Work Hard Now, Recover Later. Half of That Is True.

What you'll learn

What is already happening in the arteries of healthy people in their twenties, measured directly

Why long hours are treated as an occupational hazard rather than a lifestyle preference

Why two people working identical hours are not necessarily doing the same job

Why a thirty-second work message can outlast the interruption physiologically

What genuinely recovers when you start late, and what does not

How SmartBar applies the deficiency framework

SmartBar assesses products using three things we can verify: ingredient composition, level of processing, and nutritional data. We use two organizing terms throughout our education. Deficiency means something the body needs and isn't getting. Toxicity means a burden the body must process and clear, not a poison.

This page is about both at once, which is why it exists. Deferred sleep, movement, cooking and rest are deficiencies. Sustained hours, low control over demands, and unrelenting availability are burdens. What makes the combination unusual is the delay: the exposure and the consequence are separated by decades, and almost nothing about the first announces the second.

Start here

You already know the sentences. I'll sleep when I'm dead. Career first, health and family later. I'm young, my body can take it. Who has time to cook. I'll take a proper holiday once the promotion lands.

None of these are stupid. They are reasonable-sounding trades, and they share one assumption: that the cost is a debt, and a debt can be repaid later with interest.

Some of it genuinely is a debt. Fitness is. There is a study further down in which men in their fifties recovered, completely, thirty years of decline in six months. That is not a consolation prize. It is one of the more remarkable findings in exercise physiology.

But some of it is not a debt. It is an exposure with a clock, and the clock does not pause while you are busy. The reason nobody notices at twenty-eight is that a young body compensates extremely well, and compensation is not protection. It is what hides the accumulation.

This is not a scolding, and it is not about willpower. A large part of this exposure is not chosen: mandatory overtime, rotating shifts, a second job, caring for a parent, no paid leave. The biology is indifferent to whether the hours were chosen, which is precisely why it matters who gets to choose.

What is already happening at twenty-five

Start with the hardest evidence in the subject, because it does not depend on anyone remembering their own habits.

Between 1987 and 1994, pathologists across fifteen US centres examined the arteries of 2,876 people aged 15 to 34 who had died of accidents, homicide and suicide. Not patients. Young people who died of something unrelated to their hearts.[1]

Intimal lesions were present in every aorta and in more than half of the right coronary arteries of the youngest group, aged 15 to 19, and both prevalence and extent rose steadily with age.[1]

The ordinary risk factors were already leaving marks. Higher non-HDL and lower HDL cholesterol tracked with more extensive lesions. Smoking was associated with roughly a threefold increase in raised lesions of the abdominal aorta in the 25 to 34 group. Elevated glycohemoglobin tracked with raised lesions across the whole span. And hypertension worked in a specific way: it accelerated the conversion of fatty streaks into raised lesions beginning in the third decade of life, with its effect growing as people aged.[2]

Read that last mechanism slowly. The process is not waiting for midlife. It is being laid down through the twenties and thirties, silently, at a rate set by conditions a twenty-five-year-old experiences as an ordinary Tuesday.

Hours are an exposure, and they are measured like one

Long working hours are not treated by the relevant international bodies as a lifestyle preference. They are treated as an occupational hazard with a quantified burden.

In 2016, 488 million people, 8.9% of the global population, were working 55 hours a week or more. The World Health Organization and the International Labour Organization attribute 745,194 deaths and 23.3 million disability-adjusted life years from ischemic heart disease and stroke to that exposure, being 4.9% of all deaths from those causes.[3] Long working hours account for roughly a third of the total estimated work-related burden of disease, making them the occupational risk factor with the largest disease burden.[3]

The risk estimates come from two separate systematic reviews. Compared with a standard 35 to 40 hour week, working 55 hours or more was associated with a 35% higher risk of stroke and a 17% higher risk of dying from ischemic heart disease.[4]

And then the sentence that is this entire article in one line. Most of those deaths occurred in people aged 60 to 79, who had worked 55 hours or more per week between the ages of 45 and 74.[3]

Exposure and outcome, separated by decades. Not by a bad quarter. That is why the feedback never arrives while it would still change anyone's mind.

Three honest notes. These are modelled attributable-burden estimates, produced by applying population-attributable fractions to global health estimates, not counted deaths.[3] The underlying evidence is observational, so it establishes a consistent association rather than proving that any individual's hours caused any individual's stroke. And the estimates are contested: the ischaemic heart disease review drew published criticism from other researchers in the field, with a published reply from its authors defending the conclusions.

