## Evaluating non-randomised intervention studies JJ Deeks, J Dinnes, R D'Amico, Al Sowden, C Sakarovitch, F Song, M Petticrew, DG Altman This report evaluates methods and evidence for assessing bias in non-randomised intervention studies. It discusses the sources of bias, case-mix adjustment methods, and the use of quality assessment tools in systematic reviews of non-randomised studies. The report also presents new empirical investigations into the bias associated with non-random allocation and the ability of case-mix adjustment methods to correct for selection bias. The report finds that non-randomised studies may sometimes differ from randomised studies of the same intervention, but often do not. Non-randomised studies may still give seriously misleading results when treated and control groups appear similar in key prognostic factors. Standard methods of case-mix adjustment do not guarantee removal of bias. Residual confounding may be high even when good prognostic data are available, and in some situations adjusted results may appear more biased than unadjusted results. The report concludes that non-randomised studies should only be undertaken when RCTs are infeasible or unethical. It recommends further research into the resampling methodology, the development of quality assessment tools for non-randomised studies, and the evaluation of the role of the propensity score.## Evaluating non-randomised intervention studies JJ Deeks, J Dinnes, R D'Amico, Al Sowden, C Sakarovitch, F Song, M Petticrew, DG Altman This report evaluates methods and evidence for assessing bias in non-randomised intervention studies. It discusses the sources of bias, case-mix adjustment methods, and the use of quality assessment tools in systematic reviews of non-randomised studies. The report also presents new empirical investigations into the bias associated with non-random allocation and the ability of case-mix adjustment methods to correct for selection bias. The report finds that non-randomised studies may sometimes differ from randomised studies of the same intervention, but often do not. Non-randomised studies may still give seriously misleading results when treated and control groups appear similar in key prognostic factors. Standard methods of case-mix adjustment do not guarantee removal of bias. Residual confounding may be high even when good prognostic data are available, and in some situations adjusted results may appear more biased than unadjusted results. The report concludes that non-randomised studies should only be undertaken when RCTs are infeasible or unethical. It recommends further research into the resampling methodology, the development of quality assessment tools for non-randomised studies, and the evaluation of the role of the propensity score.