Lily E Johnston1, William P Robinson1, Margaret C Tracci1, John A Kern2, Kenneth J Cherry1, Irving L Kron2, Gilbert R Upchurch3. 1. Division of Vascular and Endovascular Surgery, Department of Surgery, University of Virginia, Charlottesville, Va. 2. Division of Vascular and Endovascular Surgery, Department of Surgery, University of Virginia, Charlottesville, Va; Division of Thoracic and Cardiovascular Surgery, Department of Surgery, University of Virginia, Charlottesville, Va. 3. Division of Vascular and Endovascular Surgery, Department of Surgery, University of Virginia, Charlottesville, Va. Electronic address: gru6n@virginia.edu.
Abstract
OBJECTIVE: Both the Vascular Quality Initiative (VQI) and the National Surgical Quality Improvement Program Procedure Targeted (NSQIP-PT) databases aim to track outcomes and to improve quality in vascular surgery. However, both registries are subject to significant selection bias. The objective of this study was to compare the populations and outcomes of a single procedure in VQI and NSQIP-PT and to identify areas of similarity and discrepancy. METHODS: Deidentified regional data were provided by VQI, and the public use files were provided by NSQIP. Patient characteristics and outcomes were compared between data sets with parametric and nonparametric statistical tests as appropriate. For variables with different definitions between VQI and NSQIP-PT, a standardized definition was created to permit comparison across databases. To account for differences in populations of patients between the data sets, VQI and NSQIP-PT records were propensity matched, allowing a comparison of outcomes between databases adjusted for case mix. RESULTS: VQI contained 1358 records from 2011 to 2015, whereas NSQIP-PT contained 5273 complete records from 2011 to 2013. Patients in VQI are younger than those in NSQIP (65 [15] vs 68 [16] years; P < .001) and were less likely to have congestive heart failure (1.7% vs 3.1%; P = .005), to be on dialysis (4.0% vs 6.1%; P = .003), or to be receiving preoperative aspirin (62% vs 79%; P < .001) or statin therapy (63% vs 68%; P < .001). Significant discrepancies were noted in preoperative angina symptoms, prior myocardial infarction, and prior percutaneous coronary intervention, with 0, 1, and 0 NSQIP patients, respectively, having these risk factors compared with 9.4%, 0.7%, and 19.5% of the VQI cohort. Approximately 20% of patients in VQI underwent surgery for acute limb ischemia, which is not a recognized indication in NSQIP-PT. Overall 30-day mortality was equivalent (2.0% vs 1.8%; P = .6), as was composite myocardial infarction/stroke (3.9% vs 3.2%; P = .2). Major amputation (3.3% vs 1.6%; P = .002), return to operating room (16.1% vs 11.5%; P < .001), and wound infection rates (12.8% vs 1.4%; P < .001) were higher in NSQIP relative to VQI. Bleeding rates were higher in VQI (36.5% vs 17.2%; P < .001). Significant differences persisted in the propensity-matched groups. CONCLUSIONS: This is the first study to compare patient characteristics and outcome reported in the VQI and NSQIP-PT registries. These data documented statistically significant differences in demographics and comorbidities as well as in outcomes between databases. Physicians, payers, and the public should consider differences between these databases when reporting on outcomes and quality. Results from these two registries should not be directly compared.
OBJECTIVE: Both the Vascular Quality Initiative (VQI) and the National Surgical Quality Improvement Program Procedure Targeted (NSQIP-PT) databases aim to track outcomes and to improve quality in vascular surgery. However, both registries are subject to significant selection bias. The objective of this study was to compare the populations and outcomes of a single procedure in VQI and NSQIP-PT and to identify areas of similarity and discrepancy. METHODS: Deidentified regional data were provided by VQI, and the public use files were provided by NSQIP. Patient characteristics and outcomes were compared between data sets with parametric and nonparametric statistical tests as appropriate. For variables with different definitions between VQI and NSQIP-PT, a standardized definition was created to permit comparison across databases. To account for differences in populations of patients between the data sets, VQI and NSQIP-PT records were propensity matched, allowing a comparison of outcomes between databases adjusted for case mix. RESULTS: VQI contained 1358 records from 2011 to 2015, whereas NSQIP-PT contained 5273 complete records from 2011 to 2013. Patients in VQI are younger than those in NSQIP (65 [15] vs 68 [16] years; P < .001) and were less likely to have congestive heart failure (1.7% vs 3.1%; P = .005), to be on dialysis (4.0% vs 6.1%; P = .003), or to be receiving preoperative aspirin (62% vs 79%; P < .001) or statin therapy (63% vs 68%; P < .001). Significant discrepancies were noted in preoperative angina symptoms, prior myocardial infarction, and prior percutaneous coronary intervention, with 0, 1, and 0 NSQIP patients, respectively, having these risk factors compared with 9.4%, 0.7%, and 19.5% of the VQI cohort. Approximately 20% of patients in VQI underwent surgery for acute limb ischemia, which is not a recognized indication in NSQIP-PT. Overall 30-day mortality was equivalent (2.0% vs 1.8%; P = .6), as was composite myocardial infarction/stroke (3.9% vs 3.2%; P = .2). Major amputation (3.3% vs 1.6%; P = .002), return to operating room (16.1% vs 11.5%; P < .001), and wound infection rates (12.8% vs 1.4%; P < .001) were higher in NSQIP relative to VQI. Bleeding rates were higher in VQI (36.5% vs 17.2%; P < .001). Significant differences persisted in the propensity-matched groups. CONCLUSIONS: This is the first study to compare patient characteristics and outcome reported in the VQI and NSQIP-PT registries. These data documented statistically significant differences in demographics and comorbidities as well as in outcomes between databases. Physicians, payers, and the public should consider differences between these databases when reporting on outcomes and quality. Results from these two registries should not be directly compared.