Hours are not the whole exposure

There is a second dimension a timesheet cannot capture, and leaving it out would misrepresent the field.

The occupational-health literature calls it job strain, and it is not simply having a demanding job. It is high demands combined with low decision latitude: how much say you have over what gets done, when, and how.

Individual records from 13 European cohort studies covering 197,473 workers, of whom 30,214 reported job strain, produced 2,358 incident coronary events over 1.49 million person-years. Job strain was associated with a 23% higher risk of incident coronary heart disease, hazard ratio 1.23, 95% CI 1.10 to 1.37.[5]

Two things must be said alongside that number. The authors were restrained about it: the attributable burden is much smaller than that of conventional risks such as smoking, and the evidence is observational.[5] And in published correspondence, other researchers argued the pooled estimate is likely biased toward the null, because the cohorts used arbitrary cut points and under-represented industrial workers, who have higher job-strain prevalence.[5] The argument in this field runs in both directions, which is itself worth knowing.

What it changes is what balance means. Ten hours you chose, can interrupt, and can recover from are not necessarily the same exposure as ten hours governed by demands you cannot refuse or predict.

Hours matter. So does whether any of them belong to you.

The message that outlasts itself

The long-hours research measures average or actual working time, mostly through large labour-force surveys of what people report, and it counts hours in second jobs.[3] So unpaid evening work can be captured when people report it as work.

What a weekly total cannot tell you is whether the remaining hours were genuinely yours: whether the phone stayed live, whether tomorrow's problem occupied tonight, or whether you could predict when work would next arrive.

That gap has a literature. Standard models of stress examined what happens during a stressful event and in doing so missed part of the exposure. Worry and rumination prolong stress-related physiological activation both before and after the event itself.[6] A meta-analysis found perseverative cognition associated with higher blood pressure, heart rate and cortisol and lower heart rate variability, with substantial heterogeneity between studies.[7]

Now the email that lands at 7pm on Friday, mentions restructuring, and attaches a Monday meeting invitation.

It takes thirty seconds to read. The meeting is sixty hours away. If the meeting keeps being rehearsed, the physiological exposure can outlast the message, because anticipation and rumination are part of the stress response rather than merely thoughts about it.

Experimental work reaches into sleep itself. A randomized study of healthy young adults found lower parasympathetic modulation during both NREM and REM sleep, and increased sympathovagal balance during NREM, under a stress condition.[8] Later controlled work found that anticipating a psychosocial stressor after sleep reduced slow-wave sleep and other measures of sleep depth, with the effects particularly apparent later in the sleep period, closer to the anticipated stressor.[9]

That does not mean a single email costs an entire weekend physiologically. It means the clock does not necessarily stop when the laptop closes, and that the recovery a weekend is supposed to provide is not guaranteed by the calendar.

Why the vacation you are saving up will not do the job

The most common deferral is the banked holiday. Two weeks somewhere warm, once the promotion lands, and everything resets.

It does work. It does not work the way the plan assumes.

The benefit is real and it fades fast. A meta-analysis of 13 studies, average vacation 11 days, found wellbeing improved overall, with gains in positive affect, negative affect, stress and exhaustion, but no significant improvement in life satisfaction. Those gains did not survive long after the return to work.[10] In one study following 54 employees through vacations averaging 23 days, health and wellbeing peaked around the eighth day and returned toward pre-vacation levels during the first week back.[11]

The reward being banked fades in weeks. The exposure it is meant to offset runs for years.

The active ingredient is not the location. Recovery research identifies psychological detachment, genuinely refraining from work-related thought during non-work time, as consistently associated with lower fatigue and exhaustion, better sleep and better wellbeing.[12] The same literature describes the interrupted evening directly: someone who cannot detach sleeps worse, shows lower nocturnal recovery of blood pressure, and carries fatigue into the next morning, compensating with additional effort until the cost surfaces anyway.[12][13]

Which is why a holiday with the laptop open is not a holiday, and why rest cannot be saved up and spent later like currency. Recovery is a process with requirements, not a quantity.

Much of this is self-report and diary-based occupational psychology, with physiological measures thinner than questionnaire measures, and the vacation meta-analysis rests on a small number of studies. It supports the claim that recovery has requirements and that interruption defeats them. It does not establish disease outcomes.

