Semaglutide vs Dulaglutide: One Trial Randomized 1,201 Patients and Compared Six Side Effects at Two Dose Levels

Ask which of the two older weekly GLP-1 injections, semaglutide (Ozempic) or dulaglutide (Trulicity), has the worse side effects, and most of what you will find splices together rates from different trials — one drug's nausea rate from one study, the other's from a different study, with a different population, a different duration and a different way of asking about symptoms. That is the confound our dose-gradient piece found for dose and our SURMOUNT-5 piece avoided for tirzepatide versus semaglutide at obesity doses. This article does the same for the diabetes doses of semaglutide and dulaglutide, from the one trial in our corpus that randomized patients between them.

SUSTAIN 7 (NCT02648204; Pratley et al., The Lancet Diabetes & Endocrinology, April 2018) randomized 1,201 adults with type 2 diabetes on metformin, 1:1:1:1, to once-weekly semaglutide 0.5 mg, semaglutide 1.0 mg, dulaglutide 0.75 mg or dulaglutide 1.5 mg, for 40 weeks, at 194 sites in 16 countries. It was open-label: patients and investigators knew which drug and dose was given. Its primary endpoint was the change in HbA1c; the trial was powered for HbA1c non-inferiority and body-weight superiority, and it was funded by Novo Nordisk, semaglutide's maker. Its adverse-event table entered our corpus on 28 September 2026, and every one of the 24 rows we stored was re-read against the registry's own posted counts the same morning: arm, term, numerator and denominator all match.

Read this as one trial, not as 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 twenty-four registry rows from a single study, useful for exactly one reason: randomization inside it removes the population and duration confounds that make every cross-trial comparison of these two drugs unreliable. It cannot tell you how common an effect is in general, and it is not a placebo comparison.

Two dose pairs, chosen by the trial, not by us

The protocol compared the drugs at two levels: semaglutide 0.5 mg against dulaglutide 0.75 mg (the "low" pair) and semaglutide 1.0 mg against dulaglutide 1.5 mg (the "high" pair). Those are the trial's own pairings, the licensed diabetes maintenance doses at the time, and they are not claimed to be equipotent. Neither is the weight-management dose of either drug: semaglutide 2.4 mg (Wegovy) and dulaglutide's later 3.0 and 4.5 mg doses were not in this trial. Per the registry, 301, 300, 299 and 299 patients were exposed in the four arms, and 279, 279, 287 and 284 completed; the abstract counts 72 withdrawals (6%), 22, 21, 13 and 16 by arm in that order.

Adverse events were counted from the first dose to week 40 plus a five-week follow-up (the registry's stated window), and the non-serious table lists a term only if it reached 5% in at least one arm. Six of the nine terms it lists are effects we track; the other three (lipase increased, nasopharyngitis, upper respiratory tract infection) are outside our fifteen-effect taxonomy and are not stored.

The low pair: semaglutide 0.5 mg against dulaglutide 0.75 mg

Every cell is the registry's count divided by the same arm's at-risk denominator, read on 28 September 2026. "Reduced appetite" is the registry term "Decreased appetite". Gaps are semaglutide minus dulaglutide, in percentage points, from the raw counts rather than the rounded percentages.

Semaglutide's arm is higher on all six. Two example rows, verbatim from our stored evidence: "In the 'Semaglutide 0.5 mg' arm (n=301 at risk), Nausea occurred in 67 participants (22.3%)"; "In the 'Dulaglutide 0.75 mg' arm (n=299 at risk), Nausea occurred in 39 participants (13.0%)."

The high pair: semaglutide 1.0 mg against dulaglutide 1.5 mg

At the high pair the two drugs sit within 1.4 points of each other on five of six effects, and the one larger gap, diarrhoea, runs the other way, with dulaglutide higher.

What is real and what is noise

Twelve gaps invite twelve stories. Running a two-proportion z-test on each row (the standard large-sample approximation, no continuity correction) and applying a Bonferroni correction for testing twelve comparisons at once (critical |z| ≈ 2.87 for a family-wise 5%, against the usual single-test 1.96):

Nausea at the low pair (z ≈ 2.96): the only gap that clears the corrected bar. 67 of 301 against 39 of 299 is not sampling noise.

Vomiting (z ≈ 2.86) and reduced appetite (z ≈ 2.81) at the low pair: both sit just under the corrected line, vomiting by a hair. They clear the single-test line comfortably, so read them as probable, not established.

Diarrhoea (z ≈ 2.58) and headache (z ≈ 2.19) at the low pair: clear the single-test line and not the corrected one. With twelve comparisons and no correction, the chance of at least one false "finding" is about 46% even if nothing genuinely differed.

Constipation at the low pair, and every gap at the high pair: within sampling noise at these sample sizes. Dulaglutide's 4.1-point diarrhoea excess at the high pair (z ≈ −1.37) is the largest of them and is not evidence of anything on its own.

We are publishing the z-values rather than a "significant" flag because the two tables will otherwise read as twelve findings when they support one without hedging and two more with it.

