A Novel EVAR Long-Term Quality Metric, Wins Above Average, Discriminates Center Performance and Predicts Future Outcomes.
Academic Article
Overview
abstract
OBJECTIVE: Endovascular AAA repair (EVAR) has low perioperative morbidity and mortality, thereby making long-term outcomes the primary determinant of overall quality. Our aim was to develop a novel, long-term EVAR quality metric, capable of discriminating center performance and predicting future outcomes, utilizing previously developed earned outcomes methods. METHODS: VQI-Medicare linked data were used to examine elective EVAR (2010-2015, with follow-up to 2019). Centers were evaluated on the primary outcome: a composite of mortality, major reintervention or aneurysm rupture. Major reintervention was defined as any EVAR-related reintervention which met one of the following criteria: (1) open surgical reintervention, (2) > 2 night hospital admission for reintervention, or (3) excessive endovascular reintervention burden (≥ 3 reinterventions). Using methods similar to those used in sports analytics, cumulative, probability-based metrics (wins above average, WAA) were calculated to generate risk-adjusted, volume-sensitive center-level rankings based on the primary outcome. Risk adjustment used Cox proportional hazards modeling. Patient and center characteristics and outcomes were compared across center quality quartiles (Q1 lowest quality quartile through Q4 highest quality quartile). The ability of 2010-2015 quality quartiles to predict outcomes in a future period (2016-2017) was determined. RESULTS: For 2010-2015, analysis included 10,081 patients treated at 198 Centers. WAA quartiles effectively discriminated 4-year composite mortality, aneurysm rupture, or major reintervention: Q1 44.5% vs Q2 40.4% vs Q3 34.0% vs Q4 28.8%, P<0.001. Quartiles also demonstrated differences in each primary outcome component: mortality (Q1 38% vs. Q2 35% vs. Q3 28% vs. Q4 24%, P<0.001), major reintervention (Q1 13% vs. Q2 11% vs. Q3 8.5% vs. Q4 7.8%, P<0.001), and aneurysm rupture (Q1 3.6% vs. Q2 2.0% vs. Q3 1.5% vs. Q4 1.8%, P<0.001). In 2016-2017, 5,153 patients underwent EVAR at centers where quality quartiles had been defined from the preceding study period. Two-year composite mortality, aneurysm rupture or major reintervention rates confirmed the predictive value of WAA quartiles: Q1 22.8% vs Q2 21.0% vs Q3 19.6% vs Q4 17.0%, P=0.002 CONCLUSIONS: WAA is a novel, long-term EVAR quality metric that evaluates aggregate aorta-related outcomes and effectively differentiates center-level performance. This innovative approach has potential for applications in quality improvement initiatives and benchmarking of EVAR care delivery.