The chemistry the arrangement requires

Sustained demand creates fatigue, disrupted sleep, meals skipped or replaced, and a nervous system that will not settle. A market exists for each.

On alcohol, the association has been quantified. A meta-analysis combining 36 published studies with unpublished individual participant data from 27 more found long working hours associated with new-onset risky alcohol use, odds ratio 1.12. Against a 35 to 40 hour week, 49 to 54 hours gave 1.13 and 55 or more gave 1.12.[14]

Two things about that deserve stating plainly. The effect is small: adjusted absolute differences under one percentage point against a baseline incidence of 6.2%.[14] This is a population-level shift, not a prediction about any individual. But it held with no meaningful difference by sex, age, socioeconomic group, geographical region, or cohort type.[14] A signal that stable across that many cuts is telling you about the exposure rather than about who happens to be exposed.

Then the workplace itself. In many organisations, after-hours socialising is not optional in practice; attendance is career currency and the drinking is part of the arrangement. That is a description of a norm, not a research finding, and we are labelling it as one. What is worth noticing is who pays for it: the person in recovery, the parent who has to be home, the person who does not drink for religious reasons. The cost of that norm falls hardest on those least able to absorb it.

On medication we will say only this. Prescribed drugs have legitimate indications and very different dependence profiles, and there is no comparable body of evidence allowing anyone to quantify medication use as a consequence of long working hours. We are not going to turn a work-stress article into a list of substances people use to stay awake or switch off.

Nothing here is a reason to start, stop or change any medication. Our addiction article sets out what actually distinguishes addiction from heavy use, and it does not turn on quantity.

Different societies have reached different answers

There is no biological reason a working week should be any particular length. It is a settlement, and different places have settled differently. Three examples, chosen because they show three different things.

France has had a right to disconnect since 2017. Its labour code requires annual negotiation over the terms on which employees may disconnect and over the regulation of digital tools, to protect rest periods, leave and personal and family life.[15]

Australia goes further and gives employees something they can act on: a right to refuse to monitor, read or respond to unreasonable out-of-hours contact, in force for larger employers from August 2024 and for small-business employees from 26 August 2025, with the workplace tribunal able to resolve disputes.[15]

Ontario shows why the label matters. Employers with 25 or more employees must have a written policy on disconnecting from work, but the province's own guidance states that the requirement does not itself create a right to disconnect.[15] A policy about a thing is not the thing.

There is still no European Union-wide right to disconnect. The issue moved through a dedicated social-partner consultation, which concluded in 2025, and into the European Commission's broader Quality Jobs Act process, whose second-stage consultation runs through September 2026, with a legislative proposal planned for later in the year.[15] The United States has no federal equivalent.

A caution, because this is where articles like this usually go wrong. It is tempting to line countries up by working hours and life expectancy and draw a conclusion. Do not. Those comparisons are confounded by healthcare, wealth, diet, smoking and inequality at once, and retirement age is worse still, because healthier populations can afford to raise it, which runs the arrow backwards. Our longevity article covers what happened when the most famous version of that reasoning was audited.

The legitimate point is narrower and better. The Friday-night email is a legal question in one place and a career expectation in another. Neither arrangement is natural. Both are recent, both are contested, and both were chosen. Once you see the arrangement as a choice somebody made, "who has time to cook" stops sounding like a fact about life and starts sounding like a fact about a schedule.

The afternoon rest common in parts of southern Europe belongs in the same frame: a culture that builds recovery into the working day rather than treating rest as a failure of commitment. What it is not is a longevity intervention, and the evidence on daytime sleep is narrower than its reputation. Our sleep article sets out what naps actually do.

The ledger: what comes back, and what does not

Everything above could leave a reader in their forties concluding it is too late. The evidence says otherwise, and it also says the opposite of "no harm done."

In 1966, five twenty-year-old men underwent detailed physiological testing before and after three weeks of bed rest and a period of heavy endurance training. Thirty years later, at fifty, the same five were tested again and put through six months of endurance training.[16]

The good news is startling. One hundred percent of the age-related decline in aerobic power accumulated over thirty years was reversed by six months of training, with maximal oxygen uptake rising 14% to the level measured three decades earlier.[16]

And a companion finding demolishes "I'm young, my body can handle it" more effectively than any warning could. Three weeks of bed rest at age twenty reduced cardiovascular capacity more than thirty years of aging did: bed rest cost 27% of VO2max, three decades of living cost 12%.[17] Inactivity was the more powerful force by a wide margin.

