2006 HSR&D National Meeting Abstract
3088 — Association Between Nurse Staffing and Education and Inpatient Mortality in VHA
Sales AE (VA Puget Sound HCS)
Greiner GT (VA Puget Sound HCS)
Sharp ND (VA Puget Sound HCS)
Li YF (VA Puget Sound HCS)
Lowy E (VA Puget Sound HCS)
Liu CF (VA Puget Sound HCS)
Sochalski JA (University of Pennsylvania)
Mitchell PH (University of Washington)
Needleman J (University of California at Los Angeles)
Several large scale studies have found associations between nurse staffing and mortality outcomes for hospitalized patients, aggregated to the facility level. In the largest study yet conducted with unit level staffing and patient data, we examined the association between nurse staffing, skill mix, educational factors, and patient in-hospital mortality.
Data came from several sources: DSS nursing labor input files (ALBCC); National Patient Care Databases at Austin for data on all patients admitted to VHA inpatient acute care between 2/03-6/03; and DSS TRTIPD files, a DSS extract file linking inpatients to nursing units. We developed a 2-step multilevel logistic regression model with patient, nursing unit, and hospital level data corrected for clustering at the unit and facility levels. The first step predicted patient probability of developing a serious complication using patient-level predictors. The second step estimated risk of dying using patient risk, nursing unit, and facility-level predictors. We stratified the analysis by whether the first inpatient unit was intensive care or non-intensive acute care.
The analyses included 126,382 patients from 463 nursing units in 119 VAMCs. 184 were intensive care, and 279 non-intensive acute care units. In all cases, patient risk was the most significant factor associated with mortality (OR 1.18 in ICU, 1.35 in non-intensive acute care). Skill mix and staffing had a non-linear, U-shaped association with mortality risk in both ICU and non-intensive care patients, and there was no significant relationship between non-RN staffing and mortality. RN educational level was not significantly associated with mortality in either type of unit.
Some of the relationships that have been shown between patient mortality and nursing factors at the facility level do not appear to hold in unit level analyses. Multi-level modeling, adequate patient risk adjustment, and careful modeling of non-linear relationships are critical. Aggregating both patients and nurse staffing to the facility level may result in biased estimates through mixing very heterogeneous groups.
Staffing variables are modifiable by managers, but their relationship to outcomes is more complex than previously reported. It may be possible to optimize the mix between costly RN and less costly staff without an adverse impact on patient mortality.