REF-1138 · Westerståhl et al., 2025 · Journal of Cachexia, Sarcopenia and Muscle · Paper · PubMed · Free full text
Rise and fall of physical capacity in a general population followed from 16 to 63
Every capacity peaked by 36 and fell 30-48% by 63, the yearly loss rising from under 1% to about 2%, while the spread between people grew 3- to 25-fold and being active at 16, or becoming active later, went with higher capacity
Westerståhl et al. followed a random sample of 427 Swedish adolescents (222 boys, 205 girls) born in 1958, testing them at 16, 27, 34, 52 and 63.
Estimated aerobic capacity (absolute and per kg), bench-press repetitions (muscular endurance) and Sargent jump height (leg power) all rose to a peak between 19 and 36 and then fell, slowly at first and faster with age: the rate of decline grew from under 1% a year in the first decade after the peak to about 2-2.5% a year by 63, with no sex difference in the rate of decline after the peak.
From peak to 63 the loss was 30-48% depending on the outcome and sex (mean 37%). The spread between people widened markedly: the variance in relative aerobic capacity was 25 times larger at 63 than at 16, jump height about 5 times and bench press about 3 times.
Being active in leisure time at 16 was associated with higher capacity on every outcome, with no interaction with age; switching from inactive to active within a person was associated with 4-11% higher capacity on every outcome (no age interaction was reported for switching).
The authors conclude that nobody escapes the decline, that the age of the peak looks fixed, and that the slope before the peak, the level reached and the rate of decline after it are modifiable by physical activity.
The general population lost more than master athletes do: at 63 they kept ~65% of peak aerobic capacity and muscular endurance, against >80% reported for master athletes.
Protocol
Cohort: SPAF, a random, representative sample of Swedish upper-secondary pupils drawn in 1974 (222 male, 205 female, aged 16). All 427 enter the mixed models, so drop-outs do not remove people from the estimates; at 63, 33 participants (8%) had died and 3% had no registered address. Participants with a university degree or who were active at 16 were more likely to attend later rounds.
Aerobic capacity: at 27, 34, 52 and 63 the Åstrand submaximal cycle test (steady-state heart rate after 5-6 min, nomogram, age-corrected); at 16 a 9-min run converted to aerobic capacity with a regression built in a subgroup that also did the Åstrand test. No maximal test at any age.
Muscular endurance: bench-press repetitions at 25 per minute with 20 kg (men) or 12 kg (women), at 16, 34, 52 and 63.
Leg power: Sargent countermovement jump with arm swing (cm), at 16, 27, 34 and 63.
Lifestyle: leisure-time physical activity as yes/no (at 16: "do you take part in leisure-time sports?"; from 34: any free-time activity "including e.g. walking or gardening"); university degree yes/no.
Analysis: log-transformed outcomes, linear mixed-effects models with splines (knots at 21-31 y), random intercepts; percentage change per year computed from the model against the year before; variance at 16 vs 63 from the predicted random effects.
Key Results
| Outcome | Peak (age) | Value at 63 | Yearly change at 40 / at 63 | Peak to 63 | Variance 63 vs 16 |
|---|---|---|---|---|---|
| Aerobic capacity, L/min — men | 3.26 (35) | 2.18 | −0.6% / −2.2% | −33% | ×5.242 |
| Aerobic capacity, L/min — women | 2.61 (36) | 1.83 | −0.5% / −2.0% | −30% | (pooled) |
| Aerobic capacity, mL/kg/min — men | 42.2 (26) | 25.2 | −1.2% / −2.2% | −40% | ×25.250 |
| Aerobic capacity, mL/kg/min — women | 39.7 (31) | 24.9 | −1.0% / −2.3% | −37% | (pooled) |
| Bench press, reps — men | 52.7 (36) | 34.3 | −0.6% / −2.5% | −35% | ×2.896 |
| Bench press, reps — women | 39.7 (34) | 27.0 | −0.6% / −2.1% | −32% | (pooled) |
| Jump height, cm — men | 45.5 (27) | 26.9 | −1.2% / −2.2% | −41% | ×4.700 |
| Jump height, cm — women | 33.0 (19) | 17.2 | −1.6% / −2.2% | −48% | (pooled) |
| Lifestyle association (all outcomes) | Result |
|---|---|
| Active in leisure time at 16 | higher capacity throughout: main effect P < 0.001 for both aerobic measures and bench press, P = 0.019 for jump; no interaction with age (P 0.214-0.450) |
| Switching inactive → active (within person) | ×1.07 absolute and ×1.06 relative aerobic capacity, ×1.11 bench press, ×1.04 jump (all P ≤ 0.016) — reported as 6-7%, 11% and 4% |
| University degree | higher absolute aerobic capacity (P = 0.044) and bench press (P = 0.024); not relative aerobic capacity or jump |
| Stable individual differences (ICC) | 0.29-0.50 of the error variance |
| EWGSOP2 sarcopenia screen at 63 | no men and four women below a cut-off (grip or chair stand) |
Two decline figures in one abstract. The abstract's results give "0.3%-0.6% per year … accelerating to 2.0%-2.5% per year" for aerobic capacity and bench press, and a later sentence "0.2%-0.5% per year … (2.2% per year)" for the rate of decline overall; Table 2's model rates at 63 run from −2.0% to −2.5% a year. They are the same curve summarised twice, not two findings.