The pattern is in the dulaglutide dose step, not the semaglutide one

Reading the four arms as two dose ladders explains why the low pair separates and the high pair does not. Moving semaglutide from 0.5 to 1.0 mg changed no effect by more than 1.3 points (nausea 22.3% to 21.0%, vomiting 10.0% to 10.3%, diarrhoea 14.3% to 13.7%). Moving dulaglutide from 0.75 to 1.5 mg moved diarrhoea by +10.0 points (7.7% to 17.7%), reduced appetite by +7.4 (3.0% to 10.4%), nausea by +7.0 (13.0% to 20.1%) and vomiting by +5.7 (4.0% to 9.7%). Those within-drug differences are descriptive here — they are not part of the twelve-test family above and we have not built a second corrected family around them — but the direction is plain: dulaglutide 0.75 mg was the gentlest arm on every one of the six effects, and its 1.5 mg arm looked like the two semaglutide arms.

Beyond the six tracked effects

The abstract's own summary line is that gastrointestinal disorders as a class occurred in 129 (43%) of 301 patients on semaglutide 0.5 mg, 133 (44%) of 300 on semaglutide 1.0 mg, 100 (33%) of 299 on dulaglutide 0.75 mg and 143 (48%) of 299 on dulaglutide 1.5 mg, and that they were the most common reason for stopping either drug. The authors' interpretation calls the safety profiles "similar". The registry's serious-events table (96 terms) lists three pancreas-related terms, each affecting one participant, all in the semaglutide 1.0 mg arm: acute pancreatitis, pancreatolithiasis, and a stage IV pancreatic carcinoma — the registry's adverse-event section defines its events as treatment-emergent (onset after the first dose, through five weeks of follow-up) and gives no further detail on the case. It also lists six gallbladder term entries (cholecystitis, acute cholecystitis, cholelithiasis, gallbladder pain), one in the semaglutide 1.0 mg arm and five in the dulaglutide 1.5 mg arm. These are single events, not rates, and our collector does not pool serious rows for effects that any trial posts in a non-serious table. The abstract reports six deaths: one in each semaglutide group and two in each dulaglutide group.

On the efficacy side, which was the trial's purpose, the registry's least-squares mean changes at week 40 were HbA1c −1.51 and −1.78 percentage points on semaglutide 0.5 and 1.0 mg against −1.11 and −1.37 on dulaglutide 0.75 and 1.5 mg, and body weight −4.56 and −6.53 kg against −2.30 and −2.98 kg. The abstract's estimated treatment differences are −0.40 and −0.41 HbA1c points and −2.26 and −3.55 kg, all p<0.0001. So at the low pair, the arm with the larger side-effect gaps was also the arm with about twice the weight loss, which is the trade the trial was designed to show; at the high pair the weight gap widened while the side-effect gaps closed.

Where this sits against the head-to-heads we have already written up

Our diarrhoea article noted that in three head-to-heads the newer or higher-exposure agent reported more diarrhoea by 4 to 10 points, and that SURMOUNT-5 did not fit. SUSTAIN 7 fits at one of its two pairs and reverses at the other: +6.6 points for semaglutide at the low pair, −4.1 at the high pair. A dated note now says so on that article. Do not read this trial beside SURMOUNT-5's rates: different drugs at one end (tirzepatide), different doses (2.4 mg semaglutide there, 0.5 and 1.0 mg here), a different population (obesity without diabetes there, type 2 diabetes on metformin here) and a different window (72 weeks there, 40 here). Each trial compares its own two arms and nothing else.

What this does not tell you

No placebo arm. SUSTAIN 7 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 a mean HbA1c of 8.2% and a mean weight of 95 kg on metformin; the doses are the 2016 diabetes maintenance doses. None of these numbers describes semaglutide 2.4 mg in people without diabetes.

One trial, open-label, sponsor-run. A single randomized trial is one sample; neither patients nor investigators were blinded, which can shape symptom reporting; and the sponsor makes one of the two drugs. None of that changes a registry count, but all of it bounds what a count can mean.

A 5% listing threshold. Acid reflux, fatigue, dizziness, abdominal pain, hair loss and injection-site reactions are absent from the non-serious table, which means none reached 5% in any arm, not that none occurred.

Whole course, 40 weeks. Each arm includes its titration weeks, and the window ends at week 45, so these are at-least-once shares over that span, not rates at a maintenance dose.

The pairs are the protocol's. 0.5 against 0.75 mg and 1.0 against 1.5 mg are how the trial was built; they are not matched for potency, and the tables should not be read as "equal dose" comparisons.

Multiple comparisons, restated. Read nausea at the low pair as established, vomiting and reduced appetite as probable, diarrhoea and headache as suggestive, and every high-pair gap as noise.

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 SUSTAIN 7's ClinicalTrials.gov results record and from the PubMed abstract of its primary publication (PMID 29397376) on 28 September 2026, the day its rows entered our corpus, and the twenty-four stored rows were verified against that record before this piece was written. They now count toward our pooled estimates for their six effects, tagged as low-tier (0.5 mg semaglutide, 0.75 mg dulaglutide) and medium-tier (1.0 and 1.5 mg) diabetes-indication registry rows; the pooled figures at the data API move with each collection run, so verify there before citing them.