Session: Nurse Staffing, Nurse Human Capital And The Quality Of Hospital Care
Room: Phillips 203
Time: Tue 10:15-11:45

Presenter: Ciaran Phibbs (Veterans Affairs Palo Alto and Stanford University. )

Discussant: Dennis Scanlon (Pennsylvania State University)

Abstract

Objective: There is growing evidence of the effect of nurse staffing levels on patient outcomes in acute-care hospitals and increased interest in mandating nurse staffing ratios, but previous work is limited by inadequate data and methods. This project extends current knowledge using monthly, unit-level data for a large panel of hospitals to estimate fixed-effects models to obtain more precise estimates of the effects of nurse staffing. It also expands previous research to consider the effects of general and specific human capital (use of contract labor, tenure on the unit, and team cohesiveness and stability).

Methods: Used: Monthly data from FY 03-06 for all VA acute inpatient care (266 general acute wards and 161 intensive care units [ICUs], 11,637 and 7,148 monthly observations, respectively). Inpatient utilization files were used to calculate length of stay (LOS) and nursing-sensitive patient outcomes from the Agency for Healthcare Research and Quality (AHRQ) Patient Safety Indicators (PSIs). Nurse staffing for each unit was obtained from the VA Decision Support System. The registered nurse (RN) human capital variables were extracted from VA payroll data. Data were aggregated by month for each inpatient unit. Fixed-effects models were used to control for unit-specific unobserved heterogeneity.

Results: Consistent with prior literature, there was only limited significance for the PSI regressions. For LOS, the results show adding one hour of nursing care per patient day, reducing non-RN nursing staff by 10 percentage points, reducing contract nurses by 10 percentage points, or increasing average unit tenure by one year all reduce LOS by about 3 percent for acute care. The effects for ICUs are smaller and not significant for reducing non-RN staff; but the ICUs used almost no non-RN staff. Using unit-level, monthly, panel data and fixed-effects models yields smaller estimated effects of nurse staffing on LOS than models using cross-sectional data and/or more aggregated data.

Conclusions: Nurse staffing does affect patient outcomes and LOS, but previous studies may have over-estimated these effects. Factors besides staffing levels, such as use of contract nurses and turnover, are also important determinants of patient outcomes. With their higher staffing levels, the effects of differences in nurse staffing are smaller in ICUs. In setting guidelines for nurse staffing, policy makers need to consider the characteristics of the nursing staff in addition to the number of nurses.

Key Terms

nurse staffing, patient outcomes, human capital

Authors:

Ciaran Phibbs (VA Palo Alto Health Care System and Stanford University) , Ann Bartel (Columbia University) and Patricia Stone (Columbia University)