Sex. Table 1 shows an age × sex interaction for every outcome (P ≤ 0.027); the authors' "no sex difference in decline" refers to the post-peak slope (similar from about 40-45), the interaction mostly to the climb to the peak.
Peak age, absolute aerobic capacity: the text says 35 in women and 36 in men, Table 2 says 35 in men and 36 in women; the table is quoted above.
Mechanisms
Measured: the shape is the same in every capacity and both sexes — a peak by 36 (jump in women at 19), then an accelerating decline; the post-peak rate of decline does not differ between the sexes.
Measured: the between-person variance widened 3- to 25-fold from 16 to 63, and 29-50% of the variance was stable individual differences: people who started high tended to stay high.
Authors' interpretation: the general shape of the curve is "innate rather than acquired through lifestyle", while its slope before and after the peak and its level depend on how much of their adaptive potential a person realises by being active. Early mitochondrial dysfunction, then extracellular-matrix and neuromuscular-junction remodelling, atrophy and anabolic resistance are offered as the drivers; none was measured.
Authors' interpretation: cross-sectional studies underestimate the decline; this longitudinal cohort lost more than cross-sectional data and master-athlete records suggest.
Inference: "becoming active" is a within-person association from a yes/no question; it is consistent with training slowing decline but cannot separate cause from the people who chose to start (health, motivation) — and it says nothing about intensity or dose.
Practical Application
Is decline with age inevitable? On average yes — in 427 Swedes every capacity peaked by 36 and fell 30-48% by 63 — but the spread between people grew up to 25-fold, and being active at 16, or becoming active later, went with higher capacity. The yearly loss rose from under 1% to about 2%.That supports "the decline is real and speeds up, but how far you fall is partly yours", not "you will lose X% a decade". The aerobic numbers are estimated from a submaximal test, the activity measure is a yes/no question, and this is the untrained population, not athletes.
Do not quote population percentages at an individual athlete. By 63 the variance in relative aerobic capacity was 25 times what it was at 16; the group mean describes almost nobody in it. An athlete's own tests over time are the only trajectory that applies (inference).
It is not too late to start. Switching from inactive to active was associated with 6-7% higher aerobic capacity, 11% more bench repetitions and 4% higher jumps, (association, not a trial; the authors read it as "at any age", but no age interaction was reported for switching). For a trial of starting at retirement age see REF-1139.
Limitations
Aerobic capacity is estimated from heart rate at a submaximal workload (Åstrand nomogram, age-corrected), and at 16 from a 9-min run converted by regression; no VO2max was measured. Heart-rate-based estimates carry error that may differ with age and medication.
Leisure-time activity is a single yes/no item ("including e.g. walking or gardening"); no intensity, volume or type, so the activity associations cannot be read as a dose or as evidence for any kind of training. The authors list this limitation themselves.
Observational: active people and degree holders differ in many ways; attendance at later rounds was more likely for the active and educated, which the mixed model only partly offsets.
Five measurement points over 47 years, with gaps of 7-18 years; the curve between them is a spline, not data. Not every test was run at every age (jump not at 52, bench press not at 27).
One Swedish birth cohort (1958); the authors note that adolescents born 20 years later were less fit at the same age, so later cohorts may start lower.
The abstract summarises the decline twice with slightly different ranges (see Key Results); the text and Table 2 disagree on which sex peaked at 35 vs 36 for absolute aerobic capacity.
The 25-fold widening for relative aerobic capacity rests on a tiny age-16 variance in a value predicted by regression from a 9-min run, which compresses the spread (later values carry nomogram error, which widens it); the directly measured bench press and jump widened about 3- and 5-fold (our inference).
Funding: Swedish Research Council, Swedish Research Council for Sport Science, Tornspiran Foundation, Lindhé's Law Firm; no conflicts declared.
How this entry was checked and changed
Why this level
large observational cohort (n=427, all original participants kept in the model), 47 years of repeated objective tests; aerobic capacity is ESTIMATED (submaximal Åstrand cycle test, a converted 9-min run at 16), leisure activity is one yes/no question, one birth cohort in one country; read from the full text
Full text (checked 2026-10-15)
427 individuals (48% women) born in 1958; 222 male and 205 female pupils aged 16 years; data collections at ages 16, 27, 34, 52 and 63 (full text).
Aerobic capacity and bench press declined "starting at 0.3%-0.6% per year and accelerating to 2.0%-2.5% per year"; overall "0.2%-0.5% per year … (2.2% per year)"; "from peak to age 63 ranged from 30% to 48%" (full text).
Variance increase from 16 to 63: 5.242 (aerobic L/min), 25.250 (aerobic mL/kg/min), 2.896 (bench press), 4.700 (jump) (full text).
Relative aerobic capacity fell ~0.4 mL·kg−1·min−1 (~1.1%) per year at 40 and ~0.6 mL·kg−1·min−1 (~2.2%) per year at 63; cumulative 37% in women and 40% in men (full text).
Jump height fell 48% during 44 years in women and 41% during 36 years in men (full text).
Switching from inactive to active: exp(b) 1.07, 1.06, 1.11 and 1.04; "6%–7% for aerobic capacity, 11% for the bench press and 4% for the vertical jump" (full text).
On average 37% lost from peak age to 63; ~65% of peak aerobic capacity and muscular endurance kept at 63, against master athletes who "maintained > 80%" (full text).
At 63, 33 participants (8%) had died; ICC 0.35, 0.29, 0.50 and 0.40 (full text).