Hospital and unit characteristics associated with nursing turnover include skill mix but not staffing level: An observational cross-sectional study

Background: Nursing turnover is expensive and may have adverse effects on patient care. Little is known about turnover's association with most hospital and nursing unit characteristics, including nurse staffing level and registered nurse skill mix. Objective: To explore associations between nursing unit turnover rates and several hospital- and unit-level variables, including staffing level and skill mix. Design: Observational cross-sectional study of longitudinal data. Settings: 1884 nursing units in 306 U.S. acute care hospitals. Methods: During a 2-year period units reported monthly data on staffing and turnover. Total nursing staff turnover and registered nurse turnover rates were modeled as dependent variables in hierarchical Poisson regression models. The following hospital characteristics were considered as predictors: Magnet (R) status, ownership (government or non-government), teaching status, locale (metropolitan, micropolitan, or rural), and size (average daily census). The U.S. state in which the hospital was located was included as a covariate. Unit-level variables included total nursing hours per patient day, size of nursing staff, registered nurse skill mix, population age group (neonatal, pediatric, or adult), and service line (critical care, step-down, medical, surgical, medical/surgical, psychiatric, or rehabilitation). Results: Government ownership, Magnet designation, and higher skill mix were associated with lower total turnover and registered nurse turnover. Neonatal units had lower total and registered nurse turnover than pediatric units, which had lower total and registered nurse turnover than adult units. Unit service line was associated only with total turnover. Psychiatric, critical care, and rehabilitation units had the lowest mean turnover rates, but most differences between service lines were not significant. The other explanatory variables considered were not significant. Conclusions: Several hospital and unit characteristic variables have significant associations with nursing turnover; these associations should be taken into account in nursing turnover research and need to be explored further. Controlling for hospital ownership, Magnet status, unit service line, and unit population age group, registered nurse skill mix is apparently more important than total nurse staffing level in predicting nursing turnover. (C) 2012 Elsevier Ltd. All rights reserved.