Eating for Better Sleep & Foods that Improve Metabolic Health | Dr. Marie-Pierre St-Onge

St-Onge explained that the thermic effect of food generates heat, and the body's natural circadian drop in core temperature is essential for initiating and maintaining sleep. If you provide a heat-producing meal late in the evening, you blunt that cooling and can wake more frequently or shift to lighter sleep stages. In her self-selected diet study, when participants ate what they wanted, they took 70% longer to fall asleep and lost 20% of deep sleep. While she acknowledges individual variation, her clinical experience and data point to a 3-hour buffer as a practical rule. She also notes that alcohol can disrupt sleep architecture, though the discussion focused mainly on food.

St-Onge's metabolic-chamber study compared an eating window starting 1 hour after waking (e.g., 8 a.m.–6 p.m.) with one starting 5 hours after waking (e.g., 12 p.m.–10 p.m.). Participants ate the exact same foods and amounts. Fat oxidation was significantly lower in the later condition. She compares this to Marta Garullet's findings that people who ate lunch earlier lost more weight. The takeaway is not about intermittent fasting per se, but about aligning food intake with the body's circadian metabolic rhythm, which favors lipid oxidation earlier. She also connects this to the sleep benefits: eating earlier creates a longer fasting window before bed, reinforcing the 3-hour rule.

In the inpatient study where diet was tightly controlled and sleep was measured with polysomnography, St-Onge found a clear within-person association: on days when a participant ate more fiber, they spent more time in deep slow-wave sleep that night. Conversely, more saturated fat meant less deep sleep, and more refined carbohydrates meant more arousals (transitions to lighter sleep). The effect sizes were meaningful—e.g., a 20% reduction in deep sleep on the self-selected high-fat diet. While the mechanism isn't fully elucidated, she hypothesizes that fiber may modulate gut-brain signaling or glycemic control in a way that stabilizes sleep. This protocol aligns with eating more unprocessed plant foods.

St-Onge's PhD work examined purified MCT oil (C8 and C10 fatty acids) compared to beef tallow (saturated long-chain fat) and later olive oil. In controlled feeding studies, she found a statistically significant increase in the thermic effect of food of 45–60 calories after a meal containing MCTs. Over time, a weight-loss study with MCT oil vs olive oil showed greater weight loss and improved body composition. She emphasizes that MCTs are saturated and should not be seen as a free pass to increase overall fat intake, but as a strategic swap. Because the effect is small, it's best combined with other healthful practices like front-loading calories and choosing high-fiber foods.

St-Onge's lab investigated ginger after a graduate student reviewed a list of available spice donations from McCormick. In a crossover design, participants drank ginger powder in warm water versus plain hot water, and the thermic effect of food was measured under a metabolic hood for several hours. Ginger significantly elevated energy expenditure. She hypothesizes it works via the capsaicin receptor (TRPV1), similar to chili peppers. She sees such functional foods as small levers that, cumulatively, can tip the energy balance in a favorable direction, given that most adult weight gain results from a tiny daily calorie surplus over years. The protocol is simple and can be combined with MCT oil or other strategies.

St-Onge expressed strong support for fermented foods, consistent with research from Justin Sonnenburg's lab at Stanford showing that such foods can reduce inflammatory markers. While her own kefir study found no cholesterol benefit, she notes that she only measured cholesterol synthesis—other outcomes like glycemia or gut inflammation weren't tested. She believes the benefits of fermented foods extend beyond any single biomarker. In the context of sleep, a healthy gut may indirectly improve sleep through reduced inflammation and better metabolic control, though this is still an emerging area.

St-Onge emphasized that sleep apnea is alarmingly underdiagnosed. She noted that apnea always involves snoring (followed by a pause and gasping), and that weight loss is the first-line treatment. She encouraged patients to proactively ask their clinicians to assess sleep, not just count hours. She also mentioned that women may show blood pressure sensitivity at lower apnea severity than men. The discussion highlighted that many people resist CPAP because of comfort or Darth Vader stigma, but newer in-home testing and lower-pressure machines have made treatment more accessible. She stressed that objective testing is critical because subjective sleep quality often doesn't capture breathing disturbances.

St-Onge's first NIH grant used a crossover design with 9h vs 4h time-in-bed opportunities for 5 nights. In the first 3 days, participants ate identical controlled diets, then on the final day they could self-select food. Appetite-regulating hormones were measured, and neuroimaging assessed brain responses to food cues. In men, short sleep significantly raised ghrelin, driving hunger. In women, GLP-1—the satiety hormone—was reduced, lowering the 'brake' on eating. No change was seen in leptin or cortisol in the full sample. This is one of the first studies to demonstrate that insufficient sleep may promote weight gain through different biological levers in males and females. The finding underscores the need for sex-specific analyses in nutrition and sleep research, and it also aligns with reports that women with poor sleep often struggle more with satiety and weight maintenance.

The study used a 9-hour time-in-bed opportunity phase during which participants were first fed a controlled diet and then allowed to self-select their meals. Under the controlled diet, sleep was stable; when participants chose their own food, they ate nearly 450 additional calories, 33% more saturated fat, and slightly more carbohydrates. As a result, time to fall asleep increased by over 70%, and slow-wave (deep) sleep decreased by about 20%. By correlating specific nutrients with polysomnography measures night-by-night, St-Onge found that fiber intake predicted more deep sleep, saturated fat predicted less deep sleep, and refined carbohydrates predicted more arousals—transitions from deeper to lighter sleep stages without full awakening. This pattern suggests a direct mechanistic link between dietary macronutrients and the brain's ability to maintain restorative sleep.

The follow-up study recruited habitual good sleepers (≥7h/night) and randomly assigned them to maintain sleep or delay bedtimes by 1.5h for 6 weeks—a reduction that mirrors real-world short sleep. In the sleep-restricted condition, insulin resistance increased, insulin sensitivity fell (more dramatically in postmenopausal women), and blood pressure rose. These changes emerged despite no alteration in cortisol in the earlier severe-restriction study. St-Onge argues that the chronic, low-grade sleep loss permits dietary indiscretions and reduced physical activity to manifest as insulin resistance and hypertension, effectively turning the combination of moderate sleep loss and a typical Western diet into a slow metabolic toxin.

Participants were kept in a metabolic chamber that measured oxygen consumption and carbon dioxide production minute-by-minute. In one condition, they began eating 1 hour after waking and had a 10-hour eating window; in the other, they started 5 hours after waking, shifting all meals later by 4 hours. Despite identical energy intake and macronutrient composition, the later-eating condition reduced fat oxidation, meaning the body burned less stored fat and preferentially stored incoming fat, especially from evening meals. This suggests that the circadian system primes lipid metabolism to be more efficient earlier in the day, and that delaying meals pushes the body toward fat storage. St-Onge interprets this as a clear signal to front-load calories and avoid large, late dinners.

After completing the kefir study, St-Onge wrote the manuscript and submitted it to multiple journals. Each rejected it because the result was not statistically significant and thus perceived as uninteresting. She also described an industry-funded trial that found no benefit of the tested product; she wrote the report, got sponsor approval to publish, but again met a wall of rejections. She noted that this is a widespread problem—null results are essential for science to self-correct, yet journals and reviewers favor positive, exciting findings. The consequence is a literature that overstates true effects and underrepresents interventions that simply don't work, making it harder for clinicians and the public to make evidence-based decisions.