A review of methodologies for assessing drug effectiveness and a new proposal: randomized database studies.

The need to evaluate the effects of health technologies in clinical practice is increasingly important. In this article, we review the advantages and limitations of naturalistic randomized clinical trials (RCTs) and database analyses, the two primary methods for evaluating treatment effectiveness. Also, we comment on a newer research strategy, cross-design synthesis, which proposes the complementary use of both experimental RCTs and observational database methodologies to avoid the main weaknesses of each: respectively, the lack of external and internal validity. Finally, we propose a new strategy--randomized database studies--capable of generating results with an acceptable balance between internal and external validity. This strategy consists of the simultaneous use of both experimental and observational tools in the assessment of drugs' effectiveness. Randomization is essential to minimize comparison bias, and one possibility for such studies is that randomization modules could be included in computer-based patient records. Although we identify some of the difficulties in implementing the process, the progressive standardization of clinical practice and the development and widespread adoption of improved computer-based patient records could facilitate the use of randomized database studies as a new method of research.