Abstract
In this study we aimed to examine the association between serum albumin concentration and mortality risk in critically ill patients.
We retrospectively studied 1003 patients admitted to our Intensive Care Unit (ICU) over an 18-month period. Serial albumin measurements over 72 hours were compared between survivors and non-survivors, and medical and surgical patients were also compared.
Our results showed that serum albumin decreased after ICU admission, most rapidly in the first 24 hours, in both survivors and non-survivors. Serum albumin was lower in non-survivors than in survivors, but albumin concentrations poorly differentiated the two groups. Medical patients had higher admission albumin levels than surgical patients, but both subgroups showed a similar albumin profile over 72 hours. We evaluated the prognostic value of serum albumin using receiver operator characteristic (ROC) curves. We constructed ROC curves for APACHE II score, admission albumin, albumin at 24 and 48 hours. We also combined APACHE II with albumin values and constructed the corresponding ROC curves. Our data showed that serum albumin had low sensitivity and specificity for predicting hospital mortality. Combining APACHE II score with serum albumin concentrations did not improve the accuracy of outcome prediction over that of APACHE II alone.