Background: Observational studies convey valuable information about the effectiveness of interventions in real-life clinical practice. Recent research has found no evidence of systematic differences in the estimated effects between observational and randomized controlled trials (RCTs). However, there is little available guidance on how to combine real-world data with evidence from RCTs, especially when there are multiple competing available treatments for the same condition.
Objectives: To assess existing methodology and develop new methods for jointly synthesizing evidence on relative treatment effects provided by RCTs as well as evidence from observational studies that report individual patient data (IPD) in a network meta-analysis (NMA).
Methods: We adjusted the relative treatment effects from the observational studies utilizing patient level covariates. We combined the adjusted estimates with the evidence provided by RCTs in a joint network meta-regression, accounting for trial-level covariates, and exploring a variety of alternative methods. We quantified the impact of different levels of confidence for the adjusted estimates on the NMA relative effects.
Results: We applied our methods in a published network of 167 RCTs, comparing 15 antipsychotics and placebo for schizophrenia. We extended our evidence base by including IPD from a cohort study involving five interventions and more than 10,000 patients. The adjusted evidence was found to be in agreement with the evidence from RCTs. Statistical inconsistency in the network was unaffected by the inclusion of observational evidence. A range of sensitivity analyses shows our results to be robust to the alternative methods used.
Conclusions: Including real-world evidence from observational studies can corroborate findings of a NMA based on RCTs alone, increase precision and enhance the decision-making process.