REF-1615 · García-Hermoso et al., 2025 · Translational Pediatrics · Paper · PubMed · Free full text

A sporty childhood goes with only a slightly higher chance of an active adulthood: across 38 studies (63,158 people, mean follow-up 20.9 years) physical activity tracked weakly from youth into adulthood (r = 0.14) and youth sport participation a little more (r = 0.26), both more strongly into the twenties than later and fading slightly with longer follow-up; mostly self-report and unadjusted correlations. Read in full

Main caveat: Mostly self-reported activity, much of it retrospective, open to recall bias over long follow-ups.
fairly strong evidenceMeasured on non-athletes and recreational peopleRead in full

García-Hermoso and colleagues pooled cohort studies that measured physical activity or sport participation at ages 6-18 and again in young adulthood (19-30) or later adulthood (over 30), using correlation coefficients as the measure of stability. Thirty-eight studies with 63,158 participants followed for an average of 20.9 years (range 5-55) were included.

Overall physical activity tracked weakly (r = 0.14), with no significant difference by sex or by whether the first measure was in childhood or adolescence, and more strongly into young adulthood than later. Sport participation tracked somewhat more (r = 0.26), with the same patterns. Longer follow-up went with slightly lower correlations for both.

The authors read the results as support for promoting activity and sport in youth, while noting that the link is weak and that factors beyond habit shape adult activity.

Participants

38 studies, 63,158 participants, baseline ages 6-18, follow-up 5-55 years (mean 20.9); 16 countries, most of them European, plus Australia, Brazil, Canada, Israel, Japan, New Zealand and the United States.

Protocol

Inclusion: prospective or retrospective cohort studies measuring physical activity or sport participation at 6-18 years and its association in young adulthood or adulthood; any measure (self- or parent-report, accelerometer, pedometer).
Synthesis: correlations (other statistics converted to r), random-effects inverse-variance pooling with HKSJ adjustment; studies analysed by sex where possible; overlapping cohorts reduced to the longest or largest.
Risk of bias: JBI checklists for cohort (11 items) and analytical cross-sectional (8 items) studies, two reviewers.
Bias across studies: Doi plots, LFK index, non-parametric trim-and-fill.

Key Results

AnalysisPhysical activity, r (95% CI)Sport participation, r (95% CI)
Overall0.14 (0.11 to 0.16); I² 85.41%0.26 (0.20 to 0.31); I² 88.57%
Men / women0.14 (0.10 to 0.18) / 0.13 (0.09 to 0.17); P = 0.720.26 (0.18 to 0.33) / 0.25 (0.16 to 0.33); P = 0.85
From childhood / adolescence0.10 (0.06 to 0.14) / 0.15 (0.12 to 0.18); P = 0.060.22 (0.13 to 0.32) / 0.27 (0.20 to 0.33); P = 0.39
To young adulthood / adulthood0.20 (0.16 to 0.24) / 0.10 (0.07 to 0.13); P < 0.0010.30 (0.22 to 0.38) / 0.20 (0.13 to 0.26); P = 0.03
Follow-up length, per yearB = −0.002 (−0.004 to −0.001); P = 0.04B = −0.006 (−0.012 to −0.001); P = 0.04
Sport vs overall activity: sport participation tracked significantly more strongly (P < 0.001).
Risk of bias: mean share of JBI items met about 51.91% (cohort) and 62.5% (cross-sectional); losses to follow-up and simple questions or diaries in some studies.
Publication bias: physical activity LFK 2.85 (major asymmetry), trim-and-fill imputed none and left r at 0.136; sport LFK −1.12 (minor), none imputed, r 0.262.
Size (our arithmetic): r = 0.26 means youth sport accounts for about 7% of the variance in the later measure, and r = 0.14 for about 2%.

Practical Application

Across 38 cohorts (63,158 people followed about 21 years), children and teenagers who did more sport were only slightly more likely to be active adults (r = 0.26 for sport, 0.14 for overall activity), and the link was weaker after 30 than in the twenties. A sporty childhood is only weakly linked to an active adulthood, not a guarantee: it accounts for well under a tenth of the difference between adults (our arithmetic); inference, not tested here: most of what makes an adult active lies elsewhere.
The pattern did not differ significantly by sex or by whether the youth measure came from childhood or adolescence.
Organised and informal sport were not separated, and most measures were self-reported, often recalled over decades.
This is the behavioural side only; for fitness carried over see REF-1616 and REF-1617, for bone and lean mass REF-1618.
The base's own practice for an adult starting or restarting: build the habit first, frequency before duration (SYNTH-75).

Limitations

Mostly self-reported activity, much of it retrospective, open to recall bias over long follow-ups.
Correlations without adjustment for confounders such as socioeconomic status or life events.
High heterogeneity; diverse measures of activity and sport.
Organised and informal sport not separated; mostly European studies.
Supplementary tables (study list, risk-of-bias detail, subgroup plots) not read.
How this entry was checked and changed
Why this level

a registered systematic review and meta-analysis with the risk of bias read; held at A- at most because it is serious: on average about 52% of checklist items were met by the cohort studies and 62.5% by the cross-sectional ones, and the authors name recall bias and unadjusted correlations; heterogeneity is high (I² 85-89%), and the physical-activity pooling showed major small-study asymmetry (LFK 2.85), although trim-and-fill imputed no studies