What evidence gaps are holding back GLP-1 receptor agonists?

Why do so many patients stop taking GLP-1 drugs within a year?

The biggest practical gap is that real-world persistence—whether patients actually keep taking the drug—is dramatically lower than what clinical trials suggest. A large 2026 study of over 33,000 commercially insured patients without diabetes found that only 33.2% of those starting semaglutide (Wegovy) in 2021 were still on it after one year [1]. Even after shortages eased and persistence improved to 60.9% by mid-2024, that still means nearly 4 in 10 patients stop within 12 months [1]. For tirzepatide (Zepbound), one-year persistence was higher at about 64% [1]. This matters because the drugs' benefits—weight loss, cardiovascular protection—depend on continued use, and high discontinuation undermines their cost-effectiveness.

A 2025 review of the evidence base for market access confirms this, reporting real-world persistence of only 32-48% at 12-24 months, which 'undermines expected long-term value' [7]. The reasons for stopping are not well understood, but likely include side effects, cost, and supply issues. The same review notes that high prices and a projected national budget impact exceeding $100 billion annually are major barriers to equitable access [7]. Until we understand why patients stop and how to keep them on treatment, the full potential of these drugs won't be realized.

What serious risks might we be missing?

While GLP-1 drugs are generally safe in the short term, there are critical gaps in our knowledge about rare but serious adverse events. A 2026 systematic review of respiratory safety signals found that common upper respiratory infections are mild and comparable to placebo, but rare events like anaphylaxis, acute eosinophilic pneumonia, and acute respiratory distress syndrome (ARDS) have been reported in case reports and pharmacovigilance data [2]. The evidence for these serious events is rated 'very low certainty' because it comes from uncontrolled designs [2]. The same review also found a potentially meaningful reduction in pneumonia risk (hazard ratio 0.60) in a large cohort study, but this finding is also low certainty due to potential confounding [2]. So we have signals of both benefit and harm, but no definitive answers.

Another major gap is thyroid effects. A 2026 systematic review of six studies encompassing over 900,000 participants found that only one study directly measured thyroid function tests, reporting a mild decrease in free thyroxine and slight thyroid nodule progression after 12 months [9]. The other five studies focused on thyroid cancer risk and found no increased risk, but they did not systematically measure thyroid hormone levels [9]. This means we cannot rule out functional thyroid effects. Similarly, a 2022 review of case reports identified gastrointestinal problems (especially pancreatitis) as the most common adverse drug reaction, but also found reports of renal, hepatic, and immunologic issues [3]. A large 2025 VA cohort study confirmed increased risks of gastrointestinal disorders, hypotension, syncope, and drug-induced pancreatitis compared to usual care [4]. These findings highlight that while the drugs are effective, we need longer-term, systematic safety monitoring to fully characterize the risk profile.

Is it safe to take GLP-1 drugs during pregnancy or while breastfeeding?

There is almost no human safety data for GLP-1 use during pregnancy or lactation, which is a critical gap given that these drugs are increasingly prescribed to women of reproductive age. A 2023 systematic review found only 7 human studies (covering 76 offspring) and 23 animal studies [8]. In animal studies, GLP-1 agonists were associated with reduced fetal weight, delayed ossification, and skeletal variants, usually linked to reduced maternal weight gain [8]. Human data are extremely limited: one study found minimal transfer of exenatide across the placenta (fetal-to-maternal ratio ≤0.017), and another found no significant transfer of liraglutide at least 3.5 hours after maternal exposure [8]. However, these are single-subject studies, so no firm conclusions can be drawn.

A 2026 commentary highlights that current manufacturer recommendations advise discontinuing GLP-1 drugs at least 2 months before conception due to their long half-lives and lack of pregnancy safety data [6]. Despite this, inadvertent periconceptional exposure is rising as use increases [6]. The same commentary notes that a nationwide Danish cohort study is beginning to address this gap, but until more data are available, clinicians have limited evidence to guide counseling [6]. The bottom line: current evidence supports discontinuing these drugs before pregnancy, but the evidence base is thin and mostly from animal studies.

Could combining GLP-1 drugs with digital monitoring improve outcomes?

A promising but unproven approach is combining GLP-1 receptor agonists with continuous glucose monitoring (CGM) to personalize treatment. A 2025 review argues that CGM could provide real-time feedback on glucose variability and behavioral responses, while GLP-1 drugs address hyperinsulinemia and appetite regulation [10]. However, the authors note that 'no randomized controlled trials have assessed their combined use in obesity without diabetes, representing a critical evidence gap' [10]. Furthermore, CGM-derived metrics are not yet standardized for this purpose, limiting their interpretive value [10]. The review suggests that AI-driven analysis of CGM data could eventually predict individual responsiveness, but this remains speculative without trial data.

A related gap is the lack of research on how these drugs affect lifestyle behaviors. A 2026 systematic review of long-acting injectables (including GLP-1 agonists) found that while all four included studies measured pharmacologic adherence, only one measured lifestyle-related outcomes, and none evaluated whether patients compensated for drug effects by reducing healthy behaviors (so-called 'behavioral moral hazard') [5]. The authors describe a 'Measurement Asymmetry Index' of 75 percentage points, indicating a pronounced imbalance between drug-focused and lifestyle-focused outcome measurement [5]. This means we don't know if patients who take GLP-1 drugs maintain or reduce their diet and exercise efforts, which is crucial for long-term weight management.

About These Sources

This answer is built on 10 peer-reviewed studies — published from 2022 to 2026, 8 from 2024 or later, 4 in Q1 journals, collectively cited 430 times — selected as the most relevant from 11 studies that passed quality screening, drawn from 58 papers retrieved from a database of over 500 million.

Sources used in this answer

Related research questions