Digbi Health reports 18.4% weight loss at 12 months among tirzepatide users in an employer-based observational analysis, sharpening the case for a prospective test of personalized support.
Digbi Health says members using tirzepatide while enrolled in its personalized nutrition and lifestyle program lost 18.4% of their body weight over 12 months. A report on the 12-month finding describes the result as comparatively strong, but its meaning depends on how the participants and benchmarks were selected.
This was an observational analysis of routine care rather than a randomized clinical trial. Digbi’s account of the analysis says it linked pharmacy claims from 10 self-insured US employers with weights recorded by connected scales between 2022 and 2025. Participants enrolled voluntarily, while the comparison group comprised people using the same medications who did not join the program.
The finding is therefore best understood as an encouraging real-world signal—not evidence that the Digbi model caused the reported weight loss. The result appears to come from an ObesityWeek 2026 poster, and the public evidence is currently a company description rather than a peer-reviewed paper or registered prospective trial.
What sits behind the 18.4% figure
The reported percentage applies specifically to members taking tirzepatide, not to every medication commonly grouped under the GLP-1 label. It represents change in body weight after 12 months among people participating in a broader care model alongside their prescription.
That distinction matters because a medication outcome can be shaped by more than the drug itself. Follow-up, dietary changes, persistence with treatment and engagement with monitoring may all influence weight trajectories. An employer-based care program can affect several of those factors at once.
Digbi’s use of connected scales gives the analysis repeated real-world weight data rather than relying solely on recalled or self-reported measurements. Pharmacy claims also make it possible to identify prescribed medication use across participating employers. But those features do not remove selection bias: people who opt into a data-intensive program may differ from non-participants in motivation, engagement or other characteristics that also affect weight loss.
The company itself acknowledges possible residual confounding and says prospective testing is needed. Without random assignment, statistical adjustment cannot ensure that the enrolled and non-enrolled groups were equivalent at the outset or remained comparable during follow-up.
Personalization is a bundled hypothesis
Digbi describes its intervention as personalized nutrition and lifestyle planning informed by genetic, gut-microbiome and glucose-response data. Meal quality is assessed using photographs logged by participants, adding another stream of information to the connected-scale and glucose data.
This design raises two separate questions. The first is whether adding the overall program to medication improves outcomes. The second is whether its biology-based personalization performs better than a well-delivered lifestyle program that does not use genetic or microbiome information.
The current analysis cannot separate those possibilities. Participants receive a bundle of planning, measurement and feedback, so any difference cannot be assigned to one component. Frequent logging and monitoring may themselves alter behavior, irrespective of whether the underlying recommendations are based on genetics, microbiome profiles or more conventional clinical information.
That does not make the model uninformative. It does mean that the 18.4% result should be treated as evidence about a package of care in a self-selecting population, rather than validation of each personalization technology within it.
How the result compares with tirzepatide trial data
The most relevant clinical-trial benchmark is SURMOUNT-1, a randomized, double-blind study of 2,539 adults with overweight or obesity who did not have diabetes. At 72 weeks, mean weight loss was 15.0% with 5 mg of tirzepatide, 19.5% with 10 mg and 20.9% with 15 mg, compared with 3.1% for placebo.
Digbi’s 18.4% figure falls within that dose-dependent trial range, but it is not a direct replication or extension of SURMOUNT-1. The follow-up periods differ, and the employer analysis did not randomly assign treatment or support. Differences in participant characteristics, dosing, persistence and statistical methods can also make percentages from separate studies look more comparable than they are.
Nor was SURMOUNT-1 a medication-only experiment. All participants received a lifestyle intervention involving a reduced-calorie diet and increased physical activity, according to the trial’s design summary. Its weight-loss estimates therefore reflect tirzepatide delivered alongside standardized lifestyle guidance.
Digbi’s internal comparison—program enrollees versus people taking the same medications without enrolling—is more directly aligned with the question of whether additional support contributes value. Yet voluntary enrollment makes that comparison less causally secure than random allocation. Trial results provide a useful clinical range; they do not prove that the personalized program accounts for Digbi’s outcome.
Average weight loss conceals different trajectories
Even in a controlled trial, response to tirzepatide varies substantially. A secondary analysis of SURMOUNT-1 found that participants categorized as late responders had lost an average of 11.0% of body weight by week 72, compared with 22.5% among early responders.
That variation helps explain why a single average should not become an expectation for every patient. It also makes longitudinal real-world datasets potentially valuable: they can show when weight changes occur and whether patterns of engagement or treatment persistence accompany different trajectories.
However, such data need carefully specified analyses. If participation, logging or continued scale use is related to success, the people with the most complete data may not represent everyone who began treatment. Prospective studies can address this by defining inclusion, follow-up and missing-data rules before outcomes are known.
The prospective test that would clarify value
For employers and health plans, the Digbi analysis offers a testable proposition: structured, individualized support may improve outcomes among people prescribed anti-obesity medication. It does not yet establish the size of that contribution or which parts of the program matter most.
A more decisive study would prospectively assign eligible participants to medication with usual support or medication with the Digbi model. Using the same follow-up schedule, weight-measurement process and medication-exposure definitions across groups would make any incremental effect easier to interpret. Predefined analyses could also examine whether engagement changes the result without confusing engagement with assignment to the program.
Until such evidence is available, 18.4% is a notable observational result rather than a new efficacy benchmark for tirzepatide. Its practical importance lies in showing that medication outcomes can be studied within the care systems surrounding a prescription—and in setting up a clear question about whether personalized support produces additional, reproducible benefit.