Two syntheses of randomized tirzepatide data, published within a year of each other, disagree about whether the drug raises the risk of acute kidney injury. A Bayesian network meta-analysis of 67 randomized trials and 199,877 participants found high-dose tirzepatide the only agent in either drug class with a significant increase, at an absolute risk difference of 0.28% and a number needed to harm of 357.[1] A meta-analysis of 15 randomized trials in 14,471 participants concluded that tirzepatide, pooled and at every separate dose, did not raise the risk of acute kidney injury against placebo, insulin or other GLP-1 drugs.[2] Same class of evidence, same molecule, opposite conclusions. The label warning that sits above both of them does not name a molecule at all.
What the warning actually says
Every label in this class carries a Warnings and Precautions section called Acute Kidney Injury Due to Volume Depletion, and the wording is close to identical across all six. There have been postmarketing reports of acute kidney injury, in some cases requiring hemodialysis. The majority of the reported events occurred in patients who experienced gastrointestinal adverse reactions leading to dehydration — nausea, vomiting or diarrhea. Monitor renal function in patients reporting reactions that could lead to volume depletion, especially during dosage initiation and escalation.[3] [4][5]
Two things about that construction are worth noticing. It is a warning about a pathway rather than a rate: the drug does not injure the kidney directly, it produces fluid losses at one end of a body whose fluid intake has already fallen at the other. And its monitoring instruction names a window, which is unusual — most label warnings say to watch, without saying when. The liraglutide label makes the chain explicit in its postmarketing list, which reads “nausea, vomiting and diarrhea leading to dehydration” as a single entry alongside acute kidney injury and increased serum creatinine.[5] The symptom at the front of that chain is covered in the vomiting article.
The label's own numbers are small, and one of them is specific
The semaglutide weight-reduction program recorded acute kidney injury in 7 treated adults — 0.4 cases per 100 patient-years — against 4 on placebo at 0.2 cases per 100 patient-years, with two further cases in other trials involving dehydration. The label adds two qualifiers that are the most useful sentences on this page: the risk of renal adverse reactions was increased in adults with a history of renal impairment, and those reactions occurred more frequently during dose titration.[3] The trials included 65 patients with moderate or severe renal impairment at baseline, which is a small enough subgroup to explain why that statement is directional rather than quantified.
The tirzepatide weight program recorded acute kidney injury in 0.5% of treated patients against 0.2% on placebo.[4] Neither program has a rate that could justify alarm in a healthy person over 68 or 72 weeks. Both have a rate roughly double placebo, in a population that trials screened before enrolling.
Why it concentrates in the titration weeks
The label says titration; a second instrument agrees from outside the trials. A pharmacovigilance analysis of 67,305 tirzepatide cases and 137,583 adverse events applied a Weibull time-to-onset model and found a median time to onset of 6.36 days(interquartile range 0.85 to 31.2), with a shape parameter of β = 0.44 — an early-failure profile, meaning the reporting rate falls as exposure continues rather than accumulating. Dehydration was among the events it flagged as carrying serious clinical outcomes.[6]
A median under a week is the arithmetic of this warning. Every dose step is a fresh escalation, the gastrointestinal effects arrive in the days after one, and the fluid deficit is the sum of what leaves and what never arrives. Someone in week two of 5 mg who has eaten little for three days and vomited twice is running a deficit from two directions at once, while their appetite for water has fallen along with their appetite for everything else. That is a different situation from the same person in month eight at a steady dose, and the titration article sets out where those steps fall.
The molecule split, three ways
Ask which drug carries the renal signal and the answer depends entirely on the instrument. The randomized network meta-analysis ranked high-dose tirzepatide most likely to induce acute kidney injury, consistently across sensitivity analyses and still so after excluding patients with baseline renal impairment; lixisenatide and three SGLT2 inhibitors came out protective.[1] The 15-trial tirzepatide synthesis found no renal excess at all, alongside meaningful reductions in urinary albumin — 26.95% at 10 mg and 18.03% at 15 mg against placebo.[2]
Postmarketing reports rank them in the reverse order. Among 133,872 reports — 92,807 tirzepatide and 41,065 semaglutide — acute kidney injury was listed in 432 (0.47%) of tirzepatide reports and 440 (1.07%) of semaglutide reports, a reporting odds ratio for tirzepatide against semaglutide of 0.44 (95% CI, 0.38 to 0.50).[7] The authors note the obvious limit: a spontaneous reporting database has no denominator, is shaped by who was prescribed what and when, and cannot establish causality. What it does show is that the instrument that reads most like real-world experience puts the signal on the molecule the randomized synthesis cleared.
