What is the 46 per cent a percentage of?
The ARCHES analysis assembled 2,939,364 people assigned female at birth from the Utah Population Database between 1996 and 2021, and followed them for a mean of 10.8 years. Over that period 99,978 received an endometriosis diagnosis, and their adjusted hazard of type 2 diabetes came out 46 per cent higher than in the rest of the cohort: an adjusted hazard ratio of 1.46, with a confidence interval running from 1.43 to 1.50. A denominator of nearly three million and an interval that narrow put the existence of the association beyond much argument.[1]
It is still a ratio between two groups, and it answers a question about groups. What it does not hand a woman with endometriosis is her own number. A hazard ratio can sit at 1.46 whether the underlying yearly risk moves from very small to slightly less small, or from moderate to noticeably higher; the ratio looks identical in both cases. Turning it into a personal probability needs the baseline rate for a woman of that age, weight and family history, and the ratio does not carry that baseline with it.[1]
Where the association is strongest
Broken out by subgroup, the association is strongest in the women a clinic would place at lower risk. Among premenopausal women the adjusted hazard ratio is 1.55. Among women with a body mass index under 30 it is 2.39. Endometriosis outside the pelvis gives the highest figures of all, 2.67 for a specified site and 2.47 for an unspecified one. The authors put this in the conclusion themselves: the associations are strongest among individuals traditionally considered at lower baseline risk.[1]
Two readings fit that pattern and this design cannot separate them. The first is biological: sustained inflammation from ectopic lesions acting on metabolism, which would suit the extra-pelvic subtype carrying the largest figure. The second is that a woman diagnosed with endometriosis is already inside the health system, being examined and having blood drawn, so an undiagnosed diabetes is more likely to be found in her than in a woman who never came in. That effect would also be largest where routine glucose testing is least expected, which is precisely the low-weight, premenopausal group. The authors name it, alongside residual confounding, as a reason to read the results cautiously. Unequal testing and a real metabolic mechanism produce the same table.[1]
What would have to be true before this changes practice?
More of the same data will not separate the two readings. The test is a cohort in which glucose is measured on a fixed schedule for everyone, so that contact with a doctor stops predicting who gets diagnosed. If such a cohort reports the subgroup with a body mass index under 30, I expect its adjusted hazard ratio to come in below 2.39, because part of the present figure is doing the work of unequal testing. If it lands near 2.39 anyway, the metabolic reading becomes much harder to argue against, and endometriosis turns into a reason to check glycaemia in a woman carrying no other flag.[1]
Until then, what the study supports is narrower than the headline and still worth having. About 11 per cent of reproductive-aged women in the US live with endometriosis, so this is a large group learning that a condition they already manage may sit alongside a metabolic risk nobody mentioned to them. The honest statement to that group is that the association holds in the data, that its size for any one woman is unknown, and that the cheapest response, a glucose test she might not otherwise be offered, carries almost no cost either way. Evidence this thin does not justify alarm. It does justify asking.[1]