Opens the fitness cluster. Cardiorespiratory fitness (CRF, peak VO2, measured in METs — 1 MET =
3.5 mL/kg/min) is one of the strongest mortality predictors in medicine — but the wiki’s other
lever, physical activity, is measured as dose (minutes); this is measured as the outcome (capacity).
The distinction is load-bearing: both sources here are observational, so CRF is a predictor, not a
proven causal lever — the general prognostic-marker-vs-modifiable-lever distinction
(Surrogate Outcomes -> Prognostic marker vs modifiable lever). The modifiable lever that acts on it
is exercise (Measuring and Raising Cardiorespiratory Fitness), which carries its own intervention
evidence; a high VO2max is otherwise a stratification metric, not a target in itself.
The dose-response — each MET matters, and no upper limit was observed over the studied range
- Per 1-MET higher CRF (Kodama meta-analysis, dose-response): all-cause mortality «RR 0.87 (95% CI, 0.84-0.90)» and CHD/CVD «0.85 (95% CI, 0.82-0.88)» — i.e. «a 1-MET higher level of MAC was associated with 13% and 15% decrements in risk of all-cause mortality and CHD/CVD». (Kodama, 2009)
- No plateau observed (Mandsager, 122,000-patient cohort): «Cardiorespiratory fitness is inversely associated with long-term mortality with no observed upper limit of benefit» — mortality keeps falling into the elite band: «elite vs high: adjusted HR, 0.77 (95% CI, 0.63-0.95)». (Mandsager et al., 2018) The elite-vs-high step is carried by subgroups: significant overall, in patients aged 70 or older (HR 0.71, 0.52-0.98) and in hypertensives (HR 0.70, 0.50-0.99), but non-significant by sex («By sex, a nonstatistically significant improved survival in elite vs high performers was present in both men and women»), and «In younger age groups, there was no difference in survival between elite and high performers.» No ceiling is a finding in a treadmill-referral cohort, not a monotonicity law. (Mandsager et al., 2018) (corrected 2026-10-06: bare no ceiling -> no upper limit observed, with the subgroup limits; self-critique) (corrected 2026-10-22: the 10-12 MET passage’s only citation identified as an activity study, per its title, below)
- A stated counterpoint, not reconciled here (Ross 2016, AHA scientific statement): «Most of the lower mortality risk associated with a higher CRF occurs by the time a CRF of 10 to 12 METs is achieved. CRF values >12 METs are associated with a relatively lower impact on risk of all-cause and CVD mortality.» (Ross et al., 2016) Ross’s display is a statement-level summary sentence (one cited reference, no curve or CI printed with it), and the only citation in the three sentences ending at ref 290 (on the closing sentence) is Arem 2015, titled as a pooled dose-response analysis of leisure time physical activity and mortality (Ross et al., 2016); by its title an activity study rather than a fitness study. Mandsager’s is a categorical HR from one treadmill-referral cohort (elite >=97.7th percentile vs high 75-97.6th). Detail and the training-dose reading: Measuring and Raising Cardiorespiratory Fitness.
- Why the pair is not a contradiction as stated, and not a resolution either. The two answer different contrasts: Mandsager asks whether the top band still differs from the band below it (yes, HR 0.77, carried by subgroups); Ross asks where most of the gradient has already been realized (by ~10-12 METs). A further step beyond 12 METs can be real and still small relative to the low-end steps, so both can hold. Ross’s 10-12 MET sentence carries no citation of its own and the nearest one (ref 290) is by its title an activity study, so that sentence gives no measured-fitness location for the flattening. In chunk 01 Ross writes that many studies find «CRF levels >8 to 10 METs are associated with relative protection», and that «Relatively less is gained by increasing CRF between moderately fit and highly fit individuals.», stated alongside its Figure, modified from Blair 1989, a study in which fitness was estimated by maximal treadmill testing (Ross et al., 2016); the references cited for the >8-10 MET sentence include Kodama 2009 and Myers 2002 (Ross et al., 2016), by title mostly fitness/exercise-capacity studies (inferred from Ross et al., 2016). Neither source tests the other’s contrast, so the wiki asserts neither a plateau nor its absence: the held statement is gradient steepest at the low end, a smaller residual step at the top observed in one referral cohort.
