ABSTRACT
Applicability of smartphone-based digital health in diabetes management still face challenges due tolow user retention or engagement. Thus, this systematic and meta-analysis aimed to estimate the dropout rate from the clinical trials. Search of literature was performedon 4 September 2023 through various databases (PubMed, Scilit, Scopus, Embase, and Web of Science). Those reporting clinical trials of smartphone apps for diabetic controls (either type 1 or type 2 diabetes mellitus) were screened and selected in accordance with PRISMA guideline. Of 5,429 identified records, as many as 36 studies were found eligible with a total of 3,327 patients in the intervention group. The overall dropout rate was 29.6 % (95 %CI: 25 %-34.3 %) with high heterogeneity (p-Het < 0.001;I2 = 84.84 %). Sample size, intervention duration, patients' age and gender, and cultural adaptation on the app appeared to be non-significant moderators (p > 0.05). In sub-group levels, notably high dropout rates were observed in studies performing cultural adaptation (34.6 %) and conducted in high-income countries (31.9 %). Given the high dropout rate, the engagement level toward diabetic management apps in real-world setting is expected to be low. High heterogeneity in this study, however, requires careful interpretation of the foregoing results. PROSPERO: CRD42023460365 (14 September 2023).
