Talk to the Veterans Crisis Line now
U.S. flag
An official website of the United States government

VA Health Systems Research

Go to the VA ORD website
Go to the QUERI website
HSRD Conference Logo

2023 HSR&D/QUERI National Conference Abstract

Printable View

4097 — Estimating the Non-linear Relationship Between Nurse Staffing and Workload Using National Barcode Medication Administration Data

Lead/Presenter: Chase Eck,  COIN - Houston
All Authors: Eck CE (Center for Innovations in Quality, Effectiveness, and Safety (IQuESt), Houston), Knox, MK (Center for Innovations in Quality, Effectiveness, and Safety (IQuESt), Houston) Mehta, P (University of Houston, Houston) Petersen, LA (Center for Innovations in Quality, Effectiveness, and Safety (IQuESt), Houston)

Prior work has linked inpatient nurse staffing and patient mortality and outcomes. Less is known about the relationship between staffing and nursing care processes, which could elucidate a possible mechanism for the link between staffing and outcomes. To address this gap, we estimated the relationship between patient per staff (PPS) and workload, as measured by the duration of peak medication pass (MEDPASS). MEDPASS is a high workload period in which the nursing staff administers medications to nearly every patient, serving as a promising marker of nursing workload and practice patterns. MEDPASS duration was measured using bar code medication administration (BCMA) data that is now available at 98% of U.S. hospitals.

We identified the most frequent, or peak, time for scheduled medications on each unit using BCMA data for inpatient medical/surgical units in the Veterans Health Administration (VHA), the largest integrated health care system in the U.S. We measured PPS as the number of patients present at MEDPASS divided by the number of staff present at MEDPASS. We measured workload as the average time between first and last medication administered for each staff member on the unit. We model the non-linear relationship between PPS and MEDPASS duration using a generalized additive model (GAM) and estimate it using penalized regression splines. We adjust for patient illness burden derived from the NOSOS Relative Risk Score (RRS) and the total number of medications administered. We also adjust for whether it is a weekday, fixed unit-level factors by including unit fixed effects, and seasonality by including year-week fixed effects.

All Veterans (n = 309,204) admitted to 113 inpatient medical units in 60 VHA hospitals in calendar year 2019 (38,704 unit-days). On average, there were 2.93 PPS at MEDPASS and each staff member spent approximately 62 minutes on MEDPASS. Average time spent on MEDPASS per staff was positively related to PPS. However, this effect was smaller at higher levels of PPS. For example, a change from 2 PPS to 3 PPS was associated with the average staff member spending 15 more minutes on MEDPASS, but a change from 3 PPS to 4 PPS was associated with spending 13 more minutes on MEDPASS.

Nurse staffing levels are associated with time spent on critical nurse processes of care, including MEDPASS. While the average workload per nurse is increasing with PPS, this relationship is non-linear—it diminishes as PPS increases. These non-linearities may reflect efficiencies of scale or factors not captured by the control variables. These results suggest that time spent on nursing processes may be a possible mechanism for the effect of staffing on patient outcomes.

Our work shows that nurse staffing decisions directly relate to how nursing care is delivered. The diminishing relationship suggests that the effectiveness of staffing-based interventions may vary based on the existing level of staffing. Although these findings address the relationship between staffing changes and workload during medication administration, more work is needed to examine how nurse staffing affects workload for other parts of nursing practice, including non-medication related tasks.