ICU nursing workload is not related to hospital survival

Introduction

A previous study in a small (six beds) ICU in the United Kingdom indicated that high nursing workload is associated with increased hospital mortality [1]. However the multiple logistic model fitted the data poorly and the use of APACHE II to assess severity of illness would have introduced confounding.

Method

We performed a retrospective analysis of prospectively collected data. The patient cohort consisted of all adult emergency admissions to our 22 bed ICU between 0800 h on December 1^st, 1997 and 0800 h on July 1^st 1998 except in-patients transferred from other hospitals directly to ICU. The following data were collected for all patients in the cohort: time between hospital and ICU admission, the data required to calculate predicted mortality based on the Mortality Prediction Model II₀ (MPM II₀) and hospital mortality. Over the same time period nursing dependency scores (=number of nurses required, according to UK Intensive Care Society recommendations [2]) and numbers of nurses on duty were recorded for every shift. From this the average and peak nursing dependency:nurses on duty ratios during the time each patient was in the ICU were calculated. Logistic regression analysis was performed using hospital mortality as the dependent variable and age, sex, MPM II₀, ICU admission on the day of hospital admission, average nursing dependency:nurses on duty ratio, peak nursing dependency:nurses on duty ratio and interactions between the latter two and MPM II₀ as potential independent variables.

Results

The cohort consisted of 388 patients. Twenty-seven patients were excluded from the analysis because of incomplete data. Independent variables in the final logistic regression model and their coefficients are given in Table 1. The Hosmer and Lemeshow goodness of fit test indicated that the model fitted the data well (?² = 4.3, P = 0.83).

Table 1

Conclusion

Our data suggest that in a large ICU age, sex and MPM II₀ but not ratio of nursing workload to nurses on duty are related to outcome. The difference between our findings and those of Tarnow-Mordi et al [1] may reflect differences in size and organization of the respective units.