OSHA Injury Data: An Opportunity for Improving Work Injury Prevention

Abstract

Millions of US workers are seriously injured on the job annually. These injuries have a significant and deleterious impact on injured workers, their families, and their communities. The limitations of the historical work injury surveillance systems have constrained research into the distribution and determinants of work injuries and efforts to improve allocation of limited injury prevention resources. Most work injury data sets suffer from significant limitations and fail to include a sizable proportion of work injuries.

In recent years, the Occupational Safety and Health Administration has begun to collect and make available to the public more detailed data on work injuries at thousands of high hazard establishments. These data sets provide the opportunity to greatly improve our work injury surveillance system. Researchers are now using these data to investigate and compare injury risk in industries and high-hazard firms where workers are at increased risk of musculoskeletal disorders. However, these rich data sets are underused.

Maintaining and facilitating access to accurate, current data can contribute to improved prevention of work-related injuries and deaths. (Am J Public Health. 2025;115(4):588–595. https://doi.org/10.2105/AJPH.2024.307934)

Workers in the United States face unacceptably high risks of work-related injury. Although most employers have the legal responsibility to provide workplaces free of recognized, serious hazards, millions of workers are injured on the job each year. Despite efforts by government agencies such as the Occupational Safety and Health Administration (OSHA), employers, labor, and public health professionals over the past decade, the rate of fatal injuries has not decreased.¹ This appears to also be true for the rate of nonfatal injuries, but these injuries are so significantly undercounted that the true number and rate of work injuries are unknown.²

The fact that we do not have even close to accurate estimates of the number and distribution of workplace injuries is a significant barrier to understanding and addressing the causes and distribution of injuries and deaths from work.³ With the exception of events such as mine collapses or explosions that kill several workers at once, workplace fatalities receive little public attention. If a workplace death receives any public notice at all, at most there will be a small media report of the event, often concluding with a statement that the police determined that the incident was an “accident” (meaning that it was unintentional) and that no criminal investigation would take place. The initial statements by employer representatives often attribute the incident to mistakes made by the worker: “human error.”

Although work injuries are rarely in the public’s consciousness, they often have a profound impact on injured workers, their families, and their communities. Beyond the human costs of injuries, the financial costs are subsidized by injured workers, their families, and other taxpayer-supported components of the social safety net.^4–6 For many injured workers and their families, a workplace injury creates a trap in which they are less able to save for the future or to make investments in skills and education that provide the opportunity for advancement, contributing to income inequality and poverty. Many injured workers, even those awarded workers’ compensation, end up disabled and receiving often meager Social Security Disability Insurance benefits, resulting in taxpayer subsidization of high-hazard employers.^7,8

A recent consensus study report issued by the National Academy of Sciences, Engineering, and Medicine recommended prioritizing and strengthening work injury and illness surveillance. The conclusion of the report was that the nation clearly needs a smarter occupational safety and health surveillance system for the 21st century.²

CURRENT INJURY SURVEILLANCE SYSTEMS

The Bureau of Labor Statistics (BLS) operates the 2 leading annual national work injury surveillance systems, the Census of Fatal Occupational Injuries (CFOI) and the Survey of Occupational Injuries and Illnesses (SOII). The BLS compiles the CFOI from many data sources and requires 3 pieces of confirmatory data for inclusion of a single case. The CFOI includes fatal injuries that employers under OSHA’s jurisdiction are required to report, as well as fatalities that occur among employees of firms covered by other agencies such as the Mine Safety and Health Administration and workers who are self-employed or whose employer has no regulatory coverage such as some state and local government agencies.⁹ During 2023, the most recent year for which statistics are available, the BLS reported that about 5300 workers were fatally injured on the job.¹

The SOII is, by definition, a survey and not a census. Thus, the SOII can provide an estimate of the annual number and rate of injuries and illnesses derived from data recorded on forms that OSHA requires employers to maintain to track workplace injuries and illnesses. To create the SOII, the BLS first identifies a stratified sample from across all industries that considers industry, ownership, and establishment size. The agency then instructs these employers to maintain and submit the relevant data after the end of the calendar year. Because OSHA does not require small establishments (10 or fewer employees), firms covered by the Mine Safety and Health Administration and other regulatory agencies, some public sector agencies and establishments, and establishments in certain low-hazard industries to maintain OSHA logs, the BLS requests a sample of establishments in these categories to maintain the logs and send the content to the agency.

