Differences in risk too small to be detected by conventional RCTs for drugs and supplements with common indication were

judged to be clinically important, and a large RCT was conducted to resolve the question about their safety. When confronted with the task of assessing the safety of a marketed drug product, the pharmaco-epidemiologist must evaluate the specific hypothesis to be tested and estimate the magnitude of the hypothesized association and determine whether confounding by indication is possible. If incomplete control of confounding is likely, it is important to recognize the limitations of observational research designs and consider conducting an RCT. There is nothing inherent in an RCT that precludes pharmacoepidemiologist from designing and carrying out these studies. To the contrary, the special skills of a pharmacoepidemiologist can be very useful in performing large-scale RCTs after a drug is marketed. In the absence of techniques that reliably control for confounding by indication in observational studies, there may be a growing need for LSTs to evaluate larger relative risks.