Faced with a shortage of RNs, many hospitals rely on nursing assistants to provide some patient care. However, there is concern about the possible adverse effects of this practice on patient safety.

In a retrospective, longitudinal, observational study, researchers examined how variation in the size and composition of the nursing workforce may influence hospital mortality rates. They assessed staffing levels, which were measured as hours per patient per day for both RNs and nursing assistants, and identified days when admissions exceeded 125% of the mean for a given unit, reflecting a mismatch between demand and available staff. The hazard of death increased by 3% for each day a patient spent on a unit with RN staffing below the mean for that unit. When admissions per RN exceeded 125% of the mean for that unit, the hazard of death increased by 5%. The hazard of death decreased by 3% for every additional RN hour per patient during the first five days of a patient's stay. There was no significant association between additional nursing assistant hours and hazard of death.

In contrast to the linear association between RN staffing and risk of death, there was evidence that the risk of death increased when nursing assistant staffing was either above or below the mean. This may be because additional nursing assistant staffing creates a division of labor that results in RNs spending less time with patients, the authors suggest. They note that although nursing assistants play an important role in hospital safety, they shouldn't be viewed as substitutes for RNs. Based on their finding of an increased hazard of death as the number of new admissions increased, the researchers suggest that this could be a result of the increased RN workload associated with each newly admitted patient. Consequently, the authors call for more research into patient turnover, in addition to patient census and acuity, as a workload variable.

The researchers caution that this study was observational, and causal inference doesn't follow directly from observed associations.

REFERENCE