Forty-seven randomised comparative studies: 8 used a double-blind design, 9 a single-blind, 9 were open-label, and 21 did not report whether or not a blinding method was used. The data required to calculate the overall sample size were not provided.
RBC-clarithromycin-amoxicillin versus RBC-clarithromycin-nitroimidazole (8 studies).
The pooled ITT cure rates for the amoxicillin and nitroimidazole arms were 81% (range: 71 to 96) and 88% (range: 78 to 94), respectively. One study reached significantly better results for the nitroimidazole-containing regimen. The weighted difference between the pooled cure rates for the two treatment arms was statistically significant in favour of RBC-clarithromycin-nitroimidazole, with a weighted difference of 6% (95% CI: 2,10). The pooled per protocol cure rates for the amoxicillin and nitroimidazole arms were 88% (range: 82 to 100%) and 91% (range: 84 to 96), respectively. The weighted difference between the pooled cure rates, in favour of RBC-clarithromycin-nitroimidazole, was not statistically significant.
There was no significant difference in the cure rates for nitroimidazole-susceptible and -resistant strains in the only study that measured antibiotic susceptibility.
No serious side-effects were reported, and the pooled drop-our rates were equal. The number and size of the studies were too small to draw any conclusions regarding publication bias.
PPI-clarithromycin-amoxicillin versus PPI-clarithromycin-nitroimidazole (40 studies). The pooled ITT cure rates for the amoxicillin and nitroimidazole arms were 79% (range: 24 to 95) and 79% (range: 42 to 100), respectively. In 5 studies, the difference was statistically significant in favour of the nitroimidazole-containing regimen, and in one study it was in favour of the amoxicillin-containing regimen. The weighted difference between the pooled cure rates, in favour of PPI-clarithromycin-nitroimidazole, was not statistically significant.
The pooled per protocol cure rates for the amoxicillin and nitroimidazole arms were 83% (range: 33 to 100) and 84% (range: 52 to 100), respectively. In 6 studies, the difference was statistically significant in favour of the nitroimidazole-containing regimen. The weighted difference between the pooled cure rates, in favour of PPI-clarithromycin-nitroimidazole, was not statistically significant.
In nitroimidazole-susceptible strains, PPI-clarithromycin-nitroimidazole performed better, and in nitroimidazole-resistant strains, PPI-clarithromycin-amoxicillin performed better.
No serious side-effects were reported, and the pooled drop-out rates were equal. The funnel plot was symmetrical, indicating that the amount of selection bias was limited.