This study is an examination of WC claim costs among career and volunteer firefighters in Ohio from 2001 to 2020. The claims were identified and characterized by their type, causes, and occupational tasks in a previously published paper that described the methods of claim identification in detail (Quinn et al., 2023). The study population for this paper was selected through a series of steps. First, all claims from the previous paper (Quinn et al., 2023) for 2001 to 2017 were included. In short, these cases were identified using the following OHBWC-specific occupational codes: 5111 (Supervisors: Firefighting and Fire Prevention Occupations), 5120 (Firefighting and Fire Prevention Occupations, Firefighters, Firemen), or 5122 (Fire Inspection and Fire Prevention Occupations). Following initial identification, claims outside of those occupational codes that included the text “fire” or “ff” in the occupation title were also included. Following this initial claim identification, the occupation title and injury narrative were manually reviewed by two independent researchers to verify that the claimants were indeed firefighters. Claims that were mutually agreed upon as not appearing to be firefighters were excluded. These same steps were followed to identify additional cases from 2018 to 2020 for the current study. Combining the 2001–2017 and 2018–2020 data identified in these processes resulted in an initial sample of n = 37,180 claims. A final step used the unique identifier for each injured firefighter included in the claims data to identify additional claims from 2001 to 2020 associated with previously identified firefighters that were not flagged in the previous steps. These cases were also manually reviewed by two independent researchers. This final step identified an additional n = 126 claims. Together, these steps resulted in a final analytic sample of n = 37,306 claims.

Both lost-time and medical-only claims were included in this analysis. Lost-time claims are defined as claims that involved indemnity payments for death, permanent disability or temporary disability (eight or more days away from work), while medical-only claims required seven or fewer days away or medical treatment only. As such, lost-time claims were generally considered to be of greater severity than medical-only, although that assumption may not always be valid. All compensable (not disallowed or dismissed) diagnoses as of October 4, 2024 were included in the current study.

The included claims and associated costs are summarized by a variety of descriptive variables including claimant demographics (sex and age), events (e.g., struck by objects or equipment, overexertion involving outside sources, falls on the same level), diagnoses (e.g., back sprain, lower extremity sprain, burn), and the task the claimant was performing.

A “task” variable was added to all claims that categorized each claim into one of the following eight task categories: (1) firefighting, (2) patient care, (3) responding to a motor vehicle incident (MVI), including patient care, (4) training tasks, (5) physical fitness/exercise, (6) worker involved in an MVI, (7) long-term or repeated chronic exposure to firefighting hazards (musculoskeletal pain, hearing loss, cancer, etc.), and (8) other/unclear. The coding method for data in the initial study has been previously described (Quinn et al., 2023). In short, the task variable was manually coded and adjudicated using an initial random sample (n = 2,000) of claims by two independent researchers based on the narrative text. These initial claims were then used as a training set to create a machine-learning algorithm to auto-code the task for the remaining claims; auto-coded tasks with < 95% prediction confidence (74% of the original claims) were reviewed and checked for accuracy by two independent researchers (Quinn et al., 2023). All new cases identified for the current study were manually reviewed and checked for accuracy by two independent researchers.

The event for each claim was defined using the Occupational Injury and Illness Classification System (OIICS) version 2.01 (BLS). These events were originally assigned a two-digit OIICS event or exposure code based on the incident narrative and the diagnosis of the claim using a previously used and described computer-assisted auto-coder developed by NIOSH (Bertke et al., 2016). This coding algorithm was improved in 2022 to use a Long Short-Term Memory (LSTM) deep learning algorithm (Meyers et al., 2022). While the previous paper (Quinn et al., 2023) used the older process to assign codes, the current analysis used the updated method. Two independent researchers conducted additional manual review on high-cost claims (≥95^th percentile) and those with the lowest auto-coder probabilities (bottom 7%) to reach positive predictive values of 95% and 85% for 1-digit and 2-digit event/exposure codes, respectively (Meyers et al., 2022).

Diagnosis group(s) were assigned to each claim based on the compensable International Classification of Diseases – Clinical Modification (Version 9 and 10) diagnosis codes. All diagnoses were considered equally. In other words, there was not a principal diagnosis assigned for each claim. There were 57 diagnosis groups, including various traumatic injuries and disease diagnoses (Supplemental Table 1), as described by Meyers et al. (Meyers et al., 2025) For claims with multiple diagnoses in the same diagnosis group, each diagnosis group was only counted once.

