Engagement in Feedback EmailIn the 3-month survey, we assessed participants’ self-reportedfrequency of reading their weekly feedback email, with thequestion “How frequently did you read your weekly ProgressReports (sent via email), on average?” and 5 response options(Several times per week,One time per week,Less than 1 timeper week,Less than 1 time per month, orNever).Sociodemographic and Clinical CharacteristicsAt baseline, we collected data on participant demographics,socioeconomic status, and type of smartphone. To assess pastMyFitnessPal use, we asked the Pew Research Center’squestion: “What kind of health apps do you currently have onyour phone?” [41]; if the “Diet, Food, Calorie Counter” responseoption was selected, then the open-ended question “What arethe names of the diet, food, or calorie-counting apps that youused on your phone?” was asked.We also assessed whether participants had ever been told by adoctor or other health professional that they had prediabetes orhypertension. Self-monitoring of weight and self-monitoringof diet in the month before baseline were each measured witha 7-point scale ranging fromseveral times per daytonever[42].Statistical AnalysisSample size was calculated based on power to detect a 3.5 kgdifference in weight change at 3 months between the Sequentialarm and the App-Only arm (our primary comparison) using3-month results from previous remotely delivered weight lossinterventions for our Sequential arm [43] (results in kilogramswere provided upon request by the author) and our App-Onlyarm [44]. Our power analysis (G*Power 3.1.9.2.) determinedthat 31 participants per group were needed to achieve 80%power for a 2-sided test with an alpha level of .05. To accountfor attrition of 10% and to obtain equal-size groups, we aimedto recruit 105 participants (35 per group). In exploratoryanalyses, we compared weight change between the Sequentialarm and the Simultaneous arm, although we were not adequatelypowered to detect a significant effect.For the baseline characteristics, we computed descriptivestatistics stratified by treatment arm. To determine whetherbaseline characteristics differed by retention status, we used thePearson chi-square test for categorical variables, analysis ofvariance for continuous variables, and Fisher exact tests withsmall cell counts. All analyses were 2-tailed. Participants whobecame ineligible during the study period up to 3 months wereexcluded from the analyses. Investigators remained blinded tooutcomes until the completion of the 6-month trial.We used intent-to-treat analyses to test our primary aim usinglinear mixed modeling with an unstructured covariance matrixand restricted maximum likelihood estimates to examine changesin weight over time by treatment arm. We did not control forany additional variables, and we assumed missing at randomand used SAS 9.4 PROC MIXED (SAS Institute) for theseanalyses. For 6-month weight values sent via photo, wesubtracted 0.172 kg (0.4 lb) to account for participants holdinga device on the scale to take the photo. To account for the6-month self-reported weight data without photos, we used aregression model to adjust for age, gender, and race/ethnicity[45]. Participants who sent a photo of their 6-month weight didnot differ on any measured sociodemographic characteristicsfrom those who did not send a photo (data not shown). We usedchi-square tests to assess proportion of participants achieving
Upload your study docs or become a
Course Hero member to access this document
Upload your study docs or become a
Course Hero member to access this document
End of preview. Want to read all 19 pages?
Upload your study docs or become a
Course Hero member to access this document
