Honest Criticisms and Limitations of the STEP-1 Trial
At a glance
| Field | Detail |
|---|---|
| N | 1,961 (semaglutide 1,306; placebo 655) |
| Intervention | Semaglutide 2.4 mg subcutaneous once weekly |
| Comparator | Matched placebo injection + lifestyle intervention |
| Duration | 68 weeks (16-week dose escalation + 52 weeks at target dose) |
| Primary endpoint | Percent change in body weight from baseline to week 68 |
| Key result | −14.9% semaglutide vs −2.4% placebo (estimated treatment difference −12.4 percentage points; P <0.001) |
| Registration | NCT03548935 |
Who Was Actually Enrolled, and Who Was Left Out
The inclusion criteria for STEP-1 required a BMI of 30 or higher, or 27 or higher with at least one weight-related comorbidity, but explicitly excluded anyone with type 2 diabetes. That single exclusion removed the population most likely to use a GLP-1 receptor agonist in practice. Patients with diabetes typically show smaller weight reductions on semaglutide (as later confirmed in STEP-2), so the headline 14.9% figure may overstate what clinicians see in mixed real-world panels.
The trial enrolled participants across 129 sites in 16 countries, yet the published demographics skew heavily: approximately 75% of participants were White, roughly 74% were female, and the mean age was 46 years. Black participants made up only about 6% of the cohort despite bearing a disproportionate burden of obesity in the United States. Hispanic representation was similarly thin. These gaps matter because pharmacokinetics, body composition trajectories, and GI tolerability can differ across ancestry groups, and the trial was not powered to detect those differences.
Baseline BMI averaged 37.9 kg/m². Participants with BMI above 50 were not excluded in the protocol but represented a small fraction of enrollment. Bariatric surgery candidates and those with recent major cardiovascular events were excluded. The result is a trial population that sits in the middle of the obesity spectrum, healthy enough to tolerate weekly injections and a structured lifestyle program, which limits extrapolation to patients with severe obesity or significant multi-organ comorbidity.
The Lifestyle Intervention Confound
Both arms received monthly counseling sessions, a 500 kcal/day deficit diet, and a recommendation for 150 minutes per week of physical activity. This co-intervention is often glossed over when the trial is cited. The published protocol confirms that adherence to the lifestyle component was monitored but not standardized across sites. Placebo-arm participants lost a mean of 2.4%, which is higher than typical placebo weight loss in obesity trials and likely reflects genuine behavioral change driven by counseling.
The clinical question a prescriber faces is rarely "semaglutide plus structured counseling versus placebo plus structured counseling." It is more often "semaglutide prescribed in a 15-minute office visit versus the patient's current habits." STEP-1 cannot answer that question. The 12.4-percentage-point treatment difference is real, but it was measured under conditions that most outpatient practices do not replicate.
A Framework for Reading STEP-1's Statistical Results
Understanding the trial's reported numbers requires attention to three analytic choices that shaped the headline figures.
1. The estimand favored treatment effect size. The primary estimand was a "treatment policy" strategy that included data from all randomized participants regardless of whether they discontinued the drug. However, missing week-68 weights (about 14% of the semaglutide arm) were handled with multiple imputation under a "missing at random" assumption. If participants who dropped out due to side effects would have regained weight, the imputed values may be optimistic.
2. Two co-primary endpoints inflated the appearance of success. STEP-1 defined two co-primary endpoints: percent body weight change and the proportion achieving 5% or greater weight loss. Both were tested at a two-sided 5% significance level with multiplicity adjustment, but the overwhelming effect size (P <0.001 for both) made the dual-endpoint structure largely cosmetic. A single, more demanding endpoint (such as 15% weight loss sustained at 12 months post-treatment) would have been more informative.
