The Slow Agers

Introduction: I am not a scientist, but I wanted to ask readers more specific questions about aging than whether they considered themselves healthy.

The questionnaire asked about the things a person can still do, the routines they maintain, their medical history, and the people they can depend on. The collection plan used Google Forms, with a Google Apps Script prepared to create the questionnaire and connect it to Google Sheets. The design excluded names and email collection, asked for age in broad bands, and made every question after the first three optional. I worked with Claude to synthesize the data and find new, careful insights.

That exercise helped me work through the comparisons the questionnaire could support and identify questions it could not answer. The further evidence about aging came from the published studies cited throughout.

The Why: Two full grocery bags and two flights of stairs are an unremarkable part of a day until someone has to consider whether they can manage both. The How We Age questionnaire asked about that task, along with income, prescriptions, sleep, and the age a person feels.

That ambition differs from reaching a particular birthday. It gives some substance to the years between now and then. Being able to travel, spend time with family, continue useful work, or simply moving through an ordinary day without help. This changes the value of having more time. A discussion of longevity should make room for the circumstances in which those additional years will be lived.

Why the affluent are worth studying

The relationship between money and longevity is substantial enough to establish the stakes. In American tax and mortality records covering 1.4 billion person-years, men in the richest one percent outlived those in the poorest one percent by 14.6 years. For women, the gap was 10.1 years. Separate work combining English and American aging cohorts found that the wealthiest third at age fifty could expect eight to nine more years free of disability than the poorest third. The advantage extended to the years people could live without disability, as well as to how long they lived. (Chetty et al., JAMA 2016; Zaninotto et al., Journals of Gerontology 2020)

Those findings are a reason to examine what happens within affluent populations. Resources can make care, good food, and places to exercise more accessible. They do not give two people identical medical histories or identical control over their days. An income category alone leaves much of a life undescribed, including the routines that survive a busy week and the relationships a person has maintained outside work.

For someone with the means to pursue better health, the question becomes how those resources are used over time. The questionnaire approaches that question through ordinary activities and recurring habits. It asks how often a person trains, how much of the day is spent sitting, whether sleep follows a schedule, and who would answer the phone during a serious problem. Those answers belong alongside the financial ones.

The separation is visible by midlife

The Dunedin Study offers a particularly useful view because its participants were born in the same place and period: 1,037 New Zealanders born in 1972 and 1973. Researchers followed change across nineteen biomarkers between ages twenty-six and forty-five. The slowest-aging participant accumulated 0.40 biological years for every calendar year, while the fastest accumulated 2.44. By forty-five, faster aging also accompanied poorer cognitive, sensory, and motor function, signs of an older brain, and more pessimistic views of getting older. (Elliott et al., Nature Aging 2021)

This gives the phrase “aging slowly” a meaning that a flattering answer on a questionnaire cannot provide. Dunedin followed biological change over time. Most of the other studies in this review examined survival, disability, cognition, or physical capability. The fourteen traits and seventeen risks below bring those findings together without treating their different outcomes as interchangeable. Most are observational associations; the treatment trials are identified where they appear.

The fourteen traits of slow agers

The physical findings begin with cardiorespiratory fitness and preserved strength. In 122,007 adults who completed treadmill testing, the least fit group had roughly five times the mortality risk of the elite group. A separate study of 139,691 adults across seventeen countries associated each five-kilogram difference toward lower grip strength with sixteen percent higher mortality. Both studies measured what the body could do. The case for training also has a practical scale: a pooled analysis of sixteen cohorts placed the largest observed benefit of muscle-strengthening activity around thirty to sixty minutes a week. (Mandsager et al., JAMA Network Open 2018; Leong et al., Lancet 2015; Momma et al., BJSM 2022)

Daily step volume extends the question beyond a scheduled workout. Across fifteen cohorts and 47,471 adults, the most active quarter, averaging about 10,900 steps a day, had less than half the mortality risk of the least active quarter, averaging about 3,600. The association leveled off around 6,000 to 8,000 steps after sixty and 8,000 to 10,000 before sixty. Walking pace supplied another view of capability. Among 474,919 UK Biobank participants, brisk walkers had the longest estimated life expectancy across body-weight categories, roughly eighty-seven to eighty-eight years for women and eighty-five to eighty-seven for men. (Paluch et al., Lancet Public Health 2022; Zaccardi et al., Mayo Clinic Proceedings 2019)

