Evaluating the Role of Antibiotics in Patients Admitted to Hospital With Decompensated Cirrhosis: Lessons From the ATTIRE Trial

Le résumé fourni par la source

INTRODUCTION: Hospital-acquired infections (HAI) are common in cirrhosis with antibiotics frequently used to prevent infections, but their efficacy for this role is unknown. To investigate this, we used Albumin to Prevent Infection in Chronic Liver Failure (ATTIRE) data to evaluate whether antibiotic use in patients without infection prevented HAI. METHODS: In ATTIRE patients without infection at baseline grouped by antibiotic prescription or not, we studied HAI during trial treatment period and mortality, with propensity score matching to account for differences in disease severity. RESULTS: Two hundred three of 408 patients prescribed antibiotics at enrollment did not have infection and they were more unwell than noninfected patients not given antibiotics. There were no differences in subsequent HAI comparing antibiotic treated (39/203, 19.2%) to nonantibiotic treated (73/360, 20.3%; P = 0.83). Twenty-eight-day mortality was higher in antibiotic-treated patients ( P = 0.004) likely reflecting increased disease severity. Matching groups using propensity scoring revealed no differences in HAI or mortality. In noninfected patients at enrollment treated with/without rifaximin, there were no differences in HAI ( P = 0.16) or mortality, confirmed with propensity matching. Patients given long-term antibiotic prophylaxis at discharge had no differences in 6-month mortality compared with nonantibiotic patients, although antibiotic-treated patients had more infections at trial entry, with numbers too small for matching. DISCUSSION: Half of antibiotics at study entry were given to patients without an infection diagnosis which did not reduce the overall risk of HAI or improve mortality. This supports prompt de-escalation or discontinuation of antibiotics guided by culture sensitivities at 24-48 hours after commencement if no infection and the patient is improving.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.