REF-1307 · Bredeweg et al., 2012 · British Journal of Sports Medicine · Paper · PubMed

Should a beginner spend a month walking and hopping before starting to run?

In 432 novice runners it did not cut injuries (15.2% vs 16.8%); the trial was too small to rule out a modest effect. Start the gentle run plan.

Main caveat: Read from the abstract. Group characteristics, adherence, exposure and injury rates per hour come from the table of the meta-analysis that pools this trial…
fairly strong evidenceMeasured on non-athletes and recreational people

Bredeweg and colleagues tested a "tolerance first" start in novice runners: before a frequently used 9-week beginner programme for the Groningen 4-Mile, half of 432 participants did a 4-week preconditioning programme of walking and hopping exercises (PRECON, 211) and half started the running programme directly (control, 221).

Running exposure and running-related injuries (RRIs) — any musculoskeletal complaint of the lower extremity or lower back that restricted running for at least a week — were registered in an internet running log.

The incidence of RRIs was 15.2% with preconditioning and 16.8% without (χ² = 0.161, df = 1, p = 0.69), and the authors conclude that the programme had no influence on injury incidence.

Our arithmetic: the two percentages match 26 of 171 and 32 of 191 runners — the analysed numbers in the table of REF-1244, which also lists 58 injuries — and those counts reproduce the published χ² of 0.161; that is a risk ratio of about 0.91 (95% CI 0.56 to 1.46) and a risk difference of −1.5 percentage points (−9.1 to +6.0).

The trial was planned for a 40% injury rate and a 25% relative reduction (40% to 30%), needing 360 runners for 80% power and 414 after 15% attrition (design paper); at the rate actually seen (about 17% in controls) a 25% relative reduction would need roughly 1,100 runners per arm (our arithmetic).

Our reading: no evidence that a short, light walk-and-hop phase prevents injuries in beginners; the interval still allows anything from a 44% reduction to a 46% increase in risk. It is a separate trial from the same group's graded-programme RCT (REF-925, also null), and it is already one of the nine trials in REF-1244.

Protocol

Design: two-arm randomised controlled trial, the authors' "level of evidence, 1"; stratified randomisation by current sporting activities, previous injury and sex (table of REF-1244); registered as NTR1906; the design paper was received in June 2010, after recruitment (April-June 2008) and the start of data collection (July 2008), so it is not a prospective protocol (whether registration preceded enrolment was not checked).
Who: novice runners preparing for the Groningen 4-Mile; n = 432 randomised, 149 men and 283 women; age 39.0 ± 10.7 (PRECON) and 37.2 ± 10.9 (control) years (table of REF-1244). Eligible: healthy, 18-65 years, not running regularly in the previous 12 months, no lower-limb or back injury in the previous 3 months, willing to start a beginners' programme; excluded with absolute ACSM contraindications to vigorous activity (design paper).
Intervention (PRECON), 4 weeks before the running programme: walking and hopping exercises (abstract).Per the design paper (Table 1): brisk walking three times a week, two of the sessions as six blocks of 5 min walking each followed by hopping in place (50-90 hops a block, rising weekly; 660-1020 hops a week) and the third a 30-60-min walk; unsupervised after an initial interview, with instruction videos; 71.6% completed 10 or more sessions, 14.2% 7-9, 6.6% 4-6 and 7.6% none.
Control: no preconditioning; started the running programme at the same time as the PRECON group finished its 4 weeks.
Both groups: the same 9-week running programme; running time over the programme 301.1 ± 184.7 min (PRECON) vs 329.7 ± 177.1 min (control), not significantly different (table of REF-1244).
Primary outcome: RRIs per 100 runners. RRI = any musculoskeletal complaint of the lower extremity or lower back restricting running for at least a week (abstract; the design paper words it as at least three training sessions).
Sample size (design paper): expected injury incidence 40%, a 25% reduction considered relevant; 360 runners (2 × 180) for 80% power at alpha 0.05 (by a logistic rank survival power analysis); 414 (2 × 207) allowing 15% attrition. Our arithmetic: a two-proportion test of 40% vs 30% needs about 356 per arm for 80% power, so 2 × 180 gives about 52%.
Funding and interests: the design paper states funding by the Netherlands Organisation for Health Research and Development (ZonMw), grant 75020009, which had no role in the design, and declares no competing interests; PubMed indexes the trial report as "Research Support, Non-U.S. Gov't". The trial report's own funding and conflict statements were not read (abstract only). No device maker, federation or data platform is named as a funder.

