Review of 38 trials puts tirzepatide at 19% and semaglutide near 15% weight loss above placebo

Researchers at McGill University and the Lady Davis Institute in Montreal have published an updated review of every randomized trial they could find of GLP-1 drugs used for weight loss in adults who have overweight or obesity but who do not have diabetes. It appeared online in Annals of Internal Medicine on September 1, 2026.

The review covers 38 randomized controlled trials and 25,816 participants. Fourteen of those trials, with about 11,000 participants, are new since the team's previous version. Trials had to last at least 16 weeks to be included. The authors searched three medical databases from October 5, 2024 through March 25, 2026. They report no funding source.

Placebo-subtracted weight loss ran from 5.8% for liraglutide to 19.0% for tirzepatide

In these trials, some people get the drug and some get a dummy injection or pill. Both groups usually lose some weight, because everyone is being watched, weighed and coached. Placebo-subtracted weight loss is the gap between the two groups. It is the part that can reasonably be credited to the drug itself, and it is always a smaller number than the total weight loss you see quoted in headlines.

Among drugs you can currently be prescribed, the review reports placebo-subtracted weight loss reaching up to:

• Liraglutide: 5.8% (95% confidence interval 3.6% to 8.0%)
• Semaglutide by injection: 14.8% (13.4% to 16.2%)
• Semaglutide by mouth: 14.3% (11.4% to 17.2%)
• Orforglipron: 12.4% (9.7% to 15.1%)
• Tirzepatide: 19.0% (16.4% to 21.6%)

A confidence interval is the range the true figure most likely sits in. A wide range, as with oral semaglutide, means the trials were less consistent with each other. These are the highest results reported for each drug, so they reflect the best performing dose in the best performing trial rather than what an average person should expect.

Two drugs that are not approved anywhere posted the largest numbers

The review lists amycretin at 23.9% (18.5% to 29.3%) and retatrutide at 22.1% (19.3% to 24.9%). Both are still in testing. Neither is approved by any regulator, neither is available by prescription, and the authors describe those reductions only as numerically greater, which is their way of saying the comparison was not made head to head.

Where trials did compare drugs directly against each other, semaglutide and an experimental drug called JNJ-64565111 produced more weight loss than liraglutide, and tirzepatide and cagrilintide plus semaglutide produced more weight loss than semaglutide alone.

Three quarters of people on the drugs reported stomach and bowel side effects

Gastrointestinal side effects, meaning nausea, vomiting, diarrhea, constipation and related complaints, were reported by 76.0% of people taking a GLP-1 drug against 40.1% of people taking placebo. That placebo figure is worth sitting with, because it shows how common these complaints are even without an active drug.

Stopping treatment because of a side effect was less common: 10.7% on a GLP-1 drug against 3.4% on placebo. The authors note this was numerically higher with some of the pills. Serious adverse events, the category that covers hospitalization, disability and other major harm, occurred in 6.5% of drug recipients and 5.2% of placebo recipients. Deaths were rare in both groups, at 0.1% and 0.0%. The authors identified no new safety signals.

The trials were too different to be pooled into one number

The most important caveat is one the authors state plainly: heterogeneity precluded quantitative synthesis. In other words, the trials varied so much in design, dose, duration and the people enrolled that the team did not combine them into a single pooled estimate. The drug-by-drug figures above are therefore best read as a summary of separate trials, not as a league table produced by comparing the drugs against each other.

The authors also report that safety outcomes were inconsistently reported across trials, which limits how firmly anyone can compare side effect rates between drugs.

Two further limits are worth keeping in mind for anyone reading these figures as a personal forecast. Trial participants get structured follow-up, free medication and regular contact with a research team, which is not how most prescriptions work in practice. And most of these trials run for a year or two, while obesity treatment is generally a long-term commitment.