Regular sleep timing introduces a different kind of consistency. In 60,977 adults who wore accelerometers, greater regularity was associated with twenty to forty-eight percent lower mortality than the most irregular fifth, and it predicted mortality better than sleep duration did in that analysis. The Mediterranean dietary pattern has evidence from a randomized trial: among 7,447 Spaniards at high cardiovascular risk, diets supplemented with olive oil or nuts produced about thirty percent fewer events in the combined outcome of heart attack, stroke, or cardiovascular death over 4.8 years. These findings concern a sleep pattern and a way of eating that recur through the week. (Windred et al., Sleep 2024; Estruch et al., NEJM 2018)

Close relationships and a sense of purpose widen the account of health considerably. An analysis of 148 studies covering 308,849 people associated stronger social relationships with a fifty percent higher likelihood of survival. The Harvard Study of Adult Development also emphasized the connection between satisfaction with relationships at fifty and health at eighty. Among 6,985 Americans over fifty in a separate study, those with the weakest sense of purpose were about 2.4 times as likely to die during follow-up as those with the strongest. A health assessment that asks only about training and diet would leave these parts of the evidence out. (Holt-Lunstad et al., PLoS Medicine 2010; Harvard Gazette 2017; Alimujiang et al., JAMA Network Open 2019)

Optimism and beliefs about aging are related, but the studies asked different questions. Research following 69,744 women and 1,429 men for ten to thirty years associated the highest optimism with eleven to fifteen percent longer lives and fifty to seventy percent higher odds of reaching eighty-five. In a cohort of 660 Ohioans over fifty, positive views of their own aging were associated with 7.5 additional years of life over twenty-three years of follow-up, after adjustment for several demographic and health factors. These are large associations. They make outlook worth examining while leaving open how much health shapes outlook, and how much outlook shapes behavior. (Lee et al., PNAS 2019; Levy et al., Journal of Personality and Social Psychology 2002)

Conscientiousness brings attention to the tendency to follow through. In 997 older Catholic clergy followed for twelve years, the highest tenth for conscientiousness had an eighty-nine percent lower risk of Alzheimer’s disease than the lowest tenth. The finding came from a restricted population, and the trait was not associated with fewer plaques or tangles at autopsy. Continued work also appeared in the research, though it presents a different interpretive problem. Among 429,803 self-employed French retirees, each additional year of age at retirement was associated with 3.2 percent lower dementia risk. Healthier people may have been able to work longer. Neither result supplies a simple prescription for personality or retirement. (Wilson et al., Archives of General Psychiatry 2007; Dufouil et al., European Journal of Epidemiology 2014)

Regular communal practice adds a setting in which relationships and routines can be sustained. Among 74,534 women in the Nurses’ Health Study, attending religious services more than once a week was associated with thirty-three percent lower mortality over sixteen years than never attending. Social support, lower smoking, and less depression explained part of the association. The population and the setting matter: this was a finding about female nurses and religious attendance, not a trial of a generic social activity. (Li et al., JAMA Internal Medicine 2016)

The fourteenth trait concerns treating risks that may receive little attention in day-to-day life. Here the evidence includes randomized interventions. SPRINT MIND enrolled 9,361 adults and found nineteen percent fewer new cases of mild cognitive impairment with intensive blood pressure treatment; its primary outcome of probable dementia did not reach statistical significance. In the 977-person ACHIEVE trial, a hearing intervention slowed cognitive decline by forty-eight percent in the higher-risk subgroup, with no benefit detected in the full trial population. The qualifications are part of the findings, and the findings give prevention a place beside the more visible work of exercise. (SPRINT MIND, JAMA 2019; ACHIEVE, Lancet 2023)

The seventeen risks that deserve attention

Reading the favorable traits as a test of character would misread much of the evidence. Depression, hearing loss, and high blood pressure are health concerns, and their presence says little about a person’s willingness to work at getting better. Several risks in this section come from the 2024 Lancet Commission, which estimated that fourteen potentially modifiable factors account for about forty-five percent of dementia cases worldwide. That is a population estimate, not the portion of any individual’s future that can be controlled. (2024 Lancet Commission summary)

