Menstrual Cycle Tied to Glucose Swings in Type 1 Diabetes

Published September 3, 2026, after new reporting on a Diabetes Care observational study.

A new observational study found that glucose metrics and model-derived insulin sensitivity varied across menstrual-cycle phases among 77 menstruating adults with type 1 diabetes who used automated insulin delivery systems. The result points to a measurable group-level pattern, but it does not establish a cycle-based insulin dosing rule.

What the researchers analyzed

Participants contributed continuous glucose monitor data from 380 menstrual cycles covering 10,107 days. The study population had a mean age of 35.1 years. Insulin-dose data were available for 64 participants, and carbohydrate-entry data were available for 61.

The researchers compared glucose measures, insulin delivery, and carbohydrate entries across six estimated cycle phases. They then used a hierarchical Bayesian state-space model to estimate insulin sensitivity from those data.

That last distinction matters. Insulin sensitivity was modeled, not measured directly with a blood test. A fasting insulin test is a separate, one-time measurement of circulating insulin and was not the method used to produce the study's phase-specific estimates.

What changed across cycle phases

At the population level, modeled insulin sensitivity was 2.6% above the population average in the early follicular phase and 2.6% below average in the midluteal phase. The study reported that 84.7% of participants had an individual pattern consistent with the broad population trend: above-average modeled sensitivity in the early follicular and periovulatory phases, followed by below-average sensitivity in the midluteal phase.

Glucose metrics shifted as well. Mean time in the 70 to 180 mg/dL range fell 2.8 percentage points from the early follicular phase to the early luteal phase. Time in the tighter 70 to 140 mg/dL range fell 7.1 percentage points over the same interval.

Those averages did not describe everyone. Sixty percent of participants had some decline in time in range between those phases, while 15.7% had a decline of more than 10 percentage points. The authors emphasized substantial variation from one person to another.

Where A1C and laboratory testing fit

An A1C test reflects average blood glucose over about three months. It does not show short-term changes, so it cannot display the cycle-phase pattern examined in this study. The researchers used continuous glucose monitoring to study that shorter-timescale variation.

The findings do not change how A1C or fasting insulin results should be interpreted. They also do not show that ordering either test at a particular cycle phase improves care. Continuous glucose data, A1C, and a fasting insulin measurement answer different questions and should not be treated as interchangeable.

What the study cannot establish

This was a voluntary, decentralized observational study. Cycle dates were self-reported, ovulation was imputed when not confirmed, and insulin sensitivity was a model-based inference rather than a direct physiological measurement. The authors also noted possible selection bias, unmeasured factors such as illness, sleep, stress, medication, and physical activity, and limited generalizability.

The results apply most directly to menstruating adults with regular cycles who have type 1 diabetes and use automated insulin delivery. They should not be generalized to people with type 2 diabetes, people with irregular cycles, or people who do not use automated insulin delivery.

Most importantly, the study does not supply a universal formula for changing insulin by cycle phase. Individual patterns varied too much for that conclusion. Nothing here should be read as guidance to change insulin, medication, monitoring, or testing without an individual's diabetes care team.

This article is editorial commentary and is not medical advice. It has not been reviewed by a physician and should not be used to make decisions about diabetes diagnosis, insulin dosing, medication, treatment, monitoring, or testing frequency.