OSHA provided data on all reports made relating to Rule 1904.39 for 6,448 workers affected by injury or illness occurring in Illinois between January 1, 2017 through December 31, 2023 including information on any related inspections-citations . OSHA defines an injury as "any wound or damage to the body resulting from an event in the work environment". OSHA does not have a singular definition for illness, but include all conditions caused by, contributed to, or significantly aggravated by work (e.g. myocardial infarction, heat related illness). All reports were included even if they were reported more than 24 hours following the incident. We excluded workers injured outside of Illinois because our hospital dataset may not capture them (n=48) and incidents occurring prior to 2017 where the inspection occurred in 2017 (n=11). The final OSHA dataset of 6,389 workers included information about the injured employee (age, gender, employment type), the employer and injury location (address, industry), narrative descriptions provided by the employer and OSHA inspector regarding the hazards, activities and contributing causes related to the incident, and information on fatalities, hospitalizations, nature and type of injury or illness, body part affected, and treating hospital.

The hospital outpatient database includes all patients treated in emergency departments (ED) for less than 24 hours who were not admitted to the hospital. The hospital inpatient database includes all admitted patients treated for 24 hours or more in Illinois hospitals for any medical reason. Both datasets include information on patient demographics (age, race, sex), clinical outcomes (diagnoses, hospital procedures, and discharge status), and financial information (hospital charges and payer source).

To identify hospitalized workers that employers are required to report to OSHA, we included all acute injuries or illnesses admitted to a hospital and amputations/enucleations treated in the ED between January 1, 2017 through December 31, 2023 that had a primary workers’ compensation payer. Per the reporting exemption, we excluded all injuries from motor vehicle crashes on roadways or public transportation (N=639), but included injuries occurring while (1) working on or near vehicles, (2) working with special vehicles used on industrial premises (e.g., forklifts), (3) off-road vehicles (non-public roadways), (4) watercraft, (5) cable cars, ski lifts, (7) construction vehicles, (8) agricultural vehicles, and (9) collisions in construction zones. In addition, we excluded cases that involved conditions that may not have occurred within 24 hours of the hospitalization. We excluded all ED visits and admissions coded for elective procedures. Elective cases are non-emergent conditions that allow the patient adequate time to schedule clinical procedures. We also excluded all secondary encounters following the initial treatment for an injury. Finally, we excluded all inpatient emergent hospitalizations involving renal failure (N=166), vertebral conditions (e.g., spondylopathies; n=222) and hernias (N=50; ICD-10 coding does not distinguish between direct and indirect hernias) in the absence of any other medical emergencies or traumatic injuries. These three conditions were consistently identified in the absence of a traumatic injury or well defined acute medical illness, and we are unable to determine if they occurred following an acute event. There were 11,795 work-related hospitalizations that met the OSHA reporting requirement.

A detailed description of the linkage methods is provided in the online supplemental material. Because neither dataset has personal identifiers, we used a multi-stage data linkage procedure to link the OSHA data with the hospital data.^20,21 In Illinois, there was an average of 2.5 reports per day made to OSHA and an average of 4.0 reportable cases per day treated in all Illinois hospitals. By linking within the incident day, we are able to reduce the overall false match rate. We standardized the format of all the linkage variables in the datasets, ensuring that the variable types, lengths, and codes were equivalent. Next, we used a combination of SQL and SAS code to link the two data files on the following variables: gender, age, incident date, vital status (alive/dead), nature of the injury/illness, body part, cause of injury, and hospital name.

In the first linkage steps, 4,057 OSHA reports matched to the hospital data file of reportable cases (Online Supplemental Figure 1). However, past research demonstrates that employers frequently report cases that do not meet the reporting requirement,^10,22 so we utilized a second data file that included all emergent ED cases with a workers’ compensation payer, but excluded elective cases, amputations, enucleations and deaths. An additional 881 reports linked to these ED visits despite not meeting the OSHA reporting requirement. Finally, research also demonstrates that injured workers frequently use alternative insurance coverage instead of workers’ compensation.^6,10 To account for this, we utilized a third data file that included all (1) hospital admissions and (2) amputations/enucleations treated in the ED that did not have a workers’ compensation payer, but we excluded all elective cases and patients under 16 or over 65 years old. In this final linkage set, 230 hospitalizations without a workers’ compensation payer matched to the remaining OSHA reports.

