What happened to older adults’ thinking as their social lives shrank in the pandemic? Among 1,679 adults, those whose participation dropped declined more than peers who stayed involved, resembling people who stayed isolated.
Earlier work tied persistent isolation and loneliness to poorer memory. This paper asks what happened when older adults’ community participation changed during the pandemic.
The analysis included 1,679 older adults for thinking scores and 2,457 for depressive symptoms. It tracked community participation from 2016 to 2020 and changes in both outcomes from 2016 to 2022, comparing people whose participation fell, rose or stayed consistent.
Those whose participation dropped declined more in thinking than consistently active peers, resembling people who stayed isolated. People who started participating held up comparably to consistently active peers. The drop did not predict worse long-term depressive symptoms. The paper does not give the size of the cognitive difference.
Health changes could explain both reduced participation and poorer thinking, despite the analysis accounting for physical health constraints.
Testing support for social activities would help answer whether participation itself protects thinking skills in later life.
🤖 Made with AI
1
1
Sources and notes
The Long-term cognitive and mental health impacts of pandemic-induced social isolation: a trajectory analysis. The Journals of Gerontology, Series B. Journal article, not a preprint. DOI: 10.1093/geronb/gbag220
agingresearch.news/s/4i3w2v
A nationally representative longitudinal analysis of older adults included 1,679 for cognition, meaning thinking skills, and 2,457 for depressive symptoms. Community participation trajectories from 2016–2020 were consistently isolated, pandemic drop-off, consistently engaged and pandemic initiators. Changes in cognition and depressive symptoms were assessed from 2016 to 2022. Models used survey weights to reflect the population and adjusted for starting scores, social and economic disadvantages, and physical health constraints. The paper does not give counts for individual participation groups.
Compared with consistently engaged peers, adults whose participation dropped had greater cognitive decline by 2022, resembling consistently isolated adults. People who began participating preserved cognition comparably to consistently engaged peers. Drop-off did not predict worse long-term depressive symptoms. Childhood adversity independently predicted poorer mental health.
Models using two-category measures detected no change in the relationships between participation and outcomes according to childhood adversity. Checks using measures across a range suggested a possible interaction between starting social participation later in life and cumulative early-life trauma, favoring cognitive preservation. Effect sizes, confidence intervals and p-values are not given.
Limit: Health changes could have contributed to both reduced community participation and cognitive decline despite adjustment for physical health constraints.
2
Keeping diabetes out of later life is a reason to care about what's on your plate. People who ate closer to a plant-rich diet had less diabetes, with the lowest risk at 84 points out of 130.
Vegetables, fruit, nuts and beans take priority over animal foods in the Planetary Health Diet. Earlier studies tied it to lower death rates and less heart and blood vessel disease.
More than 100,000 UK adults, averaging around 55 years old, completed at least two questionnaires about the previous day's food and drink. Diet was scored from zero to 130; health records tracked disease over a median 12.5 years.
Each five-point higher score went with 3% lower stroke risk and 3% lower type 2 diabetes risk, compared with a score five points lower. Diabetes risk was lowest at 84 points. There was no link with heart disease caused by reduced blood supply, or with having two or more of the diseases studied.
The catch is other health advantages: people with higher scores were more educated and less often overweight or obese. Adjustments covered these differences but could miss others; extra adjustment for medication use left the findings much the same.
The next test is whether changing the diet actually prevents a first illness.
🤖 Made with AI
1
4
Sources and notes
Association of Planetary Health Diet with decreased risk of cardiometabolic diseases, multimorbidity, and its progression: a prospective cohort study. Annals of Medicine. Not a preprint. DOI: 10.1080/07853890.2026.2718534
agingresearch.news/s/2pwseu
Prospective UK Biobank study: 108,596 to 112,323 adults across disease analyses; 105,643 in progression analyses. Diet-score quarters were compared, with the lowest as reference. Participants completed at least two dietary recalls; median follow-up was 12.5 years. Each disease analysis excluded existing cases. Adjustments covered age, sex, ethnicity, education, employment, deprivation, body mass index, smoking, alcohol, activity and energy intake.
