A newly published time-series analysis of distance runners has revisited one of the more contested ideas in endurance training: that injury risk can be predicted from the relationship between recent training load and the load a runner is accustomed to. The study drew on longitudinal data from 74 distance runners comprising 42,766 daily observations, a dataset large enough in the time dimension to model day-to-day sequences rather than the weekly averages most earlier work relied on.
That distinction matters more than it sounds. The acute-chronic workload ratio, which compares a rolling seven-day load against a rolling 28-day load, became popular partly because it is easy to calculate and easy to display on a dashboard. Its statistical foundations have been attacked steadily since 2019, largely on the grounds that the ratio bundles two quantities that behave differently, and that dividing one by the other discards information about both. Analysing days rather than weeks is one of the few ways to test whether the underlying idea survives once the arithmetic shortcut is removed.
The broader literature this study joins is not short of caution. Work on ultramarathon runners using generalised additive models has found that the relationship between load and injury is non-linear and varies substantially between individuals, which is precisely the pattern that a single population-level threshold cannot capture. Research on half marathon preparation has similarly found dose-response relationships between changes in weekly distance and injury, but with effect sizes that make individual prediction unreliable even where the group-level association is real.
For runners, the practical reading is narrower than the headlines usually allow. Large, abrupt increases in training load remain associated with elevated injury risk across almost every dataset examined, and that finding is stable enough to act on. What does not follow is that a specific ratio value marks a safe boundary, or that staying below it confers protection. Load is one input among several - sleep, surface, footwear, prior injury, and the accumulated damage a runner brings into a training block all appear in the same models with comparable weight.
The methodological direction of travel is toward individual-level modelling, and that is where the practical limitation sits. Studies of this kind require months of consistent daily data from each participant, which is expensive to collect and easy to corrupt, and the runners willing to log everything for a year are not a random sample of runners. Until the datasets grow, the honest summary is that training load is one of the better-evidenced modifiable risk factors in running, and one of the worst-calibrated.