OBJECTIVES: Assessments composed of patient-reported outcome (PRO) measures can be used in health care settings as screeners for various conditions. The objective of developing and using a PRO screening measure may be to quickly identify patients who are likely to benefit from a formal diagnostic evaluation. Alternatively, the development objective may be to avoid unnecessary diagnostic procedures, particularly when these are time or resource-intensive or invasive in nature. The PRO screener may also be administered to simply rule out the existence of a particular condition. The evaluation of a PRO screening assessment ideally occurs through analyses using a “gold standard” diagnosis of the condition of interest.
METHODS: A number of existing statistical and psychometric methods may be used in such an evaluation, including sensitivity and specificity, positive and negative predictive value, kappa, accuracy, odds ratio, and likelihood ratio.
RESULTS: The evaluation method selected depends on the objective of the screener itself. If the formal diagnostic procedures are particularly invasive or time- or resourceintensive, then screeners should minimize false positives; in contrast, diseases with exceptional risks when left undiagnosed call for screeners that minimize false negatives. In this research, we explore these methods using data from a study comparing various fibromyalgia screening instruments with the currently accepted gold standard diagnosis for fibromyalgia, namely the American College of Rheumatology 1990 diagnostic criteria (Wolfe et al., 1990).
CONCLUSIONS: Using the example application, we illustrate the pros and cons of a battery of statistical methods and how they can be used to select the “best” candidate screener.