We used hospital data for reporting information on injured workers because this information was collected directly from the worker or was based on clinical information. For all unlinked cases, we used the data provided on the OSHA report. To assess underreporting of severe injuries and illnesses to OSHA, we defined the reporting rate as shown in the formula below. We excluded from the reporting rate 881 cases reported to OSHA that did not meet OSHA reporting criteria, because they were not admitted to a hospital and did not involve amputations or enucleations.

We present reporting rates by total fatal and non-fatal cases, amputations, gender, age, OSHA area offices, Rural-Urban Continuum codes (RUCC),²³ and incidents occurring on weekends and on major U.S. holidays when administrative personnel responsible for reporting may not be working (New Years, Memorial Day, Independence Day, Labor Day, Thanksgiving and Christmas). We analyzed overall trends (increasing, decreasing) and changes in trend (joinpoints) of reporting rates (Joinpoint Regression Program v 4.9.1.0).²⁴

Because we did not have access to state vital records, we are unable to adequately evaluate the reporting rate for fatalities. For this reason, we exclude fatalities from the remaining descriptive analyses and modelling. We compare reported and unreported cases by worker demographics, injury location, incidents occurring on weekends, comorbidities of the worker,²⁵ injury severity as measured by the trauma mortality prediction model (TMPM),²⁶ course of hospital treatment and hospital charges. ICD-10 codes were used to characterize nature of injury/illness and cause of injury for all cases with hospital records, and the narratives from the OSHA reports were used for all unlinked cases. We used SAS software for statistical testing and modelling (v.9.4; Cary, NC). The chi-squared test was used for comparing categorical variables and student’s t-test for continuous variables.

Table 1 presents the distribution of fatal and non-fatal injuries and illnesses by OSHA reporting status. During the period of 2017-2023, a total of 6,389 injuries and illnesses were reported to the State and Federal OSHA plans in Illinois, of which 64.6% were made within 24 hours of the event and 91.5% were made during the first week. Reports were made via all modalities with 92.9% occurring online or by phone. Nearly all reports were made by the employer or employer representative (n=5,790 out of 6,389 reports; 90.6%) with the remainder made by an employee or employee representative (n=131; 2.1%), a government agency (n=310; 4.9%) or other source (n=158; 2.5%).

The cumulative reporting rate, including both fatal and non-fatal injuries and illnesses, was 41.6% (Table 2). The cumulative reporting rates increased between 2017-2020 from 40.3% to 47.9%, but there was a significant reversal in the trend in 2021 leading to a decline to 35.9% by 2023 (p=0.03). However, hospital data do not capture fatalities that occur at the scene or following discharge (within 30-days). Excluding fatalities, the cumulative reporting rate of non-fatal injuries was 39.7% with no significant change in trend over the seven years (Table 2). Similarly, there was no significant change in trend between 2017-2023 for non-fatal illnesses when excluding COVID cases, but the cumulative reporting rate was substantially lower (25.1%). When stratified by demographic characteristics, cumulative reporting rates were lower among females (vs males; p<0.01), older workers (35 years and up vs under 35 years; p<0.01), non-fatal cases occurring in urban regions (vs. suburban and rural; p<0.01), and injuries and illnesses occurring on weekends (vs. weekdays; p<0.01) and holidays (vs. non-holidays; p<0.01). There was no significant change in trend over the seven years for any of these subgroups.

There was a total of 1,433 deaths identified in the final linked dataset, of which 1,273 (88.8%) were reported to OSHA. In total, 809 deaths had reports made to OSHA but did not link to a hospital record, 464 deaths were reported to OSHA and linked to hospital records, and 160 deaths occurred in the hospital but were not reported to OSHA. Of the deaths identified in the hospital records, 220 (42.2%) did not have a workers’ compensation payer, but linked with an OSHA report.

Among the 160 fatalities that were not reported to OSHA (compared to all reported fatalities), these deaths disproportionately occurred on weekends (23.1% v. 16.6%; p<0.01) and 75.6% involved one the following medical emergencies (primary diagnoses only; independent counts): 43 acute myocardial infarctions and cardiac arrests, 28 cerebrovascular accidents, 7 pulmonary embolisms, 33 traumatic brain injuries, 4 severe internal injuries, and 6 exposures to toxic agents.

