Within-person associations of psychological factors with healthy or unhealthy diet in EMA studies in general adult populations: a systematic review and meta-analysis

Background: (Un)healthy eating behaviours may vary by moment and psychological context. Ecological momentary assessment (EMA) is well-suited to study within-person, momentary, and context-dependent behaviours and correlates, as it uses repeated sampling of real-time experiences. Insights in such within-person relations could inform ecological momentary interventions. A systematic review and meta-analysis of these insights in a healthy population is, however, currently lacking. The aim of this systematic review and meta-analysis was to synthesise within-person associations between psychological factors and eating behaviour in healthy adults, and summarise methodological aspects of these studies.

Methods: Observational and experimental studies on psychological predictors of (un)healthy eating behaviours in adult healthy populations were included, based on a systematic search of Ovid MEDLINE, Embase, PsycINFO and Web of Science. Meta-analysis of ≥ 5 studies per predictor/outcome was performed in metafor R-package.

Results: During title and abstract screening, n = 2046 articles were reviewed (n = 119 from overarching review, n = 1927 from updated search), n = 208 were screened in full-text screening. Forty-nine (review) and fifteen studies (meta-analysis) were included. Most studies were observational, and used fixed time-based sampling, and most assessed food eaten since the last prompt/timeframe. Various measures reported on study adherence (between 79 and 90%). Only around half reported on reliability and/or validity of the predictors or outcomes, most often referring to an adaptation of non-EMA scales, with little information of their exact adaptation. Due to high diversity in conceptualization and reporting of effect sizes, not all could be meta-analyzed. Meta-analyses showed negative psychological triggers (NPT, e.g., negative affect, daily stress, and ruminations) predicted lower levels of healthy eating behaviour (β=-0.007, p=.028, k = 8, n total = 1,489), higher levels of unhealthy eating behaviour (β = 0.054, p=.022, k = 6, n total = 1,578), but had no significant association with unhealthy behaviour event occurrence (logOR=0.469, p=.095, k = 5, n total = 934). All meta-analyses were characterized by small effect sizes and important heterogeneity.

Conclusions: Associations at the within level of NPT with (un)healthy eating highlight their potential utility for inclusion in health promoting ecological momentary interventions. Nevertheless, the methodological and conceptual heterogeneity of the studies highlight a need for harmonized practices, including broader adoption of reporting guidelines, to facilitate cumulative knowledge in the field.