October 20, 2026 · Sports, Health & Nutrition Daily Digest
A digest of sports, fitness, health, and nutrition developments compiled for October 20, 2026, with summaries, links, and commentary.
I. Nutrition and Diet
1. PLOS Medicine: In 461,682 cancer-free NIH-AARP adults, total ultra-processed food intake was not linked to colorectal cancer, while meat and whole grains pointed in opposite directions (Nutrition)
Summary:
Leila Abar, Caitlin P. O'Connell, Erikka Loftfield, and colleagues published a cohort analysis in PLOS Medicine on October 15. The NIH-AARP Diet and Health Study enrolled adults aged 50-71 in 1995-1996; the analytic sample was 461,682 people free of cancer at baseline (274,269 men and 187,413 women). Diet came from a 124-item food-frequency questionnaire classified with Nova. Median ultra-processed intake was 293 g per 1,000 kcal per day (interquartile range 198-466), about 43% of daily energy. Over a median 18.8 years, 10,075 colorectal adenocarcinomas were recorded. After adjustment for age, sex, race, education, physical activity, smoking, and alcohol, the hazard ratio for the highest versus lowest intake quintile was 0.97 (95% CI 0.91-1.03; P for trend = 0.55). Whole grain, fiber, dairy, and calcium from ultra-processed sources tracked with lower incidence, and meat from ultra-processed sources tracked with higher incidence. The design is observational, diet was self-reported, and residual confounding and misclassification remain possible.
Links:
Commentary:
Colorectal-cancer prevention still means enough whole grains, fiber, dairy, and calcium and less processed meat, rather than a single ultra-processed label; a null result for the total also does not clear every ultra-processed food.
2. Nutrition Reviews: Across 92 randomized trials, diets at or below 30% of energy from fat lowered non-HDL cholesterol by 8.00 mg/dL, with no significant change in apolipoprotein B (Nutrition)
Summary:
Sepideh Soltani, Fatemeh Meshkini, Seyed Reza Mirjalili, Kimia Torabinasab, and Nazanin Moslehi published a systematic review and meta-analysis online in Nutrition Reviews on October 16. Searches ran from inception to June 2, 2024, and were updated on August 8, 2025. Ninety-two randomized trials in adults compared low-fat diets (at or below 30% of energy from fat) with high-fat diets (above 30%). Versus high-fat diets, low-fat diets reduced non-HDL cholesterol by a weighted mean difference of -8.00 mg/dL (95% CI -13.1 to -2.89; I2 = 77.8%; high-quality evidence). Apolipoprotein A1 fell by 2.51 mg/dL at low-quality evidence. Apolipoprotein B, apolipoprotein B100, apolipoprotein B in LDL cholesterol, apolipoprotein C3, and the ApoB/ApoA1 ratio did not change significantly, at low to very low quality. The total-cholesterol/HDL-cholesterol and triglyceride/HDL-cholesterol ratios rose slightly but significantly at moderate quality, and lipoprotein(a) did not change. PROSPERO registration: CRD42023420978.
Links:
Commentary:
Keeping fat at or below 30% of energy has high-certainty evidence for lowering non-HDL cholesterol, while apolipoprotein B did not fall with it, so the lipid panel is not a blanket win for low-fat eating.
