Optimal designs of clinical studies in the context of their inclusion in systematic reviews

In this article I focus on the designs of clinical studies that have a high likelihood of producing reliable findings and those that do not, but that are still being published in large numbers in scientific journals. The former category includes randomized trials, controlled before-after studies and interrupted time series, while the latter category comprises non-randomized trials, uncontrolled before-after studies, case-control/cohort studies and other observational studies. This distinction is important, particularly when studies of inferior design are used in systematic reviews that inform clinical practice guidelines, thereby potentially exerting adverse effects on clinical practice. I also highlight the implications of failing to enrol adequate numbers of patients in clinical trials.