Do corticosteroids reduce mortality from alcoholic hepatitis? A meta-analysis of the randomized trials.

PURPOSE

To determine whether corticosteroids affect short-term mortality from alcoholic hepatitis.

DATA IDENTIFICATION

Studies published from 1966 to 1989 were identified through a MEDLINE computer search and an extensive manual search of the bibliographies of identified articles.

STUDY SELECTION

We found 11 randomized studies (10 of which were placebo controlled) that assessed mortality in hospitalized patients diagnosed with acute alcoholic hepatitis and treated with corticosteroids.

DATA EXTRACTION

Two critical appraisers independently evaluated trial quality and abstracted quantitative data on clinical characteristics of the populations, interventions, and all-cause mortality.

RESULTS OF DATA SYNTHESIS

Overall, the protective efficacy (or percent reduction in mortality) of corticosteroids was 37% (95% CI, 20% to 50%). Protective efficacy was higher among trials with higher quality scores and trials that excluded subjects with active gastrointestinal bleeding. In subjects with hepatic encephalopathy, protective efficacy was 34% overall (CI, 15% to 48%). It was also higher among trials with higher quality scores and trials excluding subjects with acute gastrointestinal bleeding, but was not present among trials with lower quality scores or trials that did not exclude subjects with acute gastrointestinal bleeding. In subjects without hepatic encephalopathy, corticosteroids had no protective effect, and this lack of efficacy was consistent across all trial subgroups.

CONCLUSIONS

These results suggest that corticosteroids reduce short-term mortality in patients with acute alcoholic hepatitis who have hepatic encephalopathy, that the protective effect depends on the exclusion criterion of acute gastrointestinal bleeding and is influenced by trial quality, and that corticosteroids are of no benefit in patients without hepatic encephalopathy.