LO85: Substantial variation in CTPE ordering patterns and diagnostic yield in a large group of specialty-trained emergency physicians

Abstract

Introduction: Computerized tomography for pulmonary embolism (CTPE) has come under increased scrutiny with recommendations for evidence-based use found on Choosing Wisely lists in both Canada and the US. However practice variation in ordering patterns and diagnostic yield have not been well-reported for the Canadian context. Our objective was to investigate practice variation in CTPE ordering and rule-in rates within a large group of specialty-trained emergency physicians. Methods: We undertook an analysis of a computerized physician order entry database from four tertiary care EDs covering a 12-month period from August 1, 2016 to July 30, 2016 with 31 419 visits for potential pulmonary embolism (PE) as determined by a previously validated algorithm based on presenting complaints. CTPE utilization and diagnostic yield were determined for 149 physicians who ordered at least 10 studies over that time period. Outcomes of interest included CT utilization as determined by electronic order entry and a confirmed diagnosis of PE based on ICD-10 coding of the emergency visit. Descriptive statistics using medians, IQR and 95% confidence intervals are reported. This study is approved through REB14-0650 and is a component of a larger cluster RCT to improve CTPE utilization. Results: During the study period 2670 CTPE studies were ordered for potential PE patients representing 8.5% of the total with relevant complaints. We observed a 10-fold variation in CTPE ordering among physicians with rates as low as 2.7% and as high as 25%. The median rate of CTPE ordering for potential PE was 8.8% with an IQR of 6.0% to 11.7%. A total of 4146 CTPE studies were ordered during the study period with physicians ordering an average of 28 CTPE studies each; range 10-90. In terms of diagnostic yield, 591/4146 studies, or 14.3% (95% CI 13.2-15.3%) were associated with a diagnosis of PE. Diagnostic yield per physician ranged from 0 to 50%, with a median of 13.5% and an IQR of 7.6% to 21.4%. Conclusion: In this large, robust administrative dataset from four Canadian urban EDs, threshold for CTPE ordering varies widely among physicians as does diagnostic yield. Efforts to improve appropriate utilization are justified with an eye to reducing unnecessary radiation, costs and incidental findings.