Now the other column. No participant reached the maximum they had achieved after training thirty years earlier, despite a comparable relative training load, and the gain came from peripheral adaptation rather than any recovery of maximal oxygen delivery.[16] The ceiling did not come back. Ten years later, the rate of decline had accelerated sharply.[18]

So the ledger reads: losses attributable to deconditioning came back, in full, at fifty, after six months. The ceiling did not. Waiting made the slope steeper.

Bone is a third case, and it is neither. Most peak bone mass is accumulated early: the peak is reached by the end of the second decade, with total mass peaking in the late twenties, and lifestyle influences somewhere between 20 and 40% of the peak a person reaches.[19] That sets the reserve carried into later decades. But adult bone remains biologically responsive. Randomized exercise trials in premenopausal women show small statistically significant gains at the femoral neck and lumbar spine.[20] A more recent analysis found the clearest gains in women under 30, and its authors state that their overall analysis did not show a significant improvement in premenopausal women as a group, explicitly contrasting this with the earlier result.[21] The size of the effect is genuinely unsettled.

Which makes bone neither a slammed door nor a debt. Building the peak and preserving or rebuilding later are different biological tasks, and only one of them has a deadline.

Five men, all male, all healthy at baseline, followed longitudinally rather than randomized. The Dallas study illustrates a mechanism vividly; it does not carry epidemiological weight alone.

Key takeaways

Some of what you are deferring is a debt and some of it is a window. Men who resumed training at fifty recovered thirty years of aerobic decline in six months. They did not recover the ceiling.

The accumulation has already started. Arterial lesions were present in every aorta examined in 15 to 19 year olds, and hypertension accelerates their progression from the third decade onward.

Hours are an occupational exposure with a measured burden, and the attributed deaths mostly occur between 60 and 79 in people who worked those hours between 45 and 74.

Control matters as well as duration. Job strain, meaning high demands with low decision latitude, was associated with a 23% higher risk of incident coronary heart disease across nearly 200,000 workers.

A saved-up holiday is not a payment plan. Measured gains in wellbeing did not outlast the first stretch back at work, and detachment rather than location does the work.

Most peak bone mass is built early, giving later life a different starting reserve, though adult bone still responds to loading.

The arrangement is a choice somebody made. In some countries after-hours contact is a legal question rather than a career expectation.

How to think about this in thirty seconds

Ask which category a trade falls into: a debt that can be repaid, or a window that closes. They do not behave the same.
Ask not only how many hours, but how many of them you control.
Notice that the absence of symptoms at thirty is not evidence of absence of accumulation. Compensation hides it.
Do not assume time off is recovery. Recovery has requirements, and repeated work intrusion can undermine them.
Remember that starting late works. Not completely, but far more than most people believe.

The spillover is the point

Work does not need a unique biological pathway to affect health if it changes the conditions under which everything else happens.

A 2026 meta-analysis of 33 studies found work–family conflict associated most clearly with more sleep disturbance, r = 0.188 across twelve studies, and with a poorer diet, though that second finding rests on only three studies. Associations with smoking and alcohol use were smaller again.[22]

The studies are predominantly cross-sectional, so this does not establish that conflict caused those behaviours, and a 2026 review restricted to longitudinal designs has criticised earlier work in this area precisely for mixing the two.[22] What it does show is narrower and still useful: these domains move together. Work, sleep, food, recovery and coping compete for the same hours.

What SmartBar does with this

This is where the rest of our library connects. Sleep, movement, what you eat, mental health and what people use to cope all turn out to sit downstream of how time is allocated, and the allocation is usually made by someone in their twenties or thirties who has been told the cost is deferrable.

We are not going to tell you to work less. For many people that is not available, and pretending otherwise would be its own dishonesty. What we can do is show what has been measured, including the parts that do not fit a tidy story: that some damage reverses remarkably well, that some does not, and that the interval between exposure and consequence is long enough to remove almost all the useful feedback.

The framework holds cleanly. Deferred sleep, movement and food are deficiencies. Sustained hours, low control and unrelenting availability are burdens. What is unusual is only the delay, and the delay is precisely what makes the deferral feel free at the time.

This article is educational and is not medical advice. Nothing here is a reason to start, stop or change any medication. Persistent exhaustion, sleep that does not restore, or a low mood that will not lift are worth discussing with a qualified clinician rather than absorbing as a normal cost of employment. If you are struggling to cope, or having thoughts of harming yourself, please contact your local crisis line or emergency services — the resource for your country should be checked and named at publication.

