JW Scientific American

JW Scientific American - AN ARTICLE FROM SCIENTIFIC...

Info iconThis preview shows pages 1–11. Sign up to view the full content.

View Full Document Right Arrow Icon
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 2
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 4
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 6
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 8
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 10
Background image of page 11
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: AN ARTICLE FROM SCIENTIFIC AMERICAN APRIL 1982 VOL. 246 NO. Q REPRINTED NITH PERMISSION. COPYRIGHT (:) 1982 BY SCIENTIFIC AMERICAN, INC ALL RIGHTS RESERVED. I Variations in Medical Care among Small Areas The amount and cost ofhospfta/ trea (“ment 1'17 22 comm unity ha ve more to do With the number ofphysr'a’ans there, their medical speCI'altres and the procedures they prefer than With the health of the residents by John Wennberg and Alan Gittelsohn - There is a city in Maine where the surgical procedure of hysterecto~ my (removal of the uterus) was done so frequently in the past decade that if the rate persists, 70 percent of the women there will have had the opera- tion by the time they reach the age of 75. In a city less than 20 miles away the rate of hysterectomy is so much lower that if it persists, only 25 percent of the women will have lost their uterus by age 75. What could account for the disparity? It seems unlikely that there would be any large difference in the general health of the populations of the two neighboring cities, and after looking into the matter we have found none. The populations are similar in economic status. Differ‘ ences in the number of physicians, the supply of hospital beds and coverage by medical-insurance plans cannot explain the difference in the rate of surgery. In- stead the most important factor in deter- mining the rate of hysterectomy seems to be the style of medical practice of the physicians in the two cities. In one city surgeons appear to be enthusiastic about hysterectomy; in the other they appear to be skeptical of its value. We have examined the rate of surgery _ and other forms of medical treatment in 193 small areas in the six states of New England. The overall rate of sur- gery varies more than twofold among the areas. The total rate in a given area is correlated strongly with the number of surgeons there and with the number of hospital beds per capita; these are factors that themselves vary substantial- ly. The amount spent per capita on treat- ment in hospitals is also quite different from one area to the next. The rates of three of the most common surgical pro- cedures (hysterectomy, prostatectomy and tonsillectomy) vary even more dra~ matically: the highest rate is six times the lowest one. Even in communities with the same overall rate of surgery the rates of individual procedures can differ greatly. Hysterectomy, prostatec~ tomy and tonsillectomy cause much con~ 120 troversy among physicians. In the ab- sence of general agreement on their val— ue for individual patients the style of practice of the individual physician ap— pears to take precedence. The substantial variation from area to area in the consumption of medical care and in its per capita cost is sustained by the policies of hospital boards and ads ministrators, regulatory agencies and providers of medical insurance. The policies seldom take into account the ex- isting level of health care in a communiw ty; a common result is an increase in medical services in areas that already have high rates of consumption. When such inequities develop, the people re— ceiving the greater number of medical and surgical procedures do not neces- sarily benefit, particularly when the pro- cedures entail substantial risk. The 193 areas employed in our work cover the states of Connecticut, Maine, Massachusetts, New Hamp- shire, Rhode Island and Vermont. Our aim in constructing the areas was to specify the population that attends one local hospital. Except in cases of disor~ ders requiring elaborate treatment (such as cardiac surgery) people are generally treated at a nearby hospital. The atti- tude of the physicians at that hospital therefore has a strong influence on the rate of a given procedure in the sur~ rounding area. The analysis of medical care in small geographic areas has been made possi— ble by the establishment of computer- encoded records of hospital admissions in specific regions. In Maine, Rhode Is— land and Vermont there are registries that include information on each patient admitted to a hospital. The registry lists the patient’s age, sex, place of residence, diagnoses, surgical procedures, dates of admission and discharge and health on discharge. For Connecticut, Massachu- setts and New Hampshire our data on hospital admissions come from studies in which only the hospital and the pa- / tient‘s place of residence are record— ed. The expansion of health-insurance coverage, particularly the passage of the Federal Medicare Act in 1966, has yielded additional information about the medical care of specific populations. To construct the geographic areas we extracted from the records each pa- tient’s residence and the community in which treatment was received. In the records we utilized, the residence of the patient is recorded in the form of his Zip Code, minor civil division (township, for example) or census tract. For each of these small units of residence we deter— mined the community in which resi- dents are most likely to be hospitalized. All townships, census tracts and so on whose residents were most likely to go to a particular community to be treated were combined to form