Abstract

Background: Albumin reduces mortality in spontaneous bacterial peritonitis (SBP) and is recommended by national guidelines. However, gaps and potential disparities in standard of care are not well-studied in large cohorts. We investigated temporal trends, center-level variation, and racial disparities in guideline-recommended albumin use for patients hospitalized with SBP in a large national cohort of Veterans with cirrhosis (VOCAL).

Methods: This was a retrospective cohort study of SBP hospitalizations from 2009-2021. SBP admission diagnosis was manually validated and defined as: an inpatient SBP diagnosis code in the presence of ≥1 paracentesis and hospital stay >2 days (positive predictive value 91%). Demographic and clinical variables including comorbidities, laboratory data, and liver disease severity were abstracted. Albumin administration was collected from inpatient (BCMA) tables; hospitalizations were classified as any albumin, guideline-recommended albumin (day 1+2 or day 1+3), no albumin. Temporal trends were evaluated using linear regression. Center-level variation was explored among the 92 centers with at least 10 SBP hospitalizations. Stratified analysis was performed by white vs. non-white race.

Results:

We identified 3742 SBP hospitalizations; 35% had Child Turcotte Pugh (CTP)-A cirrhosis, 57% CTP-B, and 8.6% CTP-C. The median MELD-Na on admission was 18 (IQR 23-33). Only 988 (26%) received any albumin; of whom 588 (60%) received guideline-recommended albumin. The proportion of patients receiving any albumin increased over time (beta 0.009, p=0.003; Figure 1A); guideline-recommended albumin use numerically increased but was not statistically significant (beta 0.004, p=0.12). There was substantial center-level variability in albumin administration (Figure 1B); 36 centers (39%) did not administer any albumin for patients hospitalized with SBP. Finally, we identified racial differences in albumin use (Figure 1C/D); white patients had a 7% higher probability of receiving any albumin versus non-white patients (beta 0.070, p=0.005).

Conclusion: A large proportion of VA patients hospitalized with SBP did not receive albumin. While utilization has increased over time, there is substantial center-level variability. Furthermore, white patients were consistently more likely than non-white patients to receive albumin. Additional research is needed to elucidate drivers of these observed disparities, and quality improvement efforts such as clinical decision support should prompt albumin use once SBP is diagnosed.