Keep exploring: Sleep · Exercise · Nutrition · Mental health · Addiction · Longevity

Medical Disclaimer: Read our full Medical Disclaimer

References

Strong, J.P., Malcom, G.T., McMahan, C.A., et al. (1999). "Prevalence and extent of atherosclerosis in adolescents and young adults: implications for prevention from the Pathobiological Determinants of Atherosclerosis in Youth Study." JAMA 281(8):727–735. PMID: 10052443. (Multi-institutional autopsy study, 2,876 subjects aged 15–34 who died of external causes and were autopsied between 1 June 1987 and 31 August 1994. Intimal lesions appeared in all aortas and more than half of right coronary arteries in the 15–19 group, increasing in prevalence and extent with age. The collection window predates current risk-factor profiles; stated rather than hidden.)
McGill, H.C. Jr., McMahan, C.A.; PDAY Research Group (1998). "Determinants of atherosclerosis in the young." American Journal of Cardiology 82(10B):30T–36T. DOI: 10.1016/S0002-9149(98)00720-6. PMID: 9860371. (Non-HDL cholesterol positively and HDL negatively associated with lesion extent; elevated glycohemoglobin associated with raised lesions across the 15–34 span; smoking associated with an approximately threefold increase in raised lesions of the abdominal aorta in the 25–34 group; hypertension augments atherosclerosis primarily by accelerating conversion of fatty streaks to raised lesions from the third decade, with the effect increasing with age. Bibliographic details supplied by editorial review; substance verified against the PDAY series.)
Pega, F., Náfrádi, B., Momen, N.C., Ujita, Y., Streicher, K.N., Prüss-Üstün, A.M., et al. (2021). "Global, regional, and national burdens of ischemic heart disease and stroke attributable to exposure to long working hours for 194 countries, 2000–2016." Environment International 154:106595. DOI: 10.1016/j.envint.2021.106595. (In 2016, 488 million people, 8.9% of the global population, exposed to ≥55 hours/week, 95% uncertainty range 472–503 million; 745,194 attributable deaths and 23.3 million DALYs, 4.9% of all deaths and 6.9% of all DALYs from these causes. Most deaths occurred at ages 60–79 in people who had worked ≥55 hours/week between ages 45 and 74. Exposure modelled from 2,324 cross-sectional surveys and 1,742 quarterly survey datasets of reported working time, including hours in secondary jobs — that is, self-reported hours rather than payroll records. These are modelled attributable-burden estimates, not counted deaths. Note that WHO's public summary page gives 479 million; the peer-reviewed analysis gives 488 million and is followed here.)
WHO/ILO systematic reviews underlying the risk estimates. Descatha, A., Sembajwe, G., Pega, F., et al. (2020), "The effect of exposure to long working hours on stroke," Environment International 142:105746, DOI 10.1016/j.envint.2020.105746; and Li, J., Pega, F., Ujita, Y., Brisson, C., Clays, E., Descatha, A., et al. (2020), "The effect of exposure to long working hours on ischaemic heart disease," Environment International 142:105739, DOI 10.1016/j.envint.2020.105739. (Pooled relative risk 1.35 for stroke and 1.17 for ischaemic heart disease, ≥55 hours/week against 35–40. The ischaemic heart disease review attracted published critical commentary from Kivimäki and colleagues, with a published reply from the review authors; the estimates are contested within the field. Both articles are in volume 142, 2020; a widespread secondary-source error assigns the stroke review the same locator as reference 3, which is incorrect.)
Kivimäki, M., Nyberg, S.T., Batty, G.D., Fransson, E.I., Heikkilä, K., Alfredsson, L., et al., on behalf of the IPD-Work Consortium (2012). "Job strain as a risk factor for coronary heart disease: a collaborative meta-analysis of individual participant data." The Lancet 380(9852):1491–1497. DOI: 10.1016/S0140-6736(12)60994-5. PMID: 22981903. (13 European cohort studies, 1985–2006, employees free of CHD at baseline; job strain measured with validated demand-control instruments and harmonised before linkage to outcome records. 30,214 of 197,473 participants reported job strain; 2,358 incident events over 1.49 million person-years, mean follow-up 7.5 years. Hazard ratio 1.23, 95% CI 1.10–1.37 after adjustment for sex and age, robust to adjustment for lifestyle and socioeconomic status. Authors note the attributable burden is much smaller than for conventional lifestyle risks. Published correspondence in the same journal argues the pooled estimate is biased toward the null through arbitrary median cut points and under-representation of industrial workers. Observational.)
Brosschot, J.F., Gerin, W., Thayer, J.F. (2006). "The perseverative cognition hypothesis: a review of worry, prolonged stress-related physiological activation, and health." Journal of Psychosomatic Research. PMID: 16439263. (A hypothesis review presenting preliminary support, not an outcome study. Argues that perseverative cognition prolongs stress-related activation both in advance of and following stressors.)
Ottaviani, C., Thayer, J.F., Verkuil, B., et al. (2016). "Physiological concomitants of perseverative cognition: A systematic review and meta-analysis." Psychological Bulletin 142(3):231–259. DOI: 10.1037/bul0000036. (Associations with higher blood pressure, heart rate and cortisol and lower heart rate variability, with important heterogeneity and moderators. Bibliographic details supplied by editorial review.)
