Abstract / Summary
Background: Fluoroquinolone secondary prophylaxis of spontaneous bacterial peritonitis (SBP) was established in a fluoroquinolone-naïve era. We re-appraised the randomized evidence comparing rifaximin with norfloxacin and its robustness to missing outcome data and competing mortality. Methods: We searched MEDLINE, Embase, CENTRAL, and Scopus through 15 July 2026. The primary estimand was the intention-to-treat risk ratio (RR) for six-month SBP recurrence, with death as a competing event. Trials were pooled with REML random-effects models and Hartung–Knapp–Sidik–Jonkman (HKSJ) intervals; missing-outcome bounding across all trials and a composite of recurrence or death were post hoc. Certainty was rated with GRADE. The review was registered retrospectively in PROSPERO (CRD420261501120). Results: In three randomized trials (398 randomized patients; none double-blind), rifaximin was associated with fewer recurrences than norfloxacin (RR 0.21, 95% HKSJ CI 0.05–0.89), but the modified HKSJ interval crossed unity (0.03–1.31). Across plausible missing-outcome scenarios, the Mantel–Haenszel RR ranged from 0.15 to 0.73 (worst case 95% CI 0.45–1.17) and for recurrence or death from 0.35 to 0.75 (0.55–1.03). Certainty was very low. Conclusions: Rifaximin may reduce SBP recurrence compared with norfloxacin, but the evidence is very uncertain and sensitive to unreported outcomes; a contemporary, blinded trial is needed before guideline change.