This nationwide cross-sectional study was conducted between February 2018 and March 2018 via a self-reported online survey among Korean firefighters. A total of 45,719 firefighters in South Korea participated in the survey and completed the self-reported questionnaire, including their suicidal ideation in the past year as well as demographic characteristics and occupational factors, such as traumatic experience in the past year, occupational stress, and emotional labor, and clinical factors, including PTSD and depression symptoms. Among the total, 21 were excluded from the final analyses due to a coding error during the survey. Thus, the final analyses included a total of 45,698 firefighters. Survey respondents were apprised of the anonymous and voluntary nature of the self-reported online survey.
Demographic and occupational characteristics
Demographic and occupational characteristics were obtained using a self-reported questionnaire. Demographic characteristics included age, sex (male or female), marital status (married, never married, or divorced/separated/widowed), and religion (yes or no). Occupational characteristics included length of work (years), and current duty. The roles of firefighters include fire suppression, special investigation of the cause of fire, paramedics providing emergency medical care, rescuing people who are trapped or in medical emergencies, training other firefighters, and others . For analysis, the roles were categorized into the following: fire suppression, emergency medical services (EMS: includes paramedics and rescue), and officers (including administrators, special investigators, trainers of firefighters, and communicational and informational system operators).
Suicidal ideation in the past year
Suicidal ideation in the previous year was assessed using an item of the Suicidal Behavior Questionnaire-Revised (SBQ-R) . The SBQ-R is a brief self-reported questionnaire to inquire about different aspects of suicidal behaviors. Item 1 explores whether the respondents have ever thought about or attempted suicide in his/her lifetime. Item 2 evaluates how often the respondents have thought about suicide over the past twelve months. Item 3 inquires about threats of suicide attempts, and item 4 explores the self-reported likelihood of suicidal behaviors in the future. In this study, we used item 2 to assess past-year suicidal ideation in participants Item 2 reads “How often have you thought about killing yourself in the past year?” Participants responded to this on a 5-point Likert scale: 1-never; 2-rarely (1 time); 3-sometimes (2 times); 4-often (3-4 times), and 5-very often (5 or more times). A score of greater than 2 on Item 2 was indicates as having suicidal ideation more than once in the past year. According to the validation study for the SBQ-R, item 2 had the largest effect size followed by item 1 for differentiating between suicidal-risk and non-suicidal participants in both clinical and nonclinical adult samples . The correlation coefficient of SBQ-R item 2 with PHQ-9 item 9, which reads “thoughts that you would be better off dead or of hurting yourself in some way” was r = 0.499 (p < 0.001). Additionally, the correlation coefficient between SBQ-R item 2 and the total score of PHQ-9 was r = 0.479 (p < 0.001).
The presence of recent exposure to traumatic events
Exposure to traumatic events during the previous year was identified by using the self-reported measure – developed by Beaton et al. – which assessed the duty-related incident stressors . Twenty-two items were selected among the original 33 incident stressors based on the previous result of rating the stressfulness of the 33 stressors . We excluded two stressors related to gunshots due to the generally low incidence of gunshot incidents in South Korea due to strict gun control laws. In South Korea, only government-authorized personnel can own or carry guns. Gun culture is notably absent outside of the military and gun ownership and death ranks among the lowest in the world . ‘Witness duty- related death of co-worker’ and ‘co-worker firefighter fire fatality (not witnessed)’ were changed to ‘witness duty-related death or suicide of co-worker’ and ‘co-worker death or suicide (not witnessed), respectively. Finally, three additional stressors, ‘remove the body of suicide victim’, ‘remove a severely decayed corpse’, and ‘involved in a safety accident that received public spotlight’ which were reported to be frequently encountered and associated with high level of stress in Korean firefighters were added (Table S1). Participants were asked whether they were exposed to each stressor in the previous year. More than one exposure to traumatic events in the previous year was regarded as having recent exposure to traumatic events.
Occupational stress from physical work environment
Occupational stress from physical work environment was measured using the subscale, ‘Difficult Physical Environment’ of the Korean Occupational Stress Scale (KOSS) , which was developed and validated using a nationwide epidemiological study to estimate the job stress of Korean employees. The KOSS was based on the most commonly used job stress questionnaires such as the Job Content Questionnaire , National Institute of Occupational Safety and Health job stress questionnaire  and Occupational Stress Index . The KOSS has eight subscales (Difficult Physical Environment, High Job Demand, Insufficient Job control, Inadequate Social Support, Job Insecurity, Organizational Injustice, Lack of Reward, and Discomfort in Occupational Climate). The “Difficult Physical Environment’ of the KOSS has three items, each of which was rated on a 4-point Likert scale (1: strongly disagree to 4: strongly agree) and higher scores represented higher levels of occupational stress from physical environment. Table S2 shows each item of the “Difficult Physical Environment” subscale of the KOSS. The internal consistency of “Difficult Physical Environment” of KOSS based on the presented sample was α = 0.464.
