Site Variability in a Multisite Geriatric Depression Trial

We studied differences in outcome and characteristics among 29 clinical sites of a multisite, double-blind antidepressant trial for geriatric depression. Six hundred and seventy-one outpatients aged 60 years or older (mean±SD=67.7±5.7) met DSM-III-R criteria for unipolar major depression, had baseline 17-item Hamilton Depression Rating Scale (HAMD17) scores ≥16 and were randomized to fluoxetine (20 mg daily) or placebo. Effect sizes (ESs, expressed as mean differences between effects divided by the pooled standard deviation of the differences) were calculated for each site using selected outcome measures. ES ranged from 1.84 (favoring fluoxetine) to −0.91 (favoring placebo) for incidence of remitters (endpoint HAMD17 total score of ≤8). A large, positive ES favoring fluoxetine for remission rates (ES≥0.65) was found at only six sites, moderate ES (0.35–0.64) at eight and small ES (0–0.34) at seven; ES favored placebo (<0) at eight of 29 sites. Private clinics showed an overall HAMD17 ES for change scores more than twice that of university sites. These results suggest that individual practitioners may have vastly different clinical experiences in large, multisite trials for geriatric depression. Interrater reliability, subject selection, recruitment, inadequate or fixed dosing, few patients per site, brief study duration, heterogeneity of geriatric depression, financial incentive and characteristics of individual sites may contribute to response variability.