In addition, the BLS codes and checks the accuracy of injury data, reporting injury numbers and rates by industry, occupation, state, establishment size, and nature and severity of the injury. Using these employer reports, the BLS estimated that, among workers employed by private sector employers in 2023, there were 2.4 million serious injuries (a rate of 2.2 cases per 100 full-time-equivalent workers) that required treatment beyond first aid; as discussed subsequently, however, this is likely a significant underestimate given that many work injuries are not recorded by employers.¹⁰ The nonmandatory injury reports of public sector and small establishments also result in significant underestimates.¹¹

To protect the confidentiality of employers that submit data, the BLS restricts researchers’ access to the CFOI and SOII data sets. Researchers must work through and closely with the BLS to access the data, and analyses can be performed only at the BLS or other tightly controlled facilities. In their publications, researchers may not reveal the identity of the employer or establishment. As a result, BLS data are useful in understanding injury incidence rates within or across industries over time but are of more limited value in conducting research on injury causation or providing information that enables microtargeting of prevention resources.

Workers’ compensation cases are another source of data on injuries, although these data are often difficult to compare because they are not collected and reported consistently across the more than 50 US workers’ compensation systems. To date, attempts to standardize and provide comprehensive comparative workers’ compensation reporting statistics on the types and causes of injuries across the United States have not been successful. Although these data are often used to estimate the number and rate of worker injuries, statistics based on injury cases that are filed or compensated also provide an incomplete picture because a sizable proportion of injured workers do not apply for compensation and their injuries are never registered in that system.^12–14

Finally, although OSHA’s requirement is for recording both injuries and illnesses and workers’ compensation insurance programs are meant to provide benefits for employees with work-related illnesses, it is widely recognized that both the recording and compensation of work illnesses are particularly incomplete.^2,15,16 Chronic disease may occur long after exposure ends, and claims are often denied by employers even if the disease manifests while the worker is employed. Musculoskeletal disorders are recorded infrequently, as are cases of hearing loss and chronic work-related asthma. Historically, employers have rarely noted the occurrence of illnesses other than the immediate effects of exposures to the skin or acute toxic inhalational injuries.

This changed somewhat during the COVID-19 pandemic. OSHA requires employers to record COVID-19 cases if the employee who had the disease while on the job had frequent, close contact with the general public in a locality with ongoing community transmission and there was no alternative explanation for the disease.¹⁷ It is clear that recording was inconsistent and that many work-related COVID-19 cases were not recorded. For these reasons, we focus here only on work injuries.

USING DATA TO PREVENT FUTURE WORK INJURIES

In contrast to the human, social, and financial impact of this large number of work injuries, the resources devoted to researching the causation and prevention of work injuries are modest. Primary responsibility for conducting and supporting workplace injury prevention research resides with the National Institute for Occupational Safety and Health (NIOSH), organizationally located within the Centers for Disease Control and Prevention (CDC). Only approximately 4% of the CDC budget is directed to NIOSH, whose overall budget in 2023 of $362.8 million was about $2.16 per worker in the nation’s civilian workforce. Within that budget, NIOSH funds intramural field and laboratory research as well as workplace injury surveillance, including supporting 23 states to conduct state-based targeted work injury and illness surveillance.¹⁸

Data collection varies widely from state to state, access to relevant statistical information in many states is challenging for outside researchers, and the completeness and accuracy of the data are rarely evaluated. Insurance companies that provide liability and compensation policies for employers often collect and analyze data relevant to workplace injuries. Access to these data, for example those compiled by the National Council on Compensation Insurance, is generally restricted to member organizations and is not easily available to independent researchers.

The Council of State and Territorial Epidemiologists has established and disseminated “occupational health indicators” to assist and guide state public health agencies that choose to track workplace injury and health information.¹⁹ Tracking and reporting of these indicators vary widely among states.

