47 GLP-1 Trials: What the Pooled Weight-Loss Result Means

The pooled GLP-1 result is smaller than the outcomes many people now expect from semaglutide or tirzepatide: across 47 randomized controlled trials, these drugs reduced body weight by an average of 4.57 kg more than placebo. They also reduced BMI by 2.07 kg/m² and waist circumference by 4.55 cm. All three pooled results were statistically significant.

Those numbers matter, but they are not a prediction for a person starting a current high-dose obesity treatment. The analysis combines 23,244 participants across older and newer drugs, different doses, treatment lengths, and populations with overweight or obesity. Averaging that much history produces a class-wide estimate, not a modern-drug forecast.

The 4.57 kg figure answers, "What did this broad GLP-1 evidence base achieve on average versus placebo?" It does not answer, "What will one person lose on one current drug?"

A consistent effect with a deliberately broad average

The 2025 Wong et al. meta-analysis pooled 47 randomized trials and found that GLP-1 receptor agonists lowered weight, BMI, and waist circumference versus placebo. The published abstract reports the point estimates but not confidence intervals, agent-specific forest plots, I² values, or the numerical meta-regression coefficients. That limits how precisely readers can compare drugs or estimate an individual response.

The authors also reported substantial statistical heterogeneity. That is unsurprising: the trials span multiple agents, doses, durations, and clinical populations. Heterogeneity does not erase the average effect, but it does make a single pooled number less useful at the prescription level. An exenatide-era study and a high-dose tirzepatide trial share a pathway-related family resemblance; they do not necessarily share an expected weight-loss result.

In practical terms, the pooled estimate is best read as a conservative cross-generational average. It confirms that this medication category reduces the three measured outcomes compared with placebo. It should not be used as the benchmark for every modern agent.

Newer drugs have changed the range of expected results

Agent-specific evidence points to a gradient across obesity treatments. Cochrane reviews covering liraglutide, semaglutide, and tirzepatide each support weight loss in adults living with obesity. Together, they support the clinical impression that later drugs generally produce larger reductions than earlier high-dose liraglutide treatment.

Tirzepatide deserves a terminology footnote large enough to affect interpretation. It is a dual GIP/GLP-1 receptor co-agonist, not a pure GLP-1 receptor agonist. Including it in a broad GLP-1 analysis can be useful for understanding incretin-based obesity treatment, but it also blends distinct pharmacology. The pooled result therefore cannot tell us how much of the average effect came from GLP-1-only drugs versus dual agonism.

A 2024 randomized trial by Malhotra et al. shows why the distinction matters. In 469 adults with both obesity and obstructive sleep apnea, tirzepatide at 10 mg or 15 mg produced approximately 18% total body-weight loss at 52 weeks. That result belongs to a specific population, drug, dose range, and one-year endpoint; it should not be pasted onto every person considering treatment. Still, it sits far above the 4.57 kg pooled average and illustrates how much a historical all-drug average can obscure contemporary results.

Who loses more weight?

Wong and colleagues used meta-regression to examine why trial results differed. Longer treatment duration predicted greater weight loss. Greater loss was also associated with not having diabetes, higher starting weight and BMI, younger age, female sex, and lower baseline HbA1c.

These findings refine expectations, but they do not yield a personalized calculator. The available report does not provide the regression coefficients or the kilogram difference associated with each characteristic. A person should not infer that a particular age, sex, BMI, or HbA1c guarantees a specific result.

The diabetes finding has support from a different angle. A 2025 network meta-analysis focused on type 2 diabetes found attenuated weight loss with dual and triple incretin agonists compared with outcomes in non-diabetic obesity populations. That fits Wong et al.'s observation that lower baseline HbA1c predicted greater loss. It also means that headline outcomes drawn from non-diabetic obesity trials should not be treated as default expectations for people with type 2 diabetes.

Neither analysis establishes why diabetes status changes the result, nor does either abstract quantify the exact penalty in kilograms. The useful conclusion is narrower: diabetes status and baseline glycemia are relevant to expected weight loss.

Weight loss is not the whole body-composition story

Scale weight does not distinguish fat mass from lean mass. That distinction matters for people concerned about strength, physical function, or preserving muscle during large weight loss. A 2025 network meta-analysis of body composition examined how GLP-1 receptor agonists and co-agonists differ in lean and fat mass changes.

The question remains active rather than settled. Total kilograms lost cannot establish whether one agent produces a more favorable fat-to-lean-mass pattern than another for every patient. For someone using these drugs while training, adequate protein intake, resistance exercise, and clinically appropriate monitoring are implementation questions worth discussing with a treating clinician; the weight-loss number alone cannot answer them.

What about dietitians, younger patients, and cancer?

Medication does not make food intake irrelevant. A 2025 systematic review of randomized trials examined registered-dietitian involvement in non-diabetic adults with obesity using GLP-1 agonists. Across 13 eligible studies, the review reported a pooled mean difference in weight change of -5.4 kg (95% CI, -5.8 to -5.0) and a waist-circumference difference of -5.5 cm (95% CI, -7.7 to -3.25). The review assessed the role of medical nutrition therapy, but the reported figures should not be read as proof that every form of dietary counseling adds the same amount of weight loss.

Evidence in youth with type 1 diabetes is much less mature. A retrospective single-center cohort study of 24 patients aged 10 to 20 with type 1 diabetes and obesity reported reductions in BMI and total daily insulin dose after GLP-1 receptor agonist prescribing. Because this was observational rather than randomized, it can generate a useful clinical hypothesis but cannot establish that the drugs caused both changes.

Cancer claims require even more restraint. A 2026 systematic review and meta-analysis of 48 randomized placebo-controlled trials concluded that the association between GLP-1 receptor agonists or dual agonists and obesity-related cancer risk remains unclear. Current evidence does not support calling these drugs cancer-protective or cancer-causing.

The useful verdict

GLP-1-based treatments consistently reduce weight, BMI, and waist circumference versus placebo across a large randomized-trial evidence base. The 47-trial average—4.57 kg of additional weight loss, 2.07 kg/m² lower BMI, and 4.55 cm less waist circumference—establishes that direction of effect.

It does not settle the more useful clinical questions: which drug, at which dose, for which patient, over how long, with what change in fat and lean mass. Modern high-dose agents, especially tirzepatide in the obesity-and-sleep-apnea trial, can produce results far above the pooled class average. People with type 2 diabetes should generally expect less weight loss than people in non-diabetic obesity trials. And cancer-risk claims remain unsupported either way.

The honest benchmark is specific: use evidence from the drug, dose, population, and follow-up period that most resemble the situation in front of you. A 47-trial average is useful context. It is not a fortune cookie.