Wage Workers’ Injury Patterns by Workplace Size and Industry across ICD-10 Categories: A Nationwide Cross-Sectional Study in Korea

Background Occupational injury surveillance typically depends on cases recognized as work-related, whereas health insurance claims capture treated injuries without prior attribution to work. Whether associations between claims-based injury prevalence and establishment size or industry differ across diagnostic categories and presumed mechanisms is not well established. Methods In this nationwide cross-sectional study, we analyzed National Health Insurance Service claims linked to establishment data for 11,212,512 Korean wage workers. Cases required ≥3 claims for the same three-character ICD-10 category in 2021. For all eligible categories, standardized prevalence and adjusted odds ratios were estimated by establishment size and industry; a representative subset was grouped by presumed mechanism after estimation. Results For ≥1,000 versus 5–29 workers, mechanical-group categories had lower odds (adjusted ORs, 0.33–0.72), including wrist/hand crushing injury (0.33) and external-eye foreign body (0.36). Ergonomic-group categories had slightly higher odds (1.15–1.17), but the corresponding prevalence gradients were absent when cases required ≥1 inpatient day for the category. The biomechanical comparator was null (1.00), and wrist/hand burns had the highest odds in Hotels and Restaurants versus Manufacturing (2.08). Conclusion Analysis at the category level identified patterns that diagnostic blocks obscured. Although claims cannot establish work-relatedness, they may help prioritize occupational investigation in smaller establishments and selected sectors. The ergonomic associations require confirmation. Linking claims with occupational exposure, injury-circumstance, and compensation data could identify work-related injuries absent from compensation records, complement official statistics, and inform policies addressing healthcare-access differences across worker and workplace characteristics.