BackgroundGrowing evidence indicates that improved nurse staffing in acute hospitals is associated with lower hospital mortality. Current research is limited to studies using hospital level data or without proper adjustment for confounders which makes the translation to practice difficult.MethodIn this observational study we analysed retrospectively the control group of a stepped wedge randomised controlled trial concerning 14 medical and 14 surgical wards in seven Belgian hospitals. All patients admitted to these wards during the control period were included in this study. Pregnant patients or children below 17 years of age were excluded. In all patients, we collected age, crude ward mortality, unexpected death, cardiac arrest with Cardiopulmonary Resuscitation (CPR), and unplanned admission to the Intensive Care Unit (ICU). A composite mortality measure was constructed including unexpected death and death up to 72 h after cardiac arrest with CPR or unplanned ICU admission. Every 4 months we obtained, from 30 consecutive patient admissions across all wards, the Charlson comorbidity index. The amount of nursing hours per patient days (NHPPD) were calculated every day for 15 days, once every 4 months. Data were aggregated to the ward level resulting in 68 estimates across wards and time. Linear mixed models were used since they are most appropriate in case of clustered and repeated measures data.ResultsThe unexpected death rate was 1.80 per 1000 patients. Up to 0.76 per 1000 patients died after CPR and 0.62 per 1000 patients died after unplanned admission to the ICU. The mean composite mortality was 3.18 per 1000 patients. The mean NHPPD and proportion of nurse Bachelor hours were respectively 2.48 and 0.59. We found a negative association between the nursing hours per patient day and the composite mortality rate adjusted for possible confounders (B = − 2.771, p = 0.002). The proportion of nurse Bachelor hours was negatively correlated with the composite mortality rate in the same analysis (B = − 8.845, p = 0.023). Using the regression equation, we calculated theoretically optimal NHPPDs.ConclusionsThis study confirms the association between higher nurse staffing levels and lower patient mortality controlled for relevant confounders.

Our intervention had no significant effect on the incidence of unexpected death, cardiac arrest or unplanned ICU admission when adjusted for clustering and study time. We found a lower than expected baseline incidence of unexpected death and cardiac arrest rates which reduced the statistical power significantly in this study.

No abstract

Aims:To investigate the impact of the national early warning score on the frequency and the quality of vital sign registration and to study the association between protocol compliance and patient mortality. Design:We conducted a post hoc data analysis of a stepped wedge cluster randomized controlled trial (RCT) in six hospitals. Methods: All adult, non-pregnant patients admitted to 24 wards were included. The intervention comprised an observation protocol using the national early warning score combined with a pragmatic medical response strategy. Data collection lasted from October 2013-May 2015. Patient comorbidity scores and vital signs were sampled every 4 months on each ward. All vital signs in the 24 hr before a serious adverse event were collected. Results: Patients (N = 60,956) were included of which 32,722 in the intervention group. Comorbidity scores were sampled in 3,600 patients and vital signs in 2,951 patients. In 668 patients, vital signs were collected before a serious adverse event. The mean number of vital signs per observation increased significantly in the intervention group. The observation frequency increased in patients with a serious adverse event and decreased in patients without a serious adverse event. Protocol compliance was negatively associated with patient mortality adjusted for comorbidity and age. Conclusion: Our intervention improved patient monitoring practice and reduced mortality.Impact: The impact of early warning scores on patient monitoring practice and patient outcomes remains unclear. Our intervention improved the observation of patients and reduced patient mortality. These results could support hospitals in their decision to implement rapid response systems.Trial Registration: We have registered this study in the clinicaltrials.gov database (identifier: NCT01949025). K E Y W O R D Searly warning score, mortality, nurses, protocol compliance, rapid response system, vital sign | 1997 HAEGDORENS Et Al.