A convergence worth naming, and it bears on the U/J-artifact prior. Self-reported activity studies show a plateau (and sometimes a U-shape) at high volumes; Mandsager, measuring fitness objectively, finds monotone benefit with no plateau over its studied range, and offers three hedged candidate explanations: «This difference may reflect the objective measurement of physical fitness in the present study, as opposed to self-reported activity levels»; «It may also reflect non–activity-related contributors to aerobic fitness, including genetic factors and unmeasured health habits, which may contribute to improved survival»; and «Lastly, it may, in part, be attributable to differences in the observed populations» (Mandsager et al., 2018). Only the first supports reading the activity-plateau as partly a self-report measurement artifact -> The U-Shaped Association Artifact; the second cuts against a causal reading of CRF’s missing plateau. (corrected 2026-10-06: attributes the discrepancy to one explanation -> one of three; self-critique) Kodama’s own categorical data agree on the shape: the steepest benefit is at the low end (low-vs-high RR 1.70 >> intermediate-vs-high 1.13). (inferred from Kodama, 2009; Mandsager et al., 2018)
A candidate role for CRF in the occupational-PA paradox (route-b, UNTESTED) [2026-08-14, Coenen].
High occupational physical activity is associated with higher mortality in men -> The Physical Activity Paradox, and Coenen reports the harm «appears to be stronger in workers with low compared
with high cardiorespiratory fitness» but «could not statistically test this in a sensitivity analysis»
for lack of data (Coenen et al., 2018). So CRF is a candidate buffer of the strenuous-work harm — mechanism-consistent (a fitter worker
operates at a lower relative workload), but it clears only the plausibility bar, not positive
effect-modification, so it is held as an open interaction, not a finding.
How big — low fitness carries hazards comparable to or greater than smoking and diabetes
Mandsager’s adjusted all-cause-mortality gradient (reference = low performers): «Low vs Elite … 5.04 (4.10-6.20)», «Low vs High … 3.90 (3.67-4.14)», «Low vs Above Average … 2.75», «Low vs Below Average … 1.95». Mandsager judges the reduced-CRF hazard «comparable to or greater than traditional clinical risk factors» in the same model: «smoking … 1.41», «diabetes … 1.40», «coronary artery disease … 1.29». The size depends on the contrast chosen: lowest quartile vs the top 2.3% gives 5.04, but the middle step (below-average vs above-average) gives 1.41, equal to smoking — and a quartile-vs-top-band fitness contrast and a binary smoking contrast are not the same quantity. (corrected 2026-10-06: larger … than / outranks -> the source’s «comparable to or greater than»; self-critique) Kodama concurs categorically: low-vs-high CRF «RR for all-cause mortality of 1.70 (1.51-1.92)», attenuated to 1.48 (1.31-1.68) after trim-and-fill (see Limits). (Mandsager et al., 2018)
The target — where a given VO2max sits, by age and sex
There are two anchors the wiki holds:
- A minimal floor (Kodama): «a minimal CRF of 7.9 METs may be important» (50-year-old man reference); age/sex-specific, approximately 9 and 7 METs (at 40), 8 and 6 (at 50), 7 and 5 (at 60) for men and women — derived by applying Kodama’s assumed inputs («we assumed that the MAC is 2 METs lower in women than in men and that for each year of aging, it decreased by 0.1 MET based on a prior study») to the single 7.9-MET threshold; these are not separate age/sex estimates. (corrected 2026-10-06: offsets stated as findings -> assumed inputs; self-critique) (Kodama, 2009)
- Percentile bands (Mandsager, age x sex MET grid): low (<25th), below-average (25-49th), above-average (50-74th), high (75-97.6th), elite (>=97.7th). E.g. men 50-59: «<8.2 | 8.2-9.9 | 10.0-11.3 | 11.4-13.9 | >=14.0» METs; women 50-59: «<7.0 | 7.0-8.0 | 8.1-9.9 | 10.0-12.9 | >=13.0». (Mandsager et al., 2018)
So a VO2max reading converts to METs (/3.5) and drops into a band — the practical is my fitness good? answer the activity-dose evidence cannot give. Caveat: Mandsager’s bands are from a referral population using estimated (treadmill) METs, so a directly-measured VO2max placed against them is approximate. The estimate is a work-rate conversion: peak METs were «determined based on treadmill grade and speed at peak exercise», with no conversion equation named (Mandsager et al., 2018). Why a cycle-test number may land low on this grid (the cycle-vs-treadmill modality gap, plus a treadmill conversion that would over-read if it resembles the ACSM steady-state equations; Mandsager names none) is worked through on Measuring and Raising Cardiorespiratory Fitness.
