The impact of nurse staffing levels and nurse’s education on patient mortality in medical and surgical wards: an observational multicentre study

Table 3 Linear mixed model estimates showing the effect of the mean nursing hours per patient day on study outcome measures

Dependent variables	Model 1				Model 2
	Estimate	95% CI of est.		p-value	Estimate	95% CI of est.		p-value	Δ AIC	Δ BIC
	Estimate	Lower	Upper	p-value	Estimate	Lower	Upper	p-value	Δ AIC	Δ BIC
Crude mortality	−3.043	−10.670	4.584	0.426	−2.458	− 10.444	5.528	0.538	−33.9	−34.1
Unexpected death	−0.848	−2.316	0.621	0.252	−1.737	−3.013	− 0.460	0.009	−32.1	− 32.4
CPR	−0.870	−2.020	0.280	0.135	−1.271	−2.454	−0.087	0.036	−15.3	−15.5
Death after CPR	−0.631	−1.363	0.100	0.089	−0.836	−1.576	−0.097	0.028	−12.4	− 12.7
Unplanned ICU admission	−0.201	−3.399	2.997	0.900	−0.063	−3.378	3.253	0.970	−30.5	−24.8
Death after unplanned ICU admission	−0.085	−0.739	0.569	0.791	0.047	−0.754	0.847	0.907	−3.2	−3.5
combined mortality rate	−1.525	−3.544	0.493	0.136	−2.771	− 4.471	−1.071	0.002	−33.9	−34.1
Hospital LOS	−0.172	− 0.955	0.612	0.661	−0.023	− 0.836	0.791	0.955	−11.5	− 11.8

Estimates of fixed effects, 95% confidence intervals and p-values resulting from two linear mixed models concerning the relation between the mean Nursing Hours Per Patient Day and the patient outcome measures in column one (dependent variables)
Model 1: adjusted for clustering (random effect) and study time (fixed effect)
Model 2: adjusted for clustering (random effect), study time, the proportion of nurse Bachelor hours, nurse assistant HPPD, logistic support staff HPPD, Charlson Comorbidity Index and age (fixed effects)
Δ AIC Akaike’s Information Criterion of model 2 - model 1, Δ BIC Schwarz’s Bayesian Information Criterion of model 2 - model 1
CPR Cardiopulmonary Resuscitation, ICU Intensive Care Unit, LOS Length of Stay (in mean days)

ISSN: 1472-6963