Cochrane

Finding good endpoints for diagnostic test accuracy reviews on substance use

Diagnostic test accuracy reviews increasingly are being used in substance use research, yet the majority of the studies considered in these reviews do not account appropriately for the primary endpoints of interest.

Recently, the DSM-5 criteria have been criticised as poorly able to properly distinguish those with problematic use from those with substance use disorders.

We reviewed the main implications of adopting alternate primary endpoints in diagnostic test accuracy reviews of substance use disorders and highlight a practical application of appropriate analytical techniques.

The application of literature-derived primary endpoints is demonstrated through the use of empirical data from two substance use diagnostic test accuracy reviews. Primary endpoints were derived from literature in a review of alcohol-withdrawal patient assessment (JAMA 320(8):825-33), and patient risks of developing prescription opioid use disorders when initiating opioid analgesics for pain among opioid naïve patients (JAMA Network Open. 2(5):e193365). Primary endpoint options in existing substance use studies were evaluated. This evaluation included surveys of the literature for endpoints and measurement approaches, followed by assessment of endpoint choices against diagnostic test accuracy issues, population characteristics, tests of sensitivity and specificity.

Inappropriate definitions of primary endpoints in diagnostic test accuracy studies can lead to the presentation of inaccurate results and hence potentially misleading conclusions.

Adjustment for literature-derived endpoints, as opposed to using the criticised DSM-5 criteria, can be useful in these studies and we encourage more judicious use of the established diagnostic categories to enhance accuracy of reviews and meta-analyses.

Klimas, J., Dormuth, C., Wood, E. (2020) Determining the primary endpoints for diagnostic test accuracy reviews of substance use disorders. In: Advances in Evidence Synthesis: special issue. Cochrane Database of Systematic Reviews; (9 Suppl 1):471

If you enjoyed reading this post, you might also enjoy reading the articles cited here

Will this patient go into severe alcohol withdrawal?

Why most opioid risk tools fail?