REF-1385 · Fokkema et al., 2019 · British Journal of Sports Medicine · Paper · PubMed · Free full text
Will following injury-prevention advice keep me from getting hurt?
Not on its own: in 2,378 event runners, a website of risk-factor advice left injuries unchanged (37.5% vs 36.7%). Tailored or coached advice was not tested.
Main caveat: Self-reported, undiagnosed injuries with no severity or time-loss measure (the authors' limitation).
Fokkema and colleagues asked whether a multifactorial online prevention programme lowers running-related injuries in recreational runners.
Runners who registered for one of three large Dutch events in 2017 (NN City Pier City The Hague, NN Marathon Rotterdam, LadiesRun Rotterdam; 5 km to 42.195 km) were randomised: 1,196 got access to a password-protected website with information on evidence-based risk factors and advice to lower them, in four topics — personal factors, training, biomechanics and equipment — in versions for novice and experienced runners, plus up to five monthly e-mail reminders that repeated one topic; 1,182 prepared as usual.
The outcome was a self-reported new running injury between registration and one month after the event that cut running for at least a week, led to a doctor or physiotherapist, or needed medication. 37.5% of the intervention group and 36.7% of the control group were injured (OR 1.08, 0.90-1.29; difference 0.8 percentage points, −3.1 to 4.8), with no difference after adjustment for age, BMI and earlier injury and none in any predefined subgroup (sex, experience, event distance, earlier injury).
Runners uninjured in the previous year were injured somewhat more often with the programme (30.9% vs 27.7%; OR 1.30, 0.99-1.70), an interval that includes no effect.
Only 62.7% read at least one topic and 44.1% said they applied one; those who reported using the biomechanics section (which advised a gradual change to a forefoot strike and minimalist shoes) had more lower-leg injuries than controls (OR 1.74, 1.28-2.37), a comparison of self-selected users, not of randomised groups.
An injury in the previous year was the one significant predictor of a new one (OR 2.21, 1.84-2.65). Our reading: by the authors' account the largest RCT on running-injury prevention so far, and a clean null for information alone — the class of intervention an advice-giving assistant belongs to. It bounds what the assistant may claim: its advice is not shown to prevent injury.
It does not show that advice is harmful, nor that tailored advice fails.
Protocol
Design: INSPIRE trial (INtervention Study on Prevention of Injuries in Runners at Erasmus MC), randomised controlled trial with a minimum follow-up of 3 months; protocol published in BMJ Open Sport & Exercise Medicine; randomisation by a computer-generated list in blocks of 10 made by someone outside the research team; run in collaboration with Golazo Sports, an organiser of large Dutch running events. Ethics: Erasmus MC (MEC-2016-292); NTR5998.
Who: runners who registered for one of the three events, were 18 or older, registered at least 2 months before the event, read Dutch and had internet and e-mail; 5,271 expressed interest and n = 2378 were included (1,196 intervention, 1,182 control).Baseline (Table 2): 1,252 men (52.6%), age 41.2 (SD 11.9), BMI 23.7 (2.9), running experience 6.5 (7.8) years, 2.5 runs, 3.1 h and 22.2 km a week, training pace 6:04 min/km; event distance 5/7.5 km 5.8%, 10/10.55 km 38.1%, half marathon 29.9%, marathon 26.3%. No baseline difference between groups.
Intervention: a website built from a literature search on evidence-based risk factors, reached with a personal username and password, unlimited access.Four topics: personal factors (age, weight, previous injuries and running experience), training (running distance, frequency, surface, overtraining and stretching), biomechanics (cadence and foot landing) and equipment (footwear, orthotics and the use of running shoes); separate versions for novice and experienced runners.Runners were "expected to work autonomously with the website" and to read and apply what they thought relevant. Up to five monthly e-mail reminders repeated one topic with a link. How often each runner opened the site was not logged.The biomechanics section explained that forefoot striking lowers impact on the knee and included a foot-and-calf strengthening programme for a gradual transition to a forefoot strike and minimalist shoes.The paper does not print the exact advice within each topic (for instance whether a percentage progression rule was given); the details are in the published protocol, which was not read.