a hospital area. The 193 hospital areas defined in “ this way generally have populations of between 10,000 and 200,000, which is large enough for them to have stable rates of medical procedures. In almost all the areas the majority of hospital treatment is provided by facilities within the area. By counting the surgical procedures . done on the population of a hospital area in a given period, the per capita rate of surgery can be calculated. Simi- lar methods give the rate of other kinds of medical treatment. Insurance—reim' bursement rates can be calculated by totaling the reimbursements received by residents and dividing by the num— ber of residents who are members of an insurance program. The number of hospital beds per capita is also readily determined. Although the hospital areas can be em— ployed in a variety of analyses, much of our work has concerned rates of sur- gery because the information on surgi- cal procedures in regional record-keep— ing systems has been shown to be more reliable than that for other forms of treatment or for diagnosis. After adjusting for differences in age 4. among populations we have calculated the rates of hospital admission for surgi— cal procedures in the 11 most populous hospital areas in each of three states: Maine, Rhode Island and Vermont. Pro- cedures done on residents of a hospital area are counted toward the area’s total whether the operation took place within the area or outside it. Among the hospi- tal areas in each state the overall rate of surgery varies by a factor of about two. The variation in the rates of certain common procedures are more dramatic than the variation in the total. The high- est hysterectomy rate in the 33 areas (some 90 procedures per 10,000 women per year in 1975) is about four times the lowest rate. The highest rate of prosta- tectomy is also about four times the lowest rate. For tonsillectomy the high— est rate (about 60 procedures per 10,000 people per year) is six times the lowest rate. In many cases the diflerence be- tween the extreme rates for a procedure and the average rate for all the areas is statistically significant, indicating that the difference is unlikely to be a result of chance variation. Because of the wide range of rates residents of different areas face very different probabilities of having surgery. In one area of Vermont the ton- sillectomy rate from 1969 through 1971 was such that if it had persisted, 60 per- cent of all children would have had their tonsils removed by age 20. In a second Vermont area the rate was such that only 8 percent would have had their tonsils removed by age 20. In the area with fewer tonsillectomies, however, the prostatectorny rate was such that 59 percent of all men would have had their prostate gland removed by age 80. In a neighboring area only 35 percent would have had a prostatectomy by age 80. It is important to note that such large disparities in the rate of surgery are not observed for all common surgical pro— ~ cedures. The rates of cholecystectomy (surgical removal of the gallbladder) and appendectomy, for example, vary by a ratio of less than three to one. In few cases is the difference between the rate for an individual area and the aver- age rate statistically significant. The rate of herniorrhaphy (surgical repair of her- nia) varies even less, and most of the variation seems attributable to chance. The ratio of hospital beds to popula- tion and the average cost of being hospi- talized also vary greatly among New England communities. In the 11 most populous hospital areas in Vermont the highest ratio of hospital beds to popula tion is 6.8 beds per 1,000 people; the 7 lowest is 3.7. In Connecticut the highest ratio is more than four per 1,000; the lowest is less than two. (The ratios have been adjusted to compensate for resi~ dents treated outside their hespital ar- eas.) The Federal Health Planning Pro~ gram has specified four beds per 1,000 people as its standard for health-care planning; the ratios in New England thus range from well below the standard to well above it. Furthermore, the difier- ence between small areas is so great that the number of beds per capita in a state or a county (a measure often used in health planning) is not a reliable indica- tor of conditions in each community. The average reimbursement per area resident by agencies that provide medi- cal insurance is another highly variable quantity. In Bostonrthe area of Massa- chusetts with the highest rate of Medi- HOSPITAL AREAS of Vermont are small areas in which most residents go to a single hospi- tal. The hospital areas are indicated by heavy lines; boundaries between townships within an area are indicated by light lines; hospitals are indicated by circles. The six New England states have been divided into 193 hospital areas for an analysis of geographic variations in medical care. Because most care in each hospital area is provided by physicians at the local hospital the attitudes of those physicians have a large effect on rates of treatment. Although the residents of most hospital areas appear to have similar health, rates of common surgical procedures vary W, as much as sixfold from one area to another. The number of hospital beds per capita and the amount spent per capita on hospital care also vary greatly. In two areas there are two hospi- tals. In the areas without hospitals the residents receive most of their care in New Hampshire. / 121 care reimbursement, an average of $640 was paid to each person enrolled in the program in 1975. Across the Charles River in Cambridge the amount was $540. In Manchester, N.H., less than 50 miles away, it was $176. . The amount spent per capita on al treatment in hospitals is also quite in- consistent. In 1975, $324 was spent per capita in Boston; in Providence, $225 was spent. In New Haven, on the other hand, per capita expenditure was $153; in Hanover, N.H., it was $120. In all these areas the majority of admissions are to major teaching hospitals. The services provided in the hospitals are probably similar; one would not expect such a disparity in the amount spent on treatment. ' i’ Our work has shown that residents of L some areas receive much more medical treatment than others and spend more on that care. Why? One might assume that such differences are caused by bet- ter health in some communities, but this does not appear to be the case.