Mental Health of Canadian Firefighters: The Impact of Sleep

Volunteer and career firefighters are at risk of major depressive disorders, posttraumatic stress disorder (PTSD), alcohol use disorder, and other mental health disorders due to the demanding and unpredictable nature of their employment. The mental health risks are exacerbated by the need to work extended hours, night shifts, and/or rotating schedules, or the competing demands of other employment, especially in volunteer firefighters. The mental health disorders and risk factors interact with altered sleeping patterns. In the current study, we examined volunteer and career firefighters regarding the association between mental health and sleep, drawing from a national Canadian mental health survey of 1217 firefighters. Most (69%) of the firefighters reported less than ideal sleep quality and 21% screened positive for clinical insomnia, with no significant difference between volunteer and career subgroups. Firefighters with insomnia had higher odds ratios (OR) and frequencies for PTSD (OR = 4.98), generalized anxiety disorder (OR = 7.15), panic disorder (OR = 6.88), social phobia (OR = 4.98), and major depressive disorder (OR = 7.91), than firefighters without insomnia. The burden of sleep disorders and their association with mental health disorders suggests that sleep should be considered in health monitoring and self-management, environmental design, fire service work-organization policies, and health programming.


Introduction
Firefighters are regularly exposed to highly stressful, unpredictable, dangerous, and potentially psychologically traumatic events (PPTE) when responding to calls on duty [1]. Some firefighters classify firefighting as their career, whereas others serve in a volunteer capacity [2]. Most firefighting (83%) in Canada is done by volunteers in rural areas [3] who are on call to provide 24-h emergency services, despite having separate careers [2,4]. Volunteer firefighters respond to overnight calls and are still expected to return to their day jobs, despite these calls impeding their cognitive performance [4]. Although some services maintain 12-h shift schedules, career firefighters typically work on rotating schedules of 24 h on/48 h off, or 48 h on/96 h off [5]; however, some firefighters still adopt a variation of 12, 10, or 8 h shifts [6]. Firefighter shifts, for both career and volunteer workers, can fluctuate frequently between day, night, and 24 h shifts; even a firefighter resting during a 24-h shift with no immediate emergency is still required to be alert, in case of an impending emergency [4,5,7]. Many fire departments allow for career firefighter shifts that last up to 72 consecutive hours [8], increasing firefighters' susceptibility to shift work such as posttraumatic stress disorder, depression, anxiety, stress, panic, social phobias, and alcohol use.

Study Aim
The study was designed to examine the association between sleep and mental health among Canadian firefighters.

Materials and Methods
In the current study, we analyzed data from individuals identifying as firefighters, collected from a larger cross-sectional survey of PSP in Canada [26]. Details of data collection are described elsewhere [26]; in brief, participation was solicited through emails to currently serving PSP employed in communication, correctional, fire, paramedic, or police services. Data in English or French were collected using web-based self-reporting survey methods. The survey included validated screening tools of mental health disorder symptoms, with evidence of diagnostic discriminant validity. Participants were not required to answer any question to proceed through the survey; however, participants were asked to confirm that questions left unanswered were done so intentionally. Participants were asked to provide demographic information including whether they were career or volunteer firefighters, their rank, and their job title. The survey was launched on 1 September 2016, and PSP could participate until 31 January 2017.

Self-Report Symptom Measures
Indications of mental disorder(s) and symptom severity were assessed using wellvalidated self-report screening tools (described below). These screening tools were not diagnostic; rather, a 'positive screen' on any of the tools indicated that an individual reported several symptoms or indicators associated with a disorder.

1.
PTSD was assessed using the PTSD Check List 5 (PCL-5) [32]. The PCL-5 is a commonly used 20-item self-report tool that assesses the 20 symptoms of PTSD outlined in the Diagnostic and Statistical Manual of Mental Disorder-fifth edition (DSM-5) (APA, 2013). Individuals were asked to rate how bothersome the 20 items were to them on a scale of 0 (not at all) to 4 (extremely). A total score greater than 32 is considered reasonable for a provisional positive screen for PTSD. Internal consistency of the PTSD check list for the current study was α = 0.95.