Control: usual preparation; reminders gave trial progress or general information on the epidemiology of running injuries.
Follow-up: questionnaires 2 weeks before the event, the day after it and 1 month after it; the intervention group was also asked whether it read and used each topic.
Outcome: a self-reported new running-related injury from registration to 1 month after the event: an injury of muscles, joints, tendons or bones in the lower back or lower limbs caused by running and meeting at least one criterion — reduced running distance, speed, duration or frequency for at least a week; a visit to a doctor or physiotherapist; or medication to relieve symptoms.Injury location was a secondary outcome. New injuries were cross-checked against earlier questionnaires.
Analysis: intention to treat; injury proportions with 95% CIs, the difference in proportions, univariate logistic regression (OR) and risk ratios; adjusted logistic regression for age, BMI and earlier injury; predefined subgroups by sex, running experience (≤1 or >1 year), event distance, injury in the previous 12 months and five location clusters.Sample size: 14% expected injured, a 5% risk difference, one-sided α 0.05, 80% power → 1,006 analysed, 1,106 with 10% loss. Multiple imputation was planned if more than 5% of data were missing; it did not change the main result and the authors report the analysis without it.
Funding and interests: "The INSPIRE trial was funded by the Netherlands Organization for Health Research and Development (ZonMW), grant number 536001001", performed in collaboration with Golazo Sports (an organiser of large running events in the Netherlands, as printed). Competing interests: none declared. No device maker or data platform is named.
Key Results
Primary outcome and injury locations (Table 3; control group is the reference):
| Outcome | Intervention (n = 1196) | Control (n = 1182) | Difference, % points (95% CI) | OR (95% CI) | Risk ratio (95% CI) |
|---|---|---|---|---|---|
| Runners with a new running injury | 449 (37.5%) | 434 (36.7%) | 0.8 (−3.1 to 4.8) | 1.08 (0.90 to 1.29) | 1.02 (0.92 to 1.14) |
| Number of new injuries | 736 | 747 | — | — | — |
| Lower back | 4.1% | 4.1% | 0.0 (−1.6 to 1.7) | 1.03 (0.68 to 1.55) | 1.01 (0.68 to 1.49) |
| Buttock, hip or groin | 9.2% | 8.8% | 0.4 (−2.0 to 2.8) | 1.07 (0.81 to 1.43) | 1.05 (0.81 to 1.35) |
| Upper leg or knee | 14.3% | 16.9% | −2.6 (−5.6 to 0.36) | 0.83 (0.66 to 1.04) | 0.85 (0.70 to 1.02) |
| Lower leg (shin, calf, Achilles, ankle) | 16.0% | 15.1% | 0.9 (−2.1 to 3.9) | 1.10 (0.88 to 1.38) | 1.06 (0.88 to 1.28) |
| Foot or toe | 6.5% | 6.3% | 0.2 (−1.9 to 2.2) | 1.05 (0.76 to 1.46) | 1.03 (0.76 to 1.40) |
Predefined subgroups (Table 4; injury proportion intervention vs control, OR):
| Subgroup | Intervention | Control | OR (95% CI) |
|---|---|---|---|
| Injured in the previous 12 months (611 vs 627) | 43.9% | 44.7% | 0.92 (0.72 to 1.19) |
| Not injured in the previous 12 months (585 vs 555) | 30.9% | 27.7% | 1.30 (0.99 to 1.70); risk ratio 1.12 (0.93 to 1.34) |
| Running experience ≤1 year (224 vs 211) | 36.6% | 39.3% | 0.90 (0.58 to 1.38) |
| Running experience >1 year (969 vs 964) | 37.8% | 36.3% | 1.12 (0.92 to 1.36) |
| Marathon (317 vs 308) | 41.3% | 41.2% | 0.97 (0.67 to 1.39) |
| Half marathon (367 vs 344) | 42.2% | 38.1% | 1.21 (0.88 to 1.66) |
| Men (623 vs 629) / women (573 vs 553) | 39.2% / 35.8% | 38.3% / 34.9% | 1.10 (0.86 to 1.41) / 1.06 (0.81 to 1.38) |
Adjusted and imputed: adjusted for age, BMI and earlier injury, OR 1.08 (0.90 to 1.30); with imputation of missing follow-up, OR 1.16 (0.93 to 1.44).