Of the 12,694 non-fatal injuries and illnesses, 8,871 (69.9%) were acute injuries, and 5,116 were reported to OSHA, but 881 (6.4%) of the reports involved cases that did not meet the reporting criteria (Table 1). Table 3 presents the demographic characteristics of the non-fatal injuries and illnesses. Compared to cases reported to OSHA, the hospital cases not reported to OSHA disproportionately were female (21.9% vs 15.4%; p<0.01), Non-Hispanic White (59.9% vs 51.5%; p<0.01), Illinois residents (91.3% vs 82.7%; p<0.01), occurred in urban counties (43.7% vs 34.1%; p<0.01), and occurred on weekends (19.9% vs 10.4%; p<0.01).

Table 4 presents the most common illnesses and traumatic injuries, comorbidities, course of hospital treatment and discharge status by OSHA reporting status. Compared to cases reported to OSHA, the hospital cases not reported to OSHA were disproportionately illnesses (34.4% vs 19.3%; p<0.01), had higher mean number of comorbidities (p<0.01) and a greater proportion were admitted to a hospital (83.8% vs 69.8%; p<0.01).

Table 5 presents the odds ratios from the final multivariable models. In the model evaluating factors associated with not reporting an injury or illness to OSHA, female workers, older workers, non-Hispanic White workers, workers with serious comorbidities, illnesses resulting in acute medical emergencies (as opposed to traumatic injuries), cases occurring in large urban counties, cases occurring on weekends and holidays, and injuries caused by animals were more likely to not be reported to OSHA. However, cases requiring surgical intervention or mechanical ventilation were more likely to be reported to OSHA.

In the second multivariable model (Table 5), factors associated with improperly reporting an injury or illness to OSHA that did not meet the reporting criteria included injuries and illnesses involving female workers, both Non-Hispanic White and Black workers (relative to Hispanic and Latino workers), illnesses, and those suffering open wounds, burns and sprains/strains. In addition, the odds of improperly reporting an injury or illness increased each year between 2017-2023 (aOR=1.14; CI95%: 1.10, 1.19). As the hospital charges increased, the odds of improperly reporting an injury or illness decreased (per $1,000 USD hospital charges; aOR=0.88; CI95%: 0.87, 0.89). Evaluation of the OSHA reports showed that 50.4% of the amputations reported by employers were in fact open wounds without loss of a part of a limb and were treated in the ED. In addition, 18.8% of cases reported as fractures by the employer were later diagnosed as sprains or strains.

Among the fatal and non-fatal reports to OSHA, OSHA initiated 2,376 inspections resulting in 2,603 violations for a total of $19,265,445 in penalties (after appeals). None of the non-fatal cases reported to OSHA that did not link to the hospital data had an inspection. The average final penalty was $16,466 for non-fatal incidents with a ratio of 0.28 violations per non-fatally injured worker (1,170 violations out of 4,235 non-fatal reports). We estimate that the failure to report the 7,578 non-fatal injuries and illnesses potentially obscured up to 2,122 violations (7,578*0.28).

This study has several limitations. First, some of the variables in the OSHA dataset had a high degree of missingness, particularly for age and gender, or had to be coded from narrative text. In addition, the information is primarily provided by persons without medical training who may not know the injured worker, which can contribute to reporting inaccuracies,²⁷ and false matches or poor linkage. However, the daily number of work-related cases meeting the reporting criteria is very small, which mitigated the risks introduced by data inaccuracies. After adjusting for cases expected to not link to hospital data (see online supplemental methods), the overall linkage rate was very high (95.6%). Second, we did not have access to state vital records to comprehensively assess reporting of fatalities. Despite missing most fatalities in the hospital data, we identified 160 deaths that met the reporting criteria but were not reported to OSHA. In fact, the combined number of deaths captured in the OSHA and hospital data for 2017-2023 was 1,433 (n=898 traumatic injuries) which is 23% higher than the 1,138 documented by the Census of Fatal Occupational Injuries during the same period (CFOI).²⁸ Even though the inclusion criteria for CFOI is more rigorous and restricted to injuries, this illustrates that similar OSHA underreporting studies evaluating fatalities should be done utilizing multiple data sources. Third, Illinois hospital records are missing information about the patient’s industry and occupation which has been associated with underreporting.^5,10,18 As a consequence, we are also unable to differentiate within hospital data between workers covered by the federal or state plans, or workers not covered by OSHA including the self-employed and those employed in mining, railroad, and energy sectors (comprise 0.5% of Illinois payroll employees annually).²⁹