3. Frontiers in Nutrition: In 25 randomized trials of four diabetes diets, pooled HbA1c fell by 0.50 percentage points and the prediction interval still crossed zero (Nutrition)
Summary:
Haojie Xu, Haotian Ma, Siyao Jin, Yanming Li, Shaolong Zhang, and Libo Zhao published a systematic review and meta-analysis in Frontiers in Nutrition on October 12. Searches through March 13, 2025, included 25 randomized trials in adults with type 2 diabetes comparing a Mediterranean diet, time-restricted eating with a 6- to 10-hour daily window, energy restriction, and a ketogenic diet. The PROSPERO record was retrospective (CRD420261478257); a planned network meta-analysis was changed to pairwise meta-analysis before formal analysis because the network was not connected. The pooled mean difference for HbA1c was -0.50 percentage points (95% CI -0.65 to -0.36), with I2 = 83.8% and a 95% prediction interval of -1.04 to 0.03 that crossed zero. The pooled mean difference for weight was -3.00 kg (95% CI -4.31 to -1.69; I2 = 93.9%; 2,159 people in the weight analyses). Confidence intervals for waist circumference, total cholesterol, LDL cholesterol, and triglycerides crossed the null. Some trials also included exercise, and the authors state that the estimates are not pure diet effects. Ketogenic and very-low-calorie ketogenic programs had more gastrointestinal symptoms and early hypoglycemia; the percentages cited come from individual trials, not a pooled rate.
Links:
Commentary:
For type 2 diabetes, choose a pattern that can be kept and check whether glucose-lowering drugs need adjustment, because this heterogeneous pooling does not crown one diet; where ketogenic numbers look larger, the adaptation period also needs closer supervision.
II. Exercise and Training Science
4. Advances in Nutrition: Across 78 randomized trials and 8,452 adults with metabolic syndrome, exercise, diet, and the combination still do not identify a preferred program (Training)
Summary:
Jingfeng Chen, Yujia Qu, Yue Cheng, Andrew Mark Edwards, Yanjie Zhang, Fenghua Sun, and Chen Zheng published a Bayesian network meta-analysis and component network meta-analysis online in Advances in Nutrition on October 17 (accepted October 16). Searches through December 2025 included 78 randomized trials and 8,452 adults with metabolic syndrome and overweight or obesity. The comparisons were structured exercise alone, diet alone, and exercise plus diet. Relative to an inactive or minimal-intervention control node, all four structured exercise-only approaches were associated with lower systolic and diastolic blood pressure, and resistance training was associated with lower fasting glucose. Among combined nodes, exercise plus a Mediterranean diet, moderate-intensity continuous training plus caloric restriction, and moderate-intensity continuous training plus a low-carbohydrate diet were associated with lower waist circumference, triglycerides, and fasting glucose, respectively. Waist circumference and triglycerides fit a simple sum of exercise and diet effects; the other four outcomes did not. Certainty was low to very low for most comparisons. The authors state that this Bayesian synthesis does not name a preferred intervention.
Links:
Commentary:
Metabolic syndrome still calls for both moving and eating, but these 78 trials do not assign a specific class or diet; blood pressure looks more exercise-linked, waist and triglycerides look more additive, and the certainty is still low.
5. Journal of Clinical Medicine: Fifty people with early Parkinson's disease who swapped 18-20 minutes of each session for turning practice were 1.79 seconds faster on the Timed Up and Go after 12 weeks, with unblinded assessment (Training)
Summary:
Kubra Uzun, Burak Bugday, Suzan Aydin, and Omer Sevgin published a randomized trial in the Journal of Clinical Medicine on October 15. Fifty people with Parkinson's disease, Hoehn-Yahr stage 1-2 and aged 50-70, were randomized to three sessions a week for 12 weeks. In the turning group, 18-20 minutes of each 45-50 minute session were structured turning practice and the rest was a reduced conventional program (n = 25); the control was the full conventional program (n = 25). The primary outcome was the Timed Up and Go. After baseline adjustment, the turning group differed by -1.79 seconds (95% CI -1.98 to -1.61), 10-meter gait speed was 0.116 m/s faster (0.108-0.124), and the Berg Balance Scale was 1.84 points higher (1.50-2.17). Everyone finished. Outcome assessment was not blinded, the effects were large, the tests overlapped with what was practiced, and falls and turning quality were not recorded. Registration: NCT07442370.
Links:
Commentary:
People with early, still high-functioning Parkinson's disease can put turning into three weekly sessions rather than practicing only straight walking; the 1.79 seconds come from unblinded tests that resemble the training and do not show a reduction in falls.
