The Community Care (CC) program allows VA to buy services to overcome capacity constraints and to improve Veteran access to care. This program has been expanded by the Veterans Choice Act. There is limited information comparing the quality, cost and accessibility of care provided by VA relative to that purchased by VA. Managed care organizations have used selective contracting to improve quality and lower costs but VA has made limited use of this approach.
We studied quality, cost, and accessibility of elective Percutaneous Coronary Intervention (PCI) and Coronary Artery Bypass Graft (CABG) surgery. We determined the number of procedures provided by VA and CC providers. Quality measures included mortality and readmissions within 30 days. Accessibility was determined by travel distance and associated time and travel cost. We considered the potential effects of selecting providers based on easily observed criteria: annual volume of procedures and hospital performance reported to Medicare. Hospital performance was measured using the 30-day risk-adjusted mortality following acute myocardial infarction in the Hospital Compare database.
We studied elective procedures provided and sponsored by VA received by patients under 65 in using data from the VA Corporate Data Warehouse (CDW). Since there is a lag in recording CC claims, the study considered the 36 month period ending September 30, 2011. Mortality was determined from the VA Vital Status file. Hospital admissions within 30 days were considered a readmission if the primary diagnosis was cardiovascular related. Costs were from the VA Managerial Cost Accounting system and the VA CC claims data and included cost of physician services.
Travel distance and travel time were estimated from patient residence to the index provider and to the closest provider in the alternative system (VA or CC). Travel cost included reimbursement for private auto travel and the value of time spent in transit. A sensitivity analysis included lodging expenses and caregiver travel expense and time.
Annual hospital-level procedure volume was from VA administrative data and for non-VA hospitals, from national reference sources. The Medicare Hospital Compare database was used to identify hospitals with the 10% highest risk-adjusted mortality for acute myocardial infarction.
Logistic models were used for mortality and readmission outcomes and gamma regression for costs. Sandwich estimators adjusted standard errors for the correlation within hospital. Propensity weighting controlled for differences in risk factors of VA and CC patients.
The study cohort included 13,227 patients who received PCI (79.1 % at VA) and 5,809 patients who received CABG (83.6% at VA). There was considerable overlap in the risk profile of PCI and CABG patients at VA and CC facilities. After applying propensity weights, standardized differences in risk factors were quite small, and well below the criteria to assess risk adjustment as adequate.
The unadjusted 30 day mortality following PCI was 1.63% in CC and 0.63% in VA. Risk adjusted mortality was significantly greater for CC providers (Odds Ratio OR 2.42, 95% Confidence Interval 1.57-3.72, p < .001). Readmission rates were not significantly different in the two systems. The mean cost of PCI was $22,032 in CC hospitals vs. $15,684 in VA, a 30% greater cost (p < .001). Obtaining PCI from VA required 65 additional miles of travel, with $72 extra cost in minimum reimbursement scenario, and $198 additional travel cost in the maximum reimbursement scenario.
CC hospitals had much higher PCI volumes than VA hospitals. Just 3.7% of index CC PCI took place in a hospital that failed to meet the recommended volume threshold of 200 cases annually, compared to 41.8% of VA provided PCI (p < .001). CC PCI was much more likely to take place in a hospital that had high AMI mortality risk (14.0% vs 9.9% in VA, p < .001). There were no significant differences in PCI mortality, readmission, or cost for sub-groups of hospitals defined by annual hospital volume or performance in Hospital Compare.
Unadjusted 30 day mortality following CABG was 1.26% in CC and 1.50% in VA. Adjusted mortality was not significantly different (OR 0.89, 95% CI 0.44-1.78). There were no significant differences in readmission rates. The mean cost of CABG was $55,531 in CC hospitals vs. $63,192 in VA, or 14% less in CC hospitals (p < .01). Obtaining CABG from VA required 57 additional miles of travel, with $63 extra cost in minimum reimbursement scenario, and $520 additional travel cost in the maximum reimbursement scenario.
CC hospitals had much higher CABG volumes than VA hospitals. Just 10.1% of index CC CABG took place in a hospital that failed to meet the recommended volume threshold of 125 cases annually, compared to 64.3% of VA provided CABG (p < .001).
There was no difference between systems in the fraction of CABG providers with high AMI mortality risk (10.1% in CC vs. 10.0% in VA, NS). The only significant difference between subgroups was the lower cost of hospitals that performed poorly on Hospital Compare measures.
Costs and mortality would decrease by shifting PCI from CC to VA. The cost savings would be sufficient to compensate veterans and their caregivers the additional travel expense. Given the current capacity constraints, VA could start this shift in patients who face little additional travel cost to use VA. Further research is needed to see if the higher mortality of purchased PCI stems from lack of continuing of care or delays in CC approval.
Costs would be lower if CABG was shifted from VA to CC, without any change in mortality rates. Shifting all elective CABG surgeries to CC providers could impair VA ability to care for emergent patients, however.
Selecting hospitals based on minimum annual volume or a threshold performance in Hospital Compare would not improve the quality of care in our analysis.
External Links for this Project
Grant Number: I01HX000630-01A2
None at this time.