Refuting the myth of non-response to exercise training: 'non-responders' do respond to higher dose of training

78 healthy young males (age 26.2 +/- 3.4), 5 dose groups of ~14-16

Healthy sedentary-to-active young MALES aged 18-35. No females, no children, no older adults.

Six weeks of identical endurance training at five doses — 1, 2, 3, 4 or 5 x 60 min/week. THREE DESIGN LIMITS THAT MATTER: (1) participants CHOSE their group — assignment was not randomized, so dose is confounded with whoever opts into 5 sessions a week; (2) there was NO non-exercise control arm; (3) VO2max was not measured twice at baseline, so the "typical error" threshold used to classify non-response (+/-3.96% in Wmax) is imported rather than estimated in-sample. The headline second-period result — non-responders re-trained at a higher dose all become responders — is also exactly what REGRESSION TO THE MEAN produces when you select people on a low change score and re-measure them. The paper does not rule this out.

Non-response to endurance training is dose-dependent, not a fixed trait. The proportion classified as cardiorespiratory-fitness non-responders fell monotonically with training dose: 69% at 60 min/week, 40% at 120, 29% at 180, 0% at 240 and 0% at 300. Non-responders who then completed a second identical 6-week block with two extra sessions per week ALL became responders — "non-response was eliminated in all individuals." Read conservatively, this shows the "non-responder" label is at minimum dose-relative and re-test-fragile; read at face value it would mean the category does not exist. The conservative reading is the defensible one given no control arm and no randomization.

Cuts against a genetic reading of non-response, but weakly. If non-response were a stable heritable trait it should not vanish by adding two sessions per week; but with self-selected groups, no control arm, and re-testing of a low-scoring selected subgroup, the design cannot separate a true dose effect from selection plus regression to the mean.

Converges with Churchward-Venne 2015 (resistance training, older adults) and with the statistical arguments of Hecksteden 2015 and Renwick 2024. Contested by the Ross 2019 consensus statement, which maintains that true response variability exists and is partly genetic.