Defining the Primary Outcomes and Justifying Secondary Outcomes of a Study: Usually, the Fewer, the Better
One of the first steps in designing and conducting a research study is identifying the primary and any secondary study outcomes. In an experimental, quasi-experimental, or analytic observational research study, the primary study outcomes arise from and align directly with the primary study aim or objective. Likewise, any secondary study outcomes arise from and directly align with any secondary study aim or objective. One designated primary study outcome then forms the basis for and is incorporated literally into the stated hypothesis. In a Methods section, authors clearly state and define each primary and any secondary study outcome variable. In the same Methods section, authors clearly describe how all primary and any secondary study outcome variables were measured. Enough detail is provided so that a clinician, statistician, or informatician can know exactly what is being measured and that other investigators could duplicate the measurements in their research venue. The authors provide published substantiation (preferably) or other documented evidence of the validity and reliability of any applied measurement instrument, tool, or scale. A common pitfall—and often fatal study design flaw—is the application of a newly created (“home-grown”) or ad hoc modification of an existing measurement instrument, tool, or scale—without any supporting evidence of its validity and reliability. An optimal primary outcome is the one for which there is the most existing or plausible evidence of being associated with the exposure of interest or intervention. Including too many primary outcomes can (a) lead to an unfocused research question and study and (b) present problems with interpretation if the treatment effect differed across the outcomes. Inclusion of secondary variables in the study design and the resulting manuscript needs to be justified. Secondary outcomes are particularly helpful if they lend supporting evidence for the primary endpoint. A composite endpoint is an endpoint consisting of several outcome variables that are typically correlated with each. In designing a study, researchers limit components of a composite endpoint to variables on which the intervention of interest would most plausibly have an effect, and optimally with preliminary evidence of an effect. Ideally, components of a strong composite endpoint have similar treatment effect, frequency, and severity—with the most important being similar severity.