Background: Fatty liver disease is potentially a reversible condition that may lead to end-stage liver disease. Since herbal medicines such as Crataegus pinnatifida and Salvia miltiorrhiza have increasingly been used in the management of fatty liver disease, a systematic review on herbal medicine for fatty liver disease is needed.
Objectives: To assess the beneficial and harmful effects of herbal medicines for people with alcoholic or non-alcoholic fatty liver disease.
Search methods: We searched The Cochrane Hepato-Biliary Group Controlled Trials Register, the Cochrane Central Register of Controlled Trials (CENTRAL) (Issue 3, 2012), MEDLINE, EMBASE, and Science Citation Index Expanded to 1 March 2012. We also searched the Chinese BioMedical Database, Traditional Chinese Medical Literature Analysis and Retrieval System, China National Knowledge Infrastructure, Chinese VIP Information, Chinese Academic Conference Papers Database and Chinese Dissertation Database, and the Allied and Complementary Medicine Database to 2 March 2012.
Selection criteria: We included randomised clinical trials comparing herbal medicines with placebo, no treatment, a pharmacological intervention, or a non-pharmacological intervention such as diet or lifestyle, or Western interventions in participants with fatty liver disease.
Data collection and analysis: Two review authors extracted data independently. We used the 'risk of bias' tool to assess the risk of bias of the included trials. We assessed the following domains: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other sources of bias. We presented the effects estimates as risk ratios (RR) with 95% confidence intervals (CI) or as mean differences (MD) with 95% CI, depending on the variables of the outcome measures.
Main results: We included 77 randomised clinical trials, which included 6753 participants with fatty liver disease. The risks of bias (overestimation of benefits and underestimation of harms) was high in all trials. The mean sample size was 88 participants (ranging from 40 to 200 participants) per trial. Seventy-five different herbal medicine products were tested. Herbal medicines tested in the randomised trials included single-herb products (Gynostemma pentaphyllum, Panax notoginseng, and Prunus armeniaca), proprietary herbal medicines commercially available, and combination formulas prescribed by practitioners. The most commonly used herbs were Crataegus pinnatifida,Salvia miltiorrhiza,Alisma orientalis,Bupleurum Chinense,Cassia obtusifolia, Astragalus membranaceous, and Rheum palmatum. None of the trials reported death, hepatic-related morbidity, quality of life, or costs. A large number of trials reported positive effects on putative surrogate outcomes such as serum aspartate aminotransferase, alanine aminotransferase, glutamyltransferase, alkaline phosphatases, ultrasound, and computed tomography scan. Twenty-seven trials reported adverse effects and found no significant difference between herbal medicines versus control. However, the risk of bias of the included trials was high.The outcomes were ultrasound findings in 22 trials, liver computed tomography findings in eight trials, aspartate aminotransferase levels in 64 trials, alanine aminotransferase activity in 77 trials, and glutamyltransferase activities in 44 trials. Six herbal medicines showed statistically significant beneficial effects on ultrasound, four on liver computed tomography, 42 on aspartate aminotransferase activity, 49 on alanine aminotransferase activity, three on alkaline phosphatases activity, and 32 on glutamyltransferase activity compared with control interventions.
Authors' conclusions: Some herbal medicines seemed to have positive effects on aspartate aminotransferase, alanine aminotransferase, ultrasound, and computed tomography. We found no significant difference on adverse effects between herbal medicine and control groups. The findings are not conclusive due to the high risk of bias of the included trials and the limited number of trials testing individual herbal medicines. Accordingly, there is also high risk of random errors.