PURPOSE: To describe the national program in Veterans Affairs (VA) intensive care units (ICUs) to measure and report outcomes and processes linked to evidenced based practices (EBP).
METHODS: For all VA ICU patients from 2002 –2004, customized programming extracted outcome (survival and ICU length of stay), laboratory data, source of admission, and ICD-9-CM coding (grouped into diagnosis and comorbid diseases). Using a validated risk adjustment method, the estimated ICU mortality and length of stay (LOS) are compared to observed (standardized mortality ratio[SMR] and observed minus predicted LOS [OMELOS]). Individual pharmacy orders were merged from a national inpatient database to assess adherence to EBPs as were survey results of ICU characteristics.
RESULTS: The 1906 ICU beds in 37 Medical/cardiac, 15 Medical, 15 Cardiac, 52 Surgical, and 73 mixed ICUs (Total 192)in 125 VA hospitals admitted 375,887 cases. 113 ICUs were Level 1, 28 - level 2, 51 - level 3 or 4. Most of the closed ICUs (51%) were medical units (70%). The cases included 40,008 cases with angina, 18758 cases with acute myocardial infarction, 13,756 with congestive heart failure, and 9056 patients with pneumonia as well as 12181 CABG patients, 8393 colectomy surgeries, 9314 carotid endarterectomies. The SMR for the national dataset was 1.0 with an OMELOS 0.0 days (-1.43 to 2.33). The models had excellent calibration and discrimination. 49.8% of cases were low severity (<2.5% predicted mortality; range 14% - 62%). 45% of cases were admitted from the ED or outpatient area, 28% from operating rooms, and 22% from wards (range 10 - 62%). 12.7% of the cohort was over 80 years on admission. The reports are benchmarked by type of ICU and complexity (levels). Web based databases and toolkits share learning among the ICUs.
CONCLUSION: Continuous tracking of ICU processes and outcomes through adaptation of existing hospital information systems may promote unidirectional variation reduction. Database analysis may lead to new “best” practices.
CLINICAL IMPLICATIONS: Information systems support analysis of ICU care to drive performance and estimates of future needs.
DISCLOSURE: Marta Render, None.