Three answers, no resolution. What follows from that is not a molecule choice but a posture: the risk is real enough to have a labeled warning on every product in the class, and not established well enough to rank them against each other. The head-to-head questions that can be settled are in the molecule comparison.
The population average points the other way, and does not cancel it
At cohort scale these drugs look protective. A global retrospective study drew 3,729,925 adults with obesity from the TriNetX network and propensity-matched 12,123 new users against 12,123 untreated controls, none with type 2 diabetes, over up to five years. Treatment was associated with a lower risk of acute kidney injury, alongside lower all-cause mortality at a hazard ratio of 0.23 (95% CI, 0.15 to 0.34) — a figure large enough to signal the residual confounding an observational design cannot remove.[8] The randomized record agrees in direction: across 21 trials in 99,599 patients, acute kidney failure fell 9%.[9]
A cohort average describes people who mostly tolerated the drug and kept drinking. The warning describes what happens to someone who has not kept anything down since Friday. Those are different questions, and the second one is not answered by the first — which is also why a kidney benefit in albuminuric chronic disease, the subject of the kidney article, does not mean the class is gentle on a dehydrated kidney.
Who the risk concentrates in
Four groups carry most of it, and three are identifiable before a first dose. The first is anyone with a history of renal impairment, which the semaglutide label names directly.[3] The second is anyone on a medication that lowers blood pressure, because volume depletion and pharmacologic hypotension compound: in the pooled tirzepatide weight trials, hypotension occurred in 1.6% of treated patients against 0.1% on placebo, and in 2.2% of those on concomitant antihypertensive therapy against 1.2% of those not on it. The label states it also occurred in association with gastrointestinal events and dehydration.[4] The liraglutide label reports the same category at 1.1% against 0.5%, with systolic pressure below 80 mmHg in four treated patients and none on placebo.[5]
The third is older adults, in whom thirst is blunter, baseline filtration is lower and the margin is thinner — covered in the older-adults article. The fourth is situational rather than fixed: an acute illness, a hot week, or a bout of something unrelated landing on top of a titration step, which is the subject of the sick-day article.
What none of those groups automatically get on a cash telehealth plan is the thing the label actually asks for: renal function monitored during initiation and escalation. That is a blood test and a clinician reading it, not a questionnaire, and whether a seller does either is recorded in each provider review.
What a compounded vial does not change, and what it does
Every rate above was produced with FDA-approved product at labeled doses. Most sellers covered here dispense compounded semaglutide or tirzepatide, which is not FDA-approved and is not reviewed by the FDA for safety, efficacy or quality before it is dispensed. The volume-depletion pathway does not care which vial the molecule came from — vomiting removes the same fluid either way.
What changes is the paperwork around it. A compounded product ships without the prescribing information that carries the warning, the monitoring instruction and the titration caveat, so the sentence telling a prescriber to check renal function during escalation may not reach anyone. A pharmacovigilance analysis of 81,078 GLP-1 reports, 707 of them compounded, found higher reporting odds of hospitalization at 2.35 (95% CI, 1.94 to 2.83) and of diarrhea at 1.59, alongside far higher odds of preparation errors and contamination.[10] A dosing error in a drug whose main renal hazard is severity of gastrointestinal effect is not a neutral error.
Where the line sits
Acute kidney injury on this class is rare, labeled on every product, roughly double the placebo rate in the two weight programs, concentrated in the escalation weeks, and worst in people whose kidneys or blood pressure were already marginal. It is also the one common side effect that can turn into an admission, which is why it earns a warning that ordinary nausea does not. The rest of the tolerability picture is in what the trials recorded.
The signs the labels tell patients to report are specific: nausea, vomiting or diarrhea that does not go away, and the symptoms of kidney injury on top of it — much less urine than usual or a stretch without urinating, lightheadedness on standing, swelling in the legs or ankles, and unusual tiredness or confusion. Inability to keep fluids down for a day is the threshold at which this stops being a tolerability question. Holding the dose rather than climbing on schedule is the intervention the label's own titration finding supports.