Occupational nonfatal-injury comparisons

The Bureau of Labor Statistics Survey of Occupational Injuries and Illnesses (SOII) supplies private-industry occupation rates for 2023–2024. DART means days away from work, restricted duty, or job transfer. Rates are annualized per 10,000 full-time-equivalent workers.

The 610-task bundle is mapped using posting-derived occupation-task profiles and O*NET assignments. Posting shares divide design-weighted bundle-task incidences by all retained task incidences; O*NET shares count distinct assigned task IDs. Both DWA shares count distinct activities. Individual work-context ratings are separate measures, not fitted combinations of the task bundle.

The SI Table S13 comparison uses the same occupations for every measure within each outcome. exposure_common_sample.csv provides those correlations and sample sizes. The additional task-only comparisons use broader samples. Event-specific comparisons appear in soii_category_comparisons.csv; their sample sizes differ because published rates are suppressed for different occupations. Categories and their parent totals overlap and cannot be summed.

These are descriptive Spearman correlations. Violent acts have negative correlations with both task mappings; some small-sample categories favor O*NET. A bundle derived from incident activities need not measure every hazard equally. Correlations neither validate every task assignment nor establish individual risk or causal effects.

SOII depends on employer reporting and does not cover all injuries or workers. The OSHA corpus includes fatal and nonfatal incidents. Posting and outcome periods differ; the comparison is not a temporal holdout. Missing and suppressed rates remain unavailable.

survey_occupation_profiles.csv supplies occupation-level injury categories and duration information for the navigator. Duration is conditional on a reported case and is not a fatality rate or general clinical severity measure.

Run reproduce_injury_correlations.py with the released occupation-level scores to recalculate correlations. Reconstructing the posting profiles requires authorized inputs. Official sources:
https://www.bls.gov/iif/nonfatal-injuries-and-illnesses-tables.htm
https://www.bls.gov/opub/hom/soii/calculation.htm