Per five-point score increase, stroke risk was 3% lower (hazard ratio 0.973, 95% confidence interval 0.951–0.997; p < 0.05 after adjustment for multiple comparisons); diabetes risk was 3% lower (0.968, 0.952–0.984; p < 0.001). Hazard ratios compare rates during follow-up. Diabetes hazard ratios for successive quarters above the lowest were 0.875 (0.805–0.951), 0.835 (0.773–0.902) and 0.881 (0.805–0.965), all with 95% confidence intervals; adjusted p < 0.01. Risk was lowest around 84 points (nonlinearity p = 0.04; trend p < 0.0001). No link emerged for heart disease caused by reduced blood supply or multiple diseases.
Highest versus lowest quarter: first-disease hazard ratio 0.89 (95% confidence interval 0.84–0.94); death before first disease 0.81 (0.73–0.89), both adjusted p < 0.001. Other disease transitions could reflect chance.
Findings stayed similar across demographic subgroups and after excluding disease during dietary assessment or the first two years, accounting for competing deaths, or adjusting for medications or menopause. Progression findings also stayed similar after excluding same-day transitions or using disease-specific deaths. E-values, measures of sensitivity to unmeasured confounding, ranged from 1.11–1.67 for disease outcomes and 1.11–1.77 for progression.
- Other health differences can remain after adjustment; diet was not assigned.
- Treatment after first disease was unaccounted for.
- Selection into the progression analysis could bias results.
- Recalled diets can be misreported; soybean intake may be missed.
- Participants were mainly White and recruited at ages 40–69, restricting generalization.
- Estimated ordering of same-day diagnoses can be inaccurate; excluding these cases left results similar.
3
For anyone worried about developing diabetes, a blood sample holds more than a sugar reading. A model using blood proteins ranked future diabetes risk better than a blood-fat measure alone in about 20,000 people.
The body's response to insulin, the hormone that regulates blood sugar, starts to falter early on the path to diabetes. Blood proteins can reflect that trouble across several organs.
The study began with 161 men and women with or without type 2 diabetes. Insulin infusions, with blood sugar held steady, measured how well their bodies responded. The model combined 13 blood proteins with sex, a longer-term blood sugar measure and a blood-fat ratio.
In about 20,000 UK Biobank participants, its score for ranking future diabetes risk was 0.86, versus 0.74 for the blood-fat ratio alone. A higher score means better ranking. But the longer-term blood sugar measure alone scored 0.88: the protein model performed similarly to that familiar test.
The initial insulin-testing groups were modest in size and mostly of European ancestry, leaving performance in other populations unsettled.
Its promise is broader risk prediction: the model also ranked fatty liver disease risk better than the blood sugar measure.
🤖 Made with AI
1
13
Sources and notes
Plasma proteomics identifies insulin resistance signatures predictive of cardiometabolic disease risk. Nature Metabolism; not a preprint. DOI: 10.1038/s42255-026-01662-1
agingresearch.news/s/he8g7b
Human protein-profiling and prospective prediction study: 161 adults with or without type 2 diabetes, in discovery (75), validation (43) and exercise (43 men) groups. Exercise lasted eight weeks. Validation matched sex, body mass index, age and smoking status. Predictions were tested in about 20,000 UK Biobank participants and 1,700 independent participants, with median follow-up of 18.1 years in the latter group.
Protein associations adjusted for sex; adding age and body fat reduced the associated set from 488 to 62 (adjusted P < 0.05). Model candidates adjusted for body mass index, age and sex.
The 13-protein model also included sex, longer-term blood sugar and a blood-fat ratio. Diabetes ranking scores (concordance indices; 95% confidence intervals): model 0.86 (0.84–0.89), blood-fat ratio 0.74 (0.71–0.77), longer-term blood sugar 0.88 (0.85–0.91). Fatty liver: model 0.78 (0.71–0.85), blood sugar 0.58 (0.48–0.69).
