Zepbound claims study found slower healthcare-cost growth but excluded the drug price

A large US claims study found slower growth in healthcare costs, excluding Zepbound's price, among adults who stayed on treatment than among matched controls.

The study excluded Zepbound's price from its cost totals. Researchers stopped follow-up after treated participants discontinued the medicine. The analysis therefore did not measure total spending for everyone who started Zepbound. It did not show that the drug paid for itself.

The study matched 15,843 adults with Zepbound claims to 15,843 controls

Researchers used the privately held Komodo Research Dataset from November 2022 through September 2025. They identified adults older than 55 who had obesity, or overweight plus a related health condition, and did not have type 2 diabetes.

The treated group had paid claims for the Zepbound autoinjector. Researchers matched 15,843 of those adults with 15,843 adults who had no claims for an incretin-based weight-management medicine. Matching was intended to make the groups more comparable, but it cannot remove every difference that can affect healthcare use.

The study compared changes from a 12-month baseline through three follow-up windows: 3 to 6, 6 to 12, and 12 to 18 months. Its main presentation used statistical weights to account for the different rates at which people left the analysis. Researchers first specified pairwise censoring as the primary method. After a post hoc review of attrition, they presented the weighted analysis as primary.

Healthcare-cost trends differed after six months

From 6 to 12 months, the estimated difference in monthly cost trends was $145 per person in favor of the Zepbound group. The 95% confidence interval ranged from $25 to $266, and the p value was 0.022.

From 12 to 18 months, the estimated difference was $319 per person per month. The 95% confidence interval ranged from $94 to $544, and the p value was 0.015.

The gap mainly developed because monthly healthcare costs rose in the matched control group while costs in the treated group stayed broadly stable. Control-group costs increased from $1,031 per person per month at baseline to $1,244 during months 12 to 18. That is different from showing that Zepbound caused a decline in healthcare spending.

At 12 to 18 months, the treated group had a lower change in combined hospital-admission and emergency-department rates than the control group. The incidence rate ratio compares changes from baseline between groups. It was 0.69, with a 95% confidence interval from 0.50 to 0.94. Changes in outpatient and office visits were not statistically significant.

The later estimate comes from a small remaining group

The primary 12-to-18-month analysis included 1,181 of 15,843 treated participants and 3,004 of 15,843 controls. The supporting pairwise analysis included 793 matched pairs and produced larger cost estimates. This attrition limits the precision and generalizability of the later estimate.

The analysis answers a narrow question about persistent use. It does not show outcomes after treatment stopped, and it did not report why people stopped.

The analysis cannot prove that Zepbound caused the difference

This was an observational claims study, not a randomized trial. Differences remained after matching in payer type, sleep apnea, BMI category, number of related health conditions, race, and region. Unmeasured differences in socioeconomic status, health literacy, and lifestyle can also affect costs and healthcare use.

A paid pharmacy claim does not prove that a person took every dose. Cash purchases, compounded products, and manufacturer-direct vials were not captured. The privately held data and imputed cost amounts are not publicly available for independent reproduction.

The study did not evaluate the Medicare GLP-1 Bridge

Only 38.5% of the treated group had Medicare. Most had commercial insurance. The claims ended in September 2025, while the Medicare GLP-1 Bridge began in July 2026.

The study therefore did not evaluate Bridge participants, the Bridge price, or Medicare's net costs. It cannot establish that the program saves money. It is evidence about healthcare-cost trends among selected older adults who remained persistent on treatment before the Bridge existed.