Peripheral scope (exercise-programming), but the outcome is patient-important, not a surrogate: injury is «virtually the sole drawback of exercise» (Lauersen et al., 2013), so preventing it is on the health axis directly (pain, lost function, time out) — and unlike almost everything on Resistance Training Prescription - Load Sets and Frequency, here the evidence is interventional RCT-MA (causal-grade), not observational surrogate transmission. Kept low in the Layer-1 ranking (it bears on the active, not the sedentary), but admitted on strong evidence: a gold meta-analysis of 25 RCTs, 26 610 participants, 3 464 injuries. (Lauersen et al., 2013)
Evidence-tier note (confidence: low). Single gold source (Lauersen 2013). Injury is a genuine
patient-important endpoint and the design is a proper RCT-MA, but held low on total web support:
single-source, the primary trials cannot blind an exercise exposure (an allocation/blinding
ceiling the authors flag), and the population is narrow (see transportability). Not a surrogate
discount — a corroboration-breadth discount.
The result: strength training is the standout; stretching does nothing
Stratified by the type of exercise (primary exposure), the injury-risk ratios rank cleanly — and the aggregate “physical activity prevents injury” number (RR 0.632, I2=70%) is too heterogeneous to use, which is why the stratification is the point, not the pooled figure:
| Intervention | Injury RR (95% CI) | I2 | k studies | Read |
|---|---|---|---|---|
| Strength / resistance training | 0.315 (0.207-0.480) | 0% | 4 (all ITT) | injuries cut to <1/3; the standout |
| Proprioception / balance | 0.550 (0.347-0.869) | 66% | 6 | effective, heterogeneous (ITT: 0.480, 0.268-0.862) |
| Multiple exposures | 0.655 (0.520-0.826) | 69% | 12 | effective, weakest of the three (ITT: 0.625) |
| Stretching | 0.963 (0.846-1.095) | 0% | 3 (all ITT) | null — CI spans 1 |
- Headline: «Strength training reduced sports injuries to less than 1/3 and overuse injuries could be almost halved.» (Lauersen et al., 2013)
- The strength and stretching subgroups were statistically homogeneous (I2=0%) despite differing in programme, population and outcome — «prove the generalisability of results» within the sport-injury domain: a real, direction-consistent effect for strength, a real absence for stretching, not a noisy average. (Lauersen et al., 2013)
The stretching null — a belief-refuting finding (symmetric standards)
Stretching (before or after exercise) is widely prescribed to prevent injury; this MA finds no effect: the stretching subgroup returns RR 0.963 (0.846-1.095), I2=0%, all-ITT — a tight null. The authors state plainly that their data do not support stretching for injury prevention, whether done before or after exercise; that discussion sentence then pivots to a generalizability caveat about the whole meta-analysis (mostly athletes — see transportability), which does not withdraw the null. (Lauersen et al., 2013) This is a no meaningful effect verdict, not insufficient evidence: the estimate is tight (RR 0.963, CI 0.846-1.095), homogeneous (I2=0%), ITT-analysed, and «corresponds to earlier reviews» — the expectancy test is satisfied, we would have seen the effect if it were there. (Lauersen et al., 2013) Note the boundary: this is injury prevention. It says nothing about stretching for range-of-motion or soreness (a different outcome the source does not adjudicate here).
Why strength beats the multi-component programmes — the dilution mechanism
Strength training «proved significantly better than multiple exposure studies, even though all multiple exposure studies included a strength training component.» (Lauersen et al., 2013) Counter-intuitive — bundling more exposures scored worse than the single best one. The stated mechanism is dilution: «each component may be reduced quantitatively and/or qualitatively» when bundled, so «multiple exposure programmes may therefore reduce the proportion of proven beneficial exposures and consequently reduce the overall preventive effect», and the added burden can erode compliance. (Lauersen et al., 2013) Decision consequence the authors draw: build prevention programmes «on the basis of well-proven single exposures» — i.e. do the thing that works (strength) at full dose rather than a diluted mix. (Lauersen et al., 2013) This is structural-leverage-adjacent for programming: a concentrated proven lever > a broad thin one.
