18 Variation in diagnostic test requests and outcomes: a preliminary metric for openpathology.net

Abstract

Efforts to reduce healthcare costs have led to the development of metrics to identify unwarranted variation in care. Previous work assessing diagnostic tests is limited, despite their substantial contribution to expenditure. We explored C-reactive Protein (CRP) and Erythrocyte Sedimentation Rate (ESR) tests ordered across Oxfordshire NHS General Practices, the proportion of tests that yielded an abnormal result, and identified practices that had a proportion of abnormal CRP and ESR results 3 standard deviations below the mean. We estimated the adjusted average proportion of abnormal CRP and ESR tests that yielded abnormal results from each practice, after adjusting for differences in practice populations. These proportions were plotted against the total CRP and ESR requests per practice. We constructed funnel plots to identify practices 3 standard deviations below the mean proportion of abnormal CRP and ESR tests. We analysed 143,745 CRP and 30,758 ESR requests from 69 practices. Twelve (17%) and 7 (10%) practices were more than 3 standard deviations below the mean for CRP and ESR testing respectively. Two practices (3%) were below the 99.8% limit for both CRP and ESR ordering. Variation in the proportion of tests with an abnormal result shows promise for auditing variation in care. Objectives To identify General Practices that order significantly more normal ESR and CRP tests compared with their peers. Method We explored C-reactive Protein (CRP) and Erythrocyte Sedimentation Rate (ESR) tests ordered across Oxfordshire NHS General Practices and generated the proportion of tests that yielded an abnormal result for each practice. We estimated the adjusted average proportion of abnormal CRP and ESR tests that yielded abnormal results from each practice, after adjusting for differences in practice populations. These proportions were plotted against the total CRP and ESR requests per practice. We constructed funnel plots to identify practices 3 standard deviations below the mean proportion of abnormal CRP and ESR tests. Results We analysed 143,745 CRP and 30,758 ESR requests from 69 practices. Twelve (17%) and 7 (10%) practices were more than 3 standard deviations below the mean for CRP and ESR testing respectively. Two practices (3%) were below the 99.8% limit for both CRP and ESR ordering. Conclusions We used robust and conservative methods to identify two practices (3%) that requested a significantly lower proportion of CRP and ESR tests yielding abnormal results. We also identified 12 (17%) and seven (10%) General Practices that requested a significantly lower proportion of CRP and ESR tests yielding abnormal results, respectively. Variation in the proportion of tests with an abnormal result shows promise as an automated tool for auditing variation in care, and may contribute to improving quality and cost effectiveness.