Abstract P138: Defining and Predicting Adherence to an Online Lifestyle Program: 12-Month Results from the PHIT Study

Abstract

BACKGROUND: Traditionally, behavioral lifestyle programs have been delivered in person with adherence defined as number of sessions attended. With growing use of online interventions, an understanding of how to assess and enhance adherence to such programs is needed. In this analysis, we examine: 1) measures of adherence to an online lifestyle program; 2) how measures differ from month 6 to month 12 of an 18-month program; and 3) predictors of continued active status. METHODS: The Pittsburgh-Hershey Internet-based Translation (PHIT) study is an online adaptation of the Diabetes Prevention Program delivered in coordination with primary care. We recruited 124 patients; 61 were randomized to an 18-month online intervention, including lessons; weight, diet, and physical activity (PA) self-monitoring; and feedback from a lifestyle coach. A participant is inactive when he/she has not logged in for ≥28 days or responded to phone calls. We abstracted program use data as potential adherence measures: active/inactive status, log-in in past 7 days, self-monitoring, and number of lessons completed. Potential predictors of active status included sociodemographic variables, computer skill, and baseline BMI. Differences between 6- and 12-month measures were calculated with t-test or Wilcoxan rank test. Predictors of active status were determined with logistic regression. RESULTS: Of 61 participants in the intervention group, 42 (69%) were women, 54 (89%) were white, and 30 (49%) had college degree or higher. At 6 months, 29 (45%) of participants were active in the program. Eighteen (30%) had log-in, 13 (26%) posted a weight , and 11 (18%) self-monitored diet and PA in past 7 days. All measures of adherence declined at 12 months, although only active status (with 17 (28%) active) was significantly lower (p=0.03) than 6 months. At 12 months, 9 (15%) had a log-in, 7 (12%) posted a weight, 7 (12%) self-monitored diet and PA in past 7 days. At 6 months, participants had completed an average of 7.5 (SD 6.2) of 16 expected lessons; at 12 months, participants had completed an average of 9.2 (SD 8.3) lessons of 24 expected lessons. None of the variables considered predicted active status at 6 months. At 12 months, men were four times as likely as women to be active (OR=3.8; p=0.03). Participants with higher baseline BMI (OR=0.88; p=0.04) and difficulty paying for basics (OR=0.12; p=0.04) were less likely to be active at 12 months. Age, education and computer skill did not predict active status at either time point. CONCLUSION: Active status declined significantly between 6 and 12 months and use of online self-monitoring tools was low at both time points. Females, heavier persons, and those with financial constraints may need additional support to stay engaged. Future research should address how measures of adherence relate to program outcomes such as weight loss and how to improve adherence to online interventions over time.