Eleven RCTs (n=3,720) were included.
6-month follow-up.
For ESO 20 compared with LAN 15 there was a significant difference in favour of esomeprazole: the RR was 1.09 (95% CI: 1.02, 1.17; only 1 study available).
For LAN 30 compared with LAN 15 there was a significant difference in favour of the higher dose (6 studies): the RR was 1.12 (95% CI: 1.06, 1.18; heterogeneity statistic 6.49, d.f.=5, p=0.26).
For OME 20 compared with LAN 15 there was a significant difference in favour of omeprazole: the RR was 1.12 (95% CI: 1.06, 1.19; only 1 study available).
There was no significant difference between OME 20 and LAN 30: the RR was 0.99 (95% CI: 0.95, 1.03; heterogeneity statistic 0.62, d.f.=1, p=0.43).
12-month follow-up.
The results of the 12-month follow-up mirrored the 6-month analyses for all PPIs. For LAN 30 compared with LAN 15, there was a significant difference in favour of the higher dose: the RR was 1.12 (95% CI: 1.05, 1.18; heterogeneity statistic 10.94, d.f.=6, p=0.09).
FOR OME 20 compared with LAN 15, there was a significant difference in favour of omeprazole: the RR was 1.19 (95% CI: 1.10, 1.30; only 1 study available).
There was no significant difference between OME 20 and LAN 30: the RR was 1.01 (95% CI: 0.96, 1.06; heterogeneity statistic 0.02, d.f.=1, p=0.9). There was also reference within the table to LAN 15, but this appears to be a misprint. Data were only available for LAN 15 compared with OME 10 at 12 months, and these showed a significant difference in favour of lansoprazole: the RR was 0.79 (95% CI: 0.64, 0.97; only 1 study available).