Tirzepatide in a UK digital cohort

A retrospective UK study of 630,545 adults who started tirzepatide through one digital prescribing service found 13.5% mean weight loss at six months among 270,309 people with a recorded weight. At 12 months, 58,434 people with a recorded weight had lost 17.4% on average^[1]. These are observational, complete-case estimates from self-reported weights. They do not describe everyone who started treatment and cannot prove that tirzepatide alone caused the change.

The missing-data pattern is central to the readout. It matters. Of 611,580 people eligible to have reached six months by the data cutoff, 44.2% had an observed weight in the six-month window. Of 202,974 eligible to have reached 12 months, 28.8% had an observed weight. People who continued ordering and reporting weights may differ from those who stopped, changed provider or did not report.

How the UK digital cohort was assembled

Rosenior-Patten and colleagues reviewed MedExpress records for adults who initiated tirzepatide for obesity from March 2024 through July 2025. The data cutoff was 15 February 2026. Eligible patients were 18 to 74 years old and generally had a body mass index of at least 30 kg/m2, or at least 27 kg/m2 with a weight-related condition. The analysis required complete baseline data and at least one later order^[2].

The cohort was 80.3% female, with a mean starting BMI of 36.1 kg/m2. It was also predominantly White, at 85.1%. Patients who switched away from tirzepatide were followed only until they stopped it. People who switched from another treatment to tirzepatide were excluded. That design gives a detailed view of one large self-pay pathway, not a representative sample of every UK patient using tirzepatide.

What the weight results actually measure

Weights were self-reported when patients ordered medication. For each milestone, the researchers used the weight closest to the target date within a defined window. The primary endpoint was percentage weight change at six months, using the observed values available in that window. Mean loss was 8.8% at three months, 13.5% at six months, 16.1% at nine months and 17.4% at 12 months.

Among the 270,309 people with an observed six-month weight, 89.9% had lost at least 5%, 72.1% at least 10%, 43.0% at least 15% and 16.4% at least 20%^[2]. Those percentages belong to the observed-weight subgroup. They should not be applied to all 630,545 starters. The authors also ran sensitivity analyses using several assumptions about missing measurements, but the headline figures remain complete-case estimates.

Ordering continuity was associated with more loss

The study used order timing as a proxy for treatment continuity. A gap longer than 84 days counted as a pause and was removed from cadence calculations. People without a recorded pause lost 14.5% at six months, compared with 11.0% among those who paused. At 12 months, the corresponding observed means were 20.0% and 13.5%. Among a more selected adherent group ordering every 25 to 35 days with no long pause, observed 12-month loss reached 21.1%.

That association does not prove that tighter ordering caused the difference. Selection is unavoidable. People who tolerate treatment, can afford repeat prescriptions and are pleased with early results may be more likely to keep ordering. Conversely, side effects, cost, poor response or care elsewhere may interrupt the record. An order also does not confirm that every scheduled injection was taken. The paper explicitly describes its predictive models as descriptive, with modest performance.

Dose patterns varied outside a trial protocol

Among 285,436 people with at least six completed orders, the researchers observed 789 dose trajectories. Only 8.5% followed the standard sequence from 2.5 mg through each step to 15 mg over those six orders. After six orders, 17.7% remained at 5 mg and 22.0% at 7.5 mg, while 21.4% had reduced their dose. This describes prescribing behavior within the service, not a test of which dose is best.

Six-month observed loss was 14.4% among people whose maximum dose was 10 mg, 14.9% at 12.5 mg and 15.1% at 15 mg. These groups were not randomized by dose. Dose reached can reflect response, tolerability, time on treatment and patient preference. The result cannot support choosing a lower dose without clinical assessment. The tirzepatide safety section covers why dose changes belong with a qualified prescriber.

Survey results came from an engaged subset

A six-month survey was completed by 134,673 people, or 21.4% of the analytic cohort. Among respondents, 46.7% reported better mobility, 45.9% better mood and 26.7% reduced or stopped alcohol use. In the prior two weeks, 66.0% reported no side effect that affected quality of life. Nausea was reported by 19.3% and constipation by 13.2%.

These are self-reported experiences from respondents who stayed engaged long enough to receive and answer the survey. That is a selected group. The authors found that respondents differed from nonrespondents mainly in treatment exposure and engagement. The findings may therefore overstate the experience of the full cohort. They also do not establish that tirzepatide directly caused changes in mood, mobility or alcohol use.

Commercial context and safety limits matter

All authors were employees of HeliosX during the study period. HeliosX operates the MedExpress pathway, has commercial relationships with Eli Lilly and Novo Nordisk, and sells Mounjaro and Wegovy. The paper reports no external funding, but the employment and commercial setting are important when interpreting study design, service-level data and the authors' conclusions.

Clinical incidents were reported per dispatched order, not captured through the active surveillance used in a randomized trial. The paper reported 17 incidents per 10,000 dispatched orders. One death due to pancreatitis occurred during the study period, but the report does not establish causation. This dataset cannot determine the incidence of uncommon harms, especially when care outside the provider may not appear in its records.

What we do not yet know

There was no untreated or placebo control group. We do not know the outcomes of people who stopped ordering, moved to another provider or omitted later weights. The service lacked granular data on coexisting conditions, other medicines and clinical biomarkers. Self-reported weight can also contain measurement or reporting error, especially at higher starting BMI.

The study shows that substantial weight loss was recorded among many continuing patients in one large UK digital pathway. It also shows wide variation in ordering and dose trajectories. The safest interpretation keeps both facts together: this is useful real-world evidence, but attrition, selection and confounding prevent a causal effectiveness estimate for every starter. It does not provide a personal forecast or a dosing plan.

Medical disclaimer: This article is for educational and journalistic purposes only and does not constitute medical advice. Peptides discussed may be classified as prescription medicines or research chemicals depending on your jurisdiction. Always consult a qualified healthcare professional before using any peptide product. PeptideMethods.com does not sell, distribute, or facilitate the sale of any peptide product.