Insulin-sensitivity prediction: validation correlation R = 0.73 versus 0.49 with clinical features alone; cross-validation R = 0.67.
Exercise improved insulin sensitivity by 30–40% and aerobic capacity by 8–15%; 22 proteins changed (adjusted P < 0.05). Responses differed across metabolic groups.
Independent follow-up hazard ratios (relative rates; 95% confidence intervals): diabetes 0.49 (0.39–0.61), obesity 0.61 (0.48–0.76), chronic kidney disease 0.5 (0.41–0.61), sleep apnea 0.68 (0.53–0.87), death 0.56 (0.45–0.71), stroke 0.66 (0.5–0.87), chronic obstructive pulmonary disease 0.63 (0.5–0.8), lung cancer 0.62 (0.46–0.84). Higher predicted sensitivity meant lower risk. A glucose-challenge index was the strongest single diabetes predictor: 0.21 (0.12–0.36). For most other outcomes, the model matched or exceeded fasting measures.
- The insulin-testing groups were modest in size and mainly of European ancestry, constraining generalizability.
- Organ assignments used gene-activity databases rather than protein databases.
- Model inputs were restricted to proteins available on one testing platform.
- Medication effects on blood proteins were not systematically studied.
- Exercise subgroup analyses were exploratory, with small groups and unadjusted P values.
- Broader populations and repeated sampling are needed to assess prediction stability over time.
12
Feeling lonely came before more trouble remembering words and getting around. Those problems also came before loneliness, but the pattern was weaker in that direction across 12 years of older adults’ lives.
Earlier findings on loneliness and later changes in thinking have been mixed. An earlier study using smartphone assessments found limited links between the two.
The new study covered 22,097 adults aged 50 and older in 13 European countries across six surveys over 12 years. Loneliness meant feeling left out, isolated or short of companionship. Memory was tested by recalling words; physical limitations included difficulty walking and climbing stairs.
The links were stronger from loneliness to later memory problems, depressive symptoms and physical limitations than in reverse. More long-term illnesses predicted later loneliness; loneliness predicted more illnesses only between the first two surveys.
Personality was not accounted for and could influence both loneliness and health. Adjusting for social relationships, health behaviors and other measured factors reduced some links, but generally preserved their direction.
The next test is whether easing loneliness helps preserve memory, mood and mobility.
🤖 Made with AI
1
2
Sources and notes
Does loneliness cause poor health or vice versa? A longitudinal analysis across European countries. European Journal of Ageing. Journal article, not a preprint. DOI: 10.1007/s10433-026-00946-5
agingresearch.news/s/hg5twg
This observational analysis included 22,097 adults aged 50 and older in 13 European countries. Six survey waves covered 12 years. Participants had to complete at least three consecutive waves. People reporting Alzheimer’s disease or dementia at baseline, or missing loneliness or health measurements, were excluded. Average age was 64; 57% were women, and 10.92% met the loneliness threshold at baseline.
Cross-lagged models compared earlier loneliness with later health and earlier health with later loneliness. Adjustments included sociodemographic characteristics, social relationships, health behaviors, stressful lifetime events and country. There were no assigned treatment groups.
Greater loneliness predicted later poorer word recall, more depressive symptoms and more physical limitations. Each also predicted later loneliness. The standardized associations were larger from loneliness to these health outcomes at every time point.
More chronic illnesses predicted later loneliness across all subsequent waves. Loneliness predicted more chronic illnesses only between the first two waves; later estimates could be chance. Comparing unadjusted and adjusted models generally preserved the direction of the relationships, although several estimates decreased after adjustment.
- Personality traits were not included and could influence both loneliness and health.
- The survey excluded people living in institutions, restricting its relevance to adults in long-term care.
- The models could not fully separate lasting differences between people from changes within the same person.
