Country- and app-level factors affecting the adoption and evaluation of COVID-19 mobile apps

Abstract

Abstract Background Countries across the globe have released many COVID-19 mobile apps. However, there is a lack of systematic empirical investigation into the factors affecting the adoption and evaluation of COVID-related apps. This study explores what factors at the country level and the app levels would influence the adoption and evaluation of COVID-19 apps. Methods We collected data on 267 COVID-19 apps in App Store and Google Play. The number of installs, ratings, reviews and rating scores were used as indicators of adoption and evaluation. Country-level predictors include the number of infected cases and the political system (i.e., democratic vs. non-democratic). App-level predictors include developer (i.e., government vs. non-government) and functions. Four app functions were coded for analysis: providing health information, contact tracing, home monitoring, and consultation. Negative binomial regression and OLS (Ordinary Least Square) regression were used to analyze the data. Results Our analyses show that apps developed by countries with more infected cases (B = 0.079, CI (Confidence Interval) = 0.000, 0.158; P = .049) and by non-governmental institutions (B=-0.369, CI=-0.653, -0.083; P = .01) received more positive rating scores. Apps with home monitoring function received lower rating scores (B=-0.550, CI=-0.971, -0.129; P = .01). Regarding adoption, apps developed by governments were more likely to be installed (IRR (Incident Rate Ratio) = 8.156, CI = 3.389, 19.626; P < .001), to be rated (IRR = 26.036, CI = 7.331, 92.468; P < .001), and to receive user comments (IRR = 12.080, CI = 3.954, 37.568; p < .001). Apps with functions of contact tracing or consultation were more likely to be installed (IRR = 4.533, CI = 2.072, 9.918; p < .001; IRR = 4.885, CI = 1.970, 12.111; p < .001), to be rated (IRR = 11.634, CI = 3.486, 38.827; p < .001; IRR = 17.194, CI = 5.309, 55.680; p < .001), and to receive user comments (IRR = 5.688, CI = 2.052, 5.770; p < .001; IRR = 16.718, CI = 5.363, 52.113; p < .001). Apps with home monitoring functions were less likely to be rated (IRR = 0.206, CI = 0.047, 0.896; P = .04) but more likely to receive user comments (IRR = 3.874, CI = 1.044, 14.349; P = .04). Further analysis shows that apps developed in democratic countries (OR (Odd Ratio) = 3.650, CI = 1.238, 10.758; P = .02) or by governments (OR = 7.987, CI = 4.106, 15.534, P < .001) were more likely to include the function of contact tracing. Conclusion This study systematically investigates factors affecting the adoption and evaluation of COVID-19 apps. Evidence shows that government-developed apps and the inclusion of contact tracing and consultation app functions strongly predict app adoption.