Assessing the relationship between training load and injury in ultramarathon runners: a novel approach using Generalised Additive Models.

Abstract

Background: Ultramarathon running presents significant injury risks, and monitoring training loads may identify risk factors for injury. Injury surveillance studies are required to better assess injury prevalence and its relationship to training loads.

Objectives: To determine the incidence and nature of running-related injuries and associated training loads in runners 12 weeks before and two weeks after the 2018 Comrades ultramarathon.

Methods: One hundred and six participants were recruited. Their weekly injury and training load data (distance, duration, frequency and acute-chronic workload ratio) were obtained retrospectively over 14 weeks. The relationship between training load variables and injury risk was modelled using Generalised Additive Models.

Results: The running-related injury incidence was 8/1000 hours. The overall injury proportion was 40%. The commonly injured structures were muscles (47%) followed by tendons (24%). Commonly reported anatomical areas of injury were the knee (26%) and hip (19%). Lower training load distance in the 12 weeks leading up to the race was linked to a higher risk of injury (p=0.02), primarily occurring during or after the race. Weekly training frequency and injury risk showed a significant heterogeneous relationship (p=0.02). The effect of the acute to chronic workload ratio on injury risk was minimal (p=0.3).

Conclusion: Lower training loads were associated with a higher risk for injury, and the frequency of running training per week influenced injury risk. Insufficient training may not prepare the runners for the demands of the ultradistance race. Sudden changes in training load (evident in the acute training load measurements) appeared to have a minimal effect on injury risk. The non-linear relationship between several training load variables and injury risk can successfully be modelled using Generalised Additive Models, which may improve the accuracy of injury prediction modelling in ultramarathon runners.