Serum Albumin Levels as a Predictor of Mortality in Patients with Sepsis: A Multicenter Study.

ABSTRACT

BACKGROUND: Low serum albumin is known to be associated with mortality in sepsis, as it reflects effects of nutrition, catabolism, and edema. OBJECTIVES: To examine the association of albumin levels with in-hospital mortality in adults with sepsis, stratified by age groups. METHODS: This nationwide retrospective cohort study comprised patients admitted with sepsis to intensive care units in seven tertiary hospitals during 2003-2011. Only patients with available serum albumin levels at hospital admission and one week after were included. Patients with an intra-abdominal source of sepsis were excluded. The association between sepsis and mortality was analyzed using multivariate logistic regression models. RESULTS: The study included 3967 patients (58.7% male, median age 69 years). Mean serum albumin levels were 3.1 ± 0.7 g/dl at admission and 2.4 ± 0.6 g/dl one week later. In a multivariate logistic regression model, serum albumin one week after admission was inversely associated with in-hospital mortality (odds ratio [OR] 0.64, 95% confidence interval 0.55-0.73 per 1 g/dl). In an age-stratified analysis, the association was stronger with younger age (OR 0.44 for patients aged < 45 years, 0.60 for patients aged 45-65 years, and 0.67 for patients aged > 65 years). Serum albumin on admission was not associated with in-hospital mortality. CONCLUSIONS: The decline in serum albumin one week after admission is a stronger predictor of mortality in younger patients. Older patients might have other reasons for low serum albumin, which reflect chronic co-morbidity rather than acuity of disease.