An empirical evaluation of the impact scenario of pooling bodies of evidence from randomized controlled trials and cohort studies in medical research

Abstract

Background Randomized controlled trials (RCTs) and cohort studies are the most common study design types used to assess treatment effects of medical interventions. We aimed to hypothetically pool bodies of evidence (BoE) from RCTs with matched BoE from cohort studies included in the same systematic review. Methods BoE derived from systematic reviews of RCTs and cohort studies published in the 13 medical journals with the highest impact factor were considered. We re-analyzed effect estimates of the included systematic reviews by pooling BoE from RCTs with BoE from cohort studies using random and common effects models. We evaluated statistical heterogeneity, 95% prediction intervals, weight of BoE from RCTs to the pooled estimate, and whether integration of BoE from cohort studies modified the conclusion from BoE of RCTs. Results Overall, 118 BoE-pairs based on 653 RCTs and 804 cohort studies were pooled. By pooling BoE from RCTs and cohort studies with a random effects model, for 61 (51.7%) out of 118 BoE-pairs, the 95% confidence interval (CI) excludes no effect. By pooling BoE from RCTs and cohort studies, the median I2 was 48%, and the median contributed percentage weight of RCTs to the pooled estimates was 40%. The direction of effect between BoE from RCTs and pooled effect estimates was mainly concordant (79.7%). The integration of BoE from cohort studies modified the conclusion (by examining the 95% CI) from BoE of RCTs in 32 (27%) of the 118 BoE-pairs, but the direction of effect was mainly concordant (88%). Conclusions Our findings provide insights for the potential impact of pooling both BoE in systematic reviews. In medical research, it is often important to rely on both evidence of RCTs and cohort studies to get a whole picture of an investigated intervention-disease association. A decision for or against pooling different study designs should also always take into account, for example, PI/ECO similarity, risk of bias, coherence of effect estimates, and also the trustworthiness of the evidence. Overall, there is a need for more research on the influence of those issues on potential pooling.

Impact of different surgical and non-surgical interventions on health-related quality of life after thoracolumbar burst fractures without neurological deficit: protocol for a comprehensive systematic review with network meta-analysis

“…For the sake of a comprehensive approach, we plan to include all existing evidence from controlled trials. However, as recommended by Bröckelmann et al , 43 the final decision on whether to include a non-RCT in the NMA will take into account similarity regarding patient population, interventions, outcome, risk of bias, coherence of effect estimates as well as the trustworthiness of the evidence and the result of the authors’ discussion. 26 …”

Evaluating the impact of including non-randomised studies of interventions in meta-analysis of randomised controlled trials: a protocol for a meta-epidemiological study

“…In a study of 118 BoE paired based on RCTs and cohort studies reported in 13 high impact factor medical journals, Bröckelmann et al 25 investigated whether inclusion of evidence from cohort studies modified the conclusion when only evidence from RCTs was considered. However, they used random and common effects models to quantitatively synthesise RCTs and cohort studies; these methods do not attempt to adjust for potential bias from cohort studies 26.…”