Can training load ratios predict injury?
Acute:chronic workload ratio and injury: a 2026 meta-analysis of 16 studies (797 athletes) found g = 0.35, with very high heterogeneity and mixed evidence.
Across 16 studies of 797 team-sport athletes, a higher acute:chronic workload ratio (ACWR) was associated with more injury, with a pooled effect size of g = 0.35 (95% CI 0.16 to 0.54). The 2026 meta-analysis also found very high heterogeneity (I-squared 95.8%), and its authors concluded that the evidence does not support using the ratio as a stand-alone predictive model.
What did the meta-analysis of ACWR and injury find?
The ratio divides recent training load (the acute load, usually the last week) by the average load over a longer period (the chronic load).
Ding et al., Frontiers in Public Health 2026 searched four databases through October 2024. Of 41 eligible studies in football, rugby, basketball and volleyball, 16 could be pooled. Effect sizes were Hedges’ g, a standardised measure, so they do not translate into “times more likely”.
The association differed by injury type. It was g = 0.54 (95% CI 0.40 to 0.67) for contact injuries and 0.55 (0.33 to 0.78) for non-contact injuries, but g = 0.01 (minus 0.34 to 0.37) for time-loss injuries. The load metric, calculation method and time window did not significantly change the result.
Why is ACWR not a reliable predictor on its own?
The included studies were observational, so they cannot show that a spike in load caused an injury. The authors note reverse causation: early symptoms or subclinical tissue impairment may change training load before an injury is formally diagnosed. Pre-injury windows varied across studies, and confounding was inadequately controlled.
The evidence is therefore mixed: a small association overall, large unexplained variation, and no support for the ratio as a trigger on its own. How intensity is spread across training is a related question, covered in our post on how to improve your running.
The bottom line
- In 16 studies (797 athletes), elevated ACWR was associated with injury (g = 0.35, 95% CI 0.16 to 0.54), with I-squared of 95.8%.
- The association was significant for contact and non-contact injuries and not significant for time-loss injuries.
- The authors say current evidence does not support ACWR as a stand-alone causal or predictive model.
Frequently asked questions
- Does a high acute:chronic workload ratio cause injuries?
- That has not been shown. The 2026 meta-analysis pooled observational studies, which can show an association but not a cause. The authors flag reverse causation, because early symptoms may alter training load before an injury is diagnosed, along with inconsistent injury windows and inadequate control of confounding.
- Can the acute:chronic workload ratio predict injury on its own?
- The authors concluded that the evidence does not support that. Elevated ACWR may be associated with injury in specific contexts, but heterogeneity was very high and time-loss injuries showed no association. They suggest interpreting it as one indicator within individualised, multi-marker load-management systems.
Sources
- Ding L, Weldon A, Xu J, Malone S, Sampaio J, Zhang Z, Zhou Z, Liu X, Han S, Xu K, Wang Y, Yuan P, 2026. Acute:chronic workload ratio and load management for team sports: a multilevel meta-analysis. Frontiers in Public Health, volume 14, article 1896651.