Episode 32, Menstrual Cycles and Type 1 Diabetes

Why this episode exists

Around half of people living with type 1 diabetes navigate the cycle every month, often for three decades or more. The pattern that comes up in DAFNE and BERTIE clinics, that comes up on patient forums, that comes up in the conversations after a podcast goes out, is the same arc. Three weeks are workable. The seven days before bleeding starts are not. The literature for decades sat in studies of six, twelve, fifteen women, and the lived experience sat in the gap between the literature and the clinic appointment.

Tatulashvili and colleagues (2022, Journal of Clinical Endocrinology and Metabolism) measured ambulatory glucose by cycle phase across hundreds of person-cycles on flash CGM. Time in range was lowest in the follicular phase (63.3%) and highest in the menstrual phase (74.8%), with the luteal phase intermediate (66.8%). Brown 2015 (a fifteen-woman pump cohort) found the strongest signal in the luteal phase. The two studies hold the same direction at the individual level, the cycle drives glucose, with different population shapes, which phase bites hardest varies. Dr Nobili’s work, drawing on observational data across multiple insulin delivery systems, is the practical bridge from the evidence to the clinic conversation.

In this episode

Dr Cecilia Nobili bridges lived experience and clinical evidence. Diagnosed at 25, she carried the cycle question through her own data into a clinical research career, and her observational work has tracked how different insulin delivery systems handle the monthly hormonal shifts.

The conversation covers what the cycle does to glucose physiology, what the evidence shows for each named UK AID system, where the hidden burden of early-follicular hypoglycaemia sits for AID users, and what the practical conversation with a diabetes team can look like when the data and the named question are both in the room.

Watch or listen

Key themes

The cycle is signal, not noise

For the forty to sixty per cent of women with T1D whose cycle shows up in CGM data, the change is real and measurable. The luteal-week glucose drift is not a discipline gap or a failure to count carbs; the insulin that worked the week before is no longer enough for the same food. That is physiology. The cohorts have, in the last few years, grown large enough for the pattern to be undeniable, and the larger systematic reviews now treat heterogeneity, that some women see it strongly and others not at all, as one of the central facts of the topic.

The early follicular phase is the hidden burden on AID

Many women rank the days when bleeding starts as equally or more burdensome than the luteal phase. As progesterone falls and insulin sensitivity returns, AID algorithms that lean heavily on recent total daily dose can carry the luteal-week insulin learning forward into the follicular phase, and the result is hypoglycaemia just as the period starts. The story has two halves; the second half is the one that does not always show up in clinic conversations.

The AID system you are on shapes how the cycle lands

The five hybrid closed-loop systems available in the UK each respond differently to the glucose drift the cycle produces. None takes cycle phase as a direct input. The differences are in how strongly each algorithm modulates basal, how it weights recent total daily dose, and what user-facing levers are available for temporary insulin resistance.

For MDI, anticipation beats reaction

For people on injections, the luteal phase typically asks for more insulin, and how that more is delivered matters. Options include strengthening carb ratios, making correction factors more aggressive, or increasing long-acting insulin. The principle is to anticipate the change in the mid luteal phase rather than wait until glucose has already drifted, and to reverse the change when bleeding starts to avoid lows.

Track two or three cycles and you have your answer

The fastest way to see whether the cycle is signal in your own data is the overlay. Your CGM already keeps the glucose log; a cycle-tracking app or paper diary records the cycle. Two or three cycles is usually enough to see whether a luteal drift, an ovulatory dip, or a menstrual-phase loosening is showing up for you. If it is, the data is what makes the team conversation specific.

One difficult day is not catastrophic

What matters is time in range averaged over weeks and months, not perfection every day. Reframing CGM data as information rather than judgement, what happened, what is the likely driver, what is one tweak for next time, reduces the paralysis that comes from fear of imperfection. Three to five difficult days a month, anticipated, is normal physiology, not failure.

Menstrual cycle tracking should be in the algorithm

Dr Nobili’s central point: cycle-tracking apps already exist on every smartphone. Linking them into AID algorithms is technically straightforward compared with the machine-learning and fully closed-loop systems presented at every diabetes conference. Until that link exists, the burden of compensating for what the algorithm could be doing automatically falls entirely on women.

The four phases mapped against insulin-need direction

The cycle is conventionally split into four phases, anchored to ovulation. Day numbers are approximate; every body runs its own clock. The hormonal arc is the same in everyone with regular cycles; the glucose response to it is heterogeneous, and the wheel below is direction, not destiny. The lift is from the GNL Menstrual Cycle guide (Part 1, What the Body Is Doing), where the full physiology walkthrough sits.

Evidence backbone

Site delivery is voice-first; the underlying source layer is research-grade and lives in the GNL Grace wiki. Tap any source for the citation detail.

Practical exploration

For people living with type 1 diabetes

The fastest path to seeing whether the cycle is signal in your own data is the overlay between CGM and cycle. Two or three cycles is usually enough. From there, the conversation with your team becomes a specific one.

• Track cycle length (most cycles run 28 to 35 days) and mark when bleeding starts in the same place you keep your CGM data.
• Review two to three cycles of CGM with the phase dates marked. If a luteal drift, an ovulatory dip, or a menstrual-week loosening shows up for you, that is your pattern.
• If a luteal drift is the pattern, anticipate it in the mid luteal phase rather than waiting for highs to appear; reverse the adjustment when bleeding starts to reduce hypoglycaemia risk.
• Pre-bolusing tends to matter more in the luteal week, because insulin acts more slowly relative to carbohydrate absorption; mixed meals (protein, fat, vegetables with carbohydrates) slow the curve and better match insulin action.
• Three to five harder days a month is not failure. Time in range averaged over weeks and months is what describes the picture.

For clinicians and educators

The cycle conversation lands more cleanly when the team is ready for it and the data is in the room. The shape of the conversation that works is consistent across CamAPS, MiniMed, Tandem, Mobi, Omnipod, and MDI.

• Ask explicitly about cycle phase in annual reviews and CGM downloads; the data is in front of you, the cycle is not, unless you ask.
• For AID users, the named system shapes the lever: CamAPS FX has the boost function; MiniMed 780G has the glucose target and active insulin time; Control-IQ and Mobi share the personal profile and exercise-mode use; Omnipod 5 has the basal programme and adaptive basal. Settings changes are clinical decisions, made on the team’s call.
• For MDI users, the lever set is carb ratios, correction factors, and long-acting basal. Anticipation in the mid luteal phase is the principle; the magnitude varies cycle to cycle and woman to woman.
• Two or three cycles of overlaid CGM and cycle data, brought to clinic with the named question, is what shifts the conversation from “we do not really adjust for that” to a settings discussion.
• The AID Algorithm Optimiser (reviewed by CamAPS, MiniMed, Tandem and Insulet global medical leads; not validated against manufacturer simulators; Grade D educational synthesis on a Grade A and B evidence base) is the shared vocabulary for talking about algorithm strength across systems.

About the guest

Dr Cecilia Nobili is a paediatric diabetology resident at Regina Margherita Children’s Hospital in Turin, Italy, and a physician-researcher living with type 1 diabetes. Diagnosed at age 25 during the COVID-19 lockdown, she transformed her own trial-and-error diabetes management into a clinical and research career. She leads a multi-centre observational study examining how menstrual cycles affect glucose control across different insulin delivery systems, funded by a Breakthrough T1D research grant. She is a graduate of the Spare Science School and an advocate for integrating menstrual-cycle awareness into diabetes technology algorithms, clinical guidelines, and structured education.

Dr Nobili on LinkedIn

Related reading on GNL