Key Results

OutcomePRECON (walk and hop, then running)Control (running only)Comparison
Randomised211221—
Analysed (table of REF-1244)17119170 of 432 (16%) not analysed (our arithmetic)
Runners with an RRI15.2%16.8%χ² = 0.161, df = 1, p = 0.69
Runners with an RRI, counts (our reconstruction)26 of 17132 of 191RR about 0.91 (95% CI 0.56 to 1.46); risk difference −1.5 points (−9.1 to +6.0), our arithmetic
Injuries per 1000 h of running (table of REF-1244)31 (95% CI 24.0-38.0)32 (24.0-37.0)—
Primary result (abstract): no significant difference in injury incidence; the authors conclude the 4-week programme "had no influence" on RRIs.
Power against what was seen: planned for 40% vs 30%; observed about 15-17%. The trial could detect only a large effect at that base rate (our arithmetic: about 1,100 per arm for a 25% relative reduction from 16.8%).
Pooled elsewhere: in REF-1244 this trial contributes to the null pooled injury risk (log RR −0.21, 95% CI −0.46 to 0.047, seven trials) and rate. Counting it again beside that pool double-counts it.

Mechanisms

Measured: injury counts and running exposure only; no tissue, strength or load-tolerance measure was taken. About seven in ten preconditioned runners completed at least 10 of the sessions (table of REF-1244).
Authors' interpretation (design paper): most running injuries come from training errors ("too much, too soon"); a preconditioning period would let runners adapt more gradually to the high mechanical loads of running. The trial did not support that hypothesis.
Inference: walking and twice-weekly hopping are a lighter and different load from running, and 4 weeks is shorter than the tissue time courses the base holds — tendon stiffness rose in interventions of 8 weeks or more (REF-517); tibial cortical density was still dipping at 8-14 weeks of new impact loading (REF-1315). The trial tests a short, light preparation phase, not "build structure for three months first".
Inference: the control group did not start recklessly — it began the same established beginner programme. The question answered is whether an extra month of preparation before an already gentle start adds anything; it did not, detectably.

Practical Application

Should a beginner spend a month walking and hopping before starting to run? In 432 novice runners it did not cut injuries (15.2% vs 16.8%); the trial was too small to rule out a modest effect. Start the gentle run plan.
Do not sell a short preparation phase as injury insurance (A-, from the abstract; novices, 4-mile goal): a 4-week walk-and-hop phase showed no evidence of fewer injuries, and a slower 13-week graded programme did not either (REF-925, A). For a beginner the time is better spent on the running programme itself, built in small steps (SYNTH-75).
No harm was shown either: a beginner who wants a few weeks of walking or hopping first can do it; it costs four weeks before running starts and buys no shown protection (inference).
Where the measured risk sits instead (not from this paper): a single run more than ~10% longer than the longest of the past 30 days (REF-1102), no running history (REF-1313; not a predictor in REF-926's 930 novices), and a previous injury (REF-564); in novices injury is also the main reason to quit running (REF-1311).
Weighing the evidence: count this trial once. In REF-1244 it is one of nine pooled exercise-programme trials; for the "tolerance first" question it sits beside REF-925 (slower start, null) and REF-1290 (intensity vs volume progression after a shared easy base, no difference) — three underpowered nulls, all in running.

Limitations

Read from the abstract. Group characteristics, adherence, exposure and injury rates per hour come from the table of the meta-analysis that pools this trial (REF-1244), not from the trial report; the analysed counts are our reconstruction, consistent with that table and the published χ².
Underpowered for the rate it saw. Planned for 40% injured in controls, it saw about 17%; the confidence interval of the difference is wide.
Missing runners. 70 of 432 randomised (16%) are not in the analysed counts; the design allowed for 15% attrition. Whether the analysis was intention-to-treat is not stated in the abstract; the design paper planned intention-to-treat, which 362 analysed of 432 randomised does not match.
Outcome. Self-reported in an internet running log, not clinically diagnosed; injuries in the preconditioning month itself are not described in the abstract.
What was tested. A light, unsupervised walk-and-hop programme for 4 weeks before a short-event beginner programme; not strength training, not a longer runway, not a cyclist or a returner.
Funding and interests: ZonMw grant 75020009 (design paper); no competing interests declared in the design paper; the trial report's statements were not read.
How this entry was checked and changed
Why this level

(from the abstract) — a randomised trial of this size would be A under the rubric; held at A- because the trial report was read from the abstract only (its full text is not open), 70 of the 432 randomised runners are missing from the analysed counts, injuries were self-reported in an online log and not clinically diagnosed, the observed injury rate was well under half the planning assumption, and the review that pools it rated it at high overall risk of bias (all its trials but two were rated high, REF-1244)