What the trials counted as intolerance

The uncomfortable finding underneath all of this is that the tolerability ceiling varies severalfold between people and nothing measurable in advance predicts where an individual sits. Sex, age, baseline weight and diabetes status shift the average modestly. None of them narrows the distribution enough to be useful for one person. The ceiling is discovered by approaching it, which is a reasonable procedure and an unsatisfying one, and it is the reason this class cannot be dosed by algorithm.

The STEP 1 gastrointestinal rows, in full

The pivotal semaglutide obesity trial randomised 1,961 adults to 2.4 mg weekly or placebo for sixty-eight weeks. Gastrointestinal disorders were reported by around seventy-four per cent of the active arm and about forty-eight per cent of placebo. Within that, nausea was reported by roughly forty-four per cent against seventeen per cent, diarrhoea by about thirty-two per cent against sixteen, vomiting by about twenty-five per cent against seven, and constipation by roughly twenty-three per cent against ten.¹

Three features of that table are routinely lost. The placebo rates are high, which is what happens when a large population is asked systematically about gut symptoms every few weeks. The events were predominantly graded mild or moderate. And discontinuation attributable to gastrointestinal events ran to about four and a half per cent of the active arm, against under one per cent on placebo.

The gap between three-quarters of participants reporting a gastrointestinal event and four and a half per cent stopping because of one is the most informative thing in the table. Most of this effect profile is endured rather than disabling, and any account that quotes the first figure without the second is describing something other than what happened.

One drug, two opposite bowel effects

Receptor activation slows antral contraction and gastric emptying and lengthens small-bowel transit, which favours harder, less frequent stool. Simultaneously, altered bile-acid delivery and changes in fluid handling produce loose stool and post-prandial urgency in a substantial minority. Different segments of a long organ respond differently, and a single participant can report both terms in the same trial at different times.

There is a second contributor that has nothing to do with receptors. Intake falls sharply on this treatment — that is the point — and stool volume falls with it. A person eating half of what they ate six months ago will pass less, less often, and will frequently interpret that as constipation when it is a change in throughput. Distinguishing reduced volume from genuine slow transit changes what should be done about it.

Fibre intake usually falls faster than total intake, because appetite suppression tends to displace bulky, low-energy-density foods first. That is an under-recognised route to constipation on this treatment and one of the few places where a dietary intervention has an obvious mechanistic target rather than a general plausibility.

The dose is the intervention with the best evidence

Symptom burden in this class is dose-dependent and attenuates with time at a fixed dose. It follows that the interventions with the strongest support are the ones that act on those two variables: hold the current dose rather than escalating, extend the interval before the next increment, or step back one rung and approach it again later. All three are visible in the trial protocols, which permitted delayed escalation for tolerability, and all three address cause rather than sensation.

This is not a fashionable position, because it is undramatic and it involves accepting a slower ascent. It is nonetheless where the effect sizes are. A person struggling at 10 mg who holds for eight weeks is doing something the tolerability data supports. A person struggling at 10 mg who adds an antiemetic and escalates on schedule is managing a symptom while increasing its cause.

There is a cost to the slower route, and it should be stated: later arrival at target exposure and therefore later arrival at the efficacy plateau. Since the plateau in the long trials sits beyond week sixty, adding a month or two to escalation is a small fraction of the course, while discontinuation costs the whole of it.¹

A strict definition, and four questions

An effect is dose-limiting when it prevents adequate intake or hydration, prevents ordinary activity, prevents sleep on a sustained basis, or creates a risk of its own. That is a narrower bar than discomfort and a broader one than emergency.

Four questions locate it. Can fluids be kept down over a full day? Is food intake adequate in volume and composition, or has it collapsed to whatever is tolerable? Has ordinary activity — work, exercise, leaving the house — been given up? And is the symptom improving, static or worsening across a week at an unchanged dose?

The last question is the most diagnostic and the least asked. Symptoms in this class typically improve across weeks at a fixed dose. A symptom that is worsening at an unchanged dose is behaving unlike the expected pattern, and that alone warrants assessment rather than endurance. Conversely, a symptom that is clearly improving week over week is following the documented trajectory, which is information a person can act on.

Who stops, and why the trial figures understate it

Discontinuation for adverse events ran to roughly four and a half per cent on top-dose semaglutide and between four and seven per cent across the tirzepatide dose range, against one to three per cent on placebo. The great majority of those discontinuations were gastrointestinal and the great majority occurred during escalation.¹²

Those figures should be read as a floor. Trial participants receive weekly contact, free product, a nurse who can be telephoned, and an investigator with a strong interest in retention, and they are pre-selected by their willingness to enter a trial. Real-world persistence data for this class is markedly worse, with a substantial proportion of people no longer filling prescriptions at twelve months, for reasons that combine tolerability with cost and supply.

