BMJ Network Meta-Analysis Compares 19 Obesity Drugs on Weight Loss, Heart Risk, and Side Effects

The obesity-drug field now has more approved and late-stage peptide options than at any point in its history, but until recently there was no single, rigorous comparison of how they stack up against each other. A systematic review and network meta-analysis published in The BMJ on July 8, 2026, led by researchers at West China Hospital, Sichuan University, attempts exactly that. The authors describe their goal plainly: to give policymakers, payers, clinicians, and patients an evidence summary of the comparative benefits and harms of drugs for adults with overweight or obesity.

The review searched Medline, Embase, and the Cochrane Library through November 12, 2025, for randomized controlled trials lasting 12 weeks or longer that compared a drug against lifestyle modification, placebo, or another drug. It ultimately combined 262 trials covering 99,791 participants and 19 drugs, with individual trial follow-up ranging from 12 to 172 weeks. The authors used frequentist random-effects and Bayesian dose-response models, graded the certainty of each finding with the GRADE framework, and assessed trial quality with the Cochrane Risk of Bias 2 tool across 24 separate outcomes.

On weight loss at one year compared with lifestyle modification alone, tirzepatide came out ahead with moderate to high certainty evidence, at a mean difference of -14.9% (95% confidence interval -16.0% to -13.9%). The cagrilintide-semaglutide combination known as CagriSema was close behind at -14.8% (-16.9% to -12.7%). Oral semaglutide followed at -10.9%, orforglipron at -9.9%, subcutaneous semaglutide at -9.8%, and the older drug combination phentermine-topiramate at -8.1%, all also rated moderate to high certainty.

A separate group of emerging agents, including ecnoglutide, mazdutide, and retatrutide, appeared to produce similar or even larger weight reductions, in the range of 13.1% to 14.6%, but the authors rated this evidence very low to low certainty given the smaller and less mature trial base behind those drugs. Retatrutide remains investigational and is not approved by any regulator; mazdutide's approval status varies by jurisdiction. Readers should treat these figures as preliminary compared with the tirzepatide and semaglutide data.

On harder clinical outcomes, the picture was more selective. Subcutaneous semaglutide was the only drug in the analysis associated with reduced all-cause mortality, at a risk ratio of 0.81 (95% CI 0.72 to 0.93, roughly a 19% relative reduction) and reduced myocardial infarction, at a risk ratio of 0.72 (0.61 to 0.85, roughly 28%), estimates the authors say were largely informed by cardiovascular-outcome trials conducted in high-risk populations. Subcutaneous semaglutide (risk ratio 0.43) and tirzepatide (0.49) both reduced heart-failure risk. No drug in the analysis convincingly reduced the risk of kidney failure.

Bigger weight-loss numbers came with bigger costs in tolerability. Discontinuation because of adverse events was highest with orforglipron, naltrexone-bupropion, liraglutide, phentermine-topiramate, CagriSema, and oral semaglutide, with risk ratios ranging from 1.9 to 4.2 versus comparators. Gastrointestinal events rose most with naltrexone-bupropion, oral semaglutide, orforglipron, and tirzepatide, at risk ratios of 3.1 to 4.2. Fatigue risk was particularly elevated with naltrexone-bupropion, orforglipron, and CagriSema. Tirzepatide reduced fat mass the most of any drug studied, by 25.7%, but it also reduced lean mass the most, by 8.3%, a trade-off the paper flags explicitly.

Perhaps the most sobering finding involves quality of life. Across 43 trials with 45,663 participants, no drug produced a quality-of-life improvement that cleared the established minimally important difference threshold of 10 points; all measured mean differences were under 5 points. The authors also found no credible differences in relative treatment effects across most dosage and patient-characteristic subgroups, with the exception of larger weight reductions in longer trials of subcutaneous semaglutide.

The authors' own conclusion is a caution against reading league tables of weight-loss percentages as the whole story. Obesity drugs produce variable weight loss at one year, they write, with larger benefits generally accompanied by greater harms and higher discontinuation, and few of the drugs studied show convincing cardiovascular benefit or meaningfully improve quality of life. Treatment decisions, the authors and an accompanying BMJ Group summary of the paper state, should be individualized, balancing expected benefits, harms, treatment burden, costs, availability, and patient preferences. This analysis synthesizes existing randomized trial data across drug classes; it does not establish new efficacy findings for any single drug, and nothing here constitutes medical or dosing advice.