Plasma Albumin Changes During Admission for Bloodstream Infection as Prognostic Predictors After Discharge-A...

Area of Science:

• Clinical Medicine
• Biochemistry
• Epidemiology

Background:

• Plasma albumin (PA) levels can rapidly decline due to increased vascular loss, often associated with inflammatory conditions.
• A half-life of 18-20 days for PA means that changes in its levels can quickly reflect acute physiological events.

Purpose of the Study:

• To investigate the predictive value of plasma albumin (PA) levels and their changes on mortality in patients with bloodstream infection (BSI).
• To assess the impact of PA dynamics on mortality prediction models using Area Under the Receiver Operating Characteristic Curves (AUROCs).

Main Methods:

• A cohort of 11,562 adult patients with first-time BSI was analyzed, with at least two PA specimens during hospitalization.
• Cox regression models were used to evaluate PA levels (first, last, and changes) as predictors of mortality at various time points (1-30, 31-90, 91-365, >365 days).
• AUROCs were calculated to quantify the improvement in mortality prediction by incorporating PA dynamics into baseline models.

Main Results:

• The last PA specimen level and the change in PA levels between the first and last specimens were significant mortality predictors.
• Lower PA levels (<25 g/L) were associated with higher mortality (HR 3.69 for 1-30 days), while higher levels (≥35 g/L) were associated with lower mortality (HR 0.31).
• Incorporating PA changes improved the AUROCs for 1-30 day mortality from 0.72 to 0.79 and for 31-90 day mortality from 0.73 to 0.76.

Conclusions:

• The last recorded plasma albumin level during bloodstream infection admission is a potent predictor of mortality, particularly within the first 90 days.
• Improved mortality prediction was observed in younger, non-comorbid patients, and those with more dynamic changes in PA levels.
• These findings support the role of hypoalbuminemia primarily as a marker of acute inflammatory events.