Study population and data
Panel Study of Workers’ Compensation Insurance (PSWCI) data from 2018 to 2021 by the Korea Workers’ Compensation & Welfare Service (KCOMWEL) were analyzed. The PSWCI is a nationwide annual panel study of industrial accident victims that collects data on industrial accident insurance coverage and socioeconomic characteristics after the termination of primary medical care. The PSWCI is administered by tablet-assisted personal interviews (TAPI) with a visitor interview. In total, 81,252 IIW terminated medical care between January and December 2017. In the PSWCI, 3,294 participants were included through proportional stratified random sampling based on disability grade (six categories), sex (two categories), age group (four categories), and systematic sampling based on area of residence (six categories) and rehabilitation service usage (two categories). At the initial stage of analysis, 202 participants were excluded because they did not answer several questions regarding the variables. As a result, 3,092 participants (2,562 males and 530 females) were analyzed and contributed 11,167 observations during the 4-year study period. As of 2021, 2,797 participants remained in the PSWCI (retention rate: 84.9%).
Measures
On-site first aid
The participants were asked whether they experienced appropriate on-site first aid by co-workers at the time of the industrial accident and the response items were “yes” or “no.” According to the Occupational Safety and Health Act, business owners are obliged to prevent industrial accidents and provide employees with information on safety and health at relevant places of business, including tips for first aid [1]. Therefore, the Korea Occupational Safety and Health Agency (KOSHA) provides professional tips for first aid according to the type of industrial accident [11]. For example, “Do not move the patient recklessly, and keep him/her in a comfortable position” is recommended as a general tip for fractured patients. In this study, industrial accidents involved accidents on duty, but not occupational diseases or commuting accidents.
Healthcare utilization: Outpatient and hospitalization
The participants were asked, “What is your history of visiting medical institutions in the past year?” Regardless of the reason for healthcare utilization, the number of outpatient visits, hospitalizations, and duration of hospitalization were evaluated. Pharmacies and nursing homes were excluded from the utilization count, and health checkups were excluded from the number of outpatient visits. Circuit outpatient visits were also excluded. For example, receiving treatment in more than two different departments at the same medical institution was only considered as one visit. If participants were admitted to the hospital for 365 days, and if they were admitted or discharged on the day of the emergency room visit, they were marked as hospitalized once.
Transfer time to emergency room
The participants were asked, “How long did it take for you to be transferred to the emergency room?” The transfer time to the emergency room was classified into four groups according to the PSWCI: less than 0.5 hour, 0.5 ~ 1 hour, 1–2 hours, and more than 2 hours.
Covariates
Several time-varying socioeconomic and health-related characteristics were adjusted. Patients were classified as either male or female. The age group was divided according to the PSWCI: age below 30s, 40s, 50s, and above 60s. Education level was divided into university or higher, and high school graduation or lower. Current economic activity was divided into employed, unemployed, and economically inactive. Current household income was divided according to the quintile of household income. Household income is the sum of earned, financial, real estate, and other income. Statistics Korea data for the entire population were used as the income quintile cutoff for each year [12]. Area of residence was divided into metropolitan and provincial (rural). Medical histories of chronic diseases such as cancer, hypertension, and diabetes before industrial accidents were divided into without chronic disease and with chronic disease. The types of injuries due to industrial accidents were divided into fracture, sprain, back pain/musculoskeletal disease, amputation, cuts, bruising/concussion, rupture/laceration, burns, and other (abrasions, stab wounds, frostbite, contagion/addiction, and internal organ damage). Disability grade was based on 14 levels of physical/mental aftereffects of industrial accidents designated by law after the termination of primary medical care and symptom fixation [1]. The higher the disability grade, the more severe the disability level. In the analysis, disability grade was divided into the following groups as given by the PSWCI: grades 1–3 (most severe), 4–7, 8–9, 10–12, 13–14, and no disability. The period of primary medical care, which refers to the initial acute management at the designated hospital for IIW, was divided into the following as given by the PSWCI: < 3 months, 3–12 months, and > 1 year. Variables including sex, age, area of residence, type of injury, disability grade, and period of medical care were administrative data from the KCOMWEL, while variables including education level, current economic activity, current household income, and past medical history before accidents were self-reported.
Statistical analysis
A generalized estimating equation (GEE) Poisson regression was applied for the analysis. Poisson regression with a log link function and an unstructured (UN) working correlation matrix, which had the lowest Quasi-likelihood under Independence Model Criterion (QIC) statistics, was used for the longitudinal data (from 2018 to 2021). The results are presented as adjusted relative risk (aRR) with 95% confidence intervals (CI). Subgroup analyses were conducted to determine the detailed effects based on transfer time to the emergency room and the covariates. As the variance inflation factors (VIF) for all variables were less than 1.6, there was no evidence of multicollinearity. Version 9.4 SAS software (SAS Institute, Cary, North Carolina, USA) was used. Statistical significance was set at P ≤ 0.05.