2.
Major depressive disorder (MDD) symptoms were assessed using the Patient Health Questionnaire 9-item (PHQ-9) [33]. In the PHQ-9 individuals were asked to consider the past two weeks and to rate nine symptoms of depression on a scale of 0 (not at all) to 3 (nearly every day). A provisional screen for MDD was based on 5 of the 9 items being rated at least a 2 or 3, or two questionnaire items; "little interest or pleasure in doing things" and "feeling down, depressed or hopeless" were rated 2 or 3. Internal consistency of the PHQ for the current study was α = 0.90.

3.
Panic disorder symptoms were assessed using the Panic Disorder (PD) Symptoms Severity Scale (PDSS) [34]. The PDSS is a 7-item severity scale where items are scored on 5-point scale from 0 to 4. The measure was designed to rate the overall severity of PD symptoms. A total score greater than 9 is considered reasonable for a provisional positive screen for PD. Internal consistency of the PDSS for the current study was α = 0.93.

4.
Symptoms of a General Anxiety Disorder (GAD) were assessed using the GAD 7-item Scale (GAD-7) [35]. The GAD-7 is a 7-item questionnaire where individuals are asked to rate how often symptoms of anxiety, such as feeling nervous, anxious, or on edge, have bothered them on a scale of 0 (not at all) to 3 (nearly every day). A total score greater than 9 is considered reasonable for a provisional positive screen for GAD. Internal consistency of the GAD for the current study was α = 0.91.

5.
The Depression, Anxiety, and Stress Scale-21 (DASS-21) was also used to measure broad symptoms of depression, anxiety, and stress relative to general population data [36]. The DASS-21 is a 21-item questionnaire divided into three subscales of depression, anxiety, and stress. Items were scored from 0 (does not apply to me at all) to 3 (applies to me very much or most of the time). Internal consistency of depression, anxiety, and stress for the current study was α = 0.86, α = 0.84, and α = 0.83 respectively. 6.
Social anxiety disorder (SAD) symptoms were assessed using the Social Interaction Phobia Scale (SIPS) [37]. The SIPS is a 14-item measure of social anxiety symptoms that can be divided into three subscales of Social Interaction Anxiety, Fear of Overt Evaluation, and Fear of Attracting Attention. Subscale scores and an overall score were calculated for the current study. A total score greater than 20 is considered reasonable for a provisional positive screen for SAD. Internal consistency of the SIPS for the current study was α = 0.94. 7.
Risky (hazardous) alcohol use with the Alcohol Use Disorders Identification Test (AUDIT) [38]. The AUDIT is consistent with ICD-10 definitions of alcohol use disorder and harmful alcohol use. The AUDIT is a 10-item list of questions relating to an individual's drinking behavior. Items were scored from 0 (no or never) to 4 (response depends on the question being asked). A total score greater than 15 is considered reasonable for a provisional positive screen for alcohol use disorder. Internal consistency of the AUDIT for the current study was α = 0.82. 8.
Sleep disturbance was measured using the Insomnia Severity Index (ISI). The ISI is a self-report questionnaire assessing the nature, severity, and impact of insomnia [39]. Individuals are asked to answer 7 questions rated on a 5-point Likert scale relating to the severity of their sleep difficulties where 0 = no problem; 4 = very severe problem, with total scores ranging from 0 to 28. The total score can then be divided into four categories: absence of insomnia (0-7); sub-threshold insomnia (8)(9)(10)(11)(12)(13)(14); moderate insomnia (15)(16)(17)(18)(19)(20)(21); and severe insomnia (22)(23)(24)(25)(26)(27)(28), or dichotomized into positive screen for clinical insomnia for scores of 15 or greater. Respondents were also asked to rate their sleep quality overall from very poor to very good, and provide an estimate of the number of hours of sleep they get during the workweek (or when they are on shift) and during the weekends (or when they are off shift). Finally, they were asked how many days per week they woke up feeling felt rested. Internal consistency of the ISI for the current study was α = 0.85.
Participants reported their symptoms in the timeframe per the instructions for each scale: PCL-5, past month; MDD, past 14 days; PDSS, past seven days; GAD-7, past 14 days; SIPS, currently, no specific time window; AUDIT, past year; DASS-21, past seven days, and ISI past month.

Statistical Analysis
All data were collected electronically and entered into SPSS v.25 (IBM, New York, NY, USA