Training load, strength training, and perceived recovery in relation to injury occurrence among competitive middle- and long-distance runners: a secondary analysis of publicly available longitudinal training logs

Background Running-related injuries interrupt training and physical-activity participation. Open athlete-monitoring logs allow transparent secondary analyses of associations between short-term training markers and injury labels, but the released competitive-runner dataset was constructed under specific eligibility rules that shape the estimand. Objective We estimated associations of preceding-week external load, logged strength-session exposure, and perceived recovery/exertion with injury labels in the publicly released week-approach event sample of competitive middle- and long-distance runners. Methods This study is a secondary analysis of the DataverseNL week-approach file (doi: 10.34894/UWU9PV ; 74 athletes; 42,798 selected event rows; 575 injury-labeled rows). Each row is an event day with week-0 exposures aggregated over the 7 days before that day. Upstream selection by the data creators required three injury-free weeks before injury-labeled events, 3 weeks of full fitness before and after healthy events, removal of incomplete/anomalous records, and exclusion of further injury flags within 3 weeks of an index injury. Binomial GEE (logit link; exchangeable working correlation; athlete clustering; robust sandwich variance) estimated population-averaged ORs per 1 SD for continuous predictors. Results Descriptively, injury-labeled rows showed higher preceding-week high-intensity distance and perceived exertion and fewer rest days than healthy-labeled rows. In the primary adjusted GEE model, high-intensity distance (adjusted OR = 1.32, 95% CI 1.18–1.48) and perceived exertion (adjusted OR = 1.35, 95% CI 1.10–1.65) were positively associated with injury labels, rest days were inversely associated (adjusted OR = 0.74, 95% CI 0.66–0.85), and strength-session count was not independently associated (adjusted OR = 1.08, 95% CI 0.98–1.18). When strength was coded as any vs. none, the adjusted association was positive (adjusted OR = 1.33, 95% CI 1.06–1.68). Conclusion Within this selected event sample, preceding-week high-intensity distance and perceived exertion were associated with higher injury-label odds and rest days with lower odds. Logged strength-session counts were coding-sensitive and do not estimate the preventive efficacy of structured strength programs. Findings should not be interpreted as cohort injury incidence or as transportable intervention effects for community runners.