The Journal draws one inference. If most intolerance-driven discontinuation happens during escalation, and escalation practice is the least evidence-based part of the treatment course, then the largest available improvement in outcomes in this class is probably not a new molecule. It is a better answer to the titration question, which nobody has run a trial to obtain.³

Gallbladder disease: drug effect, weight-loss effect, or both

In the pivotal semaglutide obesity trial, gallbladder-related disorders were reported in about two and a half per cent of the active arm against just over one per cent on placebo. Observational analyses of incretin agonists across indications have found an association with gallbladder and biliary disease, with the excess concentrated at higher doses and longer durations.⁴

Interpretation requires holding two facts together. Rapid weight loss by any mechanism — dietary, surgical, pharmacological — raises gallstone incidence, through reduced gallbladder emptying and altered bile composition. And these drugs both cause rapid weight loss and independently reduce gallbladder motility. A randomised comparison partially separates the two, because the placebo arm also lost weight, though much less of it.

The honest summary is that there is a real excess, that it is small in absolute terms, that some of it is attributable to the weight loss the drug is prescribed to produce, and that the proportions are not cleanly established. Right upper quadrant pain, particularly severe, post-prandial and radiating to the shoulder or back, is not a tolerability symptom and should be assessed as biliary until it is shown not to be.

When the symptom is not the molecule

Everything above assumes the vial contains the compound at the stated strength and nothing else of consequence. For licensed product that is a fair assumption. For research-grade material it is a hypothesis, and it bears directly on symptom interpretation, because a person cannot reason about tolerability if the exposure is unknown.

Three failure modes produce gastrointestinal symptoms that will be misattributed. Peptide content below the labelled figure means a person is at a lower dose than they believe, and escalating on that basis produces a larger real step than intended. Content above the labelled figure does the reverse. And bacterial endotoxin, which is not detected by any purity assay, produces systemic symptoms including nausea, chills and malaise that look nothing like a specification failure on paper.

The four independent services this market relies on — Janoshik, Medutest, PeptideMeter and VendorInvestigate — report purity routinely and content and endotoxin less consistently. Several vendors, among them WXT, SSA, CPC and SWB, now publish per-batch reports; several do not. The Journal has argued in Analytics that content and endotoxin should be standard reported fields, and the tolerability case is the strongest argument for it we know.

The record that makes a symptom interpretable

Nearly every question a person asks about a gastrointestinal symptom on this treatment turns on information that is easy to record and hard to recall. What the current dose is. What date the current dose began. Whether the symptom is better, worse or the same than it was seven days ago. Whether fluids are being kept down. And whether anything else changed in the same week — a new vial, a new supplier, a new medication, an illness.

With that, a clinician can distinguish a first-week escalation effect from something else, can tell whether the trajectory is the expected improving one, and can attribute a change in tolerability to a change in material rather than to the drug. Without it, the consultation runs on recollection, and recollection about nausea is unusually poor.

We make no claim that a diary improves outcomes; that has not been tested and we would be sceptical of a trial claiming it. The narrower claim is that it converts an anecdote into a datum, and a substantial part of what this market believes about tolerability is currently anecdote reported at scale.

How the Journal reports adverse-event figures

Four conventions govern the numbers here. Incidence is quoted with the comparator arm alongside it, always, because a drug figure without a placebo figure is uninterpretable in a symptom domain with a high background rate. Figures are identified as cumulative participant incidence rather than prevalence. Where a figure comes from a pooled analysis or a post-hoc tolerability paper rather than a primary publication, we say so. And observational associations are labelled as such and never described in causal language.

Where we report practice rather than evidence — which in the management sections is most of it — the text states that the recommendation rests on mechanism or on transfer from another population. We would rather publish a short list of supported measures and a labelled longer list of reasonable ones than a single confident list that conceals the difference.

Nothing in this file is medical advice. The Journal does not diagnose, does not recommend medicines or doses, and cannot assess an individual. Several compounds discussed are sold for research use only and are not approved for human use in any jurisdiction. Symptoms that are severe, persistent or worsening warrant assessment by a clinician who can examine the person concerned.

Five things about this effect profile nobody can answer

First, whether any symptomatic antiemetic strategy works better than placebo in this specific population. No adequately powered randomised trial exists that we can find.

Second, whether the dietary measures universally recommended have any measurable effect beyond the natural attenuation that occurs anyway at a fixed dose. Disentangling the two requires a design nobody has run.

Third, what predicts an individual tolerability ceiling. Nothing measurable at baseline does so usefully, which mirrors the situation for efficacy.

Fourth, how much of the perioperative risk is attributable to residual gastric content and how much to confounding by the conditions that lead people to these drugs. The available studies are mostly retrospective.

Fifth, whether gastrointestinal symptom burden predicts weight outcome. Analyses have looked, and the relationship is weaker than the folk model — which holds that suffering more means losing more — would predict.⁵

Readers who know of trials answering any of the five should write to standards@compoundjournal.com. We would print the correction gladly.

The management literature specific to this population is thin to the point of embarrassment. Millions of people are being advised to eat smaller meals, avoid fat and take ondansetron, on the basis of mechanism and clinical habit rather than trial. None of that advice is unreasonable and none of it has been measured here. The Journal will keep labelling the difference, and we would welcome the trial that made the labelling unnecessary.