Is very high volume harmful for longevity?
Against no activity, death HR 0.69 at 1-2 times the guideline minimum, lowest (0.61) at 3-10 times, 0.68 at 10+: no excess. Self-report once, 4,077 at the top. Cohort, not proof.
Main caveat: Baseline-only self-report; top-category values may be inflated, as the authors admit.
The question was the upper limit of benefit, and whether very high volume brings harm. Median follow-up was 14.2 years. In the authors' words, meeting the guideline went with nearly the maximum longevity benefit, and clinicians need not discourage very active adults. Their claim that marathon and triathlon growth should not cause alarm is theirs, and they concede the study is not positioned to examine risk around exercise or conditions such as atrial fibrillation.
Protocol
Cohorts: NIH-AARP, CPS II, CLUE II, US Radiologic Technologists, Women's Health Study, Women's Lifestyle and Health Study; baseline 1992-2003.
Exposure: self-reported leisure activity converted to MET-h/week; multiples of the 7.5 MET-h/week minimum. This is gross METs, not the marginal units of REF-1567. NIH-AARP and some others had unvalidated questionnaires.
Adjustment: age, sex, race, education, smoking, cancer and heart disease history, alcohol, marital status, BMI.
Funding: NCI, other public funders and the American Cancer Society, with no role in the study.
Key Results
| MET-h/week | Times minimum | Participants / deaths | All-cause HR (95% CI) | Cardiovascular HR | Cancer HR |
|---|---|---|---|---|---|
| 0 | 0 | 52,848 / 11,523 | 1.00 | 1.00 | 1.00 |
| 0.1-<7.5 | <1 | 172,203 / 33,511 | 0.80 (0.78-0.82) | 0.80 (0.77-0.84) | 0.87 (0.83-0.90) |
| 7.5-<15 | 1-2 | 170,563 / 28,957 | 0.69 (0.67-0.70) | 0.67 (0.65-0.70) | 0.79 (0.75-0.82) |
| 15-<22.5 | 2-3 | 118,169 / 19,979 | 0.63 (0.62-0.65) | 0.59 (0.57-0.63) | 0.75 (0.72-0.79) |
| 22.5-<40 | 3-5 | 124,446 / 21,114 | 0.61 (0.59-0.62) | 0.58 (0.56-0.61) | 0.74 (0.71-0.77) |
| 40-<75 | 5-10 | 18,831 / 1,390 | 0.61 (0.58-0.64) | 0.61 (0.55-0.67) | 0.72 (0.66-0.79) |
| 75+ | 10+ | 4,077 / 212 | 0.68 (0.59-0.78) | 0.71 (0.56-0.91) | 0.69 (0.55-0.87) |
Top two categories: not significantly different (all-cause p=0.127; cardiovascular p=0.226; men p=0.07). So a rise at the top is not shown, though neither is a true plateau; no equivalence test was done.
Intensity (5 cohorts, 348,725, mutually adjusted): moderate 0.80 below the minimum, 0.73 at 1-2 times, 0.72 at 30+; vigorous 0.80, 0.77 and 0.79.
Heterogeneity between cohorts was significant in every category.
Abstract against full text: the abstract prints the 10+ HR as 0.69 (0.59-0.78); text and analysis give 0.68. The abstract places the threshold at 3-5 times; the results describe 3-10 times. The abstract calls cause-specific relationships similar, which hides a difference: cancer keeps falling while cardiovascular turns up numerically.
Mechanisms
Measured: none.
Authors' interpretation: earlier reports of harm came from younger athletic samples, and transient cardiac risk during vigorous exercise does not contradict lower overall risk.
Inference: a person's one baseline answer may not represent the years that follow, which blurs both ends of the curve; REF-1570 finds a single measure tended toward a U-shape, and repeated measures did not.
Practical Application
Is very high volume harmful for longevity? Against no activity, death HR 0.69 at 1-2 times the guideline minimum, lowest (0.61) at 3-10 times, 0.68 at 10+: no excess. Self-report once, 4,077 at the top. Cohort, not proof.
Good for: reassurance on all-cause death at high leisure volume; for an inactive person, the largest step is from none to some.
Not good for: intensity prescriptions; the moderate and vigorous contrasts are coarse and mutually adjusted.
Limitations
Baseline-only self-report; top-category values may be inflated, as the authors admit.
Healthy-user bias: the most active were younger, leaner, more often never-smokers and with fewer comorbidities; adjustment may not remove this.
Reverse causation: illness at baseline lowers reported activity, and no exclusion of early deaths is reported in the main analysis.
Thin top: few people and deaths above 40 MET-h/week. Leisure only; 93-97% white by category.
How this entry was checked and changed
Why this level
large observational cohort; pooling adds precision, not causal strength