INJURY DATA AS AN UNDERUSED RESOURCE

Since its beginning in 1971, OSHA has required employers to report fatal injuries to the agency. Recently, OSHA has begun to collect and make public additional data that may be useful in conducting studies that can contribute to injury prevention. Although OSHA’s data are primarily self-reported by employers and have some of the limitations described earlier, they are much more extensive and easier to access than data from the BLS or state workers’ compensation systems.

In the sections to follow, we discuss 3 types of injury data available from OSHA covering fatal injuries, severe injuries (amputations and hospitalizations), and injuries recorded on OSHA logs and provide examples of how they have been used. All of these data are amenable to analysis by researchers using Excel or other spreadsheet programs.

OSHA Fatality Data

Employers covered by OSHA are required to report workplace fatalities to the agency within 8 hours of the event. The agency opens a case file for each report and in many instances conducts an inspection as well. Most work fatalities reported to OSHA are caused by traumatic injuries; the data set also includes fatalities related to extreme heat, chemical releases, and other nontraumatic causes of death. Employers are not required to report a fatality if it was caused by a motor vehicle crash unless it occurred in a construction work zone. However, employers are required to record fatal injuries on their injury log if they are among the employers required to maintain such a log. On its Web site, OSHA lists more than 7500 fatalities reported to the agency since 2017, linking each to the record of the inspection if one was conducted.²⁰ Additional information for many fatal and severe injuries can be obtained from OSHA inspections of the establishments where the worker was hurt. Reports of fatalities that occurred between 2009 and 2017 are archived elsewhere on the OSHA Web site.²¹

Researchers have used these data, often in combination with case data from other sources, to examine risk factors for high-hazard industries such as oil and gas extraction,²² tree felling, and arboricultural operations^23,24 or the impact of exposure to specific hazards such as extreme heat.²⁵ These data, although very useful, do not include deaths of workers whose employers are under the safety and health authority of other agencies or who are not covered by OSHA such as workers employed by certain state and local government agencies, by independent contractors, or through platforms such as Uber or TaskRabbit. Injuries among temporary workers employed by staffing agencies but supervised by the host employer are required by OSHA to be combined with injuries among those who are “direct hires” by the host employer; thus, injury rates among these workers may be obscured.

OSHA’s Severe Injury Reporting Program

Since 2015, employers have been required to report amputations and inpatient hospitalizations to OSHA within 24 hours of occurrence. Previously, employers were required to report nonfatal events only when 3 or more workers were hospitalized. Under its Severe Injury Reporting program, the agency received about 75 000 reports between 2015 and 2021 from employers in states under the authority of federal OSHA, covering about half of the nation’s workforce. Less than half of the reports resulted in an OSHA inspection. In most of the other cases, OSHA instructed the employer to conduct its own investigation into the incident and share its findings with the agency. The employer is required to provide an abatement certification signed by an official of the company that details the abatement measures that have been implemented to correct the conditions that caused the injury.²⁶ Complete data from these reports, including descriptions of the incident and the name and address of the establishment where it occurred, are available for downloading and analysis.²⁷

Many eligible hospitalizations and amputations are not reported to OSHA. One study of reports of worker hospitalizations in Michigan showed an overall compliance rate of 45.1%, with significant variability. For example, employers in manufacturing reported 64.7% of cases, while construction employers reported 38.2% and employers in agriculture, forestry, fishing, and hunting reported 8.2%.²⁸

These data have been useful to OSHA in identifying industries that had not previously been prioritized. For example, the SOII reports that supermarkets generally have low injury rates, but after OSHA received numerous reports of amputations among deli slicers and meat counter workers, the agency increased its attention to preventing those events.²⁹

Severe Injury Reporting program data have been analyzed for time trends and seasonality and to examine injury causation in specific industries and firms³⁰ and compare amputation and hospitalization rates among different employers and industries. For example, using the data, investigative journalists with NBC News Digital identified the United Parcel Service (UPS) as employing numerous workers who were hospitalized after exposure to extreme heat. These articles were important in encouraging the Teamsters union to make heat an issue in bargaining with UPS, resulting in UPS committing to purchasing air-conditioned trucks.³¹

