Cleveland Clinic study explains the trial-versus-real-world weight-loss gap

Researchers at a large health system in Ohio and Florida report that adults who started injectable semaglutide or tirzepatide for weight loss lost an average of 8.7% of their body weight in one year — well below what the drugs' main phase 3 trials showed. The reason, the authors say, is not that the drugs worked differently in everyday care, but that many people stopped treatment and most never moved past low doses ^[1].

The retrospective cohort study, published in Obesity, used electronic health record data to find adults with overweight or obesity and without type 2 diabetes who began injectable semaglutide or tirzepatide between 2021 and 2023. Follow-up ran through December 2024. Of 7,881 patients, 6,109 started semaglutide and 1,772 started tirzepatide ^[1].

What the numbers showed

Mean weight reduction at one year was 8.7% (standard deviation 9.6%) across the whole group. Broken out by whether people stayed on the medication, the results diverged sharply ^[1]:

• 3.6% (SD 8.1%) among those who discontinued early, meaning within three months of starting
• 6.8% (SD 9.1%) among those who discontinued later, between three and 12 months
• 11.9% (SD 9.2%) among those who did not discontinue

The differences were statistically significant (p < 0.001) ^[1]. The study defined discontinuation as a gap of more than 90 days between running out of one supply and the next fill, or between the last supply and the end of follow-up. People who switched between injectable semaglutide, tirzepatide or other obesity medications with a gap under 90 days were not counted as having stopped ^[1].

A second finding may matter just as much: 80.8% of patients were on low maintenance dosages ^[1]. In the trials that established these drugs for weight management, patients were titrated up to and held at higher doses. The authors conclude that the lower average weight reduction in this cohort was "likely because of higher rates of discontinuation and lower maintenance dosages" ^[1].

The study also looked at blood sugar in the 1,320 patients who had prediabetes when they started. Mean absolute reduction in HbA1c at one year was 0.1 percentage point (SD 0.4) with early discontinuation, 0.2 (SD 0.4) with late discontinuation, and 0.4 (SD 0.4) among those who kept taking the medication (p < 0.001) ^[1]. The same pattern held: staying on treatment tracked with bigger changes.

Why it matters for patients

The headline numbers from clinical trials describe people who were given free medication, escalated to target doses, and supported by study staff. This analysis puts a number on what happens when those conditions are absent. Average results in routine care were lower — but among people who stayed on the drug continuously for a year, average weight loss was 11.9%, roughly three times what early stoppers experienced ^[1].

That said, this is an observational study, and it cannot prove that dose or duration caused the difference. People who stop within three months may differ in important ways from those who continue — for example, in side effects, insurance coverage, drug supply, or how well they were responding. The published abstract does not report why patients discontinued, does not give separate one-year weight-loss figures for semaglutide versus tirzepatide, and does not define the specific milligram cutoff used for "low maintenance dosages." Those details are not available in the material reviewed here.

Several authors disclosed industry ties. W. Scott Butsch reported advisory board fees from Novo Nordisk, Eli Lilly, Boehringer Ingelheim and Abbott, plus research funding from Eli Lilly; Marcio L. Griebeler reported research funding from Novo Nordisk; Michael B. Rothberg reported consulting fees from the Blue Cross Blue Shield Association; Christopher B. Boyer reported consulting fees from Janssen. All were described as outside the submitted work, and the other authors declared no conflicts ^[1].

What happens next

The study was indexed in PubMed on June 10, 2025 (PMID 40491239) ^[1]. It does not follow patients beyond 12 months, so it says nothing about what happened to weight after the first year or after stopping. Whether the pattern holds in other health systems, in patients with type 2 diabetes, or with oral GLP-1 options is not addressed by this analysis.