3. The "responder" thresholds were chosen after decades of regulatory convention, not clinical evidence. The 5%, 10%, and 15% weight loss thresholds used in the FDA's obesity drug guidance are regulatory artifacts. Whether a 5% reduction in body weight translates to durable cardiometabolic benefit for an individual patient depends on starting weight, fat distribution, and comorbidity profile. STEP-1 reported that 86.4% of semaglutide patients hit the 5% threshold, but this tells a prescriber little about which patients benefited most.
Dropout Rates and the GI Tolerability Problem
Overall discontinuation of the trial drug was 11.6% in the semaglutide group versus 7.4% in the placebo group. The most common reasons in the semaglutide arm were gastrointestinal adverse events, primarily nausea (44.2% of semaglutide patients reported nausea at some point), vomiting, diarrhea, and constipation. Most events were mild to moderate and concentrated during the dose-escalation phase.
Still, a 4.2-percentage-point gap in drug discontinuation is not trivial. The per-protocol analysis (which excluded these dropouts) showed even larger weight loss, meaning the intent-to-treat result is diluted by people who could not tolerate the drug. Real-world adherence data from post-marketing analyses suggest that 12-month persistence with GLP-1 agonists for obesity hovers around 30-50%, far below the controlled trial setting.
Duration: 68 Weeks Is Not Long Enough
Obesity is a chronic, relapsing condition. The STEP-1 extension data showed that participants who discontinued semaglutide after 68 weeks regained roughly two-thirds of the lost weight within one year. This finding, published separately, reframes the entire trial: the 14.9% loss is not a permanent outcome but a drug-dependent suppression that requires indefinite treatment.
The 68-week primary endpoint was sufficient for FDA approval under existing guidance, but it tells prescribers and patients almost nothing about 5-year or 10-year outcomes. Long-term safety signals (pancreatic events, thyroid C-cell concerns flagged in preclinical rodent studies and carried as a boxed warning on the Wegovy label, gallbladder disease, potential effects on bone density during rapid weight loss) require years of pharmacovigilance data that STEP-1 alone cannot provide.
Industry Sponsorship and Conflict of Interest
STEP-1 was funded entirely by Novo Nordisk. The company designed the trial, held the data, performed the statistical analysis, and employed several of the authors. The principal investigators received consulting fees, speaker honoraria, or both from Novo Nordisk.
This does not invalidate the findings, but it warrants scrutiny. Industry-sponsored obesity trials have historically reported larger effect sizes than investigator-initiated studies. The STEP program as a whole (STEP 1 through 5) was coordinated as a regulatory package designed to support a single NDA submission. Each sub-trial answered one specific regulatory question. Independent replication with study-level data access has not occurred.
Post-publication letters to the NEJM raised questions about the generalizability of results to patients with psychiatric comorbidities (who were largely excluded), the lack of a structured exercise comparison arm, and the absence of patient-reported outcome data beyond the IWQOL-Lite-CT instrument. These critiques did not challenge the core efficacy finding but highlighted the gap between a regulatory-grade trial and the evidence clinicians need for individualized decision-making.
What STEP-1 Cannot Tell You
Several clinically important questions remain unanswered by this trial:
- Dose optimization. STEP-1 tested only the 2.4 mg target dose. Whether lower maintenance doses preserve efficacy with fewer side effects is unknown from this dataset alone.
- Combination approaches. The trial did not compare semaglutide to phentermine-topiramate, naltrexone-bupropion, or bariatric surgery, the established alternatives at the time of enrollment.
- Metabolic health independent of weight. Body composition data (lean mass versus fat mass loss) were not primary or secondary endpoints. Rapid weight loss at this magnitude raises questions about sarcopenia risk, particularly in older adults who were underrepresented in enrollment.
- Cost-effectiveness. At a U.S. list price exceeding $1,300/month for Wegovy, the trial provides no framework for evaluating value. Subsequent ICER analyses found semaglutide 2.4 mg cost-ineffective at list price without substantial discounts.
