28. Effect of Clinical Integration on Patient Outcomes in VA Cardiology
AE Sales, VA Puget Sound Health Care System; NR Every, VA Puget Sound Health Care System; L Parsons, Ovation Research Group; D Caldwell, University of Washington; S Pineros, VA Puget Sound Health Care System; A Keane, University of Washington; S Hedrick, VA Puget Sound Health Care System; E Perrin, VA Puget Sound Health Care System; J Rumsfeld, Denver VAMC; D Magid, Kaiser Permanente
Objectives: We studied VA patients with acute coronary syndrome (ACS) in 22 VA medical centers (9 tertiary with cardiology, 13 non-tertiary) across five VISNs. The primary purpose was to explore the relationship between degree of clinical integration and patient outcomes. Clinical integration is theorized to be an important determinant of improved efficiency and patient outcomes, but no studies to date have tested this relationship.
Methods: We identified all patients discharged with an ICD-9-CM code indicating acute myocardial infarction (AMI) and unstable angina (UAP) in any of the 22 VAMCs in five VISNs (13, 18, 19, 20 and 22) in a 12 month period from March 1998 through February 1999. We treated this as the index hospitalization identifying a cohort of patients with ACS. For each patient, we extracted data from the PTF and abstracted data from local VistA systems in each hospital. We included a wide variety of demographic and clinical indicators, modeling data collection after Medicare’s Cooperative Cardiovascular Program (CCP). We tracked deaths through PTF and BIRLS. Seven months after discharge, we sent all surviving patients the SF36, the Seattle Angina Questionnaire (SAQ), and a patient survey. Outcomes include death during the follow-up period and anginal frequency measured by the SAQ. We used an ordinal four level variable to measure integration, where 0 was least and 3 was most integrated. We analyzed outcomes using bivariate and multivariate analysis, correcting for cluster sampling.
Results: The study sample included 2236 veterans with (ACS) in 22 medical facilities. Of these, 1110 (49.6%) had AMI and 1126 (50.4%) had UAP. The seven month death rate was 11.7% in the sample overall, but much higher in the AMI patient population (17.6%) than UAP (5.9%). Anginal frequency scores were generally high for the entire sample, with a mean of 74 (on a 0-100 scale where higher is better), median 80, s.d. 27. 16% of the patients were discharged from least integrated facilities, 22% from Level 1, 35% from Level 2, and 27% from Level 3.
On bivariate analysis, death rates were higher in least integrated facilities than in the others (Level 0: 14.9%; Level 1: 11.4%; Level 2: 12.8% and Level 3: 8.9%). On logistic regression controlling for patient characteristics and correcting for cluster sampling, degree of integration was a significant predictor of death (OR 0.79, CI 0.67-0.93). However, degree of integration did not prove significant in predicting anginal frequency or other outcomes of the SAQ or SF36.
Conclusions: Deaths among ACS patients in VA appear to be influenced by how well facilities are integrated with cardiology services, even when we adjust for patient differences. However, we see little effect of integration on a major measure of morbidity in this population, anginal frequency. We continue to investigate factors that may have a mediating effect.
Impact: These results are likely to be of interest to clinicians, managers, particularly those with responsibility for cardiology services. As further analysis is completed, this study should provide some guidance about optimal consultation and referral arrangements.