Claims were described using six disability status categories as of December 2022: permanent total (PTD), permanent partial (PPD), permanent partial and temporary total (PPTTD), temporary total (TTD) disability, medical only, and death (Utterback et al., 2014). Information on medical only claims were excluded from analyses of the disability status category but was presented elsewhere. The 88 death claims, accounting for a total cost of $55,349 K and a mean cost per claim of $629 K, were excluded from analyses of the disability status category. PTD pertains to cases where an employee is permanently and entirely unable to engage in any form of gainful employment due to a work-related injury or illness. This category typically encompasses severe injuries or debilitating medical conditions. Individuals with PTD usually receive ongoing financial compensation and medical benefits. PPD describes situations where a worker sustains a permanent impairment or loss of bodily function as a result of a work-related incident. Unlike PTD, individuals categorized under PPD may still possess the ability to perform some work; however, their earning capacity may be reduced due to the enduring effects of their injuries. PPTTD combines elements of both permanent partial and temporary total disability. Workers in this category have experienced an injury or illness resulting in partial permanent impairment but also necessitating a temporary period of complete disability during their recovery. TTD encompasses situations where an employee is entirely unable to work for a temporary period due to a work-related injury or illness. This condition is often of a short-term nature, with the expectation that the worker will recover and resume their regular job duties once their medical condition improves.

For all claims, costs extracted by OHBWC were current through the second quarter of 2024; costs included payments for medical care and indemnity (partial wage replacement for temporary disability, permanent disability, and/or death) as well as actuarial reserves for future anticipated costs. OHBWC offers a salary continuation program where employers can continue to pay employees their salary for lost-time claims during periods where the injured worker could have received TTD indemnity payments from OHBWC (Ohio Bureau of Workers Compensation Salary Continuation, 2024). Although these payments were by employers, OHBWC provided NIOSH with estimates of employer-provided TTD indemnity payments for employers in this program. There were 367 policies (out of 1,044 policies, 35%) that were part of the salary continuation program; there were 7,154 claims from these policies, with a mean cost of $12,342 per claim. To best represent the total cost of claims, these employer-paid indemnity costs were included in the cost analyses for this study. Some claims may have zero cost in the OHBWC database, which can occur if no medical treatment was provided (Ohio Bureau of Workers, 2024). For these reasons, per claim calculations were performed for total costs and total costs excluding zero cost claims (mean, median). While costs for claims, especially those highest in severity, are paid out over several years, inflation adjustments were not applied to cost data.

Descriptive analyses were used to calculate the frequency of claims by year, claimant demographic characteristics, task, events, and diagnosis groups. A ratio of the percentage of total costs over the percentage of total claims, which represents the proportion of costs relative to the number of claims, were calculated and presented. Therefore, a higher cost/claim ratio indicates a higher cost burden relative to the number of claims in that category. Descriptive analyses of costs and claim frequencies were stratified by type of claim (medical-only or lost-time) and claim severity (disability category).

During the study period, specifically in April of 2017, Ohio implemented a firefighter cancer presumption law (Michael Louis Palumbo, Jr. Act) (Palumbo, 2017). To explore possible early impact on the number of claims, a separate exploratory analysis was conducted for claims with any neoplasm diagnosis to describe annual trends in distribution (percent) of neoplasm claims: (1) between firefighters and non-firefighters, and (2) as a percentage of all OHBWC compensable diagnosis categories.

Table 2 provides a detailed comparison of claimant demographics and cost characteristics between lost-time and medical-only WC claims. Lost-time claims, indicating more severe injuries and illnesses, accounted for 13,573 with a total cost of $513 million, a mean cost of $37,776, and a median cost of $8,605. In contrast, medical-only claims, involving less severe injuries and illnesses that required seven or fewer days off work, totaled 23,733 with a considerably lower associated cost of $29 million, a mean cost of $1,235, and a median cost of $529. A consistent sex distribution was seen across both claim types, with males comprising most claimants (95% for lost-time and 93% for medical-only). By age group, those 25–34, 35–44, and 45–54 had the highest percentages of total lost-time and medical-only claims, although the order of these three percentages differed. While the highest total cost for medical-only claims was among those aged 35–44, the highest total cost for lost time claims was observed among those 45–54 years old ($206 million, 40%).

Task and event analyses illustrated the varying nature of firefighter tasks and related events. However, the distribution of claim frequency and costs across the types of tasks and events were largely similar across the two claim types (Table 2). For example, firefighting and patient care tasks were prominently represented across both claim types. Also, overexertion involving outside sources had the highest claim frequency and total cost for both lost-time and medical only claims.

Table 3 presents costs for the most common diagnosis groups, including fatal claims. Of the 37,306 claims, 26,389 (71%) had a single diagnoses and 10,839 (29%) had multiple diagnoses (up to 25 per claim). For this reason, claim and cost counts in this table are not mutually exclusive as some claims were assigned to multiple diagnosis groups. Back sprains; disc disorders and spinal stenosis; and mental, behavioral and neurodevelopmental disorders, not elsewhere classified had the highest total cost for claims overall and among lost-time claims. Conversely, diagnoses such as open wounds and burns had lower total costs despite the relatively high proportion of claims represented. Furthermore, diseases of blood and blood forming organs; mental, behavioral and neurodevelopmental disorders, not elsewhere classified; and amputations, showed exceptionally high cost-to-claim ratios for all claims (ratio range = 40.6–51.9) and diseases of blood and bloodforming organs, amputations, and neoplasms showing exceptionally high cost-to-claim ratios for lost-time claims (ratio range = 11.7–16.8). Neoplasms had the highest cost-to-claim ratio (72.8) of all medical-only claims.

