Serum albumin as an outcome predictor in hospital emergency medical admissions

BackgroundTo examine the relationship between admission serum albumin and 30-day mortality during an emergency medical admission.MethodsAn analysis was performed of all emergency medical patients admitted to St. James's Hospital (SJH), Dublin between 1st January 2002 and 31st December 2008, using the hospital in-patient enquiry (HIPE) system, linked to the patient administration system, and laboratory datasets. Mortality was defined as an in-hospital death within 30 days. Logistic regression was used to calculate unadjusted and adjusted odds ratios (ORs) and 95% confidence intervals for defined albumin subsets.FindingsUnivariate analysis using predefined criteria based on distribution, identified the groups of < 10% and between 10 and 25% of the serum albumin frequency distribution as at increased mortality risk. Their mortality rates were 31.7% and 15.4% respectively; their unadjusted odds rates were 6.35 (5.68, 7.09) and 2.11 (1.90, 2.34). Patients in the lowest 25% of the distribution had a 30-day mortality of 19.9% and this significantly increased risk persisted, after adjustment for other outcome predictors including co-morbidity and illness severity (OR 2.95 (2.49, 3.48): p < 0.0001).InterpretationSerum albumin is predictive of 30-day mortality in emergency medical patients; mortality is non-linearly related to baseline albumin. The disproportionate increased death risk for patients in the lowest 25% of the frequency distribution (< 36 g/L) is not due to co-morbidity factors or acute illness severity.