Fifteen RCTs were included (n=796, range eight to 126); there was some discrepancy in patient numbers between the tables and the text. Allocation concealment was adequate in nine studies. Follow-up was adequate in 13 studies. Ten studies were double-blind and 10 undertook intention-to-treat analysis.
Normalisation of alanine transaminase (11 studies):
A significant effect on normalization of alanine transaminase in patients was found for those treated with: metformin compared with placebo (RR 2.00, 95% CI 1.23 to 3.24; two studies) or with vitamin E (RR 3.14, 95% CI 1.16 to 8.47; one study); and high-dose (3g) carnitine versus diet (RR 5.75, 95% CI 1.45 to 22.74; one study). There was no significant difference in the proportion of patients with a normal ALT level in those treated with vitamin E compared with placebo (three studies). There was no evidence for statistical heterogeneity.
Normalisation of aspartate aminotransferase (four studies):
Aspartate aminotransferase normalisation was significantly higher in those treated with UDCA combined with vitamin E versus UDCA alone (RR 1.72, 95% CI 1.04 to 2.84; one study) or placebo (RR 2.98, 95% CI 1.30 to 6.85; one study) and in those treated with metformin compared with no treatment (RR 2.81, 95% CI 1.16 to 6.82; one study).
Biopsy specimens (seven studies):
Seven RCTs assessed biopsy specimens, but most had methodological limitations. Pioglitazone was associated with reduced liver necrosis and inflammation (one study).
There was no significant difference for the frequency of adverse effects between the groups.