Six studies, including five RCTs (n=3,464) and one quasi-RCT (n=589), were included in the review.
One trial was rated as high quality, four as low quality and one as moderate quality.
The funnel plots did not suggest evidence of publication bias.
The rates of TB infection were similar in both the RZ and INH groups, regardless of whether individuals were infected with HIV (3 studies; RD 0%, 95% CI: -1, 2) or not (3 studies; RD 0%, 95% CI: -2, 1). No significant heterogeneity was detected.
There was no significant difference in mortality between RZ and INH groups in both HIV-infected individuals (3 studies; RD -1%, 95% CI: -4, 2) and those not infected with HIV (3 studies; RD 0%, 95% CI: -1, 1). No significant heterogeneity was detected.
Severe hepatotoxicity.
Significant statistical heterogeneity due to differences in the duration of treatment was identified for both HIV-infected and uninfected individuals. Subgroup analyses suggested that this was due to differences in the duration of treatment (HIV-infected individuals) and the quality of the trials (non-HIV-infected individuals).
All serious adverse events.
There was significant heterogeneity in the studies of non-HIV-infected individuals, which subgroup analyses suggested was due to differences in study quality. Events were significantly more likely to occur in those treated with RZ in both the moderate quality study (RD 29%, 95% CI: 13, 46; 1 study) and low-quality studies (RD 7%, 95% CI: 4, 10; 2 studies) among non-HIV-infected individuals. Similarly significant heterogeneity was detected in the studies of HIV-infected individuals (3 studies), which subgroup analyses suggested was due to differences in the duration of treatment. However, for these individuals, events were significantly more likely in RZ-treated patients only in the trial where the duration of INH therapy was 12 months.