Sensitivity and Specificity: A Review of Related Statistics and Controversies in the Context of Physical Therapist Education

Abstract

Background and Purpose. Sensitivity and specificity are important factors to consider when evaluating the validity of a diagnostic test. The 2 measures can help to determine whether test results are indicative of the actual disease status during client assessment and evaluation. Their properties, however, can be misinterpreted in clinical settings. The purpose of this paper is three-fold: (1) to review the appropriate interpretation of sensitivity and specificity, positive and negative predictive values, likelihood ratios, posttest probabilities, and receiver operating characteristic (ROC) curves; (2) to discuss limitations of sensitivity and specificity as sole indicators of test validity; and (3) to offer recommendations for educators to improve related curricular content. Position and Rationale. The position adopted in this paper is that physical therapist education programs should cover the concepts of sensitivity and specificity in depth and in combination with other elements of test characteristics prior to discussing specific clinical tests and their applications to clinical decision making. Related content should include a review of the appropriate interpretation of sensitivity and specificity as well as an introduction to positive and negative predictive values, likelihood ratios, posttest probabilities, and ROC curves. The limitations of sensitivity and specificity as sole indicators of test validity should also be discussed. The rationale for this approach is based on the potential over-reliance of students and educators on sensitivity and specificity alone to establish the validity of clinical measures. Conclusion and Recommendations. Sensitivity and specificity will continue to be very important concepts in physical therapist education. Educators, however, should thoroughly discuss sensitivity and specificity prior to using them across the curriculum and emphasize limitations of these test properties in clinical diagnostic practice. Specific recommendations include a review of the 2 x 2 table, evaluation of specific studies on diagnostic tests, using related measures to evaluate validity, class exercises using interval estimates, and a review of the Standards for Reporting of Diagnostic Accuracy Studies (STARD) and Quality Assessment of Diagnostic Accuracy Studies (QUADAS) initiatives for research on diagnostic test validity.