The load-bearing honesty — predictor, not proven cause
Both sources are observational, and both say so. Kodama: CRF is «associated with» / a «predictor», never «causes», and it «suggest[s]… a clinical trial to determine whether an intervention that improves CRF by exercise reduces the risk». Mandsager: «the association between CRF and mortality does not prove causation… The degree to which high CRF preselects patients with lower mortality vs causes a reduction in mortality is not discernible». (Kodama, 2009)
So the operative reading: CRF is a superb risk marker and the trackable outcome of the physical- activity lever -> Physical Activity Dose and Mortality (which IS an evidenced mortality lever). But raise your VO2max to live longer is an association, not a demonstrated intervention effect — the proven lever is the activity that raises fitness, and CRF is how you measure whether it worked.
The caveat is partially — not fully — lifted by within-person change (Ross et al., 2016) (Ross 2016, Measuring and Raising Cardiorespiratory Fitness): people who go from unfit to fit between exams have lower subsequent mortality («44%» in Blair’s cohort), and in «The largest randomized trial of exercise training in HF patients» (HF-ACTION) a larger CRF rise was associated with fewer CV events «after adjustment for potential confounding variables». Within-person change is a stronger design than the cross-sectional comparison here, so it narrows the reverse-causation worry — but it does not close it. No leg is a randomized CRF contrast: HF-ACTION randomized patients to exercise training, not to a CRF change, so its CRF-change result is an adjusted within-trial association carrying the same confounding risk as the cohorts, in heart-failure patients, on a hospitalization-inclusive composite. (corrected 2026-10-06: the one exercise-training RCT / only HF-ACTION is randomized -> adjusted within-trial association in HF patients; self-critique) The upgrade is real and bounded: CRF is a modifiable target whose improvement tracks benefit, not yet a proven cause of longer life.
Limits
- Not two independent methods. A meta-analysis of cohorts (Kodama) and one mega-cohort (Mandsager) share the observational, confounded, reverse-causation-prone design — this is corroboration and refinement, not an [E-independent] convergence. Kodama’s sensitivity analysis: adjusting for smoking weakened the CHD/CVD link but not the all-cause link (CRF «independently associated with longevity»). (Kodama, 2009)
- Reverse causation is the central threat — subclinical illness lowers fitness; Mandsager’s single time-point measurement cannot separate unfit because sick from sick because unfit.
- Publication bias in Kodama: Egger p=.002 (all-cause) for the per-MET estimate, where trim-and-fill «could not detect hypothetical negative unpublished studies»; the categorical low-vs-high all-cause RR attenuates from 1.70 to 1.48 (1.31-1.68) after trim-and-fill. (Kodama, 2009) (corrected 2026-10-06: attenuation was attached to the per-MET estimate -> it applies to the categorical RR; self-critique)
- Transportability of the bands — Mandsager’s referral population is not the general population, and its METs are estimated, not measured.
- Coherence, not validity (R1): a strong, graded, mechanism-plausible association — but not proof that acting on it changes a given person’s life.
