Losing weight through lifestyle change alone is hard. Over the past decade, a growing list of injectable and oral drugs has entered the picture, each promising to tip the scale further. But comparing them head-to-head in a single clinical trial is rarely practical, because there are too many drugs and too many possible combinations to test all at once.
A research team publishing in the BMJ tackled this problem with a technique called network meta-analysis. Rather than running one new trial, the team pooled data from 262 separate randomised controlled trials involving nearly 100,000 participants. By linking the results mathematically, they could compare 19 different drugs against one another even when those drugs were never tested side by side in the same study. The follow-up periods ranged from 12 weeks to more than three years.
The findings sketch a nuanced picture. Some drugs produce striking weight reductions, but the largest reductions tend to come with the most side effects and the highest rates of people stopping treatment early. A few drugs showed hints of cardiovascular benefit. None convincingly improved quality of life beyond what researchers consider a meaningful threshold.
How the analysis was structured
The researchers searched three major medical databases up to late 2025 and included only randomised controlled trials lasting at least 12 weeks. Each trial had to compare a drug either against placebo, against lifestyle modification alone, or against another drug. The team then measured 24 separate outcomes, covering weight, body composition, side effects, cardiovascular events, kidney function, and quality of life.
To rate how trustworthy each finding was, the team used the GRADE framework, a widely accepted system that scores evidence as very low, low, moderate, or high certainty. This distinction matters enormously when interpreting results: a high-certainty finding means future research is unlikely to overturn it, while a very-low-certainty finding could easily look different in the next big trial.
Weight loss results at one year
Compared with lifestyle modification alone, the review found moderate to high certainty evidence for substantial weight loss with several agents. A dual glucose-dependent insulinotropic polypeptide and glucagon-like peptide-1 receptor agonist peptide produced the largest confirmed reduction, roughly 14.9 percent of body weight. A combination of a long-acting amylin analogue and a glucagon-like peptide-1 receptor agonist came close at 14.8 percent. Oral and subcutaneous forms of another glucagon-like peptide-1 receptor agonist produced reductions in the range of 9.8 to 10.9 percent. A phentermine-topiramate combination and an oral small-molecule glucagon-like peptide-1 receptor agonist called orforglipron also showed meaningful reductions.
Three emerging agents, including retatrutide, showed preliminary reductions of 13.1 to 14.6 percent, potentially matching or exceeding the leading established drugs. However, the certainty rating for these agents was very low to low, meaning the evidence base is still thin and results could shift significantly as more trial data accumulate.
Longer trial duration appeared to matter. For at least one subcutaneous agent, trials running beyond one year recorded larger weight reductions than shorter trials, suggesting that the drugs may continue working for months after the standard one-year endpoint.
Side effects and discontinuation
Greater weight loss generally came with a higher price in tolerability. The review found moderate to high certainty evidence that several drugs led to significantly more people stopping treatment because of adverse events compared with placebo or lifestyle change. Risk ratios for treatment discontinuation ranged from 1.9 to 4.2 across the agents with the worst profiles.
Gastrointestinal problems, including nausea and vomiting, were most prominent with a naltrexone-bupropion combination, an oral glucagon-like peptide-1 receptor agonist, orforglipron, and the dual-action peptide that produced the largest weight reductions. Risk ratios for gastrointestinal events reached as high as 4.2 in some comparisons.
Fatigue stood out as a notable and sometimes underappreciated side effect. The naltrexone-bupropion combination was associated with a risk ratio of 8.9 for fatigue, translating to an absolute increase of 331 additional people per 1,000 experiencing fatigue over one year. Orforglipron and the amylin-glucagon-like peptide-1 combination also raised fatigue risk meaningfully, adding roughly 92 to 100 more cases per 1,000 people compared with control.
Body composition beyond the scale
The review also examined what happens to the body beyond simple weight. The dual-action peptide with the largest overall weight reduction also produced the largest reduction in fat mass, at 25.7 percent. However, the same drug led to the largest reduction in lean mass at 8.3 percent. Lean mass includes muscle, and losing it alongside fat is a tradeoff that researchers consider significant for long-term health, particularly in older adults.
The finding raises a question the current data cannot fully answer: whether the metabolic benefits of large fat-mass losses outweigh the potential functional costs of lean-mass losses, and whether interventions like resistance exercise could buffer the lean-mass decline. The review noted this as an area where additional research is needed.
Cardiovascular and other outcomes
Perhaps the most clinically striking finding involved all-cause mortality. Only one agent, a subcutaneous glucagon-like peptide-1 receptor agonist, was associated with a statistically meaningful reduction in all-cause mortality, with a risk ratio of 0.81. The same drug was also linked to reduced risk of heart attack (risk ratio 0.72) and heart failure (risk ratio 0.43). The dual-action peptide also showed a reduction in heart failure risk (risk ratio 0.49).
The researchers were careful to note that the mortality and cardiovascular findings were largely driven by trials conducted in high-risk populations, meaning people who already had established cardiovascular disease. Whether the same benefits would appear in lower-risk populations remains an open question.
No drug convincingly reduced the risk of kidney failure. Quality-of-life scores were measured across 43 trials involving more than 45,000 participants, and the results were similarly underwhelming. All mean differences in quality-of-life scores fell below five points, while the established threshold for a meaningful difference is ten points. In other words, despite significant weight changes, participants on average did not report feeling substantially better in their day-to-day lives.
What the review means for research
The systematic review offers the most comprehensive head-to-head comparison of obesity drugs published to date, and its conclusions are broadly consistent with the direction the field has been moving. Peptide-based agents that work on multiple hormonal pathways, including those acting on glucagon-like peptide-1, glucose-dependent insulinotropic polypeptide, and amylin receptors simultaneously, appear to push weight reduction further than older single-target approaches.
Retatrutide, a triple-receptor agonist that targets glucagon-like peptide-1, glucose-dependent insulinotropic polypeptide, and glucagon receptors at once, is among the emerging agents flagged as potentially producing reductions on par with the current leaders. The very-low-to-low certainty rating reflects the fact that the evidence base for retatrutide is still early-stage, and the review calls for more rigorous long-term trials before any firm conclusions can be drawn.
The overall message from the literature is that no single drug dominates across every outcome. Policymakers, researchers, and clinicians weighing these data must account for the consistent pattern that larger weight reductions tend to accompany larger burdens of side effects, more treatment discontinuation, and greater loss of lean tissue. The review explicitly frames these as tradeoffs requiring careful consideration rather than a simple ranking of best to worst.
