When the stomach slows down or contents reflux back into the esophagus, the consequences for a hospitalized patient can range from uncomfortable to serious. Researchers have long known that certain medications can disrupt the muscles and nerves that move food through the gut, but a clear picture of which drugs carry the highest risk, and how quickly that risk appears, has been hard to assemble. A recently published analysis in PLoS One set out to fill that gap.
The research team, led by Qian Zhiheng and colleagues, turned to two large pharmacovigilance databases: the FDA Adverse Event Reporting System, which contained more than 58 million reports spanning 2004 to 2025, and Canada's Vigilance Adverse Reaction Online Database, used as an independent check. By applying three separate statistical algorithms, the researchers looked for drug-event pairs where adverse reports clustered far more than chance would predict. Their focus was delayed gastric emptying and gastroesophageal reflux, two conditions that are common, often underrecognized, and potentially preventable.
Study design and methods
The core technique used here is called disproportionality analysis. The idea is straightforward: if a particular drug appears alongside a particular adverse event far more often than it does alongside all other adverse events, that imbalance, or disproportionality, is worth investigating. The team used three well-established measures of disproportionality: the Reporting Odds Ratio, the Proportional Reporting Ratio, and a Bayesian method called the Confidence Propagation Neural Network. Requiring a positive signal across all three algorithms simultaneously reduces the chance of a false positive.
From the top 50 drugs flagged in the initial screen, 20 cleared the threshold on all three methods. The researchers then validated those signals against the Canadian database to see whether the pattern held in a completely independent dataset. Finally, they used a statistical technique called Weibull time-to-onset analysis to estimate when, after starting a drug, adverse events tended to appear. That temporal piece turned out to reveal meaningful differences between drug classes.
GLP-1 receptor agonists and the strongest signals
The findings place GLP-1 receptor agonists at the top of the signal list by a considerable margin. As a class, these peptide-based drugs, which mimic the natural gut hormone glucagon-like peptide-1, were most strongly associated with gastric motility disorders in the dataset. The specific compound with the highest Reporting Odds Ratio for impaired gastric emptying recorded an ROR of 80.27 (95% confidence interval: 76.39 to 84.34). When the researchers cross-checked that figure in the Canadian database, the signal remained strong, returning an ROR of 54.17.
GLP-1 receptor agonists are not new to discussions about gastrointestinal tolerability. These molecules slow gastric emptying as part of their pharmacological profile, a mechanism that is central to how they influence post-meal glucose and appetite signaling. The analysis here adds population-level weight to that mechanism, suggesting that slowing is not simply a minor side note but a statistically prominent feature across a very large real-world dataset.
Importantly, this is an observational pharmacovigilance study, not a controlled clinical trial. Disproportionality signals reflect reporting patterns, not confirmed causation, and databases like FAERS are known to include incomplete information and reporting biases. The authors themselves frame the findings as evidence-based priorities for enhanced monitoring rather than definitive proof of harm.
Other drug classes with notable signals
GLP-1 receptor agonists were not the only class flagged. Insulin formulations also appeared in the analysis, with one long-acting insulin analog recording an ROR of 18.90 for gastric motility-related events. The study also identified signals for bisphosphonates, a class of drugs commonly used in bone metabolism research, angiotensin receptor blockers, which act on blood pressure regulation pathways, and trofinetide, a synthetic analog of the tripeptide glycine-proline-glutamate that has been studied in neurological contexts.
These associations are pharmacologically diverse, which underscores the researchers' broader argument: gastric motility disruption is not the exclusive territory of any one drug class. Multiple mechanisms, from direct effects on gut muscle receptors to indirect effects through the autonomic nervous system, can converge on the same clinical outcome.
Timing patterns across drug classes
One of the more clinically informative parts of the study is the Weibull time-to-onset analysis. Not all drug-associated motility problems appear on the same schedule, and the data reflected striking differences.
Trofinetide showed a median time to onset of just 6.6 days, making it an example of what the authors categorize as early-onset signal. At the opposite end, immunoglobulin G preparations showed a median onset of 535.1 days, suggesting a process that builds over a much longer exposure period. GLP-1 receptor agonists fell somewhere in between. Understanding these timelines matters because a problem appearing within days of starting a drug is more likely to be recognized and attributed correctly than one that emerges gradually over months or years.
The divergence in onset timing also hints at different underlying mechanisms. A rapid-onset signal may reflect a direct pharmacological action on gastric smooth muscle or enteric neurons. A delayed signal could point to something more structural or adaptive, such as changes in receptor expression or motility nerve remodeling over time. The study does not resolve those mechanistic questions, but the temporal data provide a framework for future investigation.
Implications for pharmacovigilance research
The authors describe their findings as providing evidence-based priorities for enhanced pharmacovigilance. In practical research terms, that means this kind of large-scale signal-detection work can highlight which drug-event combinations most deserve closer scrutiny in prospective studies, mechanistic laboratory work, or carefully designed clinical trials.
For the peptide research community specifically, the prominence of GLP-1 receptor agonist signals in this dataset is a reminder that peptide-based compounds interact with the gut in ways that extend well beyond their primary intended targets. The enteric nervous system is densely populated with peptide receptors, and modulating those receptors, whether intentionally or as an off-target effect, can ripple through gastric function in measurable ways.
The study also demonstrates the value of cross-database validation. Finding a large signal in FAERS is notable, but the fact that the signal survived a check against the Canadian database adds meaningful confidence that the pattern is not an artifact of one reporting system's quirks.
Limitations and context
Pharmacovigilance databases have well-documented limitations. Adverse events are underreported in general, and reporting rates can spike when a drug receives heightened media or regulatory attention, inflating disproportionality scores for high-profile compounds. GLP-1 receptor agonists have attracted substantial public and clinical attention in recent years, which could contribute to elevated reporting rates independent of actual risk. The authors acknowledge this possibility.
Additionally, these databases lack denominator data. Without knowing how many patients were actually exposed to each drug without incident, ROR values describe reporting proportions rather than absolute incidence rates. A high ROR means the event is disproportionately reported alongside that drug, not that a specific percentage of users will experience the outcome.
With those caveats in mind, the study represents a useful contribution to the literature on drug-induced gastrointestinal effects. The breadth of the dataset, the use of three independent algorithms, and the cross-database validation all strengthen the signal-detection methodology. Future controlled research will be needed to move from statistical signal to confirmed mechanism and quantified risk.
