Huang 2004.

Methods	Randomised clinical trial, China Parallel group design (four‐arms)
Participants	73 participants diagnosed with chronic hepatitis B according to Chinese guidelines 2000 (CMA 2001), with HBeAg positive, HBV‐DNA positive, and ALT > 100 U. Male:female: 58:15 Mean age: 35.7 years (experimental group), 41.2 years (control group) Exclusion criteria: concomitant jaundice and other viral infections such as hepatitis virus A, C, D, E, or F infections
Interventions	Experimental intervention: kushensu capsules, 200 mg, 3 times daily, oral administration, 6 months (n = 36) Control intervention: interferon‐a 2b, 5 MU, intramuscular injection, every other day, 6 months (n = 37) Cointervention: liver protective drugs (e.g. silymarin and vitamin C), oral administration, 6 months Post‐treatment follow‐up: 6 months In addition to the two groups mentioned above, there were two more groups: ‐ 37 participants (mean age 37.4 years, 5 females) receiving kushensu capsules plus interferon‐a 2b plus liver protective drugs. ‐ 36 participants (mean age 39.5 years, 6 females) receiving liver protective drugs. The dose and treatment duration were same as with the mentioned two groups above, of interest to our review.
Outcomes	HBV‐DNA; HBeAg; liver (function) tests (ALT level); TGF‐β, serum liver fibrosis biomarkers (HA, LN, IVC, PCIII); liver histology tests
Notes	Study dates: June 2001 to May 2003 Funding information: study received no funding. Notes: we contacted the authors on 28 June 2018 by telephone and received the reply on missing data.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Used random number table to generate random sequence
Allocation concealment (selection bias)	High risk	Author reported they did not conceal the allocation.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Author did not report that they blinded participants and personnel, and the administration route was different between groups, which can easily unseal the blinding.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Outcome assessors were blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	There were dropouts, but author did not remember the number and did not include those people in the analysis.
Selective reporting (reporting bias)	High risk	We could not find protocol of this study, and author did not report any data on the primary outcomes.
Other bias	Low risk	No other potential sources of bias