Introduction: Reviewing procedures inspired by Cochrane improve evaluation of treatment procedures by accessing all relevant RCTs, specifying search procedures and data used, with subjective judgements minimized by crosschecking between reviewers. The weaknesses of conventional approaches to reviewing, which led to the Cochrane movement also apply more widely when reviews attempt to evaluate hypotheses, for instance about aetiopathological bases of disease. This is closely related to treatment evaluation because many treatment procedures derive from inadequate hypotheses which cannot survive critical scrutiny in a systematic review.

Objective: This paper discusses possibilities of systematic reviewing for hypothesis testing in medicine.

Methods: The interrelation between hypothesis testing and RCT evaluation is illustrated from reviewing work in progress on efficacy of classical antipsychotic drugs.

Results: There is much scope for increased systematization of procedures for reviews which attempt to test hypotheses about disease processes. Increased systematization of reviews is also needed more widely in the basic sciences which are the foundation of scientific medicine.

Discussion: Testing disease hypotheses is more complex than evaluating RCTs, because relevant information is more diverse. Nevertheless, if a definite hypothesis is identifiable, explicit predictions and ways of falsifying it can be deduced. Specific questions can than be posed, which recognize practical limitations on available information. Then methods of systematic reviewing of existing evidence, similar to those used for RCTs can begin, with explicit statements of the literature search strategies, and where appropriate, the pooling of data across studies. Complete systematization and objectivization of hypothesis testing is less easily achieved than for RCT evaluation. A crucial final stage is thus to specifiy questions not answered by existing evidence, to which further studies should be directed.