Framework for assessing causality in disease management programs: principles

To credibly state that a disease management (DM) program "caused" a specific outcome it is required that metrics observed in the DM population be compared with metrics that would have been expected in the absence of a DM intervention. That requirement can be very difficult to achieve, and epidemiologists and others have developed guiding principles of causality by which credible estimates of DM impact can be made. This paper introduces those key principles. First, DM program metrics must be compared with metrics from a "reference population." This population should be "equivalent" to the DM intervention population on all factors that could independently impact the outcome. In addition, the metrics used in both groups should use the same defining criteria (ie, they must be "comparable" to each other). The degree to which these populations fulfill the "equivalent" assumption and metrics fulfill the "comparability" assumption should be stated. Second, when "equivalence" or "comparability" is not achieved, the DM managers should acknowledge this fact and, where possible, "control" for those factors that may impact the outcome(s). Finally, it is highly unlikely that one study will provide definitive proof of any specific DM program value for all time; thus, we strongly recommend that studies be ongoing, at multiple points in time, and at multiple sites, and, when observational study designs are employed, that more than one type of study design be utilized. Methodologically sophisticated studies that follow these "principles of causality" will greatly enhance the reputation of the important and growing efforts in DM.