Abstract
Background. Depression is a significant public health problem in older adults due to its harmful consequences and high prevalence. The fifteen-item Geriatric Depression Scale (GDS-15) is one of the commonly recommended measures for depressive symptoms in this population; however, its effectiveness is usually measured using statistical significance measures that don鈥檛 reflect clinical importance. To address this limitation, the minimal clinically important difference (MCID) on a scale is intended to represent the smallest change that would likely be perceived to be worthwhile by patients.
Objectives. (1) To evaluate the MCID using the Standard Error of Measurement (SEM) for each included primary study and synthesize the overall estimates. (2) To assess the heterogeneity of the SEM-based MCID across studies and examine whether MCIDs differ based on age, sex or gender, participant recruiting setting, major depression classification status, and depressive symptom severity.
Methods. We used a subset of studies and participants from previously conducted Individual Patient Data Meta-analyses (IPDMA) to carry out our analysis. We estimated the MCID for each of the included studies and generated pooled estimates for one standard error of measurement (SEM), 2 SEM and one standard deviation (SD) using restricted maximum-likelihood random-effects meta-analysis, as well as 95% confidence intervals and prediction intervals via bootstrap. We then assessed the differences between subgroups across age, sex, recruitment setting, and depressive symptom severity using restricted maximum-likelihood random-effects meta-analysis to generate pooled estimates for individual level variables and meta-regression for study-level variables.
Results. The final sample included 21 studies and 6,212 participants (40.5% male, 28.5% over 80 years old, 10.8% high symptom severity). The project is still ongoing, and preliminary results are under review.