III. Public Health and Disease Prevention
6. Clinical Gastroenterology and Hepatology: A model finds twice-yearly liver-cancer screening cost-effective in alcohol-associated cirrhosis once annual risk is about 0.45% (Public health)
Summary:
Ali Hajjar, Jovan Julien, Gizem Nemutlu, H. Mert Sahinkoc, Mary Linton B. Peters, Fasiha Kanwal, and Jagpreet Chhatwal used a Markov microsimulation to compare twice-yearly ultrasound plus alpha-fetoprotein with no screening in alcohol-associated cirrhosis. The paper is in Clinical Gastroenterology and Hepatology, and Mass General Brigham described the results on October 20. At a U.S. willingness to pay of $100,000 per quality-adjusted life year, screening became cost-effective when annual hepatocellular carcinoma incidence reached about 0.45 per 100 person-years, below the 1 per 100 person-years figure used in current guidance. The incremental cost-effectiveness ratio was $18,952 per quality-adjusted life year in people who continued to drink and $56,000 in people with sustained abstinence. Per 100,000 people, screening added about 2,083 life-years and 992 quality-adjusted life-years among those still drinking, and 3,357 life-years and 1,700 quality-adjusted life-years among those abstinent. This is a model, not a count of cancers prevented in a prospective cohort.
Links:
- Clinical Gastroenterology and Hepatology — Incidence threshold for liver-cancer screening in alcohol-associated cirrhosis
- Mass General Brigham — October 20 account of the model
Commentary:
For someone who already has alcohol-associated cirrhosis, twice-yearly ultrasound plus alpha-fetoprotein should not be set aside only because estimated annual risk is under 1% or drinking continues; the result is a cost-effectiveness model, and the scan decision still follows the liver specialist's staging.
7. Annals of Internal Medicine: A model projects that shared decision-making on hepatitis B vaccine would raise childhood chronic infections from 255 to 467 in one U.S. birth cohort (Public health)
Summary:
Amir M. Mohareb, Lakshmi Ganapathi, Caitlin M. Dugdale, Ruma Rajbhandari, Andrea Ciaranello, Anne M. Neilan, Ravi Jhaveri, Arthur Y. Kim, and Emily P. Hyle published a Markov cohort model in Annals of Internal Medicine on October 20. It follows one year of U.S.-born infants, about 3.6 million, for 20 years, and compares a universal birth dose, deferral to age 2 months, and shared decision-making. Assumed completion was 91.1%, 87.1%, and 50.0%, and maternal hepatitis B surface antigen prevalence was 0.4%. Against 255 childhood chronic infections under the birth-dose strategy (95% uncertainty interval 130-530), deferral produced 303 (155-610), a 19% increase, and shared decision-making produced 467 (209-1032), an 83% increase. People still susceptible at year 20 numbered 232,000 with a birth dose, 474,000 with deferral, and 1,806,000 with shared decision-making. In a scenario with 84.5% coverage, maternal prevalence of 0.72%, and higher horizontal transmission, chronic infections rose to 1,700 and 2,192. Infants whose mothers were known to be positive were still vaccinated in every scenario. Uncertainty in horizontal transmission is the stated limitation. Funding was from the National Institute of Allergy and Infectious Diseases.
Links:
Commentary:
This is a projection under assumed completion rates, not an observed rebound; at the model's own 50% completion, replacing the birth dose with parental choice increases both childhood chronic infections and the number still susceptible in adulthood.
