- Poster presentation
- Open Access
Nutritional markers as a surrogate of severity of illness: association with ICU length of stay and mortality
© BioMed Central Ltd 2006
- Published: 21 March 2006
- Linear Regression Model
- Enteral Nutrition
- Multiple Linear Regression Model
- Nitrogen Balance
- Multiple Logistic Regression Modelling
Nutritional status has been associated with several outcomes in critically ill patients, including prolonged ventilatory dependence, increased infection rate, prolonged ICU length of stay (ICU LOS) and mortality.
To study the influence of several markers of nutritional status on ICU LOS and mortality.
An observational study regarding several nutritional markers (serum albumin, nitrogen balance and caloric deficit) routinely recorded on our ICU and their association with ICU LOS and mortality. Appropriate descriptive statistics are presented. Comparisons between groups were performed using the chi-square test, Student's t test or the nonparametric Mann-Whitney U test or Kruskal-Wallis test as appropriate. Analysis of the association between nutritional markers and mortality, estimated by the OR, unadjusted and adjusted for sex and SAPS II, were calculated using logistic regression. Multiple logistic regression modelling was performed, using the stepwise forward method, with mortality as the dependent variable and age, sex, nutritional markers and ICU variables (reason for ICU admission, SAPS II, ICU LOS) as independent variables. For the analysis of ICU LOS, and because of the right skewness of its distribution, a logarithmic transformation of this variable was used. Simple and multiple linear regression models were then used to study the association between nutritional and other factors and the ICU LOS, with the log-transformation of ICU LOS as the dependent variable. For hypothesis testing a value of P < 0.05 was considered significant. Statistical analysis was performed using SPSS 13.0® software package.
Ninety-one patients were included, of which 53 were male; mean age was 62 years. The reason for admission was medical in 68 patients, surgical in 16 patients and trauma in seven patients. The mean SAPS II was 45 and the median ICU LOS was 11 days. Patients surviving, patients with serum albumin lower than 2.5 g/dl, and those with a caloric deficit higher than -500 kcal/day exhibit a significantly shorter ICU LOS. In the linear regression model, however, these variables were no longer significantly associated with ICU LOS. Severity of illness (SAPS II) (OR 1.052; 95% CI 1.014; 1.091), nitrogen balance (OR 1.115; 95% CI 1.013; 1.228) and the need for parenteral nutrition plus enteral nutrition (OR 6.255; 95% CI 1.277; 30.645), were significantly associated with mortality in the logistic regression model. Caloric deficit was also significantly associated with mortality after adjustment for sex and SAPS II.
Nutrition markers were associated with a longer ICU stay and a higher mortality. This hypothesis should be evaluated with more appropriate methodological approaches.