Sleeve surgery beat the endoscopic version on weight loss, while the endoscopic version had fewer complications

Surgical sleeve gastrectomy produced more weight loss than the endoscopic version at every time point measured out to two years, while the endoscopic procedure carried a lower rate of complications and far less new acid reflux. That is the result of a systematic review and meta-analysis published September 10, 2026 in Surgical Endoscopy.

The two procedures are often discussed together because they do a similar thing by different routes. Laparoscopic sleeve gastrectomy, or LSG, is an operation that removes part of the stomach through small incisions. It is the most common bariatric operation worldwide. Endoscopic sleeve gastroplasty, or ESG, reshapes the stomach with stitches placed through a tube passed down the throat, with nothing removed and no incisions.

Long-term comparisons of the two have been thin. The researchers searched five databases and pulled together ten studies covering 7,442 patients, 3,583 who had ESG and 3,859 who had LSG. Average age was 35.5, average starting BMI was 34.6, 86.2 percent were women, and average follow-up was 32.1 months.

Surgery led on weight loss at six, twelve and twenty-four months

The measure used was total body weight loss percentage, meaning the share of starting body weight lost. On that measure ESG trailed LSG at every point.

At six months the gap was 8.34 percentage points in favour of LSG, with a 95 percent confidence interval of 6.19 to 10.50, drawn from nine studies. At twelve months the gap widened to 12.58 points, interval 9.70 to 15.47, from six studies. At twenty-four months it was 9.81 points, interval 4.90 to 14.72, from three studies.

The gap is real and it is large. A twelve-point difference in total body weight loss is not a rounding error. Note also how the number of contributing studies falls as the timeline lengthens: nine at six months, six at twelve, three at twenty-four. The two-year figure rests on the least evidence and has the widest interval to match.

The endoscopic procedure had fewer adverse events and much less new reflux

ESG had a lower risk of adverse events overall, with a risk ratio of 0.54 across nine studies, p equals 0.03. A risk ratio of 0.54 means roughly half the risk.

The larger difference was in new acid reflux. De novo GERD, meaning reflux that appears after the procedure in someone who did not have it before, had a risk ratio of 0.33 across five studies, p equals 0.02. That is about a third of the risk.

New reflux after sleeve gastrectomy is a well-recognised issue and a common reason people end up needing further treatment or a revision. A difference of this size on that specific problem is the practical core of the paper.

The authors' conclusion is that LSG achieves greater weight loss out to two years, while ESG still delivers moderate weight loss with meaningfully lower risk, which may suit someone who wants to avoid the risks of an operation or who already has reflux.

Pooled observational studies, mostly women, average BMI 34.6

This is a meta-analysis of published comparative studies rather than of randomised trials, so people were not assigned to one procedure by chance. Patients and surgeons chose, and those choices tend to follow patterns. Someone picking the endoscopic route may differ from someone picking surgery in ways the analysis cannot fully account for, and any of those differences could influence the results in either direction.

The average starting BMI of 34.6 is at the lower end of the range usually considered for sleeve gastrectomy. These findings describe that population and may not transfer to people with substantially higher BMI.

86.2 percent of participants were women, so the data on men are comparatively thin. Average follow-up of 32.1 months is long by the standards of ESG research but short for judging durability of a weight-loss procedure, and only three studies reached the two-year weight comparison.

We read the published abstract and the indexed record rather than the full text, so we could not examine the individual studies pooled, their quality ratings, or how much the results varied between them.