- Requiring consecutive participation selected a group whose excluded peers were older and reported slightly poorer health at baseline. The authors say this could underestimate the loneliness–health relationship in the broader older population.
- Adjusting for health behaviors may also account for pathways through which loneliness might affect health.
2
Imagine heart or blood vessel disease, diabetes and fatty liver piling up in the same person. Among 375,790 people who started free of them, those with better blood sugar handling scores were less likely to develop that mix.
The buildup of these illnesses matters for health in later life. The liver condition is a buildup of fat tied to metabolic problems. Insulin is a hormone that controls blood sugar.
The study tracked 375,790 UK Biobank participants initially free of cardiovascular disease, type 2 diabetes and this liver condition for an average 13.74 years. During that time, 8,984 developed combinations of these diseases. Of nine metabolic scores, an estimate of how well the body responds to insulin best distinguished future cases from non-cases.
Each increase of one standard deviation—a measure of variation among participants—in that score went with a 66% lower relative risk of developing multiple diseases. Higher scores also went with less progression from no disease to one disease, then to several.
Everyone began free of these diseases, so the ranking may differ for people who already have one.
The finding puts an estimate of insulin response first among these scores for spotting the risk of illnesses accumulating in later life.
🤖 Made with AI

ALT Who develops several diseases? An insulin response score best flagged future risk. Lower relative risk of multiple diseases: 66%. Per standard deviation higher insulin response score. Standard deviation: variation among participants. 375,790 initially disease-free people · Prospective cohort study. Peer reviewed.
1
9
Sources and notes
Associations of nine metabolic-related indices with the incidence, progression, and mortality outcomes of cardiovascular–liver–metabolic multimorbidity: evidence from a large prospective cohort. Frontiers in Nutrition. DOI: 10.3389/fnut.2026.1988246
agingresearch.news/s/9q8jjm
This prospective UK Biobank study included 375,790 participants initially free of cardiovascular disease, type 2 diabetes and metabolic dysfunction-associated steatotic liver disease: liver fat accumulation tied to metabolic problems. Mean follow-up was 13.74 years. The comparison was between nine metabolic indices. Analyses examined disease onset, discrimination between future cases and non-cases, and transitions between disease states. Mortality risks with increasing disease burden were also examined.
During follow-up, 8,984 participants developed combinations of these diseases. All nine indices were associated with disease onset and progression. Estimated glucose disposal rate, an estimate of the body's response to insulin, best distinguished future cases from non-cases. Its AUC, a measure of that discrimination, was 0.811 (95% confidence interval 0.807–0.815), outperforming the other eight indices.
Each one-standard-deviation increase—a measure of variation among participants—went with a 66% lower relative risk of multiple diseases: hazard ratio 0.34 (95% confidence interval 0.33–0.35). A hazard ratio compares the rate at which an outcome occurs during follow-up. The corresponding reductions were 65% for developing two of the diseases and 70% for developing all three. Higher scores also went with lower risks of progressing from no disease to one disease and then to multiple diseases.
- Other health differences could have contributed to the relationships between the scores and later disease.
- The findings may not apply to people who already had one of these diseases at enrollment.
4
Why do older adults with less education get fewer years of independence? Low activity accounted for more of that gap than smoking, drinking or obesity in a group of 754 older adults.
Social disadvantage goes with shorter lives and fewer years of independence. Here, disability meant needing help with essential daily activities, so the question was about both living longer and managing daily life.
The study covered 754 adults aged 70 or older in South Central Connecticut, using records from 1998–2024. Disability was checked monthly. Models estimated lifespan and years without disability at age 70, then simulated each lifestyle risk being absent.
For education-related gaps, low activity accounted for 35.4% of men's gap in years without disability and 67.4% of women's. For total life expectancy, the figures were 54.5% and 93.4%, respectively. Smoking, drinking and obesity contributed little to these social gaps.
