Variation in placebo effect sizes in clinical trials of oral interventions for management of the behavioral and psychological symptoms of dementia (BPSD): a systematic review and meta-analysis

Increasing placebo effect sizes over time have been reported in randomized controlled trials (RCTs) for outcomes related to psychiatric symptoms. The Neuropsychiatric Inventory (NPI) is a key outcome measure in clinical trials of the behavioral and psychological symptoms of dementia (BPSD). Accurate placebo effect size estimates for NPI are needed for sample size calculations in order to adequately power future studies. This study investigated variation in placebo effect sizes for NPI in RCTs testing oral interventions for BPSD. A search of PubMed was conducted in April 2016 for two-armed, double-blinded, placebo-controlled RCTs testing any oral intervention for management of BPSD using the NPI. Meta-analysis was conducted of baseline versus end of treatment placebo group data of included studies. Twenty-five RCTs published from 2000 to 2015 were included. Substantial variation in placebo effect sizes was detected. Participants in placebo groups showed greater improvements in recent studies compared with earlier studies. Subgroup analyses indicated robustness of this finding. From 2000 to 2008 there was no significant change in total NPI scores within placebo groups (12 studies; 1,056 participants), whereas from 2009 to 2015 there was significant improvement (mean difference: -2.68; 95% confidence interval: -4.38, -0.99; z = 3.10; p = 0.002, random effects; I2 = 76%; 13 studies; 1,170 participants). This increase in NPI effect sizes in placebo groups has important implications for power calculations for future clinical trials of BPSD. Effect size estimates for NPI need to be based on more recent studies