Repetitive negative thinking is one of the more difficult findings to interpret. In 292 adults over fifty-five, habitual rumination and worry were associated with faster decline in memory and global cognition over four years and with more amyloid and tau on brain scans. Chronic patterns of thought were the concern, rather than the ordinary distress of a difficult period. The result sits alongside evidence on cynical distrust: among 622 older Finns, the highest distrust was associated with roughly three times the dementia risk over about eight years. That second finding rested on forty-six dementia cases and needs replication. Depression is another risk identified by the Commission; the negative-thinking research also associated depressive and anxious symptoms with later cognitive decline. These findings do not establish that a past diagnosis means someone has an untreated illness today. (Marchant et al., Alzheimer's & Dementia 2020; Neuvonen et al., Neurology 2014; 2024 Lancet Commission summary)

Loneliness and social isolation also need to be kept distinct. In research covering more than 3.4 million people, loneliness was associated with twenty-six percent higher early mortality among those healthy at the start. The same analysis associated social isolation with twenty-nine percent higher mortality and living alone with thirty-two percent. Those categories describe different circumstances. A person can live alone and have dependable relationships, or share a home and feel lonely. Asking both about living arrangements and available support gives a more useful account of someone’s situation than treating an address as a measure of connection. (Holt-Lunstad et al., 2015)

Long sitting hours and heavy television use return the discussion to how an entire day is spent. Among 1,005,791 adults, those who sat more than eight hours a day and were least active had fifty-nine percent higher mortality than active people sitting fewer than four hours. In that analysis, sixty to seventy-five minutes of daily moderate activity removed the excess risk associated with sitting, while only attenuating the association with five or more hours of television. Slow gait supplied a separate warning in midlife: among 904 Dunedin participants examined at forty-five, slower measured walking accompanied accelerated biological aging, weaker brain measures, and older-looking faces. (Ekelund et al., Lancet 2016; Rasmussen et al., JAMA Network Open 2019)

Central adiposity explains why body weight cannot answer every question about metabolic health. In 359,387 Europeans, the largest waist-circumference fifth had about twice the mortality risk of the smallest even after accounting for body mass index: 2.05 times the risk in men and 1.78 times in women. Type 2 diabetes and hypertension in midlife are also on the Commission’s list. High LDL cholesterol in midlife was added in 2024, with an estimated population contribution of about seven percent of dementia cases. These are distinct measurements and conditions. A record of “high cholesterol” or a body mass index leaves important details unresolved. (Pischon et al., NEJM 2008; 2024 Lancet Commission summary)

Untreated hearing loss and untreated vision loss are easier to overlook in a discussion dominated by exercise and food. The Commission’s population estimates attributed about seven percent of dementia cases to hearing loss and two percent to vision loss, the latter added in 2024. The hearing-intervention trial described above gives this part of the literature particular interest, provided its higher-risk subgroup result remains attached to the group in which it was found. (2024 Lancet Commission summary; ACHIEVE, Lancet 2023)

Smoking carries an unusually long record of follow-up. In the fifty-year study of British doctors, lifelong smokers died about ten years earlier than those who had never smoked (Doll et al., BMJ 2004). The alcohol evidence complicates the familiar idea that modest drinking is protective. A 2023 analysis of 107 cohorts and 4.8 million people found no mortality protection below twenty-five grams of alcohol a day, with significantly increased risk from twenty-five grams for women and forty-five grams for men. Those study thresholds should not be read as amounts below which drinking is harmless. (Zhao et al., JAMA Network Open 2023)

A diet built on ultra-processed food and short sleep through midlife complete the seventeen risks. Among 19,899 Spanish university graduates, more than four daily servings of ultra-processed food were associated with sixty-two percent higher mortality, with an eighteen percent increase associated with each additional serving. Among 7,959 British civil servants followed for twenty-five years, sleeping six hours or less at ages fifty, sixty, and seventy was associated with thirty percent higher dementia risk. Both findings concern exposure sustained through ordinary life, a useful reason to ask about the typical week rather than the week someone would prefer to describe. (Rico-Campà et al., BMJ 2019; Sabia et al., Nature Communications 2021)

Where men and women diverge

The exercise comparison is particularly instructive. Regular exercise was associated with twenty-four percent lower mortality in women and fifteen percent in men. Women reached the reduction associated with men’s peak benefit at about 140 minutes a week; men reached that peak at about 300 minutes. The overall percentages and the time comparison answer different questions, but together they show why a single average can conceal useful information. (Ji et al., JACC 2024)

The other comparisons below also depend on the population, exposure, and outcome being measured. A difference in estimated lifespan cannot be added to a reduction in mortality risk, and findings from separate cohorts cannot be assembled into a personal forecast. The table preserves the comparisons reported in the supplied evidence review.