Academic scientists have also used Severe Injury Reporting program data in studies on the epidemiology of specific hazards or occupations; for example, data have been collected on laborers,³² oil and gas extraction workers,³³ and operators of skid steers.³⁴ Recently, OSHA has developed a dashboard to facilitate downloading and analyzing data from the program.³⁵

Injury Data From OSHA’s Injury Tracking Application

OSHA requires most establishments with more than 10 employees to keep a record (often called the OSHA log) of serious work-related injuries and illnesses. Certain lower risk industries are exempted, and large categories of workers, including independent contractors, many state and county employees, and those employed in the gig economy, are not within OSHA’s jurisdiction. Workers employed in mines are covered by the Mine Safety and Health Administration, which requires all covered employers to report injuries, fatalities, and illnesses to the agency. Operator injury reporting data and workplace safety violations are publicly available in searchable databases maintained by that agency.

The OSHA reporting requirement is establishment or workplace based rather than employer based. Although many larger employers track injuries centrally, they must maintain a record of injuries for each establishment. These data serve as the primary basis for BLS SOII estimates.

One of the objectives of this recording requirement is to help OSHA, employers, workers, and unions prevent future injuries in each establishment. OSHA inspectors examine the logs at the start of an inspection because they identify the locations of some of the most serious hazards, ones that have already injured workers.

OSHA issued a regulation in 2016 requiring certain employers with 20 or more employees in higher risk industries to electronically report the establishment-specific summary data to the agency, along with the number of hours worked at the establishment, enabling injury rate (number of injuries per 100 full-time-equivalent employees) calculation. The agency also developed the Injury Tracking Application to facilitate employers reporting data and making the data available to the public in electronic format.³⁶

OSHA’s primary objective in making these data publicly available derived from studies done in the behavioral sciences revealing that public disclosure of some types of data can cause behavior change in the party making the disclosure.^37,38 For example, under the Environmental Protection Agency’s Toxic Release Inventory, some establishments are required to report the amounts of toxic chemicals released each year. Public disclosure of these data has contributed to significant reductions in releases.³⁹ Similarly, posting of restaurant letter grades given by public health inspectors has contributed to improved sanitary practices and reductions in food-borne illnesses.⁴⁰

There are several mechanisms through which public reporting on injury rates could lead to increased injury prevention without OSHA conducting additional inspections:

• •Lower injury rates provide a comparative advantage for employers trying to hire high-skill employees in competitive markets.
• •Employers and workers could compare establishments’ safety record with other establishments in their industry, potentially stimulating modification of safety practices and policies.
• •Researchers may be able to identify previously unrecognized patterns of injuries across establishments where workers are exposed to similar hazards and evaluate the effectiveness of workplace safety activities in reducing the number or severity of work injuries or illnesses.
• •Workplace safety professionals could identify establishments with high injury rates, which could benefit from their services.
• •Low injury rates might improve employers’ reputation among investors and consumers.

OSHA developed the Injury Tracking Application to facilitate electronic collection of these data and to enable employers to upload data files. Establishment-specific summary injury data are available for years subsequent to 2016, although they do not represent the complete set of data OSHA should have received and the agency does not check the data for accuracy or completeness before posting them.³⁶ The Government Accountability Office reported that 42% and 46% of establishments sent the required data to OSHA in 2017 and 2018, respectively.⁴¹

Although these data are useful in benchmarking overall safety performance and helping to identify facilities with particularly effective or ineffective injury prevention programs, there is compelling evidence that statistics derived from employer self-reports significantly underestimate the actual burden of injuries occurring among the nation’s workers.^13–15 The BLS commissioned a series of studies by academic researchers to estimate the percentage of work-related injuries reported by employers. All of these studies showed that a significant proportion of work injuries (30% to more than 60%) were not included in employer reports, with underreporting varying by employer and type of injury.¹³