>Surveys we made in selected hospital areas show that differences in the health of residents and in the other factors that affect the demand for medical care account for , TOWNSHIP HOSPITAL" 1 2‘ 3 4 5 , 6 7 "8 9 A 55 1,236 200 19 76 ‘4 ' s V' 1a 49 B o 82 o 16 85 826 '10 7 '13 c AND a 3 37 4 18 o 82 1,421 831 2,332 OTHER HOSPITALS 6 104 129 10 32 92 386 ' 72 ,230 GEOGRAPHIC DIVISIONS from which the 193 hospital areas were assembled included townships, census tracts and Zip Code units. Patients’ addresses and the hospitalsywhere they ‘ were treated were, taken from regional registries of hospital treatment. (Whether the town, the Zip Code or the census tract was recorded depended on the state.) For each small geographic di- vision it was determined which community the residents were most likely to go to for hospital treatment. The illustration shows nine townships whose boundaries are indicated by light lines. There are four hospitals in the townships, indicated by circles: one hospital each in townships 2 and 5 and two hospitals in township 8. The table indicates where the residents of each town- ship wcnt for hospital treatment in a single year. On the basis of this information the townships V 'were combined to form the three hospital areas whose boundaries are indicated by heavy lines. In each hospital area most of the residents are treated at hospitals in a single community. 122_ / only a small amount of the difference in the consumption of services. Ronald M. Andersen and Lu Ann Aday of the University of Chicago have listed certain characteristics that influence whether or not an individual will seek medical treatment. The most important factors are those that affect the person’s health or the perception of health. A variety of studies, however, have found that “enabling factors” such as income, health-insurance coverage and education can also have a strong effect. In addition “predisposing fac- tors" such as skepticism or faith in medi- Vcine appear to play a role. In the course of our work we have tried to determine how much of the observed difference in medical care could be explained bysuch characteristics. With Floyd J. Fowler, lid, of the University of Massachusetts at Boston we interviewed residents of six Vermont hospital areas. The residents were chosen to provide a representative sample of the local populations. The total surgery rate and the amount spent per capita on hospital treat- ment differed as much as twofold in the six areas; rates of some surgical proce— dures varied even more. The interviews, hoiivever, showed that the residents dif- fered little in the factors affecting the consumption of medical services. The average numbers of episodes of acute and chronic illness in each area were similar, as were the proportion of peo- ple with an income below the poverty level, the proportion with various kinds of health insurance and the proportion with access to a physician. Indeed, ap- proximately equal proportions Of the people in the areas visit a physician each year, as would be expected in popula- ‘ tions of similar wealth and health. The large differences in surgical rates and the amount spent on hospital care must ’therefore be’traced to factors that come into play after patients have contact with physicians. ' What is it that takes effect after the patient sees a physician to increase the patient’s chance of being hospitalized or of having surgery or a diagnostic pro- cedure? The crucial factor appears to be the system of medical care in the community. Although health and other demographic factors do not differ much among the six areas, the number of hos- pital beds and the number of physicians in proportion to the population vary widely. Moreover, the supply of hospi- tal facilities and the types of physicians who practice in the area are closely cor— related with overall consumption rates. Where there are many hospital beds per capita and many physicians whose spe- cialty or style of practice requires fre- quent hospitalization, there is more treatment in hospitals and greater ex- penditure per capita for hospital care. In hospital areas where there'are many general surgeons the surgery rate is high. The surgery rate and the rate of hospi- talization are also high in communities where a large proportion of the general practitioners do surgery. In areas where there are many internists many diagnosi- tic tests are given. ‘ That the overall rate of surgery is in— fluenced by the ratio of surgeons to pop- ulation has been known for some time. In 1970 John P. Bunker of the Stanford University School of Medicine found that the total surgery rate in the U.S. was about twice that in Britain, where there are fewer surgeons per capita. In TONSILLECTOMY RHODE ISLAND MAlNE VERMONT HYSTEFIECTOMY RHODE ISLAND MAINE VERMONT PROSTATECTOMY RHODE ISLAND MAINE VERMONT CHOLECYSTECTOMY ,RHODEISLAND MAINE VERMONT APPENDECTOMY RHODE ISLAND MAINE VEFIMONT HEFINIORRHAPHY RHODE ISLAND MAINE VERMONT 00" O 1973 a study supported by the Ameri— can College of Surgeons showed