Tirzepatide vs Semaglutide 1 mg: One Trial Randomized 1,879 Patients and Cleared No Side-Effect Gap at the Corrected Bar

Our SURMOUNT-5, SUSTAIN 7 and ACHIEVE-3 pieces each read one randomized trial's own registry counts to compare two GLP-1 drugs without splicing together rates from different studies. This is the fourth: the same drug pair as SURMOUNT-5 — tirzepatide against semaglutide — but a different trial, a different population, and three tirzepatide doses against a single semaglutide arm instead of one dose against one.

SURPASS-2 (NCT03987919) randomized 1,879 adults with type 2 diabetes on metformin, 1:1:1:1, to once-weekly tirzepatide 5 mg, 10 mg or 15 mg, or once-weekly semaglutide 1 mg. It was open-label: participants and investigators knew which drug and dose was given. Its primary outcome was change in HbA1c at week 40, and adverse events were counted from baseline through a safety follow-up window of up to 44 weeks. Of 1,879 randomized, 470, 469, 470 and 469 received at least one dose, and 452, 442, 446 and 443 completed the trial. Its results were posted to the registry, and 24 rows — six tracked effects across the four arms — entered our corpus on 27 September 2026; every row was re-read against the registry's posted counts on 28 September 2026, and arm, term, numerator and denominator all match. This is the same trial the vomiting-rates piece draws one pooled row from and the SURMOUNT-5 article mentions in passing; this piece is the trial's own four-arm table, on its own.

Read this as one trial, not our pooled estimate. Our pooled clinical rate for each effect lives at the data API and averages dozens of studies with different populations, doses and durations. This article is one study's adverse-event table, useful for one reason: randomization inside it removes the population and duration confounds that make cross-trial comparisons of these two drugs unreliable. It cannot tell you how common an effect is in general, and it is not a placebo comparison.

Three tirzepatide doses, one semaglutide arm

Unlike SURMOUNT-5, which pairs one dose of each drug, or SUSTAIN 7 and ACHIEVE-3, which pair two doses of each, SURPASS-2's own design compares three tirzepatide doses against the same semaglutide 1 mg arm — there is no per-dose semaglutide comparator in this trial. The registry's non-serious adverse-event table lists a term only if it reached 5% in at least one arm; seven terms did. Six are effects we track and are stored in our corpus. The seventh, dyspepsia (34, 29, 43 and 31 people in the four arms, 6.2–9.1%), is excluded from our taxonomy by design and is not part of any table below.

Every cell is the registry's count divided by the same arm's at-risk denominator, read on 28 September 2026.

Two example rows, verbatim from our stored evidence: "In the "15 mg Tirzepatide" arm (n=470 at risk), Nausea occurred in 104 participants (22.1%)"; "In the "1 mg Semaglutide" arm (n=469 at risk), Nausea occurred in 84 participants (17.9%)."

Eighteen comparisons, one bar

Three tirzepatide doses times six effects is eighteen tests, each tirzepatide dose against the same semaglutide arm: a two-proportion z-test, pooled standard error, no continuity correction, with a Bonferroni correction for testing eighteen comparisons at once (critical |z| ≈ 2.99 for a family-wise 5%, against the usual single-test 1.96; running eighteen uncorrected tests carries roughly a 60% chance of at least one false "finding" even if nothing genuinely differs).

No effect clears the corrected bar (|z| ≥ 2.99) at any dose. The two largest magnitudes — diarrhoea at 10 mg (z=2.17) and reduced appetite at 15 mg (z=2.15) — clear the single-test 1.96 line by about 0.2 and fall about 0.8 short of the corrected one. That is the band our ACHIEVE-3 and SUSTAIN 7 pieces call suggestive (ACHIEVE-3's low-pair constipation at 2.60 and acid reflux at 2.06; SUSTAIN 7's low-pair diarrhoea at 2.58 and headache at 2.19), and we read these two the same way: worth noticing, not established. With eighteen tests, chance alone would put about one gap in that band even if nothing differed. Read plainly: across six tracked GI/appetite effects and three tirzepatide doses, none showed a side-effect rate distinguishable from semaglutide 1 mg once multiple comparisons are corrected for, in this trial's stored rows. That is itself the finding, not an absence of one. It contrasts with SURMOUNT-5, where the same two drugs at their own maintenance doses in a different (obesity, not diabetes) population produced an injection-site-reaction gap that did clear its bar.