Hall, M., Vasko, R., Buysse, D., et al. (2004). "Acute stress affects heart rate variability during sleep." Psychosomatic Medicine 66(1):56–62. DOI: 10.1097/01.PSY.0000106884.58744.09. PMID: 14747638. (59 healthy young adults randomized; lower parasympathetic modulation during NREM and REM sleep and increased sympathovagal balance during NREM under the stress condition. Bibliographic details supplied by editorial review. Establishes altered autonomic activity during sleep, not disease outcomes.)
Beck, J., Loretz, E., Rasch, B. (2023). "Stress dynamically reduces sleep depth: temporal proximity to the stressor is crucial." Cerebral Cortex 33(1):96–113. DOI: 10.1093/cercor/bhac055. PMID: 35196708. PMCID: PMC9758584. Published online February 2022, issue dated January 2023. (67 healthy young participants across two counterbalanced 90-minute conditions involving a psychosocial stressor or relaxation; in the anticipation arm, 33 healthy women expected the stressor after sleeping. Anticipation reduced slow-wave sleep, slow-wave activity and spindle measures, with effects most apparent later in the sleep period. This is a daytime nap paradigm, not overnight sleep, which limits how far it generalises to a normal night. Bibliographic details supplied by editorial review and confirmed.)
Speth, F., Wendsche, J., Wegge, J. (2023). "We Continue to Recover Through Vacation! Meta-Analysis of Vacation Effects on Well-Being and Its Fade-Out." European Psychologist 28(4):274–287. DOI: 10.1027/1016-9040/a000518. (Reanalysis of an earlier meta-analysis plus a new search covering 2009–2020. 13 studies, N = 1,428, average vacation duration 11 days, range 4–23. Wellbeing improved, d = 0.25, across positive affect, negative affect, stress and exhaustion, ds = 0.18–0.38, but not life satisfaction, d = 0.10. Improvements faded after returning to work. Speth and Wendsche share first authorship; the authors declare no conflicts of interest. Publication-year note: the journal issue is dated 2023 and most indexes give 2023, while the article carries a 2024 Hogrefe copyright and Germany's Federal Institute for Occupational Safety and Health catalogues it as 2024. 2023 is used here as the journal year; the discrepancy is recorded rather than resolved. Rests on a small number of primary studies.)
de Bloom, J., Geurts, S.A.E., Kompier, M.A.J. (2013). "Vacation (after-) effects on employee health and well-being, and the role of vacation activities, experiences and sleep." Journal of Happiness Studies 14(2):613–633. DOI: 10.1007/s10902-012-9345-3. (54 employees, vacations averaging 23 days; health and wellbeing peaked around day 8 and returned toward baseline during the first week back. Bibliographic details supplied by editorial review.)
Wendsche, J., Lohmann-Haislah, A. (2016). "A Meta-Analysis on Antecedents and Outcomes of Detachment from Work." Frontiers in Psychology 7:2072. DOI: 10.3389/fpsyg.2016.02072. (Detachment consistently associated with fatigue, exhaustion, short- and long-term wellbeing and sleep. Predominantly self-report and diary measures.)
Sonnentag, S., Fritz, C. (2015). "Recovery from job stress: The stressor-detachment model as an integrative framework." Journal of Organizational Behavior 36(S1):S72–S103. DOI: 10.1002/job.1924. (First published online April 2014; belongs to the February 2015 issue.)
Virtanen, M., Jokela, M., Nyberg, S.T., Madsen, I.E.H., Lallukka, T., Ahola, K., et al. (2015). "Long working hours and alcohol use: systematic review and meta-analysis of published studies and unpublished individual participant data." BMJ 350:g7772. DOI: 10.1136/bmj.g7772. PMID: 25587065. Open access. (36 published studies plus unpublished individual participant data from 27 more. Prospective new-onset risky alcohol use OR 1.12, 95% CI 1.04–1.20; 49–54 hours 1.13, 1.02–1.26; ≥55 hours 1.12, 1.01–1.25, on a baseline incidence of 6.2%. Adjusted absolute differences 0.8 and 0.7 percentage points. No difference by sex, age, socioeconomic group, region, sample type or attrition.)
Right-to-disconnect frameworks. France: Code du travail, obligation to negotiate annually on the modalities of the right to disconnect and the regulation of digital tools to protect rest periods, leave and personal and family life, in force since 2017. Australia: Fair Work Act right to refuse to monitor, read or respond to unreasonable out-of-hours contact, in force for larger employers from 26 August 2024 and for small-business employees from 26 August 2025, with Fair Work Commission dispute resolution. Ontario: Employment Standards Act requirement for employers with 25 or more employees to have a written policy on disconnecting from work, with the province's own guidance stating the requirement does not create a right to disconnect. European Union: no EU-wide right to disconnect; a dedicated social-partner consultation concluded in 2025 and the issue now sits within the Commission's Quality Jobs Act process, second-stage consultation running in 2026 with a legislative proposal planned for the end of 2026. (Primary instruments: France, Code du travail Article L2242-17, which also provides that absent agreement the employer establishes a charter after consultation; Australia, Fair Work Act 2009 section 333M; Ontario, Employment Standards Act written-policy requirement together with the province's own guidance that it does not create a right to disconnect. Reduced from a longer jurisdiction list in v1 on editorial review.)