an anal- ogous relation in'regions of the U.S. The total rate of surgery and the like lihood of being admitted to a hos- , pital for treatment thus depend on the supply of physicians and hospital beds in the area. The wide variations in the rates of individual procedures, howev- er, are not caused by differences in the supply of resources alone. Our work suggests that such variations are due to differences in the style of medical prac- tice of local physicians. We examined the total surgery rate and the rate of in- dividual procedures in the five most populous hospital areas in Maine. In three of the areas the total rate is close to the average for the state, in one area the rate is above average and in one it is below average. In each area, however, a different surgical procedure is the com— monest one; all the commonest proce- dures are among those whose rates vary widely. For example, hysterectomy is the commonest procedure in one of the areas but the least common in another, o. 00:. O .0 . U... 20 30 40 50 60 7O 80 90 PROCEDURES PER 10,000 PEOPLE PER YEAR RATES OF SURGICAL PROCEDURES vary greatly among hos- pital areas. The rates shown are for the six commonest surgical pro- cedures for the repair or removal of an organ in the 11 most popu- lous hospital areas of Maine, Rhode Island and Vermont. The rate of tonsillectomy varies about sixfold among the 33 areas; the rates of hysterectomy and prostatectomy vary about fourfold. Moreover, many of the extreme rates for these procedures differ from the aver- age rate for the state by an amount that is statistically significant (col— or). There is much disagreement among physicians on the value of these procedures and the conditions for which they should be done. The rates of cholecystectomy (gallbladder removal) and appendecto- my vary less, and few of the extreme rates for these procedures differ from the state average by an amount that is statistically significant. The rate of herniorrhaphy (surgical repair of hernia) varies least. /’ . 1 ') Q TONSILLECTOMY PROBABILITY l J l l L J 30 4o 50 so 70 so AGE (YEARS) HYSTERECTOMY PROBABILITY (J I __J_ l l l O _L r | .L 50 60 70 80 O 10 20 30 40 AGE (YEARS) PROSTATECTOMY PROBABILITY Li) I o 10 20 so 40 50 ‘ AGE (YEARS) PROBABILITY OF HAVING SURGERY in 11 Vermont hospital areas has a wide range of variation as a result or the differences in surgical rates. In each graph the lines indicate the frac- tion of the population that would have had the procedure if the rates of surgery prevailing in about 1970 had persisted. The top line is for the area with the highest rate of the 11. The mid- dle line is based on the average rate for the state. The bottom line is for the area with the lowest rate. In one hospital area 60 percent of all children will have had their tonsils removed by age 20, whereas in another area fewer than 10 percent will have a tonsillectomy by that age. The fractions that will haV‘e had hysterectomy and prostatectomy also Show substantial differences. / 124. although the two areas have the same overall rate of surgery. In each of the five areas of Maine the rates of common surgical procedures constitute a “surgical signature” that tends to be consistent over many years, unless physicians leave the area or enter it. In each signature the rates of some procedures exceed the state average; those of other procedures fall below the average. Nora Lou Roos, Leslie L. Roos and their colleagues of the University of Manitoba Faculty of Medicine reached similar conclusions after analyzing var— iations in small areas in Manitoba. We have accounted for the factors that might influence the rates of surgical procedures, including the health ofhresi- dents, the supply of hospital beds and the number of physicians, Even taken together, these factors cannot explain all the variation in rates of individual procedures. The strongest remaining hypothesis is that the judgments and preferences of physicians give rise to the surgical signature. Some of the most persuasive evidence that the style of practice adopt- ed by physicians has a strong influence on surgery rates comes from studies in which physicians are told of geographic variations in the rates. The studies also show that physicians’ attitudes can be ' changed. In the 1950’s Paul Lembke of the University of Rochester employed information similar to ours to calculate per capita rates of surgery in communi- ties near Rochester, N.Y. He also per- suaded physicians there to undertake an audit of surgical procedures. Soon after- ward the rates were reduced in some areas Where they had been high. We fol- lowed a similar course in Vermont, ex- cept that no formal audit was made. In- formation on the rate of tonsillectomy in each hospital area was given to the Vermont Medical Society. In the area with the highest rate physicians estab— lished the requirement that a second opinion be obtained before a tonsillecto- my was done. As a result the probability that a child living in the area would have ' a tonsillectomy before age 20 declined from 60 percent to less than 10 percent. It had been suggested that if tonsillecto— my became much less common at the local hospital, the people of the area would go to other nearby hospitals to obtain the surgery for their children. This did not happen, implying that de- mand by residents for the procedure had not been a major factor in maintaining the high tonsillectomy rate. How can the decisions made by physi- cians vary so widely from one commu- nity to another a few miles away? It seems that the procedures whose rates vary the most are the ones whose risks and benefits are least well established in the medical profession. In some in- stances the value of the