The level of emotional labor was measured by the Korean Emotional Labor Scale (KELS) . KELS was developed to measure the emotional labor of Korean workers, and was validated with a nationwide random sample of 1,042 Korean employees by the Korean Occupational Safety & Health Agency. It was based on the literatures related to emotional labor [21-23, 39], emotional labor scales, such as Emotional Labor Inventory , Emotional Labor Scale , and Frankfurt Emotion Work Scale , as well as a focused group interview. The KELS has five subscales (Effort to Control Emotion, Organizational Monitoring System, Demands of Emotional Labor, Emotional Damage, and Organizational Support System). Each item in the questionnaire was rated on a 4-point Likert scale, from 1 (not at all) to 4 (very much), and higher scores represented higher levels of stress from emotional labor. The current study included only the subscale, “Emotional Damage” which measures the severity of emotional hurt due to emotional labor. We regarded the subscale score as proxies for emotional labor in firefighters, because the “Emotional Damage” is a factor that explained the most variance of the KELS in the results of a factor analysis in the study developing the scale . Table S2 shows each item of the “Emotional Damage” subscale of the KELS. The internal consistency of “Emotional Damage” of KELS based on the presented sample was α = 0.947.
PTSD symptoms were assessed with the Korean version of PTSD Checklist-for the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (PCL-5) . The PCL-5 is a 20-item self-reported measure evaluating the degree to which an individual has been bothered in the past month by DSM-5 (Diagnostic and Statistical Manual of Mental Disorders, 5th edition) PTSD symptoms : intrusions, avoidance, negative alteration in cognition and mood, and alterations in arousal and reactivity. We instructed the participants to choose and describe the most traumatic event from a list of traumatic events and fill out the PCL-5 with this event in mind. Each item was measured on a 5-point Likert scale (0: not at all to 4: extremely). Higher scores indicated higher severity of PTSD symptoms. Items 1- 5 correlated with symptoms within Cluster B (intrusions); items 6 - 7 with Cluster C (avoidance); items 8 - 14 with Cluster D (negative alteration in cognition and mood); and items 15 - 20 with Cluster E (alterations in arousal and reactivity). The internal consistency of PCL=5 based on the presented sample was α = 0.961. Participants were considered to be experiencing the symptom when recording a score of 2 or higher (moderately to extremely) in each item. According to an algorithm-derived PTSD diagnosis method, we defined probable PTSD as having the required number of symptoms in each cluster of the DSM-5 criteria: 1 B item, 1 C item, 2 D items, and 2 E items.
Depression symptoms were assessed using the Korean version of Patient Health Quetionnaire-9 (PHQ-9) [44, 45]. Respondents rated 9 items, based on the DSM-IV criteria of major depressive disorder, measured on a 4-point Likert scale (0: not at all to 3: nearly every day) based on their experiences during the past two weeks. The PHQ-9 total score ranged from 0 to 27; higher scores indicated a greater severity of depressive symptoms. The internal consistency of PHQ-9 based on the presented sample was α = 0.905. The total score of over 15 was defined as probable depression .
Descriptive statistics were used to analyze the demographic, occupational, and clinical characteristics of participants as well as to calculate the 1-year prevalence of suicidal ideation. Chi-square test and t-test were used to examine whether there were differences in the demographic, occupational, and clinical characteristics between firefighters with and without suicidal ideation in the past year. Multivariable logistic regression analysis was used to examine the demographic (age, sex, marital status, and religion), occupational (current duty, recent traumatic experience, occupational stress, and emotional labor), and clinical (probable PTSD and depression) characteristics as correlates of suicidal ideation in the previous year. The dependent variable was suicidal ideation in the past year. In the logistic regression analysis, 143 participants with missing data were excluded; thus, a total of 45,555 participants were included in the analysis. The results were shown as odds ratios (ORs) and 95% confidence intervals (CIs). The data were analyzed using IBM SPSS Statistics ver. 22.0 software (IBM Corp., Chicago, IL, USA). A two-tailed p-value < 0.001 was considered statistically significant.