Poor health could explain part of the pattern: it can make people less active and shorten their lives, and the simulations could not rule that out.
The practical question is whether making activity easier in later life can narrow these gaps.
🤖 Made with AI
1
7
Sources and notes
Contributions of lifestyle risk factors to socio-economic disparities in active and total life expectancy among older adults: A longitudinal cohort study. The Journals of Gerontology, Series A. DOI: 10.1093/gerona/glag260.
agingresearch.news/s/8c3zqk
This longitudinal cohort study used 1998–2024 data from 754 adults aged 70 or older in South Central Connecticut. Education, income, neighborhood disadvantage and alcohol consumption were assessed at baseline only. Smoking, low physical activity and obesity were updated at 18-month intervals. Disability was assessed monthly and meant dependence in essential daily activities.
Models estimated active life expectancy, meaning years without disability, and total life expectancy at age 70 separately for men and women. Simulations set lifestyle risks to absent, separately and together, to estimate their contributions to socioeconomic gaps.
Low activity made the largest modeled contribution to men's gaps by education and neighborhood disadvantage, and women's gaps by education, income and neighborhood disadvantage. Its estimated shares, with years without disability first and total life expectancy second, were:
- Men, education: 35.4% and 54.5%.
- Men, neighborhood disadvantage: 49.4% and 53.4%.
- Women, education: 67.4% and 93.4%.
- Women, income: 39.9% and 55.3%.
- Women, neighborhood disadvantage: 23.8% and 20.1%.
Alcohol consumption, smoking and obesity made negligible contributions to these gaps. These estimates concern differences between socioeconomic groups, rather than the overall health effects of each risk.
- Poorer health could reduce activity and affect survival; the simulations could not rule this out.
- The estimates may not apply to adults under 70 or populations outside South Central Connecticut.
For families thinking about health in later life, this pattern reaches widely: people with higher blood sugar and fat scores had higher rates of 283 diseases as the years went by.
Blood sugar and blood fat are combined in a score used to flag how poorly the body responds to insulin. Called the triglyceride-glucose index, it marks insulin resistance, when the body responds less well to insulin, without measuring it directly.
The study tracked new diagnoses in 361,175 adults, averaging 56.5 years old, for a median 13.5 years. It tested 1,127 diseases, accounting for other factors and the many comparisons.
Higher scores went with higher rates of 283 diseases. An exploratory statistical analysis attributed part of 217 of these links to chronic, low-level inflammation, though its estimated contribution was often small.
Underlying health or other differences could explain both higher scores and later illness. For urinary stones, stricter diagnosis rules and extra adjustments left the direction of the results unchanged.
Together, blood sugar and blood fat may mark a broader pattern of illness as people age.
🤖 Made with AI
2
10
Sources and notes
Triglyceride-glucose index and multiple diseases with systemic inflammation as a potential mediator in a phenome-wide association study. Scientific Reports. Published journal paper, not a preprint. DOI: 10.1038/s41598-026-74168-1
agingresearch.news/s/vw49nh
This prospective UK Biobank cohort included 361,175 participants, mean age 56.5 years; 53.2% were female. Median follow-up was 13.5 years. The study examined the triglyceride-glucose index against new diagnoses of 1,127 diseases. Cox proportional hazards models adjusted for multiple covariates, and P values were corrected for multiple comparisons.
Higher index values were associated with 283 diseases, or 25.1% of those tested, after adjustment and correction for multiple comparisons. Reported hazard ratios reached 4.61.
An exploratory mediation analysis estimated how much of each positive association might operate through a low-grade chronic inflammation score. Indirect effects passed the statistical threshold for 217/283 associations (76.7%). Estimated mediation exceeded 30% for 42 diseases, was 10–30% for 136, and was 0–10% for 39. The estimated contribution was small for many outcomes.
Targeted sensitivity analyses for anemia and urinary calculus, meaning urinary stones, kept the same direction under stricter outcome definitions and additional outcome-specific covariate adjustment.
