The number is consistent with what tirzepatide does in trials, and with women tending to lose more than men on these drugs. Mean dose at six months was 9.5 mg, below the 15 mg ceiling, because clinicians held escalation for anyone already losing about 2 kg a month. That is a sensible prescribing habit and probably a cost saving. The trajectory figures are clean, and inverse probability weighting for missing data barely moved the estimates (16.06 kg versus 16.03 kg).

This is a completer number. Women who stopped before six months averaged 5.4% loss at their last weigh in, versus 15.6% for those who stayed. A crude blend of the two groups lands near 14.7%, so the dropouts do not sink the result. The 4,025 who never finished onboarding are a different matter. They are outside every denominator. Weights came from participants' own scales, proctored over video, and nobody was randomized to anything. In SURMOUNT‑1 the placebo arm lost 3.1% over 72 weeks on diet and counseling alone, and paying customers who sign up for weekly supervised weigh ins are not a placebo arm.

Deprivation quintile made no difference to weight loss (P = 0.364), which is the paper's best equity finding. Response was broadly similar across starting weight bands in percentage terms. The 99.3% of women who lost some weight looks almost too good, but it is a completer figure and a small loss counts.

Ethnicity. Nonwhite women lost 13.5% against 15.8% for white women, and were more likely to stop treatment (hazard ratio 1.27). The abstract lists the strata where results were consistent and drops the one where they were not. Age differences were also statistically significant (16.1% in the youngest group, 14.8% in the oldest). And deprivation here is measured by postcode. A woman paying out of pocket for a GLP-1 program while living in the most deprived fifth of postcodes is not a typical resident of that postcode.

Mean HbA1c fell from 36.3 to 33.0 mmol/mol (about 5.5% to 5.2%) in a population that was mostly normoglycemic to begin with. Total cholesterol fell 0.77 mmol/L (about 30 mg/dL), LDL 0.61 (about 24 mg/dL), triglycerides 0.47 (about 42 mg/dL). Very few women started new statins or diabetes drugs, so the changes are not a medication artifact. The direction matches everything known about tirzepatide.

The 84.5% figure covers 155 women classified as diabetic on one capillary HbA1c. Selecting people on a single high value guarantees some fall back on retest, and a drop from 48 to 47 mmol/mol counts as normalization. The lipid changes are also bigger than the randomized evidence. In SURMOUNT‑3 (an imperfect comparator, since everyone first completed 12 weeks of intensive lifestyle therapy), at 72 weeks tirzepatide lowered LDL 6.1% and total cholesterol 3.0%, and HDL rose 15.4%. Here LDL fell about 16%, total cholesterol about 13%, and HDL fell about 14%. The authors attribute the HDL drop to the active weight loss phase, which is plausible. Six months versus 72 weeks explains part of the gap. Home finger prick sampling, uncertain fasting status and the absence of a placebo arm (where LDL rose 6.1% in SURMOUNT‑3) may explain more.

Retention is high by any real world standard, where GLP-1 discontinuation within a year often runs past a third. Side effects were common (62.7% reported at least one) and mostly gastrointestinal and mild or moderate. Women who used the company's symptom package of over the counter medicines and diet advice stopped treatment less often (5.0% versus 7.4%).

Discontinuation was defined as no dispensing for more than eight weeks after the last one. Unless follow up ran past six months, which the paper does not state, a woman whose last pen shipped after about week 18 could not yet count as a dropout by week 26. Safety was self reported through the app with no linkage to NHS hospital records, so a gallbladder admission at a district general hospital would surface only if the patient mentioned it. Package users chose to use the package, which makes the 5.0% versus 7.4% comparison a statement about who uses packages. Retention also means a pen was shipped, not that it was injected. Eleven women became pregnant during follow up and stopped treatment. Tirzepatide's label warns that it can reduce oral contraceptive absorption, and the main paper does not describe contraception counseling or screening for eating disorders, two standing concerns with online weight loss prescribing to women.

A remote program with identity checks, proctored weigh ins, home labs and monthly clinician review can initiate and titrate tirzepatide in thousands of women with low rates of serious trouble, at least as far as self report can see. That is real implementation evidence, and the at home lab and proctoring infrastructure is more than most direct to consumer platforms bother with.

The evidence gap the authors cite is in cardiovascular outcomes trials such as SELECT, where most participants were men. This study has no outcomes, only six months of surrogates. The weight loss trials were already mostly women (67.5% in SURMOUNT‑1). Replicating specialist care cannot be shown without a comparison to specialist care. Scalability is untested on price. The program was partly subsidized by the provider during the study, the authors call that pricing possibly unsustainable, and every participant started before Lilly raised the UK list price of Mounjaro by up to 170% on 1 September 2025.