While there is increasing evidence that establishes the positive relationship between the level of registered nurse (RN) staffing and patient outcomes in general acute care settings, there are still significant scientific knowledge gaps. For example, hospitals provide 24-hour, 7-day a week service and there is evidence that patients admitted on off-shifts (nights, weekends, and holidays) have worse outcomes. However, there is almost no research that has been conducted examining relationships between the characteristics of off-shift nurses and patient outcomes. Due to limitations in previous datasets, researchers have not been able to account for variation in patient needs at the unit-level, nor have they been able to consider other important workforce characteristics such as education and tenure. Furthermore, most nurse staffing studies have been cross-sectional, and the relationship between nurse staffing and patient costs needs additional study.
The purpose of this project is to examine how the differences in nurse staffing between regular and off-shifts affect patient outcomes, and to conduct a detailed examination of the trade-offs between the costs and benefits of increased nurse staffing. Specifically our aims are to: 1) Examine how the differences between regular shifts and off-shifts in nursing inputs (staffing levels, skill mix, contract nurses, and general, facility-specific, unit/team-specific human capital) affect nursing-sensitive patient outcomes; and 2) Analyze efficiency in providing acute care services by studying the trade-offs between nursing personnel costs associated with different staffing levels and nursing characteristics and the cost savings related to improved care.
Monthly data for all VA acute care units from FY 2003-2006 were extracted. After exclusions for units that were too small and units with incomplete data, there were 185 units from 138 VA medical centers (8,243 unit-month observations). Monthly staffing for each unit, for each type of nurse (RN, LPN, aides, contract nurses), were obtained from VA accounting data (DSS). Shift differentials from payroll data (PAID) were used to identify which shifts each nurse worked. Payroll data also provided education levels and how long each nurse had worked on the unit (unit tenure). Patient characteristics and length of stay (LOS) were obtained from VA hospital discharge records (PTF). DSS IPD data were used to link patients to units. All data were aggregated to the unit-month level.
While nursing-sensitive patient safety indicators were examined, LOS was used at the outcomes measure as all of the nursing-sensitive patient safety indicators are associated with increased LOS. Using LOS captures the combined effects of the adverse outcomes, plus those of extended stays due to delayed nursing care (e.g., patient discharge teaching).
Fixed-effects regression was used to control for unobserved heterogeneity (unobserved unit characteristics that affect outcomes; for example, quality of nursing leadership or work environment). LOS was the dependent variable, and the model controlled for patient age, expected LOS, patient co-morbidities, nurse staffing, nurse skill-mix, unit tenure, and how these nursing variables differed on the off-shifts.
Initial analysis didn't detect a weekend effect, so all analyses focused on the differences between day and night shifts (defined as those where a shift differential was paid). Staffing, as measured by nursing hours per patient day (HPPD) were higher on day shift (4.3 vs. 3.4) and the composition was fairly similar; Use of LPNs was the same, with 4% more RNs and 4% fewer Aides on nights. Day RNs were slightly more likely to have a Baccalaureate degree and the average unit tenure was higher for night RNs (5.0 vs 4.1 years).
Fixed-effects models were necessary to control for the unobserved heterogeneity, but fixed-effects results must be interpreted carefully. These models essentially use each unit as its own control, and the effects that they capture are the marginal effects when a variables deviates from each unit's mean. They do NOT specifically compare the effects of different units with higher, compared to lower staffing levels. The results show that a 1 hour increase in HPPD, compared to the unit mean, is associated with a 3% reduction in LOS and that a 10% increase in the share of nursing staff that are Aides is associated with a 2% increase in LOS. Compared to the average on day shifts, each one hour increase in the night shift HPPD is associated with a 2% decrease in LOS, and each 10% increase in the share of the nursing staff that are Aides is associated with a 2% increase in LOS. Each one year increase in the average RN unit tenure is associated with a 1% decrease in LOS, as is each one year decrease in the night shift tenure, compared to days. In summary, both better nurse staffing, and reducing the night shift deficits in staffing, compared to day shifts, are associated with shorter patient LOS. The cost models show that marginal improvements in night shift nurse staffing (both staffing levels and unit tenure) are associated with small reductions in total costs.
The results will provide data that can help nursing managers make the case for expanded nursing budgets, as prevented adverse events associated with better staffing pay for the increased staffing. They also point to the value of retaining senior nurses and reducing turnover.
- de Cordova PB, Phibbs CS, Schmitt SK, Stone PW. Night and day in the VA: associations between night shift staffing, nurse workforce characteristics, and length of stay. Research in nursing & health. 2014 Apr 1; 37(2):90-7.
- de Cordova PB, Phibbs CS, Stone PW. Perceptions and observations of off-shift nursing. Journal of nursing management. 2013 Mar 1; 21(2):283-92.
- de Cordova PB, Phibbs CS, Bartel AP, Stone PW. Twenty-four/seven: a mixed-method systematic review of the off-shift literature. Journal of Advanced Nursing. 2012 Jul 1; 68(7):1454-68.
- Phibbs CS, Bartel A, de Cordova P, Needleman J, Schmitt S, Stone P. Measuring the Impact of Nurse Staffing on Patient Outcomes: The Effect of Data Aggregation and Estimation Models. Poster session presented at: AcademyHealth Annual Research Meeting; 2014 Jun 8; San Diego, CA.
- de Cordova PB, Phibbs CS, Bartel AP, Stone PW. Does the Difference Between Regular and Off-Shift Nursing Staffing Affect Patient Outcomes? Longitudinal, Unit-Specific Evidence from VA. Poster session presented at: VA HSR&D / QUERI National Meeting; 2012 Jul 16; National Harbor, MD.
- de Cordova PB, Phibbs CS, Bartel AP, Stone PW. Twenty-Four/Seven: Off-shift Nursing and Quality Care. Poster session presented at: AcademyHealth Annual Research Meeting; 2012 Jun 24; Orlando, FL.
- de Cordova PB, Bartel AP, Phibbs CS, Stone PW. Does the Difference Between Regular and Off-Shift Nursing Staff Affect Patient Outcomes? Longitudinal, Unit-Specific Evidence from VA. Presented at: American Society of Health Economists Annual Conference; 2012 Jun 13; Minneapolis, MN.
- de Cordova PB, Phibbs CS, Stone PW. Off-Shift Nursing and Quality Patient Outcomes: A Sub-Analysis of the INQRI Project Titled The Impacts of Nurse Staffing, Skill Mix, and Experience on Quality and Costs in Long-Term Care. Presented at: RWJ Foundation Interdisciplinary Nursing Quality Research Initiative Annual National Conference; 2012 Apr 26; Washington, DC.