Randomized database studies: a combination of the strengths of observational studies and randomized clinical trials?

textabstractBefore a drug is allowed into clinical practice, randomized clinical trial (RCT) have been conducted in order to demonstrate effi cacy and safety. The RCT method uses random treatment allocation to achieve comparable study groups and is further characterized by strict patient inclusion criteria, and strict follow-up schemes of the study groups.2,3 Once a drug enters the market, the drug is prescribed to a large number of heterogeneous patients who have concomitant diseases and drug use. As a result, unknown side effects may emerge (e.g., rare side effects, side effects due to interactions or comorbidity). In addition, the effectiveness of the drug in clinical practice is not the same as the effi cacy of the drug in clinical studies, mostly because usage patterns can be quite different. Postmarketing surveillance is the fi eld that studies the benefi cial and negative effects of drugs after they have been marketed. In post-marketing studies, both experimental and observational methods can be applied to study drug safety and effectiveness. Both methods have advantages and disadvantages. We will fi rst briefl y discuss these methods. Experimental MethodsPragmatic or naturalistic trials are suggested as a possible solution to the low generalisability of pre-marketing RCTs. In a pragmatic trial treatment is assigned randomly but patient inclusion is more liberal (less strict inclusion criteria) and the treatment modalities are not organized as strictly as in RCTs (corresponding with regular patient care).5 Therefore the results of pragmatic trials show effectiveness rather than effi cacy and better refl ect routine care practice. However, pragmatic trials often fail (or are not even considered as a possible alternative) because the implementation is time consuming and disrupts daily care processes or is ethically diffi cult.6,7Observational methodsObservational post-marketing studies typically are based on data that are collected for other purposes such as claims data or electronic medical records