Assessing the methodological quality of nonrandomized intervention studies

In many areas of health care, randomized controlled trials (the best evidence regarding the effectiveness of health care interventions) are lacking and decision-makers have to rely on evidence from nonrandomized studies (NRS). We conducted a Medline search to identify English-language articles describing instruments for assessing the quality of NRS of health care interventions. These instruments varied greatly in scope, in the number and types of items and in developmental rigor. Items commonly included were those related to specification of study questions, allocation method, comparability of groups, and blinding of outcome assessment. We do not support the development of a generic scale to evaluate the methodological quality of nonrandomized intervention studies. Instead, further study should be directed to investigate the degree to which, and the circumstances under which, different methodological characteristics are associated with bias. This information will assist researchers in identifying a priori which methodological characteristics need careful evaluation in particular studies.