CRF is a measured CAPACITY, not a behaviour - and the predictor claim is cited (deliverable-critique, 2026-08-01)
Two clarifications the critique asked for. First, “CRF is one of the best-evidenced mortality predictors” is cited, not asserted: Kodama’s meta-analysis, Mandsager’s large cohort, and the Ross AHA statement are the backing (this page + Measuring and Raising Cardiorespiratory Fitness). Second, CRF is not the same object as the total-activity mortality finding, and both are real:
- CRF / VO2max = a measured fitness CAPACITY (partly trainable, partly genetic) - its per-1-MET mortality gradient is objectively measured, part of why it predicts so strongly.
- Total physical activity = a BEHAVIOUR - the HR ~0.34 lever -> Physical Activity Dose and Mortality.
- Resistance training = a behaviour whose channel is muscle/strength (Momma RR 0.85, largely independent of aerobic) -> Muscle-Strengthening Activity and Mortality; RT counts as activity but raises CRF only modestly - aerobic work is what raises CRF.
So the “two dials” are real distinct channels: aerobic -> CRF, resistance -> muscle. The low-HR total-activity number and the CRF-predictor claim are different (both valid) findings, not one restated.
”Per 1-MET” - what it means, and why you cannot compound it to 0.87^5 (deliverable-critique, 2026-08-01)
“Per 1-MET” is per 1 metabolic-equivalent of CRF capacity (VO2max; 1 MET = 3.5 ml/kg/min), NOT per MET-hour of activity - a capacity, not a dose (above). So RR 0.87 per 1-MET is a between-person association: each 1-MET-fitter stratum has ~13% lower mortality across the studied range. It does not compound to a personal promise (5 METs -> 0.87^5 ~ 0.50) for two reasons:
- The per-MET effect is not constant - it is front-loaded. The gradient is steepest at the low end (low-vs-elite HR ~5; the first step, low vs below-average, HR 1.95, is the largest single adjacent step, about 40% of the log-gradient to elite; later steps are smaller but still >1), so a MET gained from a sedentary base buys more than a MET added near elite; a single averaged coefficient hides this. (corrected 2026-10-06: low-vs-below-average is most of it -> largest single step, ~40% of the log-gradient; self-critique)
- It is observational, not a causal individual effect. Between-person fitness tracks many things besides training (baseline health, genetics; reverse causation), so causally raising your own VO2max by 5 METs does not deliver the between-person gradient -> The U-Shaped Association Artifact and the causal discount in the next section. Read it as: escaping the low-fitness bottom is where the large, decision-relevant benefit sits - not as a compoundable multiplier.
The causal discount - VO2max is part lever, part marker (deliverable-critique, 2026-08-01)
The observational gradient is real and large, but a “healthier people are fitter” tautology inflates it as a personal target - the critique is right, and it is the artifact lens applied to a protective association -> The U-Shaped Association Artifact. Three discounts sit between the between-person association and the effect of raising your own fitness:
- Reverse causation / confounding by health. Subclinical disease lowers CRF, so low CRF is partly a MARKER of ill health rather than its cause -> Measuring and Raising Cardiorespiratory Fitness. The reverse-causation / marker-vs-lever reading is the wiki’s; Ross 2016 does not make it [searched: reverse caus / subclinical / confound / causal across Ross chunks 01-03]. Ross supplies only the premise that disease lowers measured CRF: «Peak/maximal V⋅ o2 values vary widely and are influenced by age, sex, genetics, lifestyle/ exercise training habits, and varied disease states.» (Ross et al., 2016) (corrected 2026-10-06: Ross 2016 flags exactly this marker-vs-lever gap -> wiki inference; Ross gives only the disease-states premise, chunk 01)
- Genetics. A large share of VO2max is heritable and non-modifiable, so the gradient (which includes genetic high-responders) overstates trainable upside.
- Intervention < observational (expected, not measured). Given the two discounts above, causally raising CRF by training should lower mortality by less than the observational gradient implies; no held trial measures the size of that gap.
Net: VO2max is part lever, part marker. The decision-relevant claim survives but shrinks - escaping the low-fitness bottom by training plausibly helps (a likely smaller causal benefit, concentrated at the low end, not yet trial-proven on mortality) - it is NOT the between-person ~5x read as a personal promise.