A Head-to-Head Look at 19 Obesity Drugs: What a New Network Meta-Analysis Found

Most of what gets cited about GLP-1-class drugs comes from individual trial programs — STEP for semaglutide, SURMOUNT for tirzepatide, TRIUMPH for retatrutide, REDEFINE and REIMAGINE for cagrilintide-semaglutide. Each program is internally consistent but not designed to compare across agents head-to-head. A systematic review and network meta-analysis published in BMJ in July 2026 addresses that gap directly, statistically pooling 262 randomized controlled trials covering 99,791 participants and 19 different drugs for adults with overweight or obesity, with follow-up periods from 12 to 172 weeks.

Network meta-analysis is a specific methodology worth understanding on its own terms: rather than requiring every drug to have been tested against every other drug in a head-to-head trial, it uses a statistical model to make indirect comparisons through a shared reference (in this case, lifestyle modification alone), while also incorporating any direct trial comparisons that do exist. The authors applied the GRADE framework to rate the certainty of each estimate, which matters a lot for interpreting the results below.

Weight loss magnitude varied widely by drug class

At one year, compared with lifestyle modification alone, the analysis reported the largest weight reductions with tirzepatide (mean difference -14.9%) and cagrilintide-semaglutide, marketed as CagriSema (-14.8%), followed by oral semaglutide (-10.9%), orforglipron (-9.9%), subcutaneous semaglutide (-9.8%), and phentermine-topiramate (-8.1%) — all rated moderate to high certainty evidence. A second tier of emerging agents, including retatrutide, ecnoglutide, and mazdutide, showed similarly large or larger reductions (13.1–14.6%), but the authors rated this evidence very low to low certainty, reflecting the smaller and earlier-stage trial base those drugs currently have compared with the more established agents.

Tirzepatide also stood out on body composition specifically: it produced the largest reduction in fat mass (25.7%) of any drug studied, but also the largest reduction in lean mass (8.3%), a tradeoff the paper flags rather than resolves.

Larger weight loss tracked with a larger side-effect burden

One of the more useful framings in the paper is that it doesn’t treat weight loss as a standalone outcome — it pairs it with discontinuation and adverse-event data from the same trials. Discontinuation due to 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 relative to lifestyle modification. Gastrointestinal adverse events followed a similar pattern, most elevated with naltrexone-bupropion, oral semaglutide, orforglipron, and tirzepatide.

Fatigue is a less commonly reported outcome in individual trial writeups, but the pooled data made it visible: naltrexone-bupropion carried the highest relative risk (8.9, translating to roughly 331 additional cases per 1,000 people over a year), followed by orforglipron (3.4) and CagriSema (3.2, or about 92 additional cases per 1,000 people). The consistent pattern across outcomes is that the drugs producing the most weight loss tend to also carry the highest discontinuation and side-effect signals — not a universal rule, but a real correlation in this dataset.

Cardiovascular findings were narrower than the weight-loss headlines suggest

This is where the analysis is most conservative, and worth reading carefully. Subcutaneous semaglutide was the only drug in the entire comparison associated with a reduction in all-cause mortality (risk ratio 0.81) and myocardial infarction (0.72) — though the authors note these estimates are heavily weighted by the dedicated cardiovascular-outcome trials run in high cardiovascular-risk populations, not the general trial population. Subcutaneous semaglutide and tirzepatide were both associated with reduced heart failure risk. No drug in the analysis showed a convincing reduction in kidney failure, and none produced a quality-of-life improvement that cleared the pre-specified minimally important difference threshold across the 43 trials (45,663 participants) that reported it.

What this means for reading the broader literature

For a reader tracking the research on any single compound — semaglutide, retatrutide, or a cagrilintide combination — the value of a network meta-analysis like this one is context: it shows where a given drug sits relative to the rest of the field on the same standardized outcomes, rather than in isolation. It’s also a useful reminder that “emerging agent” data (retatrutide included) is still accumulating and carries wider uncertainty than the drugs with a decade of trial history behind them, even when the point estimates look comparable.

It’s worth being precise about what this paper studied: the trials in this network meta-analysis evaluated the approved, pharmacy-dispensed forms of these drugs under clinical trial conditions with medical supervision. That is a distinct research context from bulk peptide material used in laboratory settings, and nothing in this summary should be read as a claim about outcomes for any other formulation or use case.

Further reading

• Nong K, Shi Q, Xie X, et al. (2026). “Comparative effects of drugs for adults with overweight or obesity: systematic review and network meta-analysis.” BMJ, 394, e372161. pubmed.ncbi.nlm.nih.gov/42419792 · doi.org/10.1136/bmj-2026-372161