procedure itself has been questioned; in others the crite- BEDS PER 1,000 PEOPLE CONN. ME. MASS. NH. RI. VT. NUMBER OF HOSPITAL BEDS per capita and the annual amount spent per capita on hospital treatment also show the influence of geo- graphic variations in medical care. The data are for the 1 1 most popu- lous hospital areas in each of the six New England states. The num- ber of hospital beds per 1,000 people (adjusted for the number of poo. ple who leave their hospital area for treatment) ranges from about two to more than six. The ratios thus range from well below to well +100 + 01 O DIFFERENCE FROM STATE AVERAGE (PERCENT) é. o o —100 AREA 1 AREA 2 - TONSILLECTOMY _] HYSTERECTOMY VARlCOSE VEINS PROSTATECTOMY HEMORRHOIDECTOMY ALL PROCEDURES 350 300 250 200 ANNUAL PER CAPITA EXPENDITURE ON HOSPITALIZATION (DOLLARS) 150 ° 8 . ‘ 0:. :‘O . : o o .g. . o 1001—- 3 i o.- 50'- UL CONN. ME. MASS. NH RI. VT. above the four beds per 1,000 estahlished by the Federal Health Plan- ning Program as a standard. Furthermore, the variation in each state is so great that the number of beds per capita in the state or county as a whole (a measure often employed by health-planning agencies) bears little relation to the conditions prevailing in each community. The average amount spent on treatment in hospitals in 1975 ranged from less than $100 per capita to more than $300 in the 66 areas. AREA 3 AREA 4 AREA 5 “SURGICAL SIGNATURE” of a hospital area reflects the specialties and preferences of the surgeons who practice in the area. Each group of six bars shows the rates of five common surgi- cal procedures and the total surgery rate for one of the five most populous hospital areas in Maine. The rates are expressed in relation to the state average. The total surgery rate in each area is closely correlated with the number of surgeons there. The rates of individual proce- dures, however, are not determined by the number of surgeons. Areas 2 and 3 have the same total rate, but their signatures are quite different. In area 2 hysterectomy is the commonest pro- cedure; in area 3 it is the least common. A signature is generally consistent over many years. ria for selecting patients for the opera- tion are not definitive. Tonsillectomy, for example, was once done almost rou- tinely for minor inflammation but is now usually reserved for more serious cases. Some practitioners, however, re- tain the older attitude. In the case of hysterectomy there is general agreement on its necessity in uterine cancer. The operation is most often done, however, for a variety of less threatening condi- tions; the appropriateness of the proce- dure in these circumstances has been widely questioned. In contrast, the procedures whose rates vary little are those that provoke little disagreement. Inguinal hernia, for example, is easily recognized; the treat- ment of choice is surgical repair. Only where the simultaneous presence of oth« er conditions makes surgery dangerous is any other therapy employed, at least in the US. Consequently the rate of sur— gical repair of hernia is relatively con- stant in the hospital areas. - The uncertainty of the medical pro- fession about the controversial proce- dures can be great indeed. Ira M. Rut- Row and George D. Zuidema of the Johns Hopkins University School of Hygiene and Public Health and one of us (Gittelsohn) recently surveyed a group of randomly selected surgical specialists. Each surgeon was given a set of fabricated case histories and asked whether he would recommend a partic- ular surgical procedure for them. There was armarked divergence of opinion. For the three fictitious cases related to hysterectomy 25 percent of the surgeons thought none of the cases warranted sur- gery; 5 percent thought all three did. The remaining 70 percent recommend- ed surgery in one case or two cases. Sim- ilar inconsistencies appeared for breast surgery, varicose—vein surgery, tonsil- lectomy, gallbladder removal, cataract surgery and prostatectomy. Earlier studies had also demonstrated extreme conflicts of opinion.» In 1934 workers from the American Child Health Association chose 1,000 school- children to be examined by physicians, who were to determine whether or not they should have their tonsils removed. Six hundred children had already had the procedure and were removed from the sample. The remaining 400 were ex- amined by school physicians, who rec- ommended that 45 percent have a ton- sillectomy. Those that remained after the first round of examinations were ex- 200 150 100 PROCEDURES PER 10,000 PEOPLE PER YEAR 50—- ;; Manama ., mu m we” 01 1969 1971 1973 1975 TONSILLECTOMY RATE in Vermont demonstrates the strong influence of physicians’ preferences on how often a surgical procedure is done. The colored line is for the hospital area with the highest tonsillectomy rate in the state. The gray line shows the state average. The 501- id black line is for the area with the lowest rate in the state. In 1971 physicians in the area with the highest rate were told of that fact. They established the practice of obtaining a second opinion before a tonsillectomy was done, and the rate fell to a level roughly equal to the lowest in the state. (The decline before 1971 followed a single physician’s leaving the hospital area.) 126 1 977 amined by another group of physicians who recommended that 46 percent of them have their tonsils out. A third examination by still another group of physicians led to 44 percent of the re— mainder having tonsillectomy recom- mended. After three successive rounds only 65 of the original 1,000 children had not had tonsillectomy recommend- ed for them. For many common illnesses welltdea signed clinical studies to test alternative forms