Abstract
Background
Identifying prognostic factors of ADL decline in community-dwelling older adults with dementia may clarify individual variability and support person-centered prognostication. This systematic review and meta-analysis aimed to review independent prognostic factors for ADL decline in this population.Methods
We searched PubMed, EMBASE, Cochrane CENTRAL, CINAHL, and PsycINFO from inception to October 2025. Eligible longitudinal studies enrolled community-dwelling adults ≥65 years with dementia. Two reviewers independently screened studies, extracted data, and assessed risk of bias. Risk of bias was assessed using the Quality in Prognostic Studies tool. Meta-analyses and GRADE were applied to predictors reported in ≥3 studies with sufficient heterogeneity.Results
We screened 6,915 records; reviewed 100 full texts and included 19 reports from 13 studies (9,490 participants). Risk of bias was frequently high for selection (16/19) and attrition (14/19), often due to poor reporting. A wide range of prognostic factors were reported but only age, global cognitive function by MMSE, sex, education, and living alone were examined consistently enough to allow meta-analysis. Age and education showed no significant effect (SMD: 0.0, 95% CI 0.0-0.1 for age, 0.00, 95% CI: -0.02 to 0.02 for years of education; moderate certainty for both). Better baseline cognition predicted slower ADL decline (SMD -0.05, 95% CI -0.06 to -0.04; moderate certainty). Confidence in results for sex and living alone was very low and inconsistent.Conclusion
In community-dwelling older adults with dementia, baseline cognition-but not age or education-predicts ADL decline. The limited and heterogeneous evidence highlights the need for more high-quality longitudinal research.Protocol registration
PROSPERO: CRD42024502839.