Evaluation of prescriber responses to pharmacogenomics clinical decision support for thiopurine S-methyltransferase testing

Purpose. Results of a study of prescribers' responses to a pharmacogenomics- based clinical decision support (CDS) alert designed to prompt thiopurine S-methyltransferase (TPMT) status testing are reported. Methods. A single-center, retrospective, chart review-based study was conducted to evaluate prescriber compliance with a pretest CDS alert that warned of potential thiopurine drug toxicity resulting from deficient TPMT activity due to TPMT gene polymorphism. The CDS alert was triggered when prescribers ordered thiopurine drugs for patients whose records did not indicate TPMT status or when historical thiopurine use was documented in the electronic health record. The alert pop-up also provided a link to online educational resources to guide thiopurine dosing calculations. Results. During the 9-month study period, 500 CDS alerts were generated: in 101 cases (20%), TPMT phenotyping or TPMT genotyping was ordered; in 399 cases (80%), testing was not ordered. Multivariable regression analysis indicated that documentation of historical thiopurine use was the only independent predictor of test ordering. Among the 99 patients tested subsequent to CDS alerts, 70 (71%) had normal TPMT activity, 29 (29%) had intermediate activity, and none had deficient activity. The online resources provided thiopurine dosing recommendations applicable to 24 patients, but only 3 were prescribed guideline-supported doses after CDS alerts. Conclusion. The pretest CDS rule resulted in a large proportion of neglected alerts due to poor alerting accuracy and consequent alert fatigue. Prescriber usage of online thiopurine dosing resources was low.