Consumers’ Preferences for Purchasing mHealth Apps: Discrete Choice Experiment

Abstract

Abstract Background There is growing interest in mobile health apps; however, not all of them have been successful. The most common issue has been users’ nonadoption or abandonment of health apps because the app designs do not meet their preferences. Therefore, to facilitate design-preference fit, understanding consumers’ preferences for health apps is necessary, which can be accomplished by using a discrete choice experiment. Objective This study aims to examine consumer preferences for health apps and how these preferences differ across individuals with different sociodemographic characteristics and health app usage and purchase experiences. Methods A cross-sectional discrete choice experiment questionnaire survey was conducted with 593 adults living in Hong Kong. A total of 7 health app attributes that might affect consumers’ preferences for health apps were examined, including usefulness, ease of use, security and privacy, health care professionals’ attitudes, smartphone storage consumption, mobile data consumption, and cost. Mixed-effect logit regressions were used to examine how these attributes affected consumer preferences for health apps. Fixed effects (coefficient β) of the attributes and random effects of individual differences were modeled. Subgroup analyses of consumer preferences by sex, age, household income, education level, and health app usage and purchase experiences were conducted. Results Cost was the attribute that had the greatest effect on consumers’ choice of health apps (compared to HK $10 [US $1.27]—HK $50 [US $6.37]: β=−1.064; P<.001; HK $100 [US $12.75]: β=−2.053; P<.001), followed by security and privacy (compared to no security insurance—some security policies: β=.782; P<.001; complete security system: β=1.164; P<.001) and usefulness (compared to slightly useful—moderately useful: β=.234; P<.001; very useful: β=.979; P=.007), mobile data consumption (compared to data-consuming—a bit data-consuming: β=.647; P<.001; data-saving: β=.815; P<.001), smartphone storage consumption (compared to >100 MB—around 38 MB: β=.334; P<.001; <10 MB: β=.511; P<.001), and attitudes of health care professionals (compared to neutral—moderately supportive: β=.301; P<.001; very supportive: β=.324; P<.001). In terms of ease of use, consumers preferred health apps that were moderately easy to use (compared to not easy to use—moderately easy to use: β=.761; P<.001; very easy to use: β=.690; P<.001). Our results also showed that consumers with different sociodemographic characteristics and different usage and purchase experiences with health apps differed in their preferences for health apps. Conclusions It is recommended that future health apps keep their mobile data and phone storage consumption low, include a complete security system to protect personal health information, provide useful content and features, adopt user-friendly interfaces, and involve health care professionals. In addition, health app developers should identify the characteristics of their intended users and design and develop health apps to fit the preferences of the intended users.