One could summarize all scientific endeavor as seeking to understand cause and effect. Clinical trials are scientific experiments meant to answer the question, “Does this intervention cause that outcome?” Though the intervention can take many forms (e.g. diet, physical therapy, psychotherapy), in pharmaceutical drug development we most often deal with pharmacologic agents as treatments.
Much has been written in recent years regarding estimands, defined as “the treatment effect that is to be estimated.” It is now generally recognized that identifying WHAT treatment effect should be estimated must be aligned with the study objectives. While that may be considered common sense, it may not always be common practice since defining the estimand for a study requires careful thought and attention to detail. Though it may seem obvious, estimating the treatment effect primarily requires a clear and precise definition of the treatment.
This presentation will review oft-neglected considerations for what satisfies the principles of cause and effect. Furthermore, arguments will be made for greater emphasis on estimating the direct treatment effect rather than a treatment policy effect.