Fig. 1 presents neoplasm claims’ frequency over time to provide context for the changing distribution of costs. Across all OHBWC claims during the study period, neoplasm claims were rare (0.01–0.05%) as a percentage of all distinct diagnosis groups (data not shown). Nonfirefighter neoplasm claims (N = 410) as a percentage of all OHBWC distinct diagnoses, has been relatively stable per year (0.01–0.02%) throughout the study period. As mesothelioma claims declined (data not shown), the total number of neoplasm claims per year decreased from 2001 to 2012 then began increasing. Since 2017, we observed > 41 neoplasm claims per year, above the previous highest count from 2001; this trend coincided with the implementation of Ohio’s firefighter cancer presumption law in April 2017. Across 20 years, we observed 149 firefighter neoplasm claims, including 36 fatalities [most (N = 20) fatalities occurred 2017–2020]. Prior to 2016, the percentage of neoplasm claims from firefighters was unstable, ranging from 0% in 2010 to 27% (N = 3) in 2008. Beginning in 2016, the pattern changed; firefighter neoplasm claims accounted for 48% (N = 11) of all neoplasm claims in 2016 and increased to 59% (N = 26) in 2020.

Table 4 presents firefighter claims categorized by disability category (PTD, PPD, PPTTD, or TTD). A total of 197 PTD claims accounted for $121 million in costs (24% of total cost with a mean cost of $616,044 and median cost of $615,188 [data not shown]), highlighting the significant financial impact of the most severe injuries. In contrast, the 3,506 PPD claims had a lower cost of $34 million (7% of total cost with mean and median costs of $9,733 and $4,909, respectively [data not shown]). The PPTTD claims, which represent a combination of severity, were the second most frequently occurring with 4,201 claims but had the highest cost burden across all severity types with $222 million in total costs (43% of total cost), a mean cost of $52,904 and a median cost of $29,078 [data not shown]. TTD claims were the most frequent (n = 4,303) but incurred relatively lower costs of $73 million (14% of total cost, Table 4), and mean and median costs of $17,007 and $6,508, respectively [data not shown].

Like the overall distribution by sex, male firefighters accounted for most claims across all disability categories, with costs aligning closely to the distribution of total claims. Conversely, while female firefighters represented a small fraction of the total claims, they had higher cost to claim ratios in all four disability categories (ratios = 1.07, 1.05, 1.28, 1.79, respectively; Table 4). By age, claims were generally most frequent and had higher costs among those 35–44 and 45–54 years (Table 4). However, the cost/claim ratio reveals that injuries among younger firefighters (25–34 years) resulted in the highest cost burden relative to the number of claims for PTD. Conversely, firefighters 65 + years had the highest cost/claim ratio for PPD and PPTTD while firefighters 55–64 had the highest cost/claim ratio for TTD claims (Table 4).

Like the overall results, analyses by task and severity type indicated injuries that were incurred during patient care and firefighting activities were generally the most frequent and came with the highest costs, particularly for PTD, PPTTD, and TTD categories (Table 4). However, the highest cost/claim ratios among PPD, PPTTD, and TTD claims were observed for long-term exposure tasks (PPD ratio = 7.9, PPTTD ratio = 1.5, and TTD ratio = 11.5). These high ratios are likely reflective of the relatively low claim numbers in these disability categories. The least frequent event that had the highest associated cost/claim ratios was exposure to other harmful substances (PPD: n = 39, ratio = 7.3; PPTTD: n = 31, ratio = 1.8; and TTD: n = 118, ratio = 4.8) (Table 4). Supplemental Table 2 presents the claims by select diagnoses and disability category.

This study provided the first comprehensive analysis of WC claim costs among Ohio firefighters, including novel insights into worker and claims characteristics by disability categories and temporal trends in neoplasm claims following presumption law changes. The analysis of claim costs across tasks, events, diagnoses, and disability categories offers valuable data for targeting prevention efforts and allocating resources. The large sample size spanning 20 years and inclusion of both public and private firefighter claims strengthen the generalizability of findings.

Several limitations should also be noted. First, this analysis was limited to the WC data from a single state. There may be differences in the types and severity of injuries and illnesses that cannot be fully explained using data from a single state. Second, differentiation between volunteer and career firefighters was not possible in the dataset, potentially masking important differences in injury/illness patterns and costs between these groups. While the task categorization provided useful insights, many narratives were ambiguous or indicated multiple tasks, resulting in 46% being classified as “other/unclear.” The analysis was limited to incidents severe enough to warrant WC claims, likely missing minor incidents handled through first aid or those incidents that went unreported. The study’s external validity was limited by excluding self-insured employers in Ohio. Additionally, the analysis did not calculate rates due to lack of reliable denominator data for the firefighter population, particularly volunteers. This prevented comparisons of injury/illness risk across demographic groups or over time. Lastly, while the analysis captured the early impact of Ohio’s cancer presumption law for firefighter claims, longer-term cost implications may still emerge.

These limitations suggest opportunities for future research, particularly in examining differences between career and volunteer firefighters, gathering accurate workforce denominators, and evaluating long-term impacts of presumption laws on WC systems.