- The index is a surrogate marker of insulin resistance; it does not measure resistance directly.
- Underlying health or other differences could contribute to both index values and later disease.
- Findings in UK Biobank adults may not apply to other populations.
- The mediation analyses were exploratory, rather than evidence of an established mechanism.
1
7
When shoulder arthritis makes surgery a worry, a drug that helps avoid it would matter. The drugs studied here fell short: replacements were more common among users, at 2.25% versus 1.77% among non-users.
Reports of fewer hip and knee replacements had raised hopes for these drugs in arthritis. They are GLP-1 receptor agonists, drugs that activate a hormone receptor.
Medical records covered 24,669 people with shoulder arthritis in each group, matched for other medical conditions. One group had prescriptions before diagnosis; the other did not. Average follow-up was 731 days for users and 1,220 days for non-users.
Shoulder replacements occurred in 2.25% of users versus 1.77% of non-users. Users also had lower levels of two markers of inflammation, but that finding came alongside more replacements.
The groups could still differ in how severe their arthritis was or how surgery decisions were made. Matching other medical conditions could not settle that.
The shoulder results give no sign of the protection reported for hips and knees.
🤖 Made with AI
1
10
Sources and notes
Do Glucagon Like Peptide-1 Receptor Agonists Modify Progression of Primary Glenohumeral Osteoarthritis? A Propensity Score-Matched Evaluation of Shoulder Arthroplasty Incidence. The Archives of Bone and Joint Surgery. Published journal paper, not a preprint. DOI: 10.22038/abjs.2026.95223.4279
agingresearch.news/s/hkrwn9
This retrospective database study compared people with primary shoulder osteoarthritis by whether they had prescriptions for GLP-1 receptor agonists before diagnosis. People with other shoulder conditions were excluded. Groups were propensity-score matched 1:1 to balance relevant medical conditions, leaving 24,669 people per group.
All comparisons below give users first, then non-users. Mean follow-up was 731 ± 694 versus 1,220 ± 1,006 days (P < 0.001).
Shoulder replacement incidence was 2.25% versus 1.77% (P < 0.001). Estimated survival without shoulder replacement was 95.50% versus 97.23% (log-rank P < 0.001); the paper does not give the time point for these estimates.
Major joint injections were recorded in 18.93% versus 16.87% (P < 0.001), although the mean number of injections per patient was lower among users: 2.48 ± 2.86 versus 2.74 ± 3.23 (P < 0.001).
Mean C-reactive protein and erythrocyte sedimentation rate, two markers of inflammation, were lower among users (both P < 0.001). Mean cortisol, a hormone, was similar between groups.
- Matching medical conditions could not rule out differences in arthritis severity or decisions about surgery.
- Follow-up was shorter among drug users.
The authors concluded that GLP-1 receptor agonist use showed no protection against progression to shoulder replacement, contrasting with findings reported for hip and knee osteoarthritis.
11
Your grocery choices and hopes for a longer life meet an unexpected finding: over 30 years, eating more heavily processed food did not go with higher death rates. Meat and cereal went different ways.
Packaged breads, canned soups and breakfast cereals all sit in this broad food category. Earlier studies have tied higher intake to earlier death. These are ultraprocessed foods: industrial products made mainly from food-derived ingredients and additives.
This study followed 11,185 US adults, averaging 54 years old, for a median 30 years. Diet questionnaires captured what they ate at two visits; 6,656 people died during follow-up.
Overall, there was no clear difference in death rates between the highest and lowest intake quarters after accounting for smoking, activity and other factors. But in exploratory comparisons, the highest intake quarter had an 11% higher relative death risk for ultraprocessed meats, 12% higher for spirits, meaning liquor, and 13% lower for ultraprocessed cereals, each against its lowest quarter.
The catch is the calendar: eating habits were recorded in 1987–1989 and 1993–1995, leaving later diet changes unmeasured across decades.
