Patients admitted to the hospital with sepsis are eight times more likely to die than patients with other diagnoses. Sepsis is associated with an intense and persistent stress response that can become dysfunctional, resulting in disease, organ failure, and death. Allostasis theory has emerged as an influential theory in describing the biological response to stress, focusing on individual differences. There is no diagnostic test that clearly identifies the presence of the dysregulated host response that is central to sepsis.

Serum albumin is a protein produced by the liver that has been identified by researchers as a possible predictor of mortality in a number of critically ill patient populations. However, these studies primarily focus on the levels on admission, neglecting the clinically significant decrease that occurs subsequently.

The purpose of this retrospective, correlational study was to examine the relationship between the trend of serum albumin over time and mortality in adults admitted to the ICU at a Midwestern regional medical center with sepsis. Serum albumin trend, admission, average, maximum, and minimum albumin levels were evaluated for association and predictive ability to each of the outcomes (mortality, length of stay, ICU length of stay, ventilator days, progression to a state of chronic critical illness, vasopressor use, presence of ICU delirium, and readmission to the ICU).

Low serum albumin trend, low admission albumin level, and low minimum albumin level significantly predicted mortality while controlling for age. The combination of serum albumin trend and minimum albumin level significantly predicted length of stay (LOS). Mortality was a moderator of the relationship between serum albumin trend and LOS. The combination of serum albumin trend and minimum albumin level significantly predicted ICU LOS. Minimum albumin level was a significant predictor of ventilator days. Minimum albumin was identified as the best predictor of progression to a state of chronic critical illness while controlling for mortality. Minimum albumin was found to be the best predictor of vasopressor use while controlling for mortality. There were no significant predictors of ICU delirium, but there was a relationship between mortality and ICU delirium. Minimum albumin was the best predictor of ICU readmission.