Topic Abstract
Background: Understanding the heterogeneity of treatment effects (HTE) is essential for making optimal treatment choices for individuals. Randomized controlled trials (RCTs) generally provide the most reliable information on whether a treatment works on average for a collection of individuals. However, the evidence on treatment effectiveness from RCT does not necessarily apply to a different collection of individuals which is unlike the RCT subjects. For example, it is well-known that women older than 65 years are under-represented in the RCTs of many therapies. It is of major clinical importance to know how we can extend the evidence on the effectiveness of therapeutics to the care of older women.
Objectives:
1. To summarize the literature on analytic methods for addressing HTE and to develop a framework, based on the literature review, for determining sufficient conditions under which HTE can be reliably and optimally managed in comparative effectiveness research methods.
2. To develop a methodology to extend the evidence on the effectiveness of therapeutics obtained in RCTs to groups under-represented in the RCTs.
3. To apply the new methodology in a comparative effectiveness study of heart failure (HF).
4. To evaluate the quality of performance of the methodology using verification (e.g., simulation studies), internal (e.g., cross-validation) and external validation (e.g., data from another clinical trial of an ACE-inhibitor).
Study Design and Methods: The investigators will develop a novel methodology to facilitate the application of evidence on treatment effectiveness from RCTs to populations that are under-represented in the RCTs. The methodology is based on the concept of a synthetic study design ‘dyad’ that combines two traditional study designs in such a manner as to mitigate the primary weaknesses inherent in each. Here the dyad comprises an RCT and a prospective observational cohort study. The methods will facilitate the extension of the evidence provided by RCTs, under appropriate conditions, to populations not originally covered by that evidence.
Expected Outputs: Four scientific reports.
Expected date of completion: Summer 2011.
