2006 HSR&D National Meeting Abstract
3071 — Organizational Factors and ICU Patient Outcomes in VHA
Greiner GT (Northwest Ctr for Outcomes Research in Older Adults)
Sharp ND (Northwest Ctr for Outcomes Research in Older Adults)
Li YF (Northwest Ctr for Outcomes Research in Older Adults)
Lowy E (Northwest Ctr for Outcomes Research in Older Adults)
Almenoff P (VISN 15 Heartland Network)
Sales AE (Northwest Ctr for Outcomes Research in Older Adults)
Several organizational factors have been demonstrated to be significantly associated with mortality among ICU patients. We report on the association between nursing and other inpatient organizational factors and inpatient mortality for patients experiencing an ICU stay during their hospitalization.
Data for the analyses came from several sources including DSS nursing labor input files (ALBCC); National Patient Care Databases for patient characteristics and outcomes on all patients experiencing an ICU stay during hospitalization between 2/03 and 6/03; DSS TRTIPD files, a new DSS extract file linking inpatients to nursing units, for data on nursing units; and the 2004 ICU survey, an organizational inventory of intensive care within the VHA. We analyzed the data using a 2-step multilevel logistic regression model that incorporated patient, nursing unit, and hospital level data corrected for clustering at the nursing unit and hospital levels. The first step predicted ICU patient probability of developing one of five serious complications. The second step estimated mortality risk.
The analyses included 21,855 patients from 158 ICU units in 105 VAMCs. Five factors appear associated with mortality: predicted risk of complication (O.R. 1.15 (95% C.I. 1.14 to 1.17)), high proportion contract nurses in the unit (O.R. 0.48 (0.28 to 0.83)), being a medical vs. surgical patient (O.R. 1.56 (1.33 to 1.82)), ICU as initial admitting unit (O.R. 0.21 (0.07 to 0.65)), and a monitoring system that interfaces with VISTA (O.R. 0.66 (0.50 to 0.88)).
Patient risk is the dominant factor associated with mortality, highlighting the need for adequate risk adjustment. Nurse staffing as usually measured appears not to be associated, and a relatively high proportion of contract nurses in the unit appears to be associated with decreased mortality, which may indicate that ensuring adequate staff through contracting is beneficial. The importance of an integrated monitoring system is unexpected, and suggests that information technology may play an important role in protecting patients.
Contract nursing may not adversely affect patient mortality outcomes, as has been suggested previously, when other factors are controlled. More analysis will focus on the possible mechanisms of action. Integrated monitoring systems are not widespread, but appear to be beneficial.