McGuire, D.K., Levine, B.D., Williamson, J.W., Snell, P.G., Blomqvist, C.G., Saltin, B., Mitchell, J.H. (2001). "A 30-year follow-up of the Dallas Bedrest and Training Study: II. Effect of age on cardiovascular adaptation to exercise training." Circulation 104(12):1358–1366. PMID: 11560850. (Five middle-aged men originally trained in 1966, retrained at approximately 50 over six months. VO2max rose 14%, reaching the level observed thirty years earlier; 100% of the age-related decline was reversed. However no subject reached the maximal VO2 attained after training thirty years earlier, and the gain was primarily peripheral adaptation with no effective improvement in maximal oxygen delivery. Five participants, all male, healthy at baseline; illustrative rather than epidemiological.)
McGuire, D.K., Levine, B.D., Williamson, J.W., et al. (2001). "A 30-year follow-up of the Dallas Bedrest and Training Study: I. Effect of age on the cardiovascular response to exercise." Circulation 104(12):1350–1357. PMID: 11560849. (Three weeks of bed rest at age 20 had a more profound impact on physical work capacity than three decades of aging: VO2max fell 27% with bed rest against a 12% decline over the following thirty years.)
McGavock, J.M., Hastings, J.L., Snell, P.G., McGuire, D.K., Pacini, E.L., Levine, B.D., Mitchell, J.H. (2009). "A forty-year follow-up of the Dallas Bed Rest and Training study: the effect of age on the cardiovascular response to exercise in men." Journals of Gerontology Series A 64(2):293–299. DOI: 10.1093/gerona/gln025. PMID: 19196908. (Annualised VO2max decline from 50 to 60 approximately four times that from 20 to 50 in these five men. Bibliographic details supplied by editorial review.)
Weaver, C.M., Gordon, C.M., Janz, K.F., Kalkwarf, H.J., Lappe, J.M., Lewis, R., O'Karma, M., Wallace, T.C., Zemel, B.S. (2016). "The National Osteoporosis Foundation's position statement on peak bone mass development and lifestyle factors." Osteoporosis International 27(4):1281–1386. DOI: 10.1007/s00198-015-3440-3. PMID: 26856587. PMCID: PMC4791473. (Lifestyle choices influence 20–40% of adult peak bone mass. Peak reached by the end of the second decade with total mass peaking in the late twenties, then plateau and decline, though some evidence suggests earlier loss at specific sites such as the lumbar spine.)
Kelley, G.A., Kelley, K.S., Kohrt, W.M. (2013). "Exercise and Bone Mineral Density in Premenopausal Women: A Meta-Analysis of Randomized Controlled Trials." PMID: 23401684. PMCID: PMC3563173. (Randomized exercise trials of ≥24 weeks. Statistically significant improvements at femoral neck, g = 0.342, 95% CI 0.132–0.553, 466 participants, and lumbar spine, g = 0.201, 95% CI 0.009–0.394, 402 participants. Small effects. See also the same group's postmenopausal analysis, BMC Musculoskeletal Disorders 2012;13:177, PMCID PMC3489866.)
Miao, T., Li, X., Zhang, W., Liu, J., Xiao, Y., Wang, X. (2026). "Effects of exercise on bone mineral density and bone turnover markers in adults: a systematic review and meta-analysis." Frontiers in Physiology 16:1672997. DOI: 10.3389/fphys.2025.1672997. PMID: 41584801. PMCID: PMC12823962. Published 8 January 2026; the 2025 element of the DOI is an article identifier, not the publication year. (22 RCTs, 1,051 participants, 547 in exercise groups and 504 controls, participants aged approximately 19 to 43, with 15 of the 22 studies in premenopausal women. Exercise significantly increased lumbar spine BMD, SMD 0.46, 95% CI 0.16–0.76, and femoral neck BMD, SMD 0.42, 95% CI 0.13–0.71, in women under 30. The statement that the overall analysis showed no significant improvement in premenopausal women as a group is taken from the authors' own discussion in section 4.1, where they state that their overall results did not show significant BMD improvement in premenopausal women and contrast this explicitly with reference 20. Cited to show the effect size is unsettled, not to overturn reference 20.)
"The Relationship Between Work–Family Conflict and Health Behaviors: A Systematic Review and Meta-Analysis." Behavioral Sciences (2026) 16(3):386. DOI: 10.3390/bs16030386. PMCID: PMC13024141. Open access. (33 studies, all rated low risk of bias, examining five health behaviours. Sleep disturbance r = 0.188, 95% CI 0.128–0.247, k = 12. Healthy diet negatively correlated, but on k = 3 studies only; smoking k = 5; alcohol k = 11; physical activity k = 8. The diet finding therefore rests on a very small evidence base and is described as such in the body. Predominantly cross-sectional. Note also a 2026 longitudinal-only meta-analysis of work–family conflict and mental health in PLOS Medicine, DOI 10.1371/journal.pmed.1005162, which criticises earlier meta-analyses in this area for pooling cross-sectional and longitudinal designs and for heterogeneous confounding adjustment — a caution that applies to this reference.)

