Norovirus does not check what is in the refrigerator. A weekly injection is due on Sunday, the vomiting started Saturday afternoon, and the question that arrives at two in the morning is whether to inject anyway. The evidence that bears on it is real, specific, and smaller than the confidence of most answers to it. It sits mostly in the kidney literature, alongside the kidney outcome trials and pointing the other way.
Both labels warn about this, and neither tells anyone what to do
The Wegovy prescribing information carries a warning titled Acute Kidney Injury Due to Volume Depletion. It reports postmarketing cases of acute kidney injury, some requiring hemodialysis, and states that the majority of reported events occurred in patients who experienced gastrointestinal adverse reactions leading to dehydration such as nausea, vomiting, or diarrhea. Its instruction is directed at the prescriber: monitor renal function in patients reporting reactions that could lead to volume depletion, especially during initiation and escalation.[1] The Zepbound label carries the same warning, in almost the same words.[2]
The patient-facing half of each document is identical in substance: both instruct advising patients to take precautions to avoid fluid depletion, and to promptly report signs of acute kidney injury or persistent nausea, vomiting and diarrhea to a healthcare provider.[1][2] Drink, and tell someone. That is the entire labeled instruction.
What is not there matters as much. Across both full labels — every section of each, from the boxed warning to the patient medication guide — the words sick, illness, withhold and temporarily discontinue do not appear. Neither manufacturer wrote a rule about skipping a dose during an acute illness, in either direction. The labels do address a late or skipped injection on its own terms, and those windows differ by product rather than by molecule, which is the subject of the missed-dose article. Nothing in them converts a catch-up window into permission to hold a dose because of a stomach bug.
The mechanism is fluid balance, not the kidney
These drugs delay gastric emptying and reduce calorie intake by design, and the labeled adverse-reaction tables show what that costs in fluid. In the pooled Wegovy weight-reduction trials, 2,116 adults on 2.4 mg weekly reported nausea at 44% against 16% on placebo (n = 1,261), diarrhea at 30% against 16%, and vomiting at 24% against 6%.[1] In the pooled Zepbound trials, vomiting rose with dose — 8%, 11% and 13% at 5, 10 and 15 mg against 2% on placebo — and diarrhea ran 19% to 23% against 8%.[2]
Zepbound's table also records hypotension in 1% to 2% of treated patients against 0% on placebo, more often in patients taking antihypertensive therapy (2.2%) than in those not (1.2%), and states that hypotension occurred in association with gastrointestinal adverse events and dehydration.[2] That is the shape of the problem. An acute illness that adds vomiting or diarrhea is not landing on a neutral baseline; it lands on someone already eating less, drinking less and emptying the stomach more slowly, whose blood pressure has a little less headroom. The general tolerability curve, and how long it usually lasts, is in the nausea article.
The Wegovy table makes the point almost literally. Gastroenteritis appears in it as an adverse reaction at 6% against 4% on placebo, and viral gastroenteritis at 4% against 3%.[1] The exact event this page is about was recorded in the pivotal trials, at a rate high enough to print, and no dosing instruction attaches to it anywhere in the document.
The randomized evidence does not say the drug injures kidneys
This is where a page could easily mislead by quoting only the alarming half. A 2026 state-of-the-art review of GLP-1 safety concludes that large-scale randomized trials, meta-analyses and observational studies suggest these drugs do not significantly increase acute kidney injury risk, and may confer renal benefit in high-risk populations — and that the reported associations with acute kidney injury come primarily from pharmacovigilance and case-based evidence.[3]
Those pharmacovigilance figures are worth reading for their shape rather than their size. An analysis of the FDA Adverse Event Reporting System from 2004 to 2021 identified 2,670 acute kidney injury events associated with the class. Liraglutide accounted for the largest share at 34.98% of cases and carried the strongest association of any member, at a reporting odds ratio of 1.50 (95% CI, 1.41 to 1.60). Hospitalization was recorded in 45.28% of those cases and death in 4.23%. The median time to onset was 63 days, with an interquartile range of 15 to 458.5 days.[4]
A separate 2025 analysis of the same database, covering 2022 to 2025, compared the two molecules sold across this market. Acute kidney injury was listed in 432 of 92,807 tirzepatide reports (0.47%) and 440 of 41,065 semaglutide reports (1.07%), a reporting odds ratio for tirzepatide against semaglutide of 0.44 (95% CI, 0.38 to 0.50).[5] Two agents used for the same purpose, a two-fold gap in reporting frequency, and no way to tell from spontaneous reports whether that reflects biology or reporting behavior.
The 63-day median is the figure that travels into practice. It falls inside the dose climb rather than at steady maintenance, which is consistent with both labels naming initiation and escalation as the period to monitor, and with the argument for a slow ladder set out in the titration article. The interquartile range reaching 458 days says it is not only a titration event. Wegovy's own trials recorded acute kidney injury in 7 patients (0.4 cases per 100 patient-years) against 4 on placebo (0.2 per 100 patient-years), and state that the risk of renal adverse reactions was increased in patients with a history of renal impairment, of whom the trials included 65 with moderate or severe impairment at baseline.[1]
Formal sick-day rules exist, and were written for other drugs
There is a real body of guidance on medicines during acute illness. It comes out of diabetes, kidney and cardiovascular care, and it is explicit about which drugs it covers. A 2023 modified Delphi process assembled 26 clinicians from 4 countries and 10 disciplines and reached consensus — defined as agreement above 75% — on 42 recommendations: 5 on the signs of volume depletion that should trigger sick-day guidance, 6 on signs warranting urgent contact (reduced level of consciousness, severe vomiting, low blood pressure, ketones, tachycardia and fever), and 14 on self-management.[6]
The medication recommendations are the part usually flattened into "stop your pills". The panel reached consensus that renin-angiotensin system inhibitors, diuretics, nonsteroidal anti-inflammatory drugs, SGLT2 inhibitors and metformin should be temporarily stopped. It did not say the same about insulin. Insulin, sulfonylureas and meglitinides were to be held only if blood glucose was low, and basal and bolus insulin were to be increased by 10% to 20% if blood glucose was elevated. Resumption was recommended within 24 to 48 hours of symptoms resolving and normal eating and drinking returning.[6]
So the rule inside the one field that has rules is not "hold the diabetes drug". It is drug-specific, and for the most important drug in that list it points upward. GLP-1 receptor agonists appear in none of the 42 recommendations, in either direction. That panel was convened for people with diabetes, kidney disease or cardiovascular disease, which is not the population most cash-pay buyers belong to — a distinction drawn in the article on the two indications.
