Why GLP-1 prescriptions go unfilled

A research letter published in Diabetes, Obesity & Metabolism on 10 August 2026 asks a question that weight-loss trial coverage almost never touches. Among patients who already hold a written prescription order for semaglutide or tirzepatide for obesity, which ones actually collect the drug? The letter comes from researchers at the University of Arkansas for Medical Sciences and is indexed as real-world population evidence rather than a trial^[1]. Its results are not yet in the public record, so what follows is the evidence base it joins, and what that evidence can honestly support.

Roughly four in ten orders are never dispensed

The clearest US figure comes from a cohort study in JAMA Health Forum that linked electronic health records at University of Colorado Health to the Colorado All-Payer Claims Database. Across 9,848 GLP-1 receptor agonist orders written for 6,094 patients between January 2018 and September 2022, 5,915 orders (60.1%) were filled^[2]. The remaining 40% never became a dispensed box.

The split by indication is sharper than the headline. Orders for patients with obesity alone were filled 37.2% of the time. Orders for patients with diabetes only were filled 47.5% of the time. Orders for patients recorded with both conditions were filled 64.6% of the time^[2]. The same molecules in the same GLP-1 class move through the pharmacy very differently depending on what the chart says they are for.

Cost explains a lot, then stops explaining

Among orders that were filled, the mean out-of-pocket cost was $71.90 per 30-day supply^[2]. That average hides the indication gap again: $134.04 per prescription when obesity was the only recorded indication, against $70.32 when diabetes and obesity were both recorded^[2]. For anyone without insurance the comparison is starker still. The authors cite a 2024 average retail price above $900 a month^[2]. That is the figure behind most of the grey-market demand we document on the semaglutide page.

Then the pattern turns. Out-of-pocket costs were lower for non-Hispanic Black patients ($41.15) and Hispanic patients ($63.69) than for non-Hispanic White patients ($78.37), yet fill rates ran the other way, at 55.3% and 58.4% against 60.9%^[2]. Whatever stops those orders, the amount charged at the counter is not the whole of it.

The drop-off starts before anyone writes a prescription

A larger cohort study, published in the same journal that carried the new letter, followed 50,678 adults with a body mass index of 30 or above across a health system spanning Ohio and Florida, using records from January 2015 to June 2023. Only 8.0% of that cohort ever received an anti-obesity medication prescription, and only 4.4% filled one^[3]. Insurance type, sex, race and ethnicity, and living in the most deprived quartile by area deprivation index were each associated with lower odds of receiving the prescription at all, and several of the same factors tracked lower odds of the fill^[3]. Fill rate is the last gate in a funnel that has already narrowed twice.

Coverage rules move these numbers more than patients do

Payer policy sets the ceiling. KFF tracking of Medicaid reports that as of January 2026 only 13 state Medicaid programmes covered GLP-1s for obesity under fee-for-service, and that California, New Hampshire, Pennsylvania and South Carolina had recently dropped that coverage^[5]. Spending on the class still rose from about $1 billion in 2019 to almost $9 billion in 2024, on prescriptions climbing from roughly 1 million to over 8 million^[5]. Coverage for diabetes is required of states; coverage for obesity is a state choice. That is the exact line the fill-rate data keeps landing on, and it is why our tirzepatide regulatory notes treat indication and reimbursement as separate questions.

An unfilled order is not always a patient going without

Some patients obtain the drug somewhere claims data cannot see. A retrospective study of the American Family Cohort, a national US primary-care records database, covered 153,044 patients with documented semaglutide or tirzepatide use from January 2021 to December 2024 and found 8.2% with compounded formulations recorded in clinical notes^[4]. The authors note that surveys have put the share of users obtaining these drugs from compounders at roughly 23%, and conclude that many patients access them outside coordinated care^[4]. They also found compounded users skewed female, non-Hispanic White, non-diabetic and resident in less deprived areas^[4]. An order logged as unfilled may have been abandoned, delayed, switched, or quietly replaced by a compounded tirzepatide vial the prescriber never saw.

What a fill-rate study can and cannot tell you

These are retrospective observational datasets, and their authors are plain about the limits. The Colorado group could not assess reasons for non-adherence, or whether medicines were bought with cash instead of insurance^[2]. The Ohio and Florida group note that patient attitudes, knowledge and preferences are absent from electronic health records, and that shifting insurance and precertification rules could not be linked to individual cases^[3]. Both draw on single health systems. Neither establishes cause. They map where people fall out of the process, not why any one person did.

Practical context

For readers, the practical reading is that a prescription is a step and not the finish line. Coverage, prior authorisation, how the indication is coded, and the price at the counter each get a vote before the pen reaches anyone. Our access and regulation notes for semaglutide and tirzepatide track the country-level rules that decide most of this well upstream of the pharmacy. When the Arkansas letter is fully indexed with its patient-level findings, we will update this piece with the figures it reports rather than the ones around it.

This article is educational and is not medical advice. Decisions about obesity treatment, including whether a prescription is appropriate, belong with a qualified healthcare provider who knows your history.