Placing STEP-1 in the Broader Evidence Base
STEP-1 remains one of the largest and most rigorously executed obesity pharmacotherapy trials. Its limitations are not unique; they are structural features of how obesity drugs reach market. The 2022 AGA clinical practice guideline on pharmacological management of obesity cited STEP-1 prominently but also noted the need for longer-term data, head-to-head comparisons, and studies in more diverse populations.
The trial proved that GLP-1 receptor agonism can produce surgically competitive weight loss in a selected population over 68 weeks. It did not prove that this weight loss is durable, universally tolerable, cost-effective, or superior to existing alternatives. Clinicians citing the 14.9% figure should pair it with the extension data showing rapid regain, the demographic narrowness of enrollment, and the lifestyle intervention that accompanied every injection.
Frequently asked questions
Did STEP-1 include patients with type 2 diabetes?
No. Patients with type 2 diabetes were explicitly excluded. The STEP-2 trial addressed that population separately and found smaller mean weight loss (approximately 9.6% at the 2.4 mg dose).
How were missing data handled in the primary analysis?
The primary estimand used multiple imputation under a missing-at-random assumption. About 14% of semaglutide-arm participants lacked week-68 body weight data. If dropouts regained weight, the imputed values could overestimate the treatment effect.
What happened when participants stopped semaglutide after the trial?
Extension data published in 2022 showed participants regained approximately two-thirds of the weight they had lost within one year of stopping semaglutide, indicating that continued treatment is necessary to maintain results.
Was the trial independently funded?
No. Novo Nordisk funded the trial, designed the protocol, held the data, and performed the statistical analysis. Several authors had financial relationships with the company.
How diverse was the trial population?
Approximately 75% of participants were White and 74% were female. Black participants represented about 6% of enrollment. The mean age was 46 years. These demographics limit generalizability.
Why does the lifestyle intervention matter?
Both arms received counseling, a 500 kcal/day deficit diet, and exercise recommendations. This co-treatment likely inflated the placebo response and means the trial does not reflect typical real-world prescribing conditions.
Is 14.9% weight loss comparable to bariatric surgery?
The magnitude approaches some surgical outcomes (e.g., adjustable gastric banding), but STEP-1 did not include a surgical comparison arm. Bariatric procedures typically produce more durable loss over 5-10 years.
What gastrointestinal side effects were most common?
Nausea was reported by 44.2% of semaglutide participants. Diarrhea, vomiting, and constipation were also frequent. Most events were mild to moderate and clustered during the 16-week dose-escalation period.
Did STEP-1 measure body composition changes?
No. Body composition (lean mass vs. fat mass) was not a primary or secondary endpoint, leaving open questions about sarcopenia risk during rapid pharmacologically induced weight loss.
Is semaglutide 2.4 mg considered cost-effective?
Independent analyses by ICER found semaglutide 2.4 mg cost-ineffective at U.S. list price (over $1,300/month) without significant discounts or rebates. STEP-1 did not include cost-effectiveness endpoints.
References
- Wilding JPH, Batterham RL, Calanna S, et al. Once-weekly semaglutide in adults with overweight or obesity. N Engl J Med. 2021;384(11):989-1002. PubMed
- Davies M, Færch L, Jeppesen OK, et al. Semaglutide 2.4 mg once a week in adults with overweight or obesity, and type 2 diabetes (STEP 2). Lancet. 2021;397(10278):971-984. PubMed
- Wilding JPH, Batterham RL, Davies M, et al. Weight regain and cardiometabolic effects after withdrawal of semaglutide. Diabetes Obes Metab. 2022;24(8):1553-1564. PubMed
- Wegovy (semaglutide) prescribing information. U.S. Food and Drug Administration. 2021. FDA Label
- Amaro A, Sugimoto D, Cleman J, et al. AGA clinical practice guideline on pharmacological interventions for adults with obesity. Gastroenterology. 2022;163(5):1198-1225. PubMed
- Institute for Clinical and Economic Review. Obesity management: effectiveness and value. 2022. ICER