JMIR Publications

ABSTRACT

Background:

Smartphone-based ecological momentary assessment (EMA) enables real-time capture of alcohol use in naturalistic settings, overcoming limitations of retrospective methods in terms of ecological validity and recall bias. However, EMA data quality depends on participant compliance, and non-compliance may be highest during drinking episodes themselves. Despite rapid growth in this field, a quantitative synthesis of compliance rates and their protocol-level predictors in smartphone-based alcohol EMA is lacking.

Objective:

This review aimed to (1) characterise EMA design and protocol features used in smartphone-based alcohol studies in adults; (2) evaluate reporting quality and transparency; (3) examine the extent to which participant experience outcomes were assessed; and (4) estimate pooled compliance and test whether prompting frequency and assessment duration predict it.

Methods:

Six databases (PubMed, Embase, Web of Science, PsycINFO, CINAHL, Scopus) were searched for studies published January 2015–March 2025 with an updated search in July 2026. Eligible studies involved adults using smartphone-based EMA to measure alcohol consumption and reported compliance data. Study quality was assessed using an adapted CREMAS checklist. Random-effects meta-analysis (REML) pooled compliance across studies with standardised data; meta-regression tested prompting frequency and assessment duration as moderators; publication bias was assessed via trim-and-fill.

Results:

Thirty-eight studies (6,594 participants) met inclusion criteria. EMA protocols predominantly used interval-contingent designs (1–9 prompts/day, 7–156 days). Reporting quality was moderate (mean CREMAS: 73.0%), with response latency and number of data collection waves the most poorly reported items. Only five studies formally assessed usability, acceptability, or burden. Pooled compliance across 21 studies was 81% (95% CI: 75%–87%; I²=94.36%). Prompting frequency significantly predicted lower compliance (b=−0.039, P=.016); assessment duration did not (P=.795). No evidence of publication bias was found.

Conclusions:

Smartphone-based alcohol EMA achieves a pooled compliance rate of 81%, though heterogeneity was substantial. Prompt burden, more than study duration, shapes engagement. Standardized compliance reporting, routine participant experience assessment, and broader population representation are needed to advance the field.