMW >> 81540 Fall, 2009
martinezrs39334.pdf
Path: Texas >> MARTINEZRS >> 39334 Fall, 2009
elshayebta87380.pdf
Path: Texas >> ELSHAYEBTA >> 87380 Fall, 2009
cowmeadowr17589.pdf
Path: Texas >> COWMEADOWR >> 17589 Fall, 2009
schougaardsb029.pdf
Path: Texas >> SCHOUGAARD >> 029 Fall, 2009
kordoskyma87090.pdf
Path: Texas >> KORDOSKYMA >> 87090 Fall, 2009
metcalfets016-x.pdf
Path: Texas >> METCALFETS >> 016 Fall, 2009
bocknackbm84986.pdf
Path: Texas >> BOCKNACKBM >> 84986 Fall, 2009
mahdjoubid26824.pdf
Path: Texas >> MAHDJOUBID >> 26824 Fall, 2009
vanderveenaa029.pdf
Path: Texas >> VANDERVEEN >> 029 Fall, 2009
crabtreejc17037.pdf
Path: Texas >> CRABTREEJC >> 17037 Fall, 2009
steubingdm73657.pdf
Path: Texas >> STEUBINGDM >> 73657 Fall, 2009
johnsonhl692102.pdf
Path: Texas >> JOHNSONHL >> 692102 Fall, 2009
quintopozosd022.pdf
Path: Texas >> QUINTOPOZO >> 022 Fall, 2009
micklerpj516685.pdf
Path: Texas >> MICKLERPJ >> 516685 Fall, 2009
00000011.pdf
Path: Carnegie Mellon >> TERA >> 05102571 Fall, 2009
00000011.pdf
Path: Carnegie Mellon >> DISK >> 05102571 Fall, 2009
strycharskiat042.pdf
Path: Texas >> STRYCHARSK >> 042 Fall, 2009
podorozhnyr48572.pdf
Path: Texas >> PODOROZHNY >> 48572 Fall, 2009
alexandermw25054.pdf
Path: Texas >> ALEXANDERM >> 25054 Fall, 2009
batchelorme80690.pdf
Path: Texas >> BATCHELORM >> 80690 Fall, 2009
franzosajw504611.pdf
Path: Texas >> FRANZOSAJW >> 504611 Fall, 2009
brumbaughms81936.pdf
Path: Texas >> BRUMBAUGHM >> 81936 Fall, 2009
abplanalpbs52539.pdf
Path: Texas >> ABPLANALPB >> 52539 Fall, 2009
notes_05_02.doc
Path: Penn State >> ME >> 481 Fall, 2009
ulloabarbaran022.pdf
Path: Texas >> ULLOABARBA >> 022 Fall, 2009
poppendieckdg026.pdf
Path: Texas >> POPPENDIEC >> 026 Fall, 2009
abelturbyms87316.pdf
Path: Texas >> ABELTURBYM >> 87316 Fall, 2009
adejumobisa73660.pdf
Path: Texas >> ADEJUMOBIS >> 73660 Fall, 2009
mckelveyme504487.pdf
Path: Texas >> MCKELVEYME >> 504487 Fall, 2009
sankaralingamr029.pdf
Path: Texas >> SANKARALIN >> 029 Fall, 2009
santamariarh60629.pdf
Path: Texas >> SANTAMARIA >> 60629 Fall, 2009
ritchiedunhamj022.pdf
Path: Texas >> RITCHIEDUN >> 022 Fall, 2009