of therapy have not been done. For this reason there is conflicting infor- mation on whether a particular proce- dure will improve a patient’s health or the quality of his life. Many diagnostic and therapeutic techniques are adopted or discarded on the basis of fashion or a physician’s personal experience rath- er than on more reliable grounds. In the absence of authoritative stan-J dards differences among physicians in perceptions of illness and preferences for treatment appear to be the cause of much variation in rates of surgery and other kinds of treatment. The variation is perpetuated by regulatory agencies and providers of medical insurance. In determining whether new hospital facil. ities should be constructed, for example, regulatory agencies often rely on the oc- ~v cupancy rate of local hospitals and the average cost of treating a particular kind of case there. Such measures of econom- ic efficiency are not strongly correlated with the per capita consumption of med- ical services in the community or with the amount spent per capita on medical services. Utilization of narrow econom- ic criteria may lead to an increase in medical care in areas where rates are already well above the average. Our colleagues Richard J. Greene, Harvey M. Sapolsky and Drew Alt- man of the Massachusetts Institute of Technology ‘ have examined all deci~ sions by regulatory agencies concerning the building of new hospital facilities in Maine, Rhode Island and Vermont from 1975 through 1978. Hospitals in areas where the ratio of hospital beds to population was above the standard of four per 1,000 were just as likely to seek permission to add beds as those in areas where the ratio was below the standard. Moreover, the likelihood that health-planning officials would agree that new beds were needed had no rela~ . tion to the ratio. The cause of such irrational decisions is the tendency to consider a hospital in isolation from the level of medical care in the surrounding community. An ex- treme example is a medium-size facility in Maine that cares for acutely ill pa- tients. In 1975 hospital ofiicials applied for permission to add 11 beds. The ofii~ cials stated that more physicians were being recruited to the area and that their patients would need the additional fa- cilities. Because of the recruitment of physicians the number of admissions to the hospital had already begun to rise. The occupancy rate was then 75 percent (close to the recommended standard). It was estimated that the recruitment of more physicians would raise the occu-i pancy rate above the acceptable level. At the time the application for new beds was made the area from which ' most of the hospital’s patients came had 6.3 beds per 1,000 people; 80 percent of the beds were in the hospital whose management had made the request. With the addition of 1] beds the ratio would be 7.0 per 1,000, which is 75 per~ cent higher than the Federal'standard. Furthermore, the rate at which people in the area were admitted to hospitals was 246 per 1,000 per year, the second—high- est rate in the state and 50 percent higher than the state average. The total surgical rate was one-third higher than the state average. The per capita rates of four of the six common surgical procedures were more than double the state average and were increasing as the supply of physicians grew. These facts were not considered by the state health—planning ofi’icials, who approved the application for new beds. This case is one of several we have seen in which regulatory deci- sions contributed to disparities among small areas without evidence of a need for more medical care. Paradoxically, a low hospital-occu- pancy rate may also contribute to in- creases in medical care in areas with hospitalization rates that are well above average. Low occupancy rates are taken to be a sign of inefficiency. A hospital with a rate of less than 75 percent often comes to the attention of regulatory agencies and health planners. Hospital managers in such a situation fear they will be compelled to reduce the number of beds in the hospital; a common re- sponse is to recruit additional physi— cians. Such decisions are generally made without considering the need for the physicians, entire need for the services the physicians are likely to provide. I The measures that provide the only direct estimate of the level of health care in a community are population- based: the number of hospital admis— sions per capita, the number of occupied beds per capita and the amount spent per capita on hospital care. The last two measures give the best estimate of the consumption of medical services in a community. They depend much more strongly on the per capita admission rate than on the average length of stay in local hospitals or the average cost of a stay. The narrow measures of efficiency, however, are the ones more commonly employed by health planners. The popu- lation—based rates can be derived from studies of small areas. Their employ- ment would make analysis of variation an explicit consideration in regulatory " decisions, and it could help to reduce §§§§r§r§r§r§r§i§ §§§§§§r§§§i§ §§t§r§r§r§r§t§§i§ r5r§s§r§ @Efifiéé‘fificfié seesssesss sasssssaas §§§§§§§§§§ aseaaseeaa §§§§§§§§§§ eaasaeeaee §§§§§§§§§§ asageaeaeé seaeeeasaa aseaeessee seesasegss sssessasse sssaaasess sssssgsses sesegaggas gasseseaaé sassseaeas §§§§§§§§§§ §§§§§§§§§§ §§§§§§§fififi §§§§§§§§5§ §§§§§§§§§fi #§§§§§§#§§ §r§§§§§i§t§s§5 §§r§§r§r§§t§s§§ i§i§i§r§ r @ fififl§§§§§§§ §§§§§§§§§§ §§§§§§§§§§ [email protected] §§§§§§é§§fi §§§§§§&§§§ §§§§§§§§§§ §§§§§§§§§§ §§§§§§§§§§ fi§§§§§§§§§ §§§§§§§§§§ §§§§§§§§§§ /( v OCCUPANCY RATE of local hospitals has little relation to the per capita level of hospital