RESUMO
BACKGROUND: Spontaneous bacterial peritonitis (SBP) is a common infection in patients with cirrhosis and ascites. Currently, the accuracy of the model for end-stage liver disease (MELD) and MELD-sodium (MELD-Na) as prognostic scores in this cohort is unclear. This study aimed to evaluate and compare the accuracy of MELD and MELD-Na for predicting 90-day mortality and determine whether the mortality risk estimates they provide accurately reflect the poor prognosis of patients with SBP Methods: Patients with cirrhosis and SBP were retrospectively identified from ascitic fluid samples sent for microscopy, culture and sensitivity analysis (1/1/18-31/12/20) and a previous audit. MELD and MELD-Na scores at diagnosis were calculated and associations with 90-day mortality were assessed using univariate analysis. Receiver operator characteristic curves were compared, and standardised mortality ratios (SMRs) were calculated by comparing the number of deaths observed to the number predicted by MELD and MELD-Na. RESULTS: Of the 567 patients identified, 15 patients with cirrhosis and SBP were included. The 90-day mortality rate was 66.7% (10/15). Only concurrent hyponatremia (<135mmol/L) was associated with mortality (6/10 non-survivors vs 0/5 survivors, p=0.04). The difference in MELD and MELD-Na's C-statistic was not significant: 0.66 (95% Cl:0.35-0.98) vs 0.74 (95% Cl:0.47-1.0) respectively (p=0.72). Patients with a MELD-Na >18.5 had significantly higher 90-day mortality than patients with MELD-Na ≤18.5 (88.9% (8/9) vs. 33.3% (2/6), p=0.05). The SMR (95% Cl) for each MELD decile evaluated was 33.3 (0-79.5), 11.1 (0.2-22.0) and 3.4 (0-7.0) for scores ≤9,10-19 and 20-29 respectively. For each MELD-Na tertile, these were: 25 (0-59.6), 5.2 (0.1-10.3) and 2.7 (0.1-8.1) for scores <17,17-26, ≥27 respectively. CONCLUSION: In a small cohort of patients with cirrhosis and SBP, the MELD's accuracy in predicting 90-day mortality was limited. MELD-Na's accuracy was higher but not significantly. Both scores consistently underestimated participants' mortality, therefore future studies could evaluate the accuracy of alternative prognostic scores in this patient group.