Acute vs overuse — both reduced, overuse slightly more
Stratified by injury type (all exposures pooled): acute injuries RR 0.647 (0.502-0.836; ITT 0.615) and overuse injuries RR 0.527 (0.373-0.746, I2=19%) — «overuse injuries could be almost halved». (Lauersen et al., 2013) Caveat on attribution: five of six overuse analyses and six of nine acute analyses were multiple-exposure studies, so this split cannot say which component did the preventing — a G-gap the authors name («not possible to derive which parts of these interventions manifested the preventive effect»). Read the type-stratified table above for “which exercise”, this split for “which injury”. (Lauersen et al., 2013)
Robustness — the ITT direction is the reassuring one
The compliance worry runs: real-world adherence is worse than trial adherence, so effects should shrink. The opposite showed: «Contrary to the expected more conservative effect esti-mate, the intention-to-treat sensitivity analyses revealed even more beneficial effect estimates.» (Lauersen et al., 2013) So the estimates are not compliance-inflated — if anything the better-conducted (ITT) trials showed larger benefit. A small-study (publication-bias) effect was significant for the total estimate and the multiple-exposure subgroup only, and was attributed largely to heterogeneity; strength and stretching subgroups showed no significant small-study effect — the two cleanest results are also the least bias-suspect. (Lauersen et al., 2013)
Transportability boundary — who was actually studied
The population is mostly young athletes: 11 of 25 studies in adolescents; the exposures are sport-specific (eccentric/Nordic hamstring in male soccer, neuromuscular ACL programmes in female youth soccer, balance training in basketball/handball), with only two army-recruit and one internet-general-population study. (Lauersen et al., 2013)
- Support-factor caveat. The mechanism is loading a sport-specific tissue at risk (hamstring, ACL, ankle) to tolerate sport-specific forces. Whether the ~0.3 injury-RR transfers to a recreational or older exerciser doing general gym training — a different exposure, different injuries, different movement demands — is not established here ( — the studied support-factor profile is young-athlete + sport-specific loading; the transfer is a named gap, not a finding).
- Distinct decision from Exercise for Preventing Falls in Older Adults — different population (young athletes vs frail elderly), different outcome (sport injury vs fall/fracture), different mechanism. Not the same question; kept separate deliberately.
Synthesis — what this adds to the resistance-training picture
(inferred from Lauersen et al., 2013) — the Resistance Training Prescription - Load Sets and Frequency page is honest that its endpoints (1RM strength, muscle size) are surrogates with no RCT closing the health-outcome link. Lauersen supplies one patient-important outcome that resistance-type training moves at causal (RCT-MA) grade — injury reduction — which no observational strength/mortality association can claim. The composite: strength training’s case is strongest for injury (interventional, RR ~0.3), moderate for mortality (observational, via Muscle-Strengthening Activity and Mortality), and weakest for hypertrophy-as-health (surrogate for a surrogate). But the exposure is not identical — Lauersen’s strength arms are eccentric/sport-specific injury-prevention protocols in athletes, not the hypertrophy-oriented general RT of the prescription page; reading one as the other is the transportability gap above.
Decision relevance
- If the goal is preventing sport injury: do strength training, at full single-exposure dose. It is the one intervention that reliably works (RR ~0.32), and it beats bundled multi-component programmes even though those contain strength — so do not dilute it into a grab-bag.
- Do not rely on stretching to prevent injury — the effect is a tight null. (Stretch for mobility or preference if wanted; just not as injury prophylaxis.)
- Proprioception/balance training is a real second lever (RR ~0.48-0.55), useful where the at-risk injury is ankle/knee instability.
- Adherence is built in, not a discount — the ITT-robustness means the benefit survives realistic compliance; the bigger threat to it is over-designing the programme.
- Scope the claim to active/athletic populations — the transfer to general or older recreational training is unproven (an-class gap), so do not oversell “lifting prevents injury” as a universal.