The prognostic value of serial measurements of serum albumin concentration in patients admitted to an intensive care unit.

The prognostic value of serial measurements of serum albumin concentration during the first 72 h after admission to a general adult intensive care unit was retrospectively reviewed in 348 consecutive critically ill patients over a one year period. The accuracy of the admission APACHE II (Acute Physiology And Chronic Health Evaluation) score in correctly predicting patient outcome was compared with the serum albumin concentration measured at different times after intensive care unit admission. Multiple logistical regression analyses were performed to evaluate whether combining APACHE II and serum albumin into a unified risk index improved prognostic accuracy. Serum albumin concentration on admission was lower in non-survivors than in survivors and decreased more rapidly in non-survivors (p < 0.001). The admission serum albumin concentration was found to be an insensitive prognostic indicator. However, serum albumin measured after 24 h was as accurate as the admission APACHE II score in correctly classifying patients according to outcome. There was a good correlation between the admission APACHE II score and serum albumin measured after 24 h but not between the admission APACHE II and the admission serum albumin. Combining the APACHE II score and serial albumin concentrations into a unified risk of death equation did not improve the accuracy of outcome prediction.