IV. Sleep and Mental Health
8. BMC Psychiatry: Fifteen adults with ADHD and a known good response to stimulants slept in a more fragmented way and rated sleep worse during a week off medication (Sleep)
Summary:
Kevin Mammeri, Sebastien Weibel, Eva Rufenacht, Aurore A. Perrault, Laurence Bayer, and colleagues published a randomized crossover trial in BMC Psychiatry on October 16. Thirty-six adults with attention-deficit/hyperactivity disorder who had previously responded well to methylphenidate or lisdexamfetamine had clinical screening and an adaptation night of polysomnography; 15 then entered the crossover. Each spent one week on stimulant medication and one week off it, in random order, followed by polysomnography at the sleep center, plus questionnaires, sleep diaries, and actigraphy. Relative to the off-medication week, stimulants reduced self-reported fatigue, improved subjective sleep quality, reduced transitions from deeper to lighter sleep, and reduced arousals in deep sleep, and they raised peak frequencies of fast sleep spindles and slow oscillations. The authors read this as no adverse sleep outcome in this group. The sample kept only known good responders, and the paper states that there is no clinical registration number. Funding was from Foundation TDAH-InSight.
Links:
Commentary:
Adults with ADHD who already take a stimulant their clinician considers effective need not stop it on the belief that stimulants ruin sleep; these 15 people do not stand in for people without that diagnosis or for someone just starting treatment.
V. Evidence and Clarification
9. Cell: With a lipopolysaccharide background, extracellular L-carnitine amplified inflammation in mice through OCTN2, and Yale says red meat and supplements were not tested (Evidence)
Summary:
Eric I. Elliott, Tomomi M. Yoshida, Andrew Wang, and colleagues published a paper online in Cell on October 19, and Yale School of Medicine explained it on October 20. In mice given lipopolysaccharide, and in mouse and human macrophages primed with lipopolysaccharide, extracellular L-carnitine entered cells through the high-affinity transporter OCTN2, suppressed mevalonate synthesis, and activated the pyrin inflammasome. IL-1-beta-dependent IL-17a increased death in the endotoxemia model. Inflammation after skeletal-muscle injury in a mouse rhabdomyolysis model also required pyrin and OCTN2. First author Elliott told Yale that the experiments used L-carnitine alone, not red meat or exercise supplements as wholes; carnitine by itself did not induce inflammation, a substantial inflammatory background was required, and it is not yet known which human inflammatory setting would reproduce the effect.
Links:
- Cell — The pyrin inflammasome senses extracellular L-carnitine
- Yale School of Medicine — October 20 note on what the diet finding does not show
Commentary:
People taking an L-carnitine supplement or eating red meat cannot use this mouse and cell work to stop or increase the dose; it describes amplification when inflammation is already present, not a rate of disease from ordinary meals.
Today's Summary
- Diet: In 461,682 NIH-AARP adults, the highest versus lowest quintile of ultra-processed food had a colorectal-cancer hazard ratio of 0.97; across 92 trials, fat at or below 30% of energy lowered non-HDL cholesterol by 8.00 mg/dL without a significant apolipoprotein B change; for four diabetes diets, the HbA1c prediction interval crossed zero.
- Training: Seventy-eight trials and 8,452 adults with metabolic syndrome still do not rank exercise, diet, or the combination; 50 people with early Parkinson's disease who gave 18-20 minutes of each session to turning practice were 1.79 seconds faster on the Timed Up and Go, with unblinded assessment.
- Public health and sleep: A model put cost-effective liver-cancer screening in alcohol-associated cirrhosis at about 0.45 cases per 100 person-years; shared decision-making on hepatitis B raised modeled childhood chronic infections from 255 to 467 in a birth cohort of about 3.6 million; 15 adults with ADHD slept in a more fragmented way during a week off stimulants.
- Evidence limit: L-carnitine amplified inflammation in mice and macrophages that were already exposed to lipopolysaccharide, and the authors say red meat and supplements themselves were not tested.
Daily Framing:
Today was a disaggregation day in the sports, health, and nutrition cycle: ultra-processed food, low-fat eating, and diabetes diets broke into specific foods, lipid markers, and side effects, screening and the hepatitis B birth dose moved as model thresholds, and carnitine supplements remain inside experiments that already had inflammation.
This digest is compiled from real-time search results and is for reference only.