Introduction Severe alcoholic hepatitis (SAH; mDF ≥ 32) is often complicated by infection. In the recent trial STOPAH, prednisolone was associated with a doubling of the numbers of infections classified as serious adverse events (SAEs). We sought to identify the determinants of infection and the impact of infection on mortality.
Methods Data regarding infection were collected weekly during admission and at all trial visits; SAE reports were included. Incident infection was defined as new infection occurring after treatment start. The follow-up period considered was 120 days.
Results Data from 1093 cases were available. 127 patients were infected at baseline; this was not associated incident infection (p = 0.51) or 90 day mortality (p = 0.19). 308 patients developed an incident infection; 100 were reported as SAEs. As in other studies in SAH median time to infection was 13 days.
Age (p = 0.015), WHO performance status (PS; p = 0.007), WBC (p = 0.002), INR (p = 0.023) and creatinine (p = 0.02) at baseline were associated with incident infection. On multivariate analysis PS (p = 0.02) and WBC (p = 0.00016) remained associated. DF (p = 0.011) and GAHS (p = 0.003) but not MELD (p = 0.41) associated with incident infection.
Lille score was associated with incident infection after day 7 (p = 0.013). However, infection within 7 days was itself associated with Lille non-response (p = 0.008, OR 1.78 [95%CI 1.32–2.40]), independent of prednisolone treatment.
Although infections classified as SAEs were more common in the prednisolone treated group (p = 0.001, OR 2.51 (95%CI 1.46–4.30). There was no association between prednisolone and all incident infections occurring during treatment (p = 0.88), but prednisolone was associated with infection in the post-treatment period (p = 0.023, OR 1.92 [95% CI, 1.10–3.35]).
Incident infection was associated with 28 (p < 0.001) and 120 day mortality (p = 0.001). This was independent of prednisolone therapy and Lille response. Prednisolone-treated Lille non-responders had a higher rate of incident infection at day 120 (49% vs 27%; p < 0.0001).
Conclusion In SAH incident infection is associated with mortality independently of early improvement in liver function measured by Lille. Very early onset of infection is associated with classification as a Lille non-responder. Thus in some instances Lille non-response may represent untreated infection not steroid treatment-failure. An association between prednisolone and post-treatment infection may explain the catch-up mortality after 28 days and fleeting benefit of prednisolone seen in STOPAH.
Disclosure of Interest None Declared