The association between nursing factors and patient mortality in the Veterans Health Administration - The view from the nursing unit level

Context: Nurse staffing is not the same across an entire hospital. Nursing care is delivered in geographically-based units, with wide variation in staffing levels. In particular, staffing in intensive care is much richer than in nonintensive care acute units. Objective: To evaluate the association of in-hospital patient mortality with registered nurse staffing and skill mix comparing hospital and unit level analysis using data from the Veterans Health Administration (VHA). Design, Settings, and Patients: A retrospective observational study using administrative data from 129,579 patients from 453 nursing units (171 ICU and 282 non-ICU) in 123 VHA hospitals. Methods: We used hierarchical multilevel regression models to adjust for patient, unit, and hospital characteristics, stratifying by whether or not patients had an ICU stay during admission. Main Outcome Measure: In-hospital mortality. Results: Of the 129,579 patients, mortality was 2.9% overall: 6.7% for patients with an ICU stay compared with 1.6% for those without. Whether the analysis was done at the hospital or unit level affected findings. RN staffing was not significantly associated with in-hospital mortality for patients with ail ICU stay (OR, 1.02; 95% Cl, 0.99-1.03). For non-ICU patients, increased RN staffing was significantly associated with decreased mortality risk (OR, 0.91; 95% CI, 0.86-0.96). RN education was not significantly associated with mortality. Conclusions: Our findings suggest that the association between RN staffing and skill mix and in-hospital patient mortality depends on whether the analysis is conducted at the hospital or unit level. Variable staffing on non-ICU units may significantly contribute to in-hospital mortality risk.