The marked increase in spending on drugs has led payers such as provincial governments to restrict funding for many drugs to specific clinical indications that are thought to be cost-effective. Some have argued that this unreasonably deprives patients of access to beneficial drugs. This article argues for a new approach to drug evaluation in Canada that combines the strengths of randomized trials and observational studies, and places more emphasis on the use of drug evaluation after marketing for decision-making.

At least 4 different types of clinical studies are required to inform rational drug policy:

1. Randomized trials to determine efficacy and safety (which are required for licensing)
2. Real-world randomized trials to determine effectiveness and safety in regular practice
3. Observational studies that use administrative databases
4. Targeted primary data collection.

Currently, most randomized trials are done to determine efficacy and safety under ideal conditions, whereas the other designs, which are less frequently used, attempt to determine a drug's pattern of use and effectiveness under real-world conditions. For some drugs, the results of randomized trials of efficacy will be so straightforward and the possibility of real-world use outside the conditions of the trial so small that no other study designs will be required. However, for other drugs there may be concern about the impact of the drug upon clinically important outcomes (e.g., if surrogate outcomes were used in the efficacy trials) or concern that the drug will be used for patients on whom it has not been studied or for whom it is not cost-effective. In these circumstances, two or more of the study designs would be required to adequately assess the impact of the drug in the real world and to guide clinical and policy decisions.