Customize your JAMA Network experience by selecting one or more topics from the list below.
Chong J, Lee TC, Attarian A, et al. Association of Lower Diagnostic Yield With High Users of CT Pulmonary Angiogram. JAMA Intern Med. 2018;178(3):412–413. doi:10.1001/jamainternmed.2017.7552
Pulmonary embolism (PE) can be life-threatening and, when suspected, is usually investigated by computed tomographic pulmonary angiogram (CTPA). Concerns related to overutilization and harmful ionizing radiation have identified CTPA as an area in need of resource stewardship.1,2 The purpose of this study was to explore interphysician variability in CTPA diagnostic yield and to identify any associated physician characteristics that could inform an intervention to reduce overuse in our institution.
We retrospectively reviewed all CTPAs at an academic teaching hospital in Montreal, Quebec, Canada, from September 2014 to January 2016. Studies were classified as positive or negative; indeterminate examinations, and those performed for chronic pulmonary emboli were excluded. A total of 1394 examinations ordered by 182 physicians were included, of which 199 (14.3%) were positive and 1195 (85.7%) were negative. A multivariable logistic regression analysis was performed to explore whether physician specialty, years in practice, physician sex, or total numbers of studies ordered per physician were associated with CTPA diagnostic yield. We used a generalized estimating equations (GEE) approach to account for patients who underwent repeated examinations over the study period.3 Statistical tests of hypothesis were 2-sided with a significance of P ≤ .05. All analyses were performed using SAS statistical software (version 9.4, SAS Institute Inc). The McGill University Health Centre research ethics board approved this study.
According to our analysis using GEE logistic regression, the odds of a positive CTPA decreased as the total number of scans ordered per physician increased (Table). For each additional 10 studies ordered, the odds of a positive result decreased (odds ratio [OR], 0.76; 95% CI, 0.73-0.79). Increasing patient age was associated with a higher diagnostic yield (OR, 1.02 per year; 95% CI, 1.01-1.03). Physician years of experience (OR, 1.01; 95% CI, 0.99-1.02; P = .39), physician sex (OR, 1.14; 95% CI, 0.79-1.63; P = .49), and studies originating from the emergency department (ER) (OR, 1.11; 95% CI, 0.75-1.65; P = .60) did not show a statistically significant association. When restricting the analysis to those studies performed by ER physicians, 123 of 974 (12.6%) were positive, and the OR for positivity for each additional 10 scans ordered decreased in a similar manner (OR, 0.74; 95% CI, 0.71-0.78).
Our institutional yield of positive CTPA was 14.3%, which is similar to prior reported studies.4-6 However, closer inspection demonstrated that there was substantial interphysician variability, with individual positivity rates ranging between 0% to 33.3%. Our study suggests that individual demographic features, such as specialty and professional experience are not significantly associated with diagnostic yield; however, physicians who ordered a greater volume of scans compared with their peers had a markedly reduced diagnostic yield.
Limitations of our study included its retrospective design and the inadequate charting of parameters, which precluded the derivation of a preexamination clinical probability score. In addition, owing to limitations in data collection, we were unable to determine the denominator of patients presenting with PE-related symptoms examined by each physician. Notably, in the cohort of ER physicians shift allotment was generally randomized and we did not observe any systematic biases.
Among physicians who ordered a greater number of CTPA scans, we observed a statistically significant decrease in the proportion of positive results. This association may reflect a fundamental relationship between individual physician overutilization and decreasing diagnostic yield, and is deserving of greater attention. Peer-relative rates of utilization are easily quantified from electronic databases, and can identify physicians most likely to benefit from individual performance feedback and decision support tools. Based on these findings, our group has designed automated yield monitoring and feedback, with the aim of closing the gap between individual physician performance in our institution. We hope this will translate to a substantial reduction in unnecessary CTPA scans along with the associated complications that may occur owing to unnecessary radiation, overdiagnosis, and overtreatment.
Corresponding Author: Emily G. McDonald, MD, Clinical Practice Assessment Unit, Department of Medicine, McGill University Health Centre, 1001 Blvd Décarie, Montréal, QC, H4A 3J1, Canada (firstname.lastname@example.org).
Accepted for Publication: October 30, 2017.
Published Online: January 8, 2018. doi:10.1001/jamainternmed.2017.7552
Author Contributions: Drs Chong and McDonald had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs McDonald and Gallix contributed equally to the work.
Study concept and design: Chong, Lee, Attarian, McDonald, Gallix.
Acquisition, analysis, or interpretation of data: Chong, Lee, Attarian, Sivakumaran, Troquet, Gallix.
Drafting of the manuscript: Chong, Attarian.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Chong, Attarian.
Obtained funding: Gallix.
Administrative, technical, or material support: Chong, Lee, Attarian, McDonald, Gallix.
Study supervision: Chong, Troquet, McDonald, Gallix.
Conflict of Interest Disclosures: None reported.
Create a personal account or sign in to: