Two treatments for the same condition, and no trial has ever assigned a person to one or the other. That is the first thing to know about this comparison, and it governs everything below it. Every figure putting an operation next to a weekly injection comes either from separate trials with separate populations, or from database studies that watched people who chose. The size of the numbers in the tirzepatide trials is not in dispute. What each one is comparable to is.
The closest thing to a direct comparison, and what it cost in rigor
A single center in Chengdu ran the comparison prospectively, allocating adults with obesity to sleeve gastrectomy (n = 36) or once-weekly semaglutide (n = 35) by shared decision rather than by randomization. At twelve months, mean total weight loss was 28.6% after surgery against 11.3% on the drug (p < 0.001). Mean body-mass index fell from 32.7 to 23.8 in the surgical group and from 32.0 to 28.4 in the drug group.[1]
Read the design before the gap. Semaglutide was planned for 24 weeks, after which continuation was left to the patient, so the twelve-month figure on that side is a mixture of people still treated and people who had stopped six months earlier. Among the 25 who did stop, mean weight rose from 74.6 kg to 83.2 kg by month twelve, and 32% were back at or above their starting weight.[1] A surgical comparison in which one arm was withdrawn halfway through is not a fair contest, and it is also an accurate picture of what happens on a cash plan that lapses.
The drug figures under their own protocols
Held at full dose for the full duration, the numbers are larger. STEP 1 randomized 1,961 adults without diabetes to semaglutide 2.4 mg or placebo for 68 weeks and reported a mean change in body weight of −14.9% against −2.4% on placebo.[2] SURMOUNT-1 randomized 2,539 adults to tirzepatide or placebo for 72 weeks and reported −20.9% at the 15 mg dose against −3.1%.[3] Those are the ceilings the class has published, and the expected-loss tool works from the same distributions.
Surgery’s long-horizon figures are reported in a different unit, which is part of why the comparison is hard to hold in one head. A meta-analysis of every dataset with ten or more years of follow-up found weighted means of 56.7% excess weight loss across 18 reports of gastric bypass, 58.3% across two reports of sleeve gastrectomy and 45.9% across 17 reports of adjustable gastric banding. A single center followed banding patients for two decades: at twenty years, 30.1 kg lost, 48.9% of excess weight, 22.2% of total body weight.[4]
What has actually been randomized
The strongest randomized evidence in this area does not involve a GLP-1 at all. ARMMS-T2D pooled four United States single-center trials that randomized adults with type 2 diabetes to bariatric surgery or to medical and lifestyle management between 2007 and 2013, then followed them observationally to 2022 — a median of eleven years. At seven years, glycated hemoglobin had fallen 1.6 percentage points in the surgical group against 0.2 in the medical group, a between-group difference of −1.4 points (95% CI, −1.8 to −1.0; P < .001), still −1.1 points at twelve years.[5]
Two details limit how far that travels. The comparator was the medical management available when the trials enrolled, which predates both approvals discussed in the two-approvals article. And a quarter of the participants assigned to medical management went on to have surgery anyway.[5] No trial has randomized anyone to an operation or to semaglutide or tirzepatide.
Diabetes remission is where the gap is widest
A 2026 systematic review compared the two paradigms outcome by outcome across 14 studies and 25,566 participants. On remission of type 2 diabetes it recorded 23% to 70% after surgery against 0% on incretin therapy, with sustained five-year weight loss of 25% to 32% against 14.9% to 20.9%.[6] Remission and control are different endpoints, and no dose of either drug has produced the first.
Remission also decays. In ARMMS-T2D it was 18.2% in the surgical group at seven years and 12.7% at twelve, against 6.2% and 0.0% in the medical group.[5] The surgical advantage is real and it is not permanent, which is the same sentence the drug side earns for a different reason.
The observational record contradicts itself on mortality
Here the obvious reading falls apart. One 2026 meta-analysis of five cohort studies (N = 39,569) found surgery associated with a 43% lower risk of death (HR 0.57; 95% CI, 0.34 to 0.95), 35% lower risk of major adverse cardiovascular events (HR 0.65; 95% CI, 0.51 to 0.83) and 55% lower risk of heart failure (HR 0.45; 95% CI, 0.39 to 0.51) than GLP-1 therapy.[7] A second, covering six cohorts and 208,751 patients, reported a pooled relative risk of 0.59 for cardiovascular events (95% CI, 0.46 to 0.77) and a non-significant trend on mortality (RR 0.66; 95% CI, 0.41 to 1.07).[8]
Then the largest single study in the field reverses it. A multi- institutional analysis matched 84,884 pairs of adults with obesity starting tirzepatide against adults undergoing metabolic surgery. All-cause mortality ran 0.19 per 100 person-years on tirzepatide against 0.57 after surgery, a hazard ratio of 0.311 (95% CI, 0.257 to 0.375; p < 0.0001), with cardiovascular events also favoring the drug (HR 0.743).[9]
The glycemic side of that second meta-analysis is less contested. Pooled across its six cohorts, final glycated hemoglobin favored surgery by a weighted mean difference of −1.31 percentage points (95% CI, −1.66 to −0.96).[8] That is roughly the size of the between-group difference ARMMS-T2D measured under randomization, which is the one place the observational and randomized records agree.
Both readings of the mortality question cannot be right about the same world. All of them are retrospective, all of them compare people who chose rather than people who were assigned, and the direction of the answer moves with the database and the matching. The honest summary is that the mortality question is open, and any page that resolves it in either direction is quoting one study and hiding the other.
The harms are different in kind, not only in size
ARMMS-T2D recorded four deaths across both arms, two in each, and no difference in major cardiovascular adverse events. What it did find more often after surgery was anemia, fractures and gastrointestinal adverse events.[5] The ten-year meta-analysis adds the quieter one: reoperation was common across every procedure type, and remains so.[4]
On the drug side the adverse events are predominantly gastrointestinal and predominantly during escalation, catalogued in the side-effect article, and the failure mode is discontinuation rather than complication. Lean tissue leaves on both routes, and the pooled body-composition figures are in the muscle article. An operation front-loads its risk into one admission; a prescription spreads a smaller risk across every month it is refilled.
There is a third column nobody advertises, which is what each route asks of a person afterward. Surgery imposes permanent changes to how food is eaten and how micronutrients are absorbed, and the ten-year review found reoperations common in every procedure group and likely to remain so.[4] A drug imposes a weekly injection and a bill, and stops working the month the bill stops. Those are not the same kind of commitment even when the weight curves look comparable.
What the comparison does not settle
Nothing above establishes which route a given person should take. It establishes that surgery’s weight and remission advantages are large, measured over decades, and undisputed; that the drug figures come from placebo-controlled trials of 68 to 72 weeks with no comparable long-horizon record; and that the mortality comparison between them is unresolved.
One more limit applies to every number here. Each trial and cohort above used an approved product. Compounded semaglutide and tirzepatide are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before dispensing, and whether a compounded preparation performs like the studied one has not been established — the distinction is set out in the compounding article, and the prices for that route are on the semaglutide board. A surgical candidacy assessment belongs to a surgical team, and how figures on this site are established before publication is described in the methodology.