treatment in a community. Regulatory agencies and hospital administrations that utilize the oc- cupancy rate as a measure of the efficiency of the hospital contribute unwittingly to excessive medical care. The illustration shows two hospital areas, each of which has 25,000 residents and a 100-bed hospital. Each area thus has the ideal ratio of beds to people (four per 1,000) set by Federal agencies. One-fifth of the hospitalized patients from area A go to the hospital in area B, which has an occupancy rate of 90 percent, above the recommended level. The manage- ment of that hospital asks permission to add 20 beds, a request granted because of the high oc- cupancy rate. At the hospital in area A the occupancy rate is 60 percent, often taken as a Sign of inefl‘icieney. Hospital ofi‘icials recruit physicians, whose patients raise the rate to 85 percent. Application is then made to add 20 beds and is approved. As a result the two areas now have a total of 240 hospital beds, or almost five beds per 1,000 people, well above the standard. some of the wide variation in services ‘ among hospital areas. The policies of agencies that provide medical insurance also contribute to dis- parities in medical care. As a result of their policies residents of areas where rates of consumption are low subsidize residents of areas where rates are high. Again, the reason for the imbalance is that the local level of medical care is overlooked. The contributions to insur- ance pools (in the form of private insur- ance premiums or taxes) are determined by rates of utilization of medical care averaged over large regions. In areas where per capita expenditures for medi- cal care are lower than the average, part of the premium goes to pay for services received by people in the areas with high rates of consumption. We examined Medicare contributions and reimbursements in 16 Vermont hos- pital areas. The contributions of those enrolled in the Medicare Part B pro- 200 150 136W UNDER MEDICARE PART B (DOLLARS) 8 o REIMBURSEMENT PER ENROLLEE O POLICIES OF INSURERS that do not take into account local variations in medical practice effectively subsidize treatment in areas with a high rate of care. The illustration shows the aver- age reimbursement per enrollee in Medicare Part B in 16 Vermont hospital areas in 1972. En- rollees paid $68 each; the amount was matched by the Federal Government. In one area the average reimbursement was less than $68; residents of that area received less than they had paid out to the Medicare program. In 10 areas reimbursement was between $68 and $136. In five areas an average of more than $136 was paid to each enrollee. The medical care of enroll- ees in those five areas was thus subsidized by Medicare recipients and taxpayers in other areas. 1.00 ._ MEDICARE RECIPIENTS 65 THROUGH 69 MEDICARE RECIPIENTS 75 THROUGH 79 MEN HAVING PROSTATECTOMY 65 THROUGH 69 .95 to o MEN HAVING PROSTATECTOMY 75 THROUGH 79 MEDICARE RECIPIENTS 85 AND OLDER . PROPORTION SURVIVING 3 b0 . o .75 MEN HAVING PROSTATECTOMY 85 AND OLDEFI .70 J I 1_ OJ... 0 60 120 180 240 300 360 DAYS FOLLOWING PROSTATECTOMY COST IN LIVES of high rates of surgery may be considerable. Of the procedures for which the authors found substantial geographic variation prostatectomy is the riskiest. The graph compares the proportion of men who have had a prostatcctomy (for a reason other than can— cer of the bladder or the prostate gland) who survive for a year after the procedure with the proportion of all male Medicare recipients of the same age who survive for a year. In the three age groups studied the risk of death is much higher among those who have had a pros- tatectomy; moreover, the added risk of death persists throughout the year after the procedure. ' f/ 132 gram are uniform: in 1972 they were $68 per capita. The average reimburse- ment per capita in the US. that year was $136; Federal taxes paid the $68 differ- ence. In the area where reimbursement was lowest, the average amount re- ceived was less than $68 per person en- rolled; Medicare recipients in the area were thus receiving less in services than . they had paid in premiums. In 10 areas the average recipient got more than $68 but less than $136. In five areas recip- ients were getting more than $136. in those communities the physicians who treated Medicare recipients and the hos- pitals inwhich the care was provided were in effect receiving contributions from enrollces and taxpayers in regions where less medical treatment was given. When all forms of insurance are tak- en into account, the value of such subsidies can be large even in small communities. We examined two neigh- boring hospital areas in Vermont. Both areas have roughly 11,000 residents, dis- tributed about equally by age and sex. Surveys of households in the two areas showed that residents had similar rates of illness and amounts of medical insur- ance, and they visited their physicians about as often. They had, however, quite different rates of admission to hos- pitals. We calculated how much would have been spent on hospital care by resi- dents of the two areas if their rates of hospitalization had been the average for the state. In one area 33 million more than the average had been spent over a 15-year period. In the other area $4 mil- lion less than the average had been spent. Most of the excess $3 million in the area where the rate was high was spent at the local hospital, benefitting the local economy. In the area where the rate was low, on the other hand, the sav- ings were not realized in the local econo- my. Much ofl‘the difference was trans- ferred to other communities (including the neighboring one) in the form of in- surance premiums and taxes. Such