GLP-1 persistence in the US military

A retrospective database study published 9 July 2026 in Military Medicine tracked 10,649 active duty US service members who started a GLP-1 receptor agonist for weight management between 2021 and 2025^[1]. The study measured persistence (defined as continuing therapy without a gap of more than 90 days) and medication adherence across three agents: liraglutide, semaglutide, and tirzepatide. The findings add a distinct population angle to the growing body of real-world evidence on how well patients actually stay on these medications over time.

What the study measured and how

The research team, led by Luke B Miller and colleagues at the Military Health System, drew on a claims-level database covering the Defense Health Agency network. Persistence was the primary outcome: the time from therapy initiation to the first treatment gap of more than 90 consecutive days. Adherence was calculated as the proportion of days covered (PDC) over each year of follow-up. The authors also fitted a Cox proportional hazards model to identify patient characteristics associated with discontinuation^[1].

Active duty service members are a distinct clinical population. They face weight and fitness standards set by each branch of the military, they access care through a single-payer system (TRICARE) rather than fragmented commercial insurance, and their occupational demands can interact with medication side effects in ways that differ from civilian patients. These features make the cohort informative precisely because it removes some of the insurance-access noise that complicates civilian real-world data.

The one-year persistence and adherence numbers

One-year persistence rates separated sharply across the three agents. Tirzepatide showed the highest continuation at 81.9%, followed by semaglutide at 70.7%, with liraglutide substantially lower at 34.2%^[1]. Mean annual adherence as measured by PDC followed the same ordering: tirzepatide 78.2%, semaglutide 71.6%, liraglutide 48.7%. The differences between agents were statistically significant on log-rank testing (P less than .001).

The Cox model estimated tirzepatide carried a 33% lower hazard of discontinuation compared to semaglutide (hazard ratio 0.67, 95% CI 0.61 to 0.74), while liraglutide carried a 26% higher discontinuation hazard than semaglutide (hazard ratio 1.26)^[1]. Male service members showed a 12% higher discontinuation hazard than female service members (hazard ratio 1.12, 95% CI 1.02 to 1.22), a finding the authors flagged for further investigation given that males make up a larger share of the active duty force.

How this fits with broader real-world evidence

The military study sits alongside a growing international dataset on GLP-1 real-world persistence. A 2026 narrative review in Nutrients, drawing on cohorts across Europe and Latin America, found one-year persistence as high as 75 to 80% in reimbursed type-2 diabetes cohorts in Scandinavia, but below 10% in obesity-focused settings with high out-of-pocket costs^[2]. The military cohort's single-payer structure and relatively high persistence numbers are consistent with the Scandinavian end of that spectrum, suggesting that cost and access barriers drive a meaningful share of real-world discontinuation in civilian settings.

A separate 2026 narrative review in Diabetes, Obesity and Metabolism found that discontinuation in civilian cohorts is often nonpermanent and is driven primarily by gastrointestinal tolerability and economic constraints^[3]. The military study does not break out discontinuation reasons, but the access stability of TRICARE coverage removes the economic variable, making tolerability and occupational factors the most likely remaining drivers of any attrition.

Limits to read before drawing conclusions

Three caveats apply. First, this is a retrospective claims analysis, not a randomised trial. Service members were not randomly assigned to agents; prescriber preference, formulary access, and patient characteristics all influenced which drug was used. Tirzepatide became available later in the study window than liraglutide, so its cohort is likely subject to a maturation effect: the sickest or least-adherent patients may have already cycled through the earlier agents before tirzepatide was an option.

Second, the outcome definition matters. A 90-day gap threshold for defining discontinuation is a common convention but is not universal. Studies using 30-day or 60-day gaps would report lower persistence numbers for the same underlying behaviour. Readers comparing the 81.9% tirzepatide figure to numbers from other real-world studies should check whether those studies used the same gap definition.

Third, active duty personnel are not a representative sample of the general adult population. They are younger on average, pass fitness standards to enter and remain in service, and have stable health care access. Generalising the persistence rates to civilian weight management programmes should be done with care.

What the gender finding adds

The 12% higher discontinuation hazard for male service members warrants attention. Active duty males outnumber females and face the same weight standards, but occupational exposures, side-effect tolerance thresholds, and help-seeking behaviours differ across the force. The authors describe this as a finding requiring further study rather than an established explanation^[1]. If replicated, it could inform how weight management programmes are structured within military branches that are predominantly male.

What this means in practice

The study does not recommend specific agents or doses. The literature reports doses spanning the titration schedules described in the prescribing information for each drug. Decisions about which GLP-1 receptor agonist is appropriate, at what dose and for how long, belong with a clinician who knows the individual patient's history, occupational demands, and comorbidities. For approved indications and regulatory status by country, the tirzepatide overview page and the semaglutide overview page cover the current picture. This article is for educational and journalistic purposes only.