Beyond the six tracked effects

The registry's adverse-event module also reports how many participants in each arm had at least one serious adverse event: 33 of 470 (7.0%) on tirzepatide 5 mg, 25 of 469 (5.3%) on 10 mg, 27 of 470 (5.7%) on 15 mg, and 13 of 469 (2.8%) on semaglutide 1 mg. These are whole-arm serious-event counts, not per-term rates, are not part of the eighteen-test family above, and this article does not have the per-term serious-events table to say what drove them — read them as descriptive, not as a nineteenth test.

A published abstract of this trial (PMID 34170647) states, for tirzepatide's three doses and semaglutide 1 mg respectively: "nausea, 17 to 22% and 18%; diarrhea, 13 to 16% and 12%; and vomiting, 6 to 10% and 8%, respectively" — ranges across the three tirzepatide doses, matching the spread in the table above (nausea 17.4–22.1%, diarrhoea 13.2–16.4%, vomiting 5.7–9.8%) and a single semaglutide figure in each case, matching the semaglutide column here (17.9%, 11.5%, 8.3%) to the abstract's whole-percent rounding. Because the abstract and the registry describe the same people, only the registry rows count toward our pooled estimates; the abstract's semaglutide figures are held in our corpus but withheld as duplicates.

What this does not tell you

No placebo arm. SURPASS-2 is drug versus drug, so nothing here separates either drug's effect from a background symptom rate, unlike our constipation or headache pieces.

A diabetes trial at diabetes doses. The population had type 2 diabetes on metformin. It says nothing directly about tirzepatide or semaglutide in obesity without diabetes, and the two drugs' maintenance doses in that population (up to 15 mg and 2.4 mg) differ from the doses tested here.

One trial, open-label. Neither patients nor investigators were blinded, which matters most for symptom-type events like nausea: knowing which drug you are on can change what you notice and report. None of that changes a registry count, but it bounds what a count can mean.

A 5% listing threshold. Fatigue, dizziness, hair loss and the other tracked effects absent from the non-serious table did not reach 5% in any arm; that is not the same as not occurring.

Up to 44 weeks. Each arm includes its titration weeks, so these are at-least-once shares over the whole safety-follow-up window, not rates at a steady dose.

Three doses, one comparator. Semaglutide 1 mg is not that drug's top maintenance dose (2.4 mg, tested in SURMOUNT-5); this trial cannot say what a higher semaglutide dose would have shown against tirzepatide.

Multiple comparisons, restated. All eighteen gaps here are noise at the corrected bar; the two closest (10 mg diarrhoea, 15 mg reduced appetite) are worth watching in a future trial, not treating as established.

If you are on a GLP-1 now

This is educational content, not medical advice, and it does not tell you which drug is right for you — that depends on your own history, your prescriber's judgement, and factors a single trial's adverse-event table cannot capture. Nothing here should be used to start, stop or switch a medication without your prescriber.

Every figure in this article was read from SURPASS-2's ClinicalTrials.gov results record and from PMID 34170647's abstract on 28 September 2026, and the twenty-four stored rows were verified against that record before this piece was written. They count toward our pooled estimates for their six effects, tagged as medium-tier (tirzepatide 5 mg, semaglutide 1 mg) and high-tier (tirzepatide 10 and 15 mg) diabetes-indication registry rows; the pooled figures at the data API move with each collection run, so verify there before citing them.