A study published in PLOS ONE on 19 August has examined how training habits relate to injury among performance-oriented parkrunners. Researchers at Loughborough University, led by Han Wu with Katherine Brooke-Wavell, Clare Stevinson, Hui Fang, Keyi Zhong and Richard Blagrove, surveyed 1,203 runners and analysed the 570 who met the study's definition of performance-oriented.
Injuries were common. About 73% of participants reported at least one lower limb injury in the previous 12 months, and around 75% had missed a parkrun because of a physical problem. The survey covered personal characteristics, training habits, strength and conditioning activities and injuries, and the team combined conventional statistical tests with eight machine learning classifiers.
The most striking result concerns strength work. Resistance training and stretching or yoga were positively associated with injury, but those links disappeared once the authors excluded people who had started strength and conditioning to manage an earlier injury. Running frequency and weekly running duration showed no association with injury. The authors concluded that strength and conditioning did not appear to be protective in this group.
That finding sits awkwardly beside recent meta-analyses, including one covered on this site that reported a large reduction in injury risk from strength training. The designs differ, however. This was a retrospective, cross-sectional survey, so causation cannot be established, and runners who were already hurt may have taken up strength work as rehabilitation. The authors also note that the sample may lean towards injured runners and that training data were self-reported typical values.
The machine learning models showed only moderate predictive ability, with average AUC values between 0.620 and 0.694, and the number of weekly interval sessions had a non-linear relationship with injury. One practical signal did emerge: following a plan prescribed by a qualified running coach was linked to fewer injuries, while following online plans or having no plan was linked to more. The authors call for prospective studies to test these associations.