Opportunities and challenges in using studies without a control group in comparative effectiveness reviews

When examining the evidence on therapeutic interventions to answer a comparative effectiveness research question, one should consider all studies that are informative on the interventions' causal effects. "Single group studies" evaluate outcomes longitudinally in cohorts of subjects who are managed with a single treatment strategy. Because these studies are "missing" a direct, concurrent comparison group, they are typically deemed non-informative on comparative effectiveness. However, in principle, single group studies can provide information on causal treatment effects by extrapolating expected outcomes in the "missing" untreated arm. Single group studies rely on before-after, implicit, or historical comparisons as a proxy for an ideal comparison group. The validity of these comparisons must be carefully examined on a case-by-case basis. While in many cases, researchers will disagree on whether such extrapolations are reasonable; circumstances exist where such studies are generally acceptable as a source of evidence. This article provides an overview of issues related to the interpretation of single group studies with a focus on the assumptions required to support their consideration in comparative effectiveness reviews. We discuss the various settings in which single group studies are employed, common research designs that systematic reviewers need to interpret, and challenges associated with using these designs to inform comparative effectiveness questions.