subsidies might be more trou- bling to residents of areas with low rates of hospitalization if they realized that the need for services is no greater in ar- eas where rates are high. Moreover, the imbalance may not be beneficial to the recipients of the greater medical care. Without controlled clinical trials it is not possible to determine conclusively which populations are getting better medical care. Nevertheless, it is clear that more medical treatment is not nec- essarily better treatment. Some of the procedures whose rates vary widely are quite risky. Of those we considered the riskiest is prostatecto- my. Some 1.3 percent of patients who have the procedure die in the hospital. To get an idea of the risks associated with variations in medical care we cal- culated how many people would have died in the US. in l975 from prostatec- JAPANESE CHINESE RUSSIAN GERMAN Mail correspondence courses. with tape assignments. Send $10 for Starter package, speci~ tying language. 3-week seminars in Smithers,‘ with language lab: skill-ori- ented crash'courses in June—— ,German. July—Russian, Au- gust—Chinese, September—Ja rates and accommodation information. FAR EAST BUSINESS LANGUAGE SERVICES Dr. Peter Leimbigler, President and Chief Instructor Box 3922, Smithers, B.C. Canada VOJ 2N0. Phone (604) 8472563 A Canadian Language School Serving the Western World panese. Write for free timetable, SCIENTIFIC AMERICAN is now available to the blind and physically handi— capped on cassette tapes. All inquiries should be made directly to RE, CORDED PERIODI- CALS, Division of Volunteer Services for the Blind. 919 Walnut Street, 8th Floor, Philadelphia, PA 19107. ONLY the blind or handicapped should apply for this service. There is a nominal charge. COUNCIL OF BIOLOGY EDITORS 26th ANNUAL MEET/NC LOUISVILLE, KENTUCKY, at the HYATT REGENCY HOTEL 4-7 MAY 1982 ’ PROG RAM 4 May, P.M.: Keynote Address 5 May, A.M.: Issues for Editors P.M.: 6 May, A.M.: Proceedings Volumes— Publish or Let Perish Computerization for the Small Editorial Office: Does it Make Economic Sense for Your Operation? P.M.: Workshops 1. Marketing The Ailing journal: Diagnosis and Treatment 2. What does a professional editor do? 3. Retreading an editor: how to freelance A 4. Design critique of journals 5. The professional biology editor 6. Economics of editing/publishing and publication- quality 7 May, A.M.: Ethics of Scientific Publication For additional information, please contact: of” 1:,0 f, E I'. a» g. :. ,6 a» ’Ttm'é ‘ 1 111’ Philip L. Altman, Executive Secretary Council ofBiology Editors, Inc. 9650 Rockville Pike Bethesda, Maryland 208M Telephone: (301) 530-7036 tomy if the lowest rate in Maine had been the national average and if the highest rate had been the national av- erage. Under the low rate there would have been 1,900 deaths from prostatec- tomy; under the high rate there would have been 6,800 deaths. The difierence of almost 5,000 deaths suggests the pos- sible cost of geographic variations in health care. Under the high Maine rate one man in 100 who had reached the age of 65 would die in the hospital following a prostatectomy. To the person who must choose whether to have a prostatectomy the most important consideration is the, bal- ance of risks and benefits from the ‘pro- cedure. In evaluating the balance claims information from Medicare is useful; it enables us to follow for long periods pa- tients who have had a prostatectomy. Their chances of surviving can then be compared with the chances of those who have not had the operation. We exam- ined records of about 1,500 men who had had a prostatectomy in Maine in 1976 and 1977. On the basis of this sam- ple we estimate that a 67-year-old man who has a prostatectomy for a condition other than cancer of the prostate or bladder has about a 9 percent chance of dying in the year after the procedure. For men in this age group who have not had the procedure the risk of dying with- in a year is about 3.5 percent. The savings in lives and money that would result from making rates of medi- cal care correspond to the health needs and preferences of informed consumers might be considerable. How could this be achieved? If physicians in an area where rates are high are made aware of that fact, rates may fall, eliminat- ing some unnecessary surgery. Reliable studies of the effects of various kinds of treatment might lead to a consensus on their value and could provide consum- ers with more information on which to base their decisions. If regulatory agen- cies consider the level of health care in the community, some of the tendencies toward excess may be restrained. If pre- miums for medical insurance reflect the amount of treatment per capita in a giv- en area, subsidization could be reduced. The most important factor, however, may be the emergence of an informed consumer of medical services. When pa— tients are aware that different forms of treatment are available, they can de— mand information on risks and benefits and make their own preferences known. If they know that rates of surgery are high at the local hospital, they may choose another. If they realize that a particular operation is a controversial one, they may seek the opinion of a second and even a third physician. In- formed patient's’ma'y therefore be the most important factor in making rates of treatment reflect health needs and eliminating unnecessary medicine. ...
View Full Document

This note was uploaded on 05/19/2010 for the course PBPL 26 taught by Professor Welch during the Spring '10 term at Dartmouth.

Page1 / 11

JW Scientific American - AN ARTICLE FROM SCIENTIFIC...

This preview shows document pages 1 - 11. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online