To date, summary injury data have been used by researchers, journalists, and public health activists to better understand the relative risk of injuries in several high-hazard industries. Studies incorporating these data have shown, for example, that workers at Amazon fulfillment centers are at greater risk of injury than workers at comparable facilities operated by other firms^42,43 and that injury risk increases significantly during high-volume events such as Prime Day and Cyber Monday.⁴⁴ Researchers have also used information collected in the CFOI, SOII, and Injury Tracking Application data sets to examine injury trends among parcel delivery drivers.⁴⁵ Elevated worker injury rates gleaned from Injury Tracking Application data, as well as the firm’s failure to report injuries in some years, were central in a recent report by Reuters on safety hazards at SpaceX⁴⁶ and a Wall Street Journal investigation into injuries suffered by aviation industry ground workers who refuel planes, handle baggage, and help move aircraft around the tarmac.⁴⁷

Because many employers who are required to send their data to OSHA fail to do so, examining the employers that report their data to the agency and the ones that do not can also provide useful insights. Researchers have looked at the impact of unionization on establishments reporting injury data to OSHA, finding that unionization increases injury reporting.⁴⁸

The availability of establishment-specific injury rates and granular information on large numbers of establishments provides researchers with the ability to conduct more in-depth analyses of a wide range of topics. It might be productive, for example, to link the Injury Tracking Application data set with other establishment- or corporation-specific data sets that focus on union status, management structure and characteristics, amount of overtime work, corporate financial health, ownership (public vs private), workers’ compensation insurance carrier, chemical emissions, Environmental Protection Agency violations, or other variables.

To further increase the value of these data in injury prevention, it would be worthwhile for OSHA to prioritize issuing citations and fines to employers that do not report and publicizing their names. Although the monetary penalties for not reporting are relatively low, this may encourage some firms to report to avoid reputational damage.

AVAILABILITY OF INJURY CASE DATA

OSHA now requires establishments in higher hazard industries with 100 or more workers to electronically submit case data on each injury.⁴⁹ Data from almost 900 000 injuries and illnesses recorded and reported by more than 350 000 establishments in 2023 have been posted by OSHA. These data include the date and time of the incident (if known) and text fields that provide additional information about the injury or illness, the parts of the body affected, and the objects or substances that directly injured or made the person ill. OSHA has redacted certain information from the text fields to protect the confidentiality of injured workers.³⁶

For research purposes, the data in the deidentified text fields will become much more useful when they are coded into the variables commonly used in injury categories. The BLS has developed an automated coding system for converting text descriptions of injuries into coding text in the standard categories used in reporting injury data (e.g., nature of injury, part of body injured).⁵⁰ OSHA has plans to provide tools to make the data easier to search and download and is working with the BLS to develop a system to code the Injury Tracking Application data. NIOSH has developed an automated system to code records for industries and occupations.^51,52

Establishment-specific data also have significant limitations. Injury rates are lagging indicators and provide limited information about the presence of hazards that could contribute to future injuries. In addition, most injuries reported by employers are caused by ergonomic hazards or are types of personal injuries sometimes called “slips, trips, and falls.” Injuries associated with process safety hazards (those that occur when there is an explosion, fire, or chemical release) are far less frequent events. Low injury rates are sometimes misinterpreted to suggest that process safety hazards are well controlled. Famously, BP executives gave an award for low injury rates to the team operating the Deepwater Horizon drilling rig the day before the rig exploded, killing 11 workers, injuring 17, and releasing the millions of barrels of oil that devastated the ecosystem of much of the Gulf of Mexico.⁵³

Even with these caveats, the existence and accessibility of the new OSHA data sets provide an important opportunity to improve our understanding and prevention of work injuries. Establishment-specific data could be cross-linked with information from other data sets to compare the characteristics of establishments that that do similar work but have dissimilar rates of different types of injuries. Employers can now benchmark their injury rates with other firms in their industry. Establishment-specific data can also be used to improve the targeting of prevention efforts. These efforts would be strengthened if OSHA is able to increase compliance with its regulations on recording and reporting work injuries and to make efforts to improve the accuracy and completeness of data.

In summary, when identifying areas in which additional research could have a significant public health impact, work injuries are low-hanging fruit. We believe that increased research into the distribution and determinants of work injuries, especially using data sets that OSHA now makes publicly available and linking them with other databases, can make an important contribution to increased prevention of work-related injuries and fatalities. Maintaining and promoting access to these data are critical.