Article révisé par les pairs
Résumé : Objective: As the assessment of nutritional status using plasma albumin level (ALB) can be confounded by the infusion of albumin solutions, the predictive value of ALB on the mortality rate in the intensive care unit (ICU) may vary before (period PRE) and after (period POST) a restriction of the use of albumin solutions (July 2012). Material and methods: Demographic, anthropometric, clinical and therapeutic data, including the proportion of patients who received albumin were recorded from consecutive patients who stayed at least 3. days in the medico-surgical ICU of the Erasme university hospital. The predictive value of mean ALB was compared between the two periods. Results: The two groups (PRE [. n= 118] and POST [. n= 104]) were similar for age, gender, severity of illness and BMI. A significantly higher proportion of patients were treated with albumin in the group PRE (14.4%) than in the group POST (2.9%, P<. 0.01). The mean ALB value was stable over the first 3. days in the PRE group (from 3.0 ± 0.8 to 3.0 ± 1.8. g/dL) but decreased in the POST group (from 3.1 ± 0.9 to 2.7 ± 0.8. g/dL, P<. 0.05). The predictive value of ALB for mortality was lower in the group PRE (0.52, 95% confidence interval [CI] 0.38-0.66) than in the group POST (0.60, 95% CI 0.47-0.73, P< 0.05). Conclusion: The predictive value for mortality in the ICU of ALB is low, and the infusion of albumin further decreases this value. Severity scores or nutritional assessment using ALB should be adjusted for the infusion of albumin.