The Weight Comparison Used 257 People

A study of 30,458 patients reported that bariatric surgery produced 28.3% weight loss against 10.3% on GLP-1 medications, at lower two-year cost.^[1] The cost half rests on the whole cohort and is worth taking seriously. The weight half rests on 257 of the 16,357 people in the drug arm, and that is a different kind of number.

What the study compared

Researchers linked an insurance claims database to a health system’s medical records and identified adults with class II or III obesity who had either metabolic bariatric surgery — sleeve gastrectomy or gastric bypass — or a GLP-1 prescription. Propensity score weighting balanced the groups on prior spending, health care use, age, sex, comorbidities and smoking. The result was 14,101 surgical patients and 16,357 on medication, followed for a little under three years on average.

Who those 257 people are

Which direction that biases the result is not knowable from the outside — and that is the problem. The 9-in-10 whose weights are missing could be doing better than the 257 or worse. What can be said is that a comparison built on 1.6% of one arm and 9.2% of the other is not the comparison the propensity weighting was designed to protect, because the weighting balanced the full cohort and not these two fragments.

The careful matching was done on 30,458 people. The headline number was computed on 1,548 of them.

And the drug arm is not one drug

The GLP-1 group pooled dulaglutide, exenatide, liraglutide, lixisenatide, semaglutide and tirzepatide. Those agents differ enormously in what they do to weight — the older ones, such as exenatide, were never marketed for weight loss and produce a fraction of the effect of the newest two.

So 10.3% is an average across a class whose members are not interchangeable, in a period during which prescribing shifted heavily toward the strongest agents. It is a fair description of “people who took some GLP-1 during these years.” It is not a fair description of what someone starting tirzepatide today should expect, and the figure will be quoted as though it were.

The cost finding is the solid half

Two-year costs came to $63,483 for the medication group against $51,794 for surgery, with the difference driven by pharmacy spending that continued undiminished through the second year. That analysis used the entire cohort, the difference is large, and it makes an obvious point: an operation is paid for once, and a prescription is paid for every month for as long as it works.

The authors conclude by asking whether surgery should still be treated as a last resort. That is a reasonable question and the cost data support raising it. What the cost data cannot address is the thing that actually makes surgery a last resort for most people: it is irreversible, it carries operative risk, and it changes how someone eats permanently. Those are not costs and do not appear in a claims database.

One disclosure worth noting: among the authors’ declared interests are personal fees from Medtronic — a manufacturer of surgical devices — alongside Novo Nordisk and Eli Lilly. That is disclosed in the paper, it is unremarkable in this field, and it is context for a conclusion favoring surgery.

What to take from it

• Surgery produces more weight loss than these drugs. That is established elsewhere by better evidence — we cover a cleaner head-to-head in surgery versus GLP-1 drugs. Nothing here is surprising, and this study’s particular weight numbers are weakly supported.
• The ongoing cost difference is real and is the genuinely useful contribution.
• Do not carry the 10.3% anywhere. It averages six drugs of very different potency, measured in 1.6% of the people taking them.
• The decision is not primarily arithmetic. Irreversibility, operative risk and permanent changes to eating are the substance of it.