Who used it: 62.7% of the intervention group read at least one topic (8.2% one, 11.0% two, 4.7% three, 38.8% all four); 44.1% said they applied at least one topic in training. Runners injured in the previous year read more often (65.6% vs 59.7%, p=0.033).
Biomechanics users: intervention runners who said they used the biomechanics section reported more lower-leg injuries than the control group (OR 1.74, 1.28 to 2.37; online appendix 2, not read) — a comparison of self-selected users with all controls, not a randomised contrast.
Previous injury: an injury in the 12 months before the trial was the only significant predictor of a new one (OR 2.21, 1.84 to 2.65); 76.6% of new injuries were at a different site from the previous one.
Follow-up: mean 4.5 (SD 1.6) months; 81.1% completed at least one follow-up questionnaire and 60.0% all; 28.4% of follow-up questionnaires were not completed. 883 runners (37.1%) reported 1,483 new injuries; most were at the knee (10.8% and 12.5% of the groups), calf and foot.
Not reported: injury rates per 1,000 h of running, injury severity or time loss.
A number that does not add up: the abstract gives the univariate OR as 1.08 (0.90 to 1.30); the Results and Table 3 give the univariate OR as 1.08 (0.90 to 1.29) and the adjusted OR as 1.08 (0.90 to 1.30). The difference is in the last digit only.
Mechanisms
Measured: no difference in injured runners overall or in any subgroup; a non-significant shift from upper-leg/knee injuries (−2.6 points, −5.6 to 0.36) toward lower-leg injuries; more lower-leg injuries in the self-selected users of the biomechanics section; only about two thirds of the intervention group read any of the material.
Authors' interpretation: runners train alone and had to extract and apply generic information themselves; low engagement, information that was hard to translate into training, and a static website may explain the null — they suggest more personalised information or "more directed practical information (eg, detailed day-to-day training schedules)".Runners with no recent injury may already have trained in a way that suited them, so changing something produced injuries. Changing to a forefoot strike may move load from the knee to the lower leg and foot; they advise not giving biomechanics advice without personal guidance, for example from a physiotherapist. Prevention should target runners with an earlier injury.
Inference: knowing a risk factor is not the same as changing the load that produces an injury. The trial shows that information on risk factors and how to lower them, offered to runners to apply on their own, did not change injury rates; a load change, a supervised programme or an intervention the runner actually carries out is a different thing, tested elsewhere (REF-1384, REF-1119).
Practical Application
Will following injury-prevention advice keep me from getting hurt? Not on its own: in 2,378 event runners, a website of risk-factor advice left injuries unchanged (37.5% vs 36.7%). Tailored or coached advice was not tested.
What the assistant may claim: this is the assistant's own class of intervention, and in a large registered RCT it did not lower injuries (A). The assistant must not say or imply that its advice prevents injury. It can say what lowered injuries in trials when runners did it — a supervised hip-and-core programme in novices (REF-1384), impact-feedback gait retraining in novices (REF-1119) — and that no load rule is a guarantee.
The 10% figure is a risk marker, not protection: the programme's training topic covered running distance and frequency (the paper does not print whether a percentage rule was given), and risk-factor advice built this way did not prevent injury.The base's line "Runners: no run over ~10% longer than your longest of the past 30 days (5,205-runner cohort)" (SYNTH-29, from REF-1102, one observational cohort; repeated as an instruction in SYNTH-47, SYNTH-64, SYNTH-65 and SYNTH-75) should be offered as what the cohort associated with fewer injuries, not as a safe line; REF-1367 found no consistent link between training parameters or their changes and injury across 36 studies.This trial did not test the single-run 10% figure itself.
