Listen in the Podli app 🎧
Follow your favourite podcasts, listen offline and in the car with CarPlay and Android Auto, and always pick up where you left off. Free to try.
Episode Summary
Mikki has said many times — on this podcast, on other podcasts, and repeatedly on Instagram — that eating carbohydrate at dinner helps calm the nervous system, and that this is why people who feel wired at night sleep better when they add some back in. In this episode she corrects herself. Not on the advice, which she still gives and still sees work, but on the mechanism she has been attaching to it: insulin clearing branched-chain amino acids out of the bloodstream, tryptophan winning the competition at the blood-brain barrier, and the resulting serotonin and melatonin sedating you. Every step in that chain is real biochemistry, which is exactly why it sounds right. It just does not happen at the doses found in any meal she would actually recommend.
From there she works through what is more likely going on. She covers the head-to-head trial that put carbohydrate, two proteins and a placebo up against each other before bed and found nothing; why the pre-sleep protein literature is muscle research rather than sleep research; the difference between an energy deficit and low energy availability; and the body of work on cognitive dietary restraint — the mental effort of not eating, which carries measurable physiological consequences independent of how much is actually eaten. She is careful throughout about how messy cortisol data is, and about which parts of it she thinks still stand. The practical conclusion is unchanged but better founded: if you are wired and under-eating relative to your training, it is probably the calories, and carbohydrate is usually the easiest thing to add back.
Key Topics
- Why the mechanism matters even when the advice is fine — Mikki's argument is that people do not just follow advice, they extend it using the mechanism you gave them, so an explanation that isn't real can cause problems in situations you never had in mind.
- The tryptophan pathway, and where it breaks — the effect requires carbohydrate high enough and protein low enough for tryptophan to win the competition at the blood-brain barrier. Add meaningful protein and it disappears, and the melatonin generated downstream is nowhere near a dose that influences sleep.
- Lower carbohydrate was associated with more slow-wave sleep, not less — a meta-analysis of polysomnography studies found lower carbohydrate intake linked to more slow-wave sleep, less REM and a shorter time to fall asleep. That is the opposite direction to what Mikki had been saying.
- The head-to-head trial in elite female athletes — 40 g of alpha-lactalbumin, 40 g of casein, 40 g of carbohydrate or a non-caloric placebo, taken 30 minutes before bed for three nights each. No difference in total sleep time, sleep onset latency, subjective sleep quality, resting heart rate or heart rate variability, and no difference in overnight glucose on CGM.
- Pre-sleep protein is muscle research, not sleep research — protein before bed does do what it says for the overnight muscle protein synthetic response. That is a separate question from whether it sedates you, and when sleep is the outcome tested directly it is no better than carbohydrate or nothing.
- Why it still works in clinic: it is the calories — the null trial was run in adequately-fed people. Mikki's clinical population is in a deficit and often training hard, so adding food at dinner adds total energy to the day. Carbohydrate is simply the lever she usually reaches for, because protein is normally already handled in a fat-loss meal.
- Energy deficit is not the same thing as low energy availability — two women in the same deficit can be in very different positions depending on training volume. Mikki cites the 30 kcal/kg fat-free mass threshold and is explicit that it is not a number everyone agrees on.
- Cognitive dietary restraint has physiological consequences — over two years in 123 healthy young women, higher restraint was associated with more cycles showing subclinical ovulatory disturbances (56% versus 34%), and physical activity, general stress, BMI and energy intake did not explain the difference. Related work found higher 24-hour urinary cortisol in high-restraint women, replicated in postmenopausal women aged 49 to 75.
- A one-off cortisol test tells you very little — within-person variability is large, the cortisol awakening response has test-retest reliability too poor for individual comparisons, and the research Mikki is drawing on compared groups using 24-hour urinary collections rather than single spot samples. Noise that swamps one person averages out across sixty or a hundred.
- What to actually do — if you are wired, under-eating and training, add the calories back. Put a small portion of carbohydrate on a plate whose protein is already covered; if protein is the thing that is low, add protein instead, particularly through midlife. And if fear of evening carbohydrate is the real barrier, add it back and watch what happens — evidence builds confidence.
Chapters
00:00 Where I got the mechanism wrong
02:59 Benton, dose and the sleep studies
04:49 Carbs, casein or placebo before bed
07:12 Why it still works in clinic
08:56 Energy deficit, training load and cortisol
11:39 Fear of carbs and dietary restraint
14:19 How reliable is cortisol testing
16:09 What to actually do
Guest / Resources
This is a solo episode. No guest.
Researchers and work referred to in the episode
- David Benton and colleagues, Swansea University — work evaluating the putative mechanisms linking carbohydrate to sleep, and a meta-analysis of polysomnography studies fin
