Why do I lose power with age?
In 129 healthy adults, power tracked thigh lean mass and evoked twitch force rise; isometric voluntary activation stayed near maximal. Cross-sectional knee extension, not change over time; no training.
Main caveat: Cross-sectional. Age differences between people are not change within a person.
Wrucke and colleagues measured maximal knee-extensor power at 14 velocities, plus voluntary activation with transcranial magnetic stimulation, twitch properties with femoral nerve stimulation, surface EMG and thigh lean mass by DXA, in 31 young, 82 old and 16 very old men and women.
Peak power was much lower with age and the gap was greatest at fast velocities; the velocity giving peak power fell from 355 to 289 to 238°/s. Thigh lean mass and the rate of torque development of the electrically evoked twitch went most closely with power, and the twitch association remained after power was normalised to lean mass.
Voluntary activation did not differ with age or sex; it was weakly related to power in women and not in men. EMG during the power effort explained almost nothing.
The authors conclude that the loss comes mainly from factors within the muscle, at or distal to the neuromuscular junction, with a possible small neural part in older women; the fast-fibre explanation they favour was not measured. Our reading: a strong method and a useful answer to "is it my nerves or my muscle?". It is not evidence of what training slows the loss.
Protocol
Design: single-session cross-sectional comparison; Marquette University Institutional Review Board; written consent.
Who: n = 129: 31 young (19-33 y; 16 men, 15 women), 82 old (61-79 y; 44 men, 38 women), 16 very old (80-93 y; 7 men, 9 women). Healthy, community-dwelling, free of known neurological, musculoskeletal and cardiovascular disease; mini mental state ≥ 26 for older participants.
Activity: steps a day by waist accelerometer. Men: young 8,623, old 8,494, very old 4,134. Women: 9,825, 7,556, 3,455. Training history is not reported.
Power: concentric isokinetic knee extension on a Biodex System 4, 15 sets of 4 contractions over 14 velocities (30-450°/s), peak power counted only where the target velocity was reached; dominant leg except where surgery, pain or osteoarthritis applied.
Neural and contractile: transcranial magnetic stimulation to the motor cortex during maximal and submaximal isometric contractions (voluntary activation by two formulas); femoral nerve stimulation for the potentiated twitch; surface EMG of vastus lateralis, vastus medialis and biceps femoris during the power efforts.
Mass: DXA thigh lean tissue mass (the authors note DXA underestimates age-related loss of thigh muscle).
Function (older adults only): Berg balance, 6-min walk, stair climb, sit-to-stand.
Analysis: simple regressions of power on age and on each neuromuscular measure; two-factor ANOVA or nonparametric equivalents; linear mixed models across velocities. No multivariable model.
Funding and interests: National Institute of Aging grants (as printed) (AG052313, AG045766, AG048262) and an American Heart Association fellowship (19POST34380411). No conflicts declared.
Key Results
| Measure, means ± SD | Young | Old | Very old |
|---|---|---|---|
| Peak power, men, W | 673.4 ± 163.0 | 398.5 ± 118.7 (−42%) | 225.1 ± 51.4 (−67%) |
| Peak power, women, W | 428.0 ± 79.2 | 222.6 ± 56.6 (−49%) | 169.5 ± 39.7 (−60%) |
| Power per kg thigh lean mass, men, W·kg−1 | 93.1 ± 14.5 | 60.4 ± 14.1 (−35%) | 43.4 ± 9.8 (−53%) |
| Power per kg thigh lean mass, women, W·kg−1 | 79.9 ± 8.4 | 49.5 ± 11.7 (−38%) | 35.7 ± 9.4 (−55%) |
| Isometric MVC torque, men, N·m | 274.6 ± 65.7 | 186.0 ± 48.5 (−32%) | 114.5 ± 28.0 (−58%) |
| Isometric MVC torque, women, N·m | 184.4 ± 43.2 | 108.0 ± 25.3 (−49%) | 90.1 ± 23.9 (−51%) |
| Voluntary activation (eRT formula), men, % | 96.4 ± 3.3 | 95.7 ± 5.3 | 95.6 ± 3.2 |
| Voluntary activation (eRT formula), women, % | 97.2 ± 2.7 | 95.1 ± 4.0 | 90.4 ± 12.0 |
| Velocity at peak power, °/s | 355 ± 42 | 289 ± 46 | 238 ± 34 |
Percent differences are compared with the same-sex young group, as printed. Association with absolute peak power:
| Measure | Men | Women |
|---|---|---|
| Rate of twitch torque development | R² = 0.69, P < 0.001 | R² = 0.57, P < 0.001 |
| Thigh lean tissue mass | R² = 0.53, P < 0.001 | R² = 0.34, P < 0.001 |
| Voluntary activation (eRT formula) | R² = 0.03, P = 0.191 | R² = 0.12, P = 0.012 |
| Rate of EMG rise at peak power | R² = 0.01, P = 0.456 | R² = 0.07, P = 0.042 |
Rates across the groups, estimated from a cross-section: ∼6.5 W·yr−1 in men and ∼4.2 W·yr−1 in women (P = 0.002); relative to young, ∼1.0%·yr−1 in both (P = 0.705). Older groups only: ∼10.7 W·yr−1 in men and ∼3.8 W·yr−1 in women (P = 0.002); ∼1.5%·yr−1 vs ∼0.9%·yr−1 (P = 0.070).