The walking comparison includes a slow, underweight reference group. The alcohol figures are study results, not safe-intake limits. These are published-research comparisons, not simulated results.

The five-habit estimate needs an additional qualification because that study counted moderate drinking as a favorable habit, while the later alcohol analysis found no mortality protection at low intake. It would be a mistake to use the earlier estimate as a reason to start drinking. Some of the other influential findings also come from one sex alone: the religious-attendance study followed female nurses, and the original Harvard cohort comprised 724 men. Their results do not establish an identical effect in everyone else. (Li et al., Circulation 2018; Zhao et al., JAMA Network Open 2023; Li et al., JAMA Internal Medicine 2016; Harvard Gazette 2017)

The six-state review

The accompanying case study is an exercise in what the questionnaire can describe. Its 1,000 records were generated to model a relatively affluent, health-conscious audience; nobody was interviewed. The assigned locations are Ohio, 158; Texas, 217; Pennsylvania, 156; Florida, 169; California, 162; and New York, 138. The file contains 341 female records, 644 male records, and fifteen with sex undisclosed. Ages forty to sixty-nine account for 71.5 percent of the model. Household-income categories of $250,000 or more account for 58.1 percent, with another 5.5 percent declining to specify income. These are construction choices, not estimates of the readership. [D]

Within the model, 68.5 percent have excellent or very good health assigned, 12.0 percent have fair or poor health, and 72.0 percent describe their aging as somewhat or much slower than their peers’. The share answering “Yes, easily” to all four functional questions is 49.2 percent. Those questions cover getting up from the floor without hands, jogging a mile, carrying groceries upstairs, and doing ten push-ups. The difference between the two percentages cannot establish overconfidence; perceived aging and performance on four tasks are different measures. It does show why the questionnaire should keep both answers visible. [D]

The assigned habits are favorable in several respects. Exercise averages 4.7 days a week, and 84.8 percent of records have strength training at least twice weekly. Yet 49.2 percent have fewer than seven hours of weeknight sleep assigned, and 34.2 percent have negative thoughts replaying often or more frequently. Recent blood pressure checks appear in 94.4 percent of records, compared with 60.6 percent that specify knowledge of both blood pressure and LDL cholesterol. These figures illustrate the different questions a real survey could examine; their values come from the model’s assumptions and generated answers. [D]

The same caution applies to the comparison between female and male records. The columns below describe this file, without age adjustment. The fifteen records with sex undisclosed remain in the overall totals. Nothing in these small model differences establishes a biological difference between men and women. [D]

Each percentage uses the denominator shown.

The most pronounced comparison is also the one most clearly shaped by the generation method. Among the 720 records describing slower aging, 77.8 percent also have excellent or very good health assigned and 62.2 percent can perform all four tasks easily. Among the remaining 280 records, those shares are 44.6 percent and 15.7 percent. Shared traits were deliberately built into the model to connect outlook, health, and function. Finding that connection again in the output is not an independent discovery about how people age. A real survey would need to allow the answers to contradict the assumptions that motivated it. [D]

The mindset thread

The research on outlook deserves attention without becoming a promise that the right attitude will prevent illness. The Ohio cohort associated positive views of aging with longer life; Dunedin found pessimism alongside faster biological aging by forty-five; the repetitive-negative-thinking study associated chronic rumination with cognitive decline and brain markers. The relationship appears in different kinds of evidence, but its direction remains unresolved. Declining health can give someone sound reasons to worry about the future. (Levy et al., Journal of Personality and Social Psychology 2002; Elliott et al., Nature Aging 2021; Marchant et al., Alzheimer's & Dementia 2020)

Behavior is one plausible connection. In the religious-attendance study, social support, lower smoking, less depression, and greater optimism explained part of the mortality association. In the optimism cohorts, the most optimistic participants were also the most physically active. My reading is that expectations may help shape what people continue to make time for. A person who expects to remain engaged in the coming decade has a reason to protect the activities and relationships that would make that decade worth anticipating. That interpretation is narrower than claiming that optimism itself adds years to a life. (Li et al., JAMA Internal Medicine 2016; Lee et al., PNAS 2019)

It also leaves room for a less flattering answer. Someone can feel younger while finding ordinary tasks harder, or feel discouraged while retaining considerable physical capacity. Asking what the person can do gives the answer about feeling young some context. Asking about mood and expectations makes it harder to treat a physical result as the whole account of their life. Neither answer needs to be forced into a reassuring score.

