BACKGROUND: Network meta-analyses (NMAs) are an increasingly important tool in comparative psychopharmacology. As do traditional metaanalyses, these studies assume that there are no unaccounted effect modifiers that could confound comparisons across time. Research methodologies, however, have changed markedly in recent decades. This is clearly seen in rising placebo response rates in modern antipsychotic and antidepressant trials (placebo inflation). The current study aimed to evaluate if NMAs display evidence of a confounding bias that varies with time.
METHODS: Efficacy rankings from 2 landmark antipsychotic meta-analyses were cross-referenced and regressed against FDA approval dates. Two prominent NMAs of antidepressants were analyzed for comparison because these 2 drug classes display distinct patterns of rising placebo response rates.
RESULTS: Newer antipsychotic medications consistently rank worse than older agents (rho = 0.49, 0.74; P = .066, .0016). This trend remains robust when excluding outliers (ie, first-generation antipsychotics and non-FDA– approved medications) and analyzing effect sizes rather than efficacy rankings (r = 0.83, 0.77; P = .0028, .0051). Antidepressant meta-analyses do not display a similar temporal pattern.
CONCLUSIONS: Rankings of antipsychotics, but not antidepressants, show evidence of a confounding temporal bias. Poorly compensated placebo inflation is one potential explanation for this finding.