Welcome to the module on adverse effects and safety evidence. GLP-1 receptor agonists can change both eating and digestive sensations, but those effects are not interchangeable. This lesson examines a specific claim: that nausea or other gastrointestinal symptoms show whether a medicine is suppressing appetite or “working” for weight loss. We will use trial data to separate an effect that co-occurs with weight loss from one that explains it.
Appetite change is not the same experience as nausea
Feeling satisfied after a smaller meal differs from feeling too unwell to eat. GLP-1 receptor activation can influence appetite regulation and food intake while also producing gastrointestinal adverse effects such as nausea, vomiting, diarrhoea, or constipation. These effects may overlap in one person, but one is not a required sign of the other.
There is a plausible reason to suspect a connection: nausea can temporarily reduce how much someone eats. Yet that possibility does not establish that nausea drives most weight loss during treatment. A symptom might last days, whereas weight change is measured over months. Nor does the absence of a reported adverse event mean that appetite signalling has not changed.
The useful distinction is between two questions. Do people who report symptoms sometimes lose more weight? They do in some analyses. Does having those symptoms account for most of the medicine’s weight-loss effect? That requires a different analysis.
What the STEP trials actually show
The following paper examines semaglutide 2.4 mg in three randomized weight-management trials. Read its results before interpreting the accompanying chart.
Read the STEP 1–3 analysis to see how weight change differed between participants who did and did not report gastrointestinal adverse events.
In Results, Section 3.5, “Impact of gastrointestinal adverse events on weight loss (STEP 1–3),” read the results, starting with the observed weight changes and ending just before the Figure 4 caption. Examine the Figure 3 and Figure 4 captions in the same section. Focus on the distinction between the observed symptom-group comparisons and the later estimate of how much weight loss symptoms mediated.

In STEP 1, the semaglutide group with a reported gastrointestinal adverse event had lost an average of 17.3% of starting body weight; the group without one had lost 15.7%. STEP 2 showed 11.4% versus 9.6%, and STEP 3 showed 17.7% versus 17.1%. The groups reporting symptoms sometimes had somewhat greater average loss, but substantial loss occurred without reported symptoms in every trial. The analysis limited to nausea, diarrhoea, and/or vomiting tells a similar story.
Read the chart as a comparison of group averages, not as a prediction for an individual. Its labels show that the groups differ in size. Moreover, these are observed measurements from people on treatment at week 68 who had a weight assessment—not a record of everyone originally randomized. The chart establishes neither that nausea caused the difference between groups nor that a particular person without nausea will lose weight.
How much of the effect might symptoms explain?
The paper also reports a mediation analysis. Rather than merely comparing people who did and did not report symptoms, this statistical approach estimates how much of semaglutide’s weight-loss effect relative to placebo could be attributed to the occurrence of those symptoms.
Across the three STEP trials, semaglutide’s observed placebo-corrected weight loss ranged from 7.6 to 14.4 percentage points. The estimated portion mediated by gastrointestinal adverse events was less than one percentage point, whether the analysis counted any gastrointestinal event or specifically nausea, diarrhoea, and/or vomiting. “Less than one percentage point” refers to the estimated share of the weight-change difference; it does not mean that fewer than 1% of participants had symptoms.
The result supports a careful conclusion: common gastrointestinal adverse events appear to explain little of the average weight-loss benefit in these semaglutide trials. It does not mean symptoms never affect anyone’s food intake. Equally, the weight-loss effect not explained by reported symptoms cannot all be assigned to one precisely identified brain pathway. Appetite regulation is a plausible major contributor, but this analysis does not measure every part of that mechanism directly.
Return to the paper for the authors’ interpretation of the results and their proposed explanation involving appetite regulation.
In the Discussion, find the paragraph beginning “GI disorders are well-known side effects.” Read that paragraph through its closing reference to the hypothalamus and hindbrain. Notice that the authors connect the mediation result to reduced energy intake and central appetite regulation; they do not use nausea as a measure of how strongly appetite was suppressed.
This distinction matters because an adverse event is an imperfect proxy. It is recorded when a participant reports a symptom; it is not a direct measurement of appetite, energy intake, receptor activation, or future weight change. Someone can notice earlier fullness without classifying it as an adverse event. Someone else can report nausea while experiencing a different degree of appetite change.
A second dataset—and a limit on causal interpretation
Analyses of semaglutide trials in people with type 2 diabetes offer a useful check. They involve different doses and populations from STEP, so the numbers should not be treated as interchangeable, but they address the same proposed link between symptoms and weight loss.
Read the SUSTAIN trial analysis for a second estimate of symptoms’ contribution, then read the authors’ warning about what a post-hoc mediation analysis can establish.
In Results, read Sections 3.4, “Weight loss by GI AEs—post hoc analysis,” and 3.5, “Direct and indirect effects on weight loss—mediation analysis,” from the symptom comparisons through the mediation finding. Then, in the Discussion, locate the final paragraph beginning “Key limitations include” and read the limitations. Focus on why random assignment to a drug does not also randomize who develops nausea.
In SUSTAIN 1–5, average weight loss was generally somewhat greater among participants reporting nausea or vomiting, yet weight loss also occurred without those events. The estimated amount explained by nausea or vomiting was only 0.07–0.5 kg of the greater weight loss seen with semaglutide versus comparators. Many events were transient, another reason not to equate a short-lived symptom with the full course of weight change.
There is an important statistical boundary here. Participants were randomized to treatments, not to having or avoiding nausea. Once researchers divide a trial into symptom groups after treatment begins, those groups may differ for other reasons. A mediation model tries to account for such differences, but its causal interpretation depends on assumptions, including that relevant confounding has not been missed. The STEP chart also excludes people without the specified on-treatment week-68 measurement. These limitations counsel against saying that the analysis has proved symptoms contribute nothing. They do not erase its consistent central finding: reported gastrointestinal events account for little of the average weight-loss effect studied.
The interpretation to carry forward
For the semaglutide populations studied, not feeling nauseated is not evidence that appetite effects are absent or that weight loss cannot occur. Conversely, experiencing nausea is not a dependable sign of stronger appetite suppression or greater eventual weight loss. Average differences between symptom groups cannot serve as an individual efficacy test.
This is not a reason to dismiss gastrointestinal symptoms. They matter as adverse effects in their own right, regardless of what they imply about efficacy. It is also not a license to assume identical findings for every GLP-1 receptor agonist or dual-agonist medicine: the strongest data examined here concern particular semaglutide regimens and trial populations.
Conclusion
Appetite change and gastrointestinal discomfort can accompany the same treatment without one reliably measuring the other. In STEP and SUSTAIN analyses, substantial weight loss occurred without reported gastrointestinal adverse events, while mediation estimates attributed only a small part of the average treatment effect to those events. Because symptom groups were formed after randomization, the estimates warrant care rather than an absolute claim of zero contribution.
The next lesson turns from these common symptoms to serious safety concerns, where the challenge is different: judging how certain the evidence is for an uncommon or potentially consequential harm.