Higher cumulative training loads measured by wearable inertial sensors may be associated with musculoskeletal injury in athletes, although the current evidence could be insufficient to establish causality, predictive performance, or injury-prevention thresholds, according to a systematic review and meta-analysis.
Investigators searched 6 databases from inception through June 2026 for longitudinal observational studies evaluating associations between accelerometer-derived training-load measures and musculoskeletal injury. They included 11 studies involving 508 athletes across 6 sports cohorts and 1 military cohort. The studies used wearable inertial sensors to quantify external training load through cumulative load magnitude and acute-to-chronic workload ratio (ACWR).
The investigators evaluated the association between inertial-sensor-derived training load and musculoskeletal injury across all included studies as well as cumulative load magnitude and acute-to-chronic workload ratio (ACWR) separately ,and sensitivity analyses by effect measure and study characteristics. Because the studies reported different effect measures, the investigators characterized their overall pooled analysis as exploratory rather than as an estimate of a common causal effect.
Across all 11 studies, higher sensor-derived training loads were associated with musculoskeletal injury, with an exploratory pooled effect estimate of 2.74. However, the investigators cautioned that combining different effect measures limited interpretation of that estimate and placed greater weight on analyses stratified by effect type.
Across 5 studies, the pooled effect estimate for cumulative load magnitude was 2.86, with no statistical heterogeneity detected. The small number of studies meant the absence of detected heterogeneity should not be interpreted as evidence that the underlying populations or methods were homogeneous.
In 5 studies, the pooled effect estimate was 2.64 for ACWR, but heterogeneity was substantial, and the association was not statistically significant under the conservative primary inference method. The investigators therefore characterized the evidence for ACWR as equivocal.
Analyses separated by effect type further demonstrated the uncertainty. The 3 studies reporting relative risks produced a pooled relative risk of 2.98, whereas the pooled result from 5 studies reporting odds ratios was not statistically significant. Several studies had injury proportions above 10%, which could cause odds ratios to diverge meaningfully from relative risks and complicate comparisons across studies.
Despite the more consistent results for cumulative load magnitude, the investigators cautioned against concluding that it was superior to ACWR. The 2 constructs were assessed in different populations and study designs rather than compared directly within the same cohorts, preventing a valid comparison of their relative performance.
The analysis had limitations. The 11 studies had a median sample size of 45 athletes, and none had a low overall risk of bias; 6 had some concerns and 5 had high risk. Confounding was generally inadequately controlled, some studies may have included overlapping cohorts, and differences in sensors, exposure definitions, injury definitions, and data-processing approaches limited comparability. Publication bias could not be excluded. All outcomes were rated very low certainty.
The review suggested that wearable inertial-sensor training-load measures may capture information associated with musculoskeletal injury, particularly for cumulative load magnitude, but the evidence did not establish that the measures can predict which athletes will be injured or that modifying training load will prevent injury. The investigators found insufficient evidence to recommend specific numerical thresholds for clinical or training decisions.
“The available evidence indicates a possible association between inertial-sensor-derived training-load metrics and [musculoskeletal] injury,” wrote lead study author Yang Zhang, of the School of Physical Education and Health at East China Jiaotong University, and colleagues.
The study authors reported no conflicts of interest.
Source: BMC Musculoskeletal Disorders