What holds, and what to hold loosely

The size of a reported association is not enough to decide how much weight to place on it. The cynical-distrust study had forty-six dementia cases, and its mortality association disappeared after adjustment. The conscientiousness finding came from Catholic clergy and did not correspond to fewer plaques or tangles at autopsy. Those are reasons to be restrained about assigning a health meaning to someone’s personality. The retirement finding has a comparable limitation: it may partly describe who remained healthy enough to keep working. (Neuvonen et al., Neurology 2014; Wilson et al., Archives of General Psychiatry 2007; Dufouil et al., European Journal of Epidemiology 2014)

The trials deserve scrutiny too. The Mediterranean-diet trial was retracted and republished in 2018 after randomization problems at some sites; the corrected analysis reached a similar conclusion. SPRINT MIND’s nineteen percent reduction concerned mild cognitive impairment, a secondary outcome. ACHIEVE’s forty-eight percent figure belonged to a higher-risk subgroup. Keeping those details attached to the results allows the evidence to be useful without making it larger than it is. (Estruch et al., NEJM 2018; SPRINT MIND, JAMA 2019; ACHIEVE, Lancet 2023)

Most of the research here is observational, and many cohorts were predominantly white, drawn from wealthy countries, or composed of health professionals, civil servants, and university graduates. Their findings may not travel unchanged to other populations. Relative risks also need a starting risk before they can tell someone much about their own circumstances. The percentages in this review cannot be added together to calculate either protection or years gained.

I would judge an optimization routine by the life it allows a person to keep participating in. Clinical measurements matter, and so does whether a familiar activity now requires help. That standard leaves room for improvement without treating illness as a verdict on discipline. For someone who has spent a career earning more control over their time, retaining the capacity to use that time is a substantial part of the return. The grocery bags and the stairs belong in that accounting.

Appendix: the complete six-state model

Every figure in this appendix describes the same 1,000 records. I then tried to measure prevalence among residents of the six states. The numbered rows retain the fourteen-trait and seventeen-risk organization used in the original review. Definitions and the questionnaire’s measurement limits follow the tables. [D]

Fourteen traits: corresponding answers

Seventeen risks: corresponding answers

The diagnosis, nicotine, food, and sleep categories cannot reproduce several of the studies’ more specific exposures. [D]

Data and measurement notes

[D] The figures are carried forward from the completed analysis of the raw data.

Percentages use all 1,000 records unless another denominator is shown. Checkbox answers are counted separately and can sum to more than one hundred percent. “All four tasks easily” requires “Yes, easily” on every functional question. “Perceived slower aging” combines “Somewhat more slowly” and “Much more slowly.” Four-or-five ratings combine the upper two responses on a five-point item. No subgroup with fewer than five records is reported.

The questionnaire did not include measured aerobic capacity, grip strength, training duration, a timed gait assessment, an accelerometer-derived sleep measure, or a validated dietary score. The relationship, purpose, optimism, and routine questions are brief indicators rather than reproductions of the studies’ assessments. General trust is not a validated measure of cynical distrust. Employment status does not supply retirement age or a measure of cognitive engagement.

A diagnosis recorded at any point does not establish current or untreated illness. Blood pressure checks, cholesterol panels, and eye exams do not establish that a condition is controlled. Body mass index cannot substitute for waist circumference. Diabetes and prediabetes are combined, as are cigarettes and other nicotine products. Weekly alcohol bands cannot reproduce the studies’ sex-specific daily doses. The food question is broader than a formal ultra-processed-food classification. The six-to-6.9-hour sleep category prevents an exact count of six hours or less, and the questionnaire contains no decades-long sleep history.

Age appears in bands, so no exact mean chronological age or felt-age gap is reported. The supplied body mass index values are retained rather than recalculated from rounded height and weight. Generated free-text answers are not member quotations. The model supplies no measured biological aging rate, mortality follow-up, dementia outcome, probability sample, response rate, population weighting, margin of error, or clinical prediction.

Research, Data, and Writing by Web Smith