In clinical trials, people taking a GLP-1 drug for weight stayed on treatment more reliably than the people who got a comparator. They also dropped out because of side effects more than twice as often. Both are true. The gap between them is what a new meta-analysis set out to explain.
The paper, published online July 26 in Obesity Reviews ↗, pooled randomized controlled trials that tested GLP-1 receptor agonists for overweight or obesity and reported how many participants stopped taking the drug. Efrata Shegena of the University of Arkansas for Medical Sciences, with coauthors at North Dakota State University and the University of Texas at El Paso, ran the search through PubMed, Embase, and Web of Science up to October 2024. Nonadherence here means a participant came off the assigned treatment before the trial ended, for any reason.
The headline number points one way. Overall nonadherence was 37 percent lower in the GLP-1 groups than in the control groups, a pooled risk ratio of 0.63, meaning the drug groups were about two-thirds as likely to fall off. Part of that is loss to follow-up, people who simply drift out of a study; that was less than half as common on the drug (risk ratio 0.43). A participant who is watching the scale move has a reason to keep showing up.
Then the number that points the other way. When the authors isolated dropouts caused by adverse drug reactions, the GLP-1 groups quit 2.3 times as often (risk ratio 2.29), driven mostly by gastrointestinal effects: nausea, vomiting, diarrhea, the familiar tax of these drugs. So the same treatment that keeps most people engaged pushes a specific subset out through the side-effect exit.
That reconciles two claims that usually get argued past each other. The optimistic take, that people love these drugs and stay on them, and the pessimistic take, that people cannot tolerate them and quit, are both describing real subgroups of the same trial populations. Net persistence is higher. The people who leave leave for a sharper, more clinical reason.
Two cautions sit on top of the numbers. First, this is trial data, not the pharmacy counter. Randomized trials prop up adherence with reminders, free drug, and staff who call when a dose is missed. Real-world persistence on GLP-1 drugs is famously worse, with a large share of patients off the drug within a year. The meta-analysis frames trial adherence as a ceiling for what to expect, not a forecast. Second, the pooled trials used different drugs at different doses over different lengths, and the abstract does not report a single count of studies or patients, so the summary figure smooths over real variation between, say, a low starting dose and a high maintenance one.
The practical read is narrow and useful. If the reason people quit in the cleanest possible setting is gastrointestinal, then tolerability is the adherence lever, not motivation. Slower dose escalation, anti-nausea support, and formulations that go easier on the gut are what move the number. Cost is the other exit, and it lands hardest where the drugs cost the most: an earlier look at pricing found a month of GLP-1 therapy can run past a minimum-wage worker's monthly pay in poorer countries ↗.
The drugs tested across these trials are the same two that anchor most GLP-1 coverage: semaglutide ↗ and tirzepatide ↗, both built as cards against the GLP-1 receptor ↗.