Eight controlled clinical trials (n=44,022 patients; 14,052 surgery and 29,970 controls) were included in the review. All trials were reported to be of low quality. Follow-up ranged from 2.5 to 12 years.
Risk of mortality was significantly reduced in patients who underwent bariatric surgery compared to controls for global mortality (OR 0.55, 95% CI 0.49 to 0.63, I2=91.4%; eight trials), cardiovascular mortality (OR 0.58, 95% CI 0.46 to 0.73, I2=73.6%; four trials) and non-cardiovascular mortality (OR 0.70, 95% CI 0.59 to 0.84, I2=60.4%; four trials).
Subgroup analysis by trial size did not significantly alter the results, but reduced statistical heterogeneity and showed greater risk of mortality in large trials compared to small trials. Risks of non-cardiovascular and global mortality were similar for gastric banding and gastric by-pass techniques. Risk of cardiovascular mortality was higher in patients who underwent banding versus bypass (OR 0.71, 95% CI 0.51 to 1.00 versus OR 0.48, 95% CI 0.35 to 0.66). There were no significant differences when trials were subgrouped by type of control (data not presented).
Meta-regression showed a significant association between global mortality and increasing body mass index when one trial was excluded.
Results that compared fixed-effect versus random-effects models were reported in the review.