Nonrespiratory predictor of mechanical ventilation dependency in intensive care unit patients.

OBJECTIVE: To determine the role of serum albumin concentration as a predictor of mechanical ventilation dependency.

DESIGN: Prospective, observation trial.

SETTING: Multidisciplinary intensive care unit (ICU) in a university hospital.

PATIENTS: One hundred forty-five consecutive patients who required mechanical ventilation for > 72 hrs.

INTERVENTIONS: Patients were classified into five different groups based on the cause of respiratory failure. The following parameters were recorded daily: serum albumin concentration; Acute Physiology and Chronic Health Evaluation II (APACHE II) score; and fluid balance. Using multiple regression, multiple logistic regression analysis, and the Anderson-Gill proportional hazards model, we determined the metabolic factors that could help predict weaning success.

MEASUREMENTS AND MAIN RESULTS: The mean length of ICU stay was 12.3 +/- 1.0 days. The duration of mechanical ventilation dependency was 10.5 +/- 1.0 days. The initial mean serum albumin concentration was 25.2 +/- 0.6 g/L. The APACHE II score on the first day of ICU stay was 19.1 +/- 0.6. Although albumin concentration was significantly lower and the APACHE II score was significantly higher in ICU nonsurvivors than in ICU survivors, albumin concentration on ICU admission was not a predictor of the length of time spent receiving mechanical ventilation. The profile of albumin concentration changes was different between weaned and mechanical ventilation-dependent patients. At the time of weaning patients from the ventilator, the median albumin concentration was higher than in those patients who continued to be supported by mechanical ventilation. This effect of albumin could not be attributed to patient fluid balance or to the severity of illness since each factor had an independent influence in predicting weaning, using the Anderson-Gill proportional hazards models.

CONCLUSIONS: Initial serum albumin concentration did not necessarily predict weaning success. However, when serum albumin concentration was assessed on a daily basis, its trend was important in determining the relative chance of being successfully weaned from the ventilator. This finding suggests that albumin may be an index of the metabolic status of the patient, which could be important in determining the weanability of the patients who are mechanically ventilated for prolonged periods of time.