Do not push a strike change: the users of the biomechanics section, which advised a gradual move to a forefoot strike and minimalist shoes, had more lower-leg injuries (OR 1.74, 1.28-2.37; self-selected users, not randomised). That fits the base's guide on foot strike (SYNTH-86: leave a pain-free runner's landing alone).
Previous injury is the one consistent signal: a running injury in the past year went with about twice the odds of a new one (OR 2.21, 1.84-2.65), mostly at another site; the authors would aim prevention at these runners. Not from this paper; the base's own practice: "A previously injured site gets the slowest ramp" (SYNTH-73, inference from a risk factor).
What it is good for / what it does not show: good for bounding claims — information alone, self-applied, did not prevent running injuries in experienced recreational road runners.It does not show that advice is harmful (the subgroup OR 1.30 has an interval from 0.99 to 1.70), that tailored, conversational or coach-led advice fails (a smaller trial of injury-status-tailored online advice in trail runners, Hespanhol 2018, reported fewer injuries in its abstract; it is not in the base and its full text was not read), or which piece of advice is useless.It is not a novice trial, and it did not measure whether runners changed their training.
Limitations
Self-reported, undiagnosed injuries with no severity or time-loss measure (the authors' limitation); the injury definition predates the 2015 consensus definition, which by randomisation should not bias the comparison.
Missing data. 28.4% of follow-up questionnaires were not completed and 18.9% of runners completed none (our arithmetic from 81.1%); imputation did not change the result.
Use not measured objectively. Reading and applying the material was self-reported; website visits were not logged. A null for "offered" advice is not a null for advice that was followed.
One delivery format. A static website plus monthly e-mails, without personalisation; the authors themselves suggest that personalised or day-to-day practical guidance might do better (untested here).
Who. Experienced recreational road runners entered for Dutch mass events, half of them injured in the previous year and a fifth still injured at entry; few novices.
Exploratory secondary findings. The biomechanics-user comparison is not randomised; the previously-uninjured subgroup is one of several predefined subgroups and its interval includes no effect.
Funding and interests: ZonMw (Netherlands Organization for Health Research and Development, grant 536001001); run in collaboration with the event organiser Golazo Sports; no competing interests declared. No device maker or data platform funded it.
How this entry was checked and changed
Why this level
a large registered RCT (rubric A), read from the full text (Europe PMC, open access, CC BY 4.0). The null is informative because the trial was about twice its planned size (1,106) and the interval for the difference in injured runners is −3.1 to 4.8 percentage points, ruling out a benefit larger than about 3 points; its limits are self-reported, undiagnosed injuries, 28.4% of follow-up questionnaires missing (imputation did not change the result) and self-reported, unlogged use of the website
Full text (checked 2026-10-28)
This trial included 2378 recreational runners (1252 men; mean [SD] age 41.2 [11.9] years), of which 1196 were allocated to the intervention group and 1182 to the control group (full text)
The website contained information on four main topics: personal factors (age, weight, previous injuries and running experience), training (running distance, frequency, surface, overtraining and stretching), biomechanics (cadence and foot landing) and equipment (footwear, orthotics and the use of running shoes) (full text)
New running injury: intervention 37.5% (95% CI 34.8 to 40.4) vs control 36.7% (95% CI 34.0 to 39.6); OR 1.08; 95% CI 0.90 to 1.29 in Table 3, risk ratio 1.02 (0.92 to 1.14) (full text)
Runners with no injury in the 12 months before the trial: injury proportion (%) 30.9 vs 27.7, OR 1.30; 95% CI 0.99 to 1.70 (full text)
62.7% of the intervention group read at least one topic of the injury prevention programme; 44.1% applied the information of at least one topic into their training (full text)
Users of the biomechanics section reported more lower leg injuries than the control group (OR 1.74; 95% CI 1.28 to 2.37) (full text)
An RRI in the 12 months before the study was the only factor with a significant effect on the occurrence of new RRIs (OR 2.21; 95% CI 1.84 to 2.65) (full text)
Funded by the Netherlands Organization for Health Research and Development (ZonMW), grant number 536001001; Competing interests: None declared (full text)