When the Weight Comes Back: A Second Operation or a Weekly Injection?

Weight regain is the quiet majority experience of bariatric surgery, and until recently it had roughly one answer. If a sleeve or a bypass stopped working, the patient was offered another operation — a conversion, a revision, a re-do. Now there is a second door, and it opens from a pharmacy.

Which door leads somewhere better is the question the specialty is living through in 2026. The evidence arriving on it is new enough that three of the papers below were published within the past five months, and unsettled enough that they disagree with one another in ways worth naming.

The second operation is no longer rare

A national analysis published in Surgery for Obesity and Related Diseases puts hard numbers on how far the revisional trend has gone. Researchers at Jefferson Health examined the MBSAQIP participant use file for 2020 through 2024 — 949,507 bariatric operations in total, of which 94,828 were conversions. Conversions, the authors note, now make up nearly one in ten procedures nationally.

They then asked a narrower question than most: not whether conversions work, but what they cost in the first thirty days. Stratifying by the anatomy the patient ends up with, they found conversion consistently carried more risk than a primary operation of the same final shape. The adjusted excess of a seventeen-component composite of serious adverse events was 18.2 per 1,000 cases in the sleeve stratum, 22.1 per 1,000 after gastric bypass, and 17.5 per 1,000 for duodenal switch or single-anastomosis duodeno-ileal bypass. Translated into the language clinicians use at the bedside, that is a number needed to harm of 55, 45, and 57 respectively.

The excess clustered where surgeons would expect it — reoperation, reintervention, readmission — and in specific technical failures: leak, organ-space infection, bowel obstruction. Patients whose first operation was a gastric bypass fared worst, with an adjusted odds ratio of 2.54 compared with a prior sleeve. And the penalty did not shrink over the study years, which argues against the comfortable assumption that experience is quietly erasing it.

This is a registry analysis with statistical adjustment, not a randomised trial, so it measures association rather than cause. Its conclusion is nonetheless blunt: conversion should be treated as its own risk category, not benchmarked against primary surgery.

What the pharmacological door actually delivers

The case for medication rests on two kinds of evidence, and they point in the same direction more than the surgical numbers do.

A systematic review and meta-analysis in Frontiers in Endocrinology pooled 27 studies of incretin receptor agonists given to people who had already had metabolic-bariatric surgery, 19 of them contributing to the quantitative synthesis. At twelve months, total weight loss averaged 9.22% on liraglutide and 9.02% on semaglutide. At six months, tirzepatide outperformed semaglutide by a mean difference of 4.23 percentage points. Serious adverse events were absent; most reported effects were mild gastrointestinal symptoms. The authors frame the drugs as a possible alternative rescue therapy to revisional surgery — then add the caveat that matters, noting substantial heterogeneity across studies.

The largest real-world cohort to date agrees, with detail. A study in EClinicalMedicine drew on electronic records from 208,155 people who had undergone metabolic-bariatric surgery, 39,750 of whom later received semaglutide or tirzepatide. Among those who stayed on therapy for a year, tirzepatide was associated with 17.2% total weight loss against 12.0% for semaglutide — an adjusted difference of 5.16 percentage points — and the effect scaled with dose. Timing mattered too: starting later in the postoperative course was associated with greater loss than starting within the first two years. On the cardiovascular side, pooled incretin use showed no association with major adverse cardiovascular events, with a hazard ratio of 0.91 and a confidence interval that crossed one.

That last result is a genuine null, and it deserves to be read as one. In a retrospective cohort it means no signal was detected, not that the drugs protect the heart after surgery — and not that they fail to.

The one head-to-head comparison, and its warning label

Only one study so far has put the two options side by side in the same cohort. Montefiore researchers followed 250 patients who had regained weight after a primary operation: 120 treated with semaglutide, 130 with revisional surgery — conversions of a sleeve to gastric bypass, or revisions of an existing bypass.

At two years the gap was wide. Total weight loss was 17.47% after revisional surgery and 2.8% on semaglutide; excess weight loss was 45.76% against 6.1%. The revisional group’s average BMI fell from 41.14 to 34.93, while the medication group’s drifted back up from 39.13 to 40.36.

The authors are unusually candid about what this does not establish. It is a single-centre retrospective cohort from patients operated between 2015 and 2020, an era before tirzepatide and before contemporary semaglutide dosing. They note that they could not report safety and complication data for the revisional arm, which limits any definitive comparison. Two corrections to the paper were subsequently issued.

There is a further problem worth flagging to anyone reading the figure at face value: the semaglutide arm’s result sits far below the roughly 9% one-year loss reported in the pooled literature. A single-centre series and a meta-analysis can describe different worlds — different adherence, different dosing, different definitions of the starting point. The discordance is itself the story. Nobody has run the randomised trial that would settle which door is better for whom, and the specialty is working without one.

Starting the drug years later

A smaller study in the International Journal of Obesity complicates the picture from a third angle. Yale surgeons reviewed 111 patients who had a sleeve gastrectomy at their institution more than a decade ago. Sixty-seven were not taking anti-obesity medication at their most recent visit; forty-four were taking semaglutide or tirzepatide. Ten-year total weight loss was 14.1% in the first group and 25.7% in the second.

The detail that deserves attention is the timing: patients had started medication an average of 7.6 years after their operation, and their weight trajectories before that point were similar. With 111 patients at one institution and no randomisation, this is a signal, not proof. It does suggest, though, that the moment when medication is added may matter less than a framing of surgery as either successful or failed would imply — and that long-horizon follow-up, the kind a prospective cohort such as the LABS study, funded by NIDDK, is designed to deliver, is what would convert that signal into an answer.

What the disagreement adds up to

Count the evidence levels honestly. There is no randomised trial comparing revisional surgery with incretin therapy for weight regain. There is one national registry analysis of conversion risk, one meta-analysis of drugs after surgery, one large real-world cohort, one single-centre head-to-head series, and one single-centre ten-year review. Every one of them is observational.

Read together, they sketch a rough trade. The second operation produces larger average weight loss in the only direct comparison available, and charges a 30-day risk premium that is real but modest and concentrated in a minority of patients. The medication route produces smaller average loss with a mild side-effect profile and, so far, unknown durability and no demonstrated cardiovascular benefit in this population — alongside the practical questions of cost, supply and adherence that shape whether a prescription actually becomes a year of treatment.

Which makes the more useful question a familiar one. This archive has written before about how the choice of ruler — percent of excess weight versus percent of total, baseline from the first preoperative weigh-in or from the day of surgery — can flip which patients count as responders. The same caution applies here: a comparison that measures only kilograms will favour the operation, because that is what operations are good at. Glycaemic control, sleep apnoea, blood pressure, nutrition, bone health and quality of life move on their own schedules, and the two doors do not move them identically.

For anyone outside the operating theatre, including people weighing their own BMI against eligibility criteria, the practical takeaway is a reading habit rather than a recommendation. When a claim about weight regain arrives, ask what kind of study it came from and how long it ran: the field’s most confident-sounding statements currently rest on the weakest designs, and its most careful ones are the ones flagging what they could not measure.

This archive is independent of NIH and NIDDK, whose LABS consortium funded much of the long-term literature referenced here; it reports what studies and agencies have published and does not recommend surgery or medication for or against anyone.