For diet and longevity, the useful next question is which foods explain those different patterns.
🤖 Made with AI

ALT Do heavily processed foods predict death? Total intake did not predict higher death risk. Higher risk: meats: 11%; Higher risk: spirits: 12%; Lower risk: cereals: 13%. Highest versus lowest intake quarter; relative death risk. Food-category comparisons were exploratory. 11,185 US adults · Observational study. Peer reviewed.
1
14
Sources and notes
Ultraprocessed Food Consumption and Risk of All-Cause Mortality in the Atherosclerosis Risk in Communities Study (ARIC). Current Developments in Nutrition. DOI: 10.1016/j.cdnut.2026.109554
agingresearch.news/s/ae3h5c
Prospective US cohort: 11,185 adults, mean age 54; 55% women, 24% Black. Intake quarters were compared using 66-item diet questionnaires from 1987–1989 and 1993–1995, updating intake with their cumulative average. Median follow-up was 30 years, through 31 December 2022; 6,656 deaths occurred. People with diabetes, cancer, chronic kidney disease or cardiovascular disease at enrollment were excluded.
The main model adjusted for age, sex, race/site, calories, education, neighborhood deprivation, smoking, physical activity and drinking. Further adjustment included body mass index, hypertension, cholesterol and kidney function. Food-category comparisons also adjusted for the other categories.
HR means hazard ratio, comparing death rates over time; all intervals below are 95% confidence intervals.
- Total intake, highest versus lowest quarter: HR 0.99 (0.92–1.06), encompassing no difference. Quartile and continuous analyses gave the same overall conclusion across models; risk was lowest in the middle of intake.
- Exploratory category comparisons: meats, 11% higher risk, HR 1.11 (1.01–1.21), P-trend = 0.04; spirits, 12% higher, HR 1.12 (1.02–1.23), P-trend < 0.001; cereals, 13% lower, HR 0.87 (0.80–0.96), P-trend = 0.002. Beverages, margarine, baked goods, fried foods, sugary snacks and ice cream showed no clear associations.
- Modeled replacement with unprocessed/minimally processed foods: HR 1.00 (0.99–1.01). Exploratory replacement of 1 daily serving of meat with cereal: HR 0.82 (0.75–0.90); spirits with cereal: HR 0.82 (0.76–0.88).
- Interaction P-values: sex 0.45, body mass index 0.57, neighborhood deprivation 0.78, age 0.02. At age 60 or older: HR 1.05 (0.93–1.19); neither age subgroup had a clear association.
- Reclassifying ambiguous foods, excluding deaths within the first 5 years, and imputing missing covariates each left the overall conclusions unchanged.
Authors’ limits:
- Missing brand and preparation details allowed classification errors; classification was not validated.
- Stress, occupational exposures and income instability were not fully captured and could affect both diet and mortality.
- Illness could influence food choices despite disease exclusions and the early-death check.
- Later dietary changes were unmeasured; historical products differed from today’s.
- Cereal details were insufficient to separate whole-grain and added-sugar contributions.
24
If you want more healthy years, headlines say Yale scientists have revealed what really slows aging: prescription drugs, healthy diets and exercise, but not popular supplements. That is partly hype, not a settled answer to which treatments will help people stay healthier for longer.
An aging test could shorten the wait to assess treatments. This peer-reviewed analysis compared 16 epigenetic clocks, aging estimates based on DNA markings, across 51 intervention studies.
Drug and lifestyle interventions produced the largest responses in those DNA measurements. A changing test score does not by itself establish slower aging or longer life.
Clocks trained to predict death risk or the pace of aging responded most and gave consistent results with one another.
Who was studied and how long the study lasted also affected how the DNA measurements responded.
The supplied abstract gives no supplement-specific results, so it cannot settle the claim that popular supplements are largely ineffective.
The next result I want is evidence that treatment-driven clock changes reliably predict longer life in people.
agingresearch.news/s/qgr966
11