After normalising power to thigh lean mass, the twitch rate association stayed: R² = 0.38 in men and in women (both P < 0.001). In very old adults lean mass was not associated with power (R² = 0.08, P = 0.279), which the authors put down to DXA; twitch rate was (R² = 0.64, P = 0.004).
Very old women, one outlier: removing one very old woman removed the voluntary-activation link in the very old group (R² = 0.16, P = 0.140) and left it in women overall by one formula (R² = 0.10, P = 0.021) but not the other (R² = 0.06, P = 0.083).
Function: peak power went with 6-min walk distance (R² = 0.28, P < 0.001) and stair climb time (R² = 0.17, P < 0.001) in older adults.
Physical activity: steps went with power in women (R² = 0.12, P = 0.012), not in men (R² = 0.00, P = 0.698). Steps do not grade training.
Mechanisms
Measured: peak power lower with age; lean mass and twitch kinetics associated with it; voluntary activation near maximal in isometric contractions at all ages; slower rate of EMG rise in older groups during dynamic efforts, not associated with power in men.
Authors' interpretation: factors at or distal to the neuromuscular junction, including loss of muscle mass and altered contractile function, are the primary cause in both sexes, with a small neural part possible in older women. They favour a smaller share of fast (MyHC II) fibres, citing modelling and other work; fibre type was not measured in this study.
Inference: near-maximal voluntary activation concerns descending drive in slow or isometric contractions (the technique cannot be used during fast ones). It does not contradict the motor-unit loss and remodelling in REF-596, which sit at a different level and would show up here as muscle-side change.
Inference: associations of R² in the 0.5-0.7 range between power and twitch force are partly built in, since both are output of the same muscle; they locate the problem rather than explain it.
Practical Application
Why do I lose power with age? In 129 healthy adults, power tracked thigh lean mass and evoked twitch force rise; isometric voluntary activation stayed near maximal. Cross-sectional knee extension, not change over time; no training.
What it is good for: explaining why power falls further than isometric strength with age (men of about 89 y −67% vs −58%, women of about 83 y −60% vs −51%; at about 70 y men −42% vs −32%, women −49% for both) and why it shows most at fast velocities: this points to the muscle, not to a loss of will or drive.
Do not read a yearly rate for one athlete: the ~1% a year is an estimate across different people, with the young group flattening the slope; the authors expect it understates a real loss of about 2-4% a year in older adults, and survivorship may flatten the oldest groups. Masters athletes were not studied.
Women and men (inference from the authors): absolute loss is larger in men, relative loss alike; mass-normalised rates did not differ. A single-sex rule has no support here.
Limitations
Cross-sectional. Age differences between people are not change within a person; the design also carries survivorship and a possible floor effect in the oldest groups, which the authors name, and they say the rates likely underestimate loss in the general older population.
Selected sample. Healthy, mobile, above clinical cut-offs on function; findings do not extend to mobility-impaired or frail people. Old adults' steps matched the young. No masters athletes identified, no training records.
One muscle group, one mode. Dominant-leg knee extension on a dynamometer; not whole-body or cycling power. Not all participants reached velocities above 300°/s, particularly older ones, so sample shrinks at the fast end.
Measurement. Voluntary activation cannot be assessed during fast contractions; surface EMG is limited; DXA underestimates loss of thigh muscle; fibre type not measured. Some participants did not complete stimulation (six old women and two young men).
Small very old group (7 men, 9 women) with one influential outlier; many simple regressions without correction or a joint model.
Reporting. The abstract gives R² = 0.13 for voluntary activation and power in women where the Results give R² = 0.12. The text says the age differences in absolute torque were greater at slower velocities while the Fig. 3 caption says faster. The discussion's "~3%" of variance for steps has no matching pooled value in the Results (women 12%, men 0%).
Interests. Public funding (NIA, AHA); no conflicts declared.
How this entry was checked and changed
Why this level
read from the full text (CC BY 4.0). The rubric places a cross-sectional measurement study with at least 100 participants at B+. It holds for the group differences and associations it measured, not for rates of change or any training conclusion; the very old group is small (16 people)
Full text (checked 2026-10-31)
Participants: 31 young (19–33 yr), 82 old (61–79 yr), 16 very old (80–93 yr), healthy and community-dwelling (full text)
Peak power in men was 42% and 67% lower in old and very old than young; in women 49% and 60% lower (full text)
Age-related loss ∼6.5 W·yr−1 in men, ∼4.2 W·yr−1 in women, P = 0.002; relative rate ∼1.0%·yr−1 in both, P = 0.705 (full text)
Rate of twitch torque development and peak power: R2 = 0.69 in men, R2 = 0.57 in women; thigh lean tissue mass R2 = 0.53 and R2 = 0.34 (full text)
Voluntary activation did not differ with age or sex, and was not associated with power in men (R2 = 0.03, P = 0.191) but weakly in women (R2 = 0.12, P = 0.012) (full text)
The authors state the cross-sectional rates most likely underestimate loss after about 40–50 yr, against other studies' ∼2%–4% a year (full text)
Fibre type composition was not measured in this study (full text)
Grants from the National Institute of Aging and the American Heart Association; no conflicts declared (full text)