Designing a text messaging intervention to improve physical activity behavior amonglow-incomelatino patients with diabetes: A discrete-choice experiment, Los Angeles, 2014-2015

Abstract

Introduction Automated text messaging can deliverself-management education to activate self-care behaviors among people with diabetes. We demonstrated how adiscrete-choiceexperiment was used to determine the features of a text-messaging intervention that are important to urban,low-incomeLatino patients with diabetes and that could support improvement in their physical activity behavior. Methods In adiscrete-choiceexperiment from December 2014 through August 2015 we conducted a survey to elicit information on patient preferences for 5 features of atext-messagingintervention. We described 2 hypothetical interventions and in 7 pairwise comparisons asked respondents to indicate which they preferred. Respondents (n = 125) were recruited in person from a diabetes management program of a safety-net ambulatory care clinic in Los Angeles; clinicians referred patients to the research assistant after routine clinic visits. Data were analyzed by using conditional logistic regression. Results We found 2 intervention features that were considered by the survey respondents to be important: 1) the frequency of text messaging and 2) physical activity behavior-change education (the former being more important than the latter). Physical activity goal setting, feedback on physical activity performance, and social support were not significantly important. Conclusion Adiscrete-choiceexperiment is a feasible way to elicit information on patient preferences for a text-messaging intervention designed to support behavior change. However, discrepancies may exist between patients' stated preferences and their actual behavior. Future research should validate and expand our findings.