Holding medicines has been tested once, and did not work
A 2022 pragmatic randomized trial enrolled 315 veterans with stage 3 to 5 chronic kidney disease who were taking a renin-angiotensin-aldosterone blocker, a diuretic, an anti-inflammatory or metformin. 159 were assigned to a sick-day protocol delivered by interactive voice response and 156 to usual care, with six months of follow-up and a mean age of 70.1 and 69.2 years.[7]
Adjusted change in filtration rate was −0.71 (95% CI, −2.11 to 0.69) in the protocol group and −0.72 (95% CI, −2.12 to 0.68) in usual care, P = 0.99. Hospitalizations ran 11.5 per 100 against 8.4 per 100 person-months, an adjusted prevalence ratio of 1.30 (95% CI, 0.96 to 1.76) — not statistically significant, and pointing the wrong way for the intervention. Execution was the other finding: across 33 genuine sick days, participants followed the protocol correctly in 14.[7] The trial's own stated limitation is that sick days were uncommon over six months, so this is an underpowered null rather than a demonstration of harm.
Nephrology has not settled the argument. A 2026 perspective makes the case for sick-day guidance as routine care while naming the objections squarely: confusion over what counts as a sick day, no clarity on how long to hold or when to restart, concern that the instruction breeds hesitancy about drugs with proven benefit, and a lack of data showing that holding helps.[8] Those objections concern classes with far more evidence behind them than a weekly incretin has.
Hypoglycemia is a question about the rest of the prescription
The generic warning that these drugs cause low blood sugar is true of a minority and misleading for everyone else. Zepbound's labeling reports hypoglycemia, defined as plasma glucose below 54 mg/dL, in 4.2% of treated patients against 1.3% on placebo in a trial of patients with type 2 diabetes. Within that trial, patients also taking a sulfonylurea reported hypoglycemia at 10.3% against 2.1% in those who were not.[2] The co-prescription moved the rate roughly five-fold; the GLP-1 alone did not.
Wegovy's adverse-reaction table names its row "Hypoglycemia in T2DM" and reports it at 6% against 2% on placebo. Its labeling states that risk increased when semaglutide was used with insulin or a sulfonylurea, and that use in type 1 diabetes and use in combination with insulin have not been evaluated.[1] A buyer taking a compounded GLP-1 with no insulin and no sulfonylurea in the regimen is not the person those sentences describe. Who does belong in a higher-risk group is set out in the contraindications article.
A 2026 case report shows what the combination looks like when it goes wrong outside a clinic. A man in his mid-30s with long-standing type 1 diabetes presented with severe vomiting, diarrhea, hyperglycemia, ketonemia and acute kidney injury shortly after self-administering a product bought online as retatrutide; his partner took the same preparation and also developed symptoms. Peak ketonemia reached 4.3 mmol/L with dehydration and insulin omission, and recovery was complicated by recurrent hypoglycemia requiring intravenous dextrose. Stool culture then grew Shigella flexneri, so causation cannot be assigned.[9] The report's own framing is the transferable part: products obtained outside a clinical pathway arrive without assessment, without counseling on adverse effects or sick-day management, and without follow-up.
A compounded vial inherits the physiology and none of the paperwork
Every sentence quoted above belongs to two approved products. Compounded semaglutide and tirzepatide are prepared by a pharmacy against a prescription; they are not FDA-approved, and the agency does not review them for safety, efficacy or quality before they are dispensed. No prescribing information ships with the vial, so the warning about volume depletion and the instruction to report persistent vomiting reach the buyer only if the seller chose to reproduce them.
That converts a clinical question into a service question, answerable before the first shipment rather than during a fever. Whether a prescriber can be reached on a Saturday, and whether anyone reads a report of three days of vomiting, are the kind of thing recorded in the provider write-ups, and their absence is one of the patterns catalogued in the telehealth article.
What none of this decides
Whether a particular dose should be taken, delayed or held during a particular illness is a question for the prescriber who wrote it, and there is no published rule that answers it from a distance. The labels describe a risk and ask for a phone call. The one randomized test of deciding in advance to hold medicines did not improve kidney outcomes. The consensus guidance that does exist names five drug groups and adjusts a sixth upward, and does not name this class at all.
What the surrounding literature does suggest is that the decision is rarely as urgent as the hour makes it feel. A 2024 case series of 75 medication-related hospital admissions for acute kidney injury or dehydration in patients aged 65 and over found that 80% had a non-acute onset of symptoms and 60% had contacted their general practitioner within the two weeks before admission. Around 40% of those admissions — 29 of 75 — were judged potentially preventable had medication been adjusted in time.[10] The failure mode is a slow one with a contact point in it, which argues for making contact early rather than adjudicating a dose alone. How every figure here was checked is described in the methodology.