Generalizing RCT Efficacy Evidence: Application to NIDA Clinical Trials Network

点击翻译按钮获取中文摘要

Randomized controlled trials in which individuals are randomly assigned to receive an experimental or a control treatment are generally considered the gold standard for establishing the efficacy of new treatments. Evidence-based practice guidelines and policy decisions often rely on the results of such trials. Nevertheless, participants in randomized controlled trials are often different from individuals who would eventually be the recipients of new treatments ("target population") when these treatments are disseminated in usual practice settings. This is a cause for concern because the results of randomized controlled trials may not generalize to these populations. This is partly due to the fact that individuals who are recruited into randomized controlled trials are often different from the target population in a number of ways. Strict eligibility criteria of randomized trials exclude many potential participants with severe health or psychiatric conditions or individuals who may not be able to comply with treatment or to show up for follow-up appointments. This study uses data from 23 randomized controlled trials of substance use disorder treatments currently available in the NIDA Clinical Trials Network to assess how representative the randomized controlled trial samples are of the target population. Data on the target population are obtained from other publicly available data sources on individuals entering substance use disorder treatments in the US. The study also uses new statistical methods to measure the degree of deviation from the composition of the target population and also uses methods to adjust the samples of the randomized trials to better match the target populations. As a result, subgroups that are under-represented in the randomized trial sample will be given a higher weight in the analyses. After such weighting of data, the randomized controlled trial samples will be more similar to the target population with regard to socio-demographic and clinical characteristics and their results will more generalizable to the target population. We also plan to combine randomized trials of similar types of treatments to estimate the efficacy of interventions with more accuracy and to assess whether variations in the settings where trials were conducted had an impact on the outcomes of these studies. Overall, these analyses will provide more generalizable data from the available randomized trials in the NIDA Clinical Trials Network and will provide a framework for future randomized controlled trials to use at the design stage to optimize generalizability of their results.