Sourcing note: this article cites occupational-health agency estimates, systematic reviews and meta-analyses of individual participant data, autopsy-based epidemiology, controlled experimental sleep research and longitudinal physiological studies, with DOI/PMID where available. Study designs, sample sizes, funding where relevant and limitations are stated in the reference notes rather than left implicit, and where a body of evidence is self-report based, small, modelled rather than counted, or contested within its own field, the body text says so. This article makes no recommendation and provides no protocol, by editorial policy. It attributes no individual's illness to their working hours, treats no country comparison as causal, and discusses medications at class level only, naming none. Cross-link targets (#) resolve once the site structure is live. Reading time is measured, not asserted: body words excluding the reference list and the metadata header, counted at 225 words per minute and rounded to the nearest minute.

PRODUCTION NOTE — DELETE BEFORE PUBLICATION

All 22 references independently verified against openly accessible sources. Zero bibliographic red flags. Citation integrity audited: every marker in the body resolves to a reference, every reference is cited, no duplicates.

Two publication/volume-year discrepancies are recorded rather than silently resolved, in references 10 and 21. Reference 10's journal issue is dated 2023 while the article carries a 2024 copyright and is catalogued as 2024 elsewhere. Reference 21 was published in January 2026 but is assigned by its publisher to a volume labelled 2025. In each case the reference states which year is used and why the other exists.

One freshness-sensitive item. Reference 15's European Union status is current as of drafting: second-stage consultation running through September 2026 with a proposal expected later in the year. This is the single element of this article most likely to be out of date by publication and it must be rechecked immediately before it goes live. Unlike the rest of the library, this page will need a recurring freshness review after publication, because legal status changes on its own schedule.

Open items, neither bibliographic:

Crisis resource not named, per the standard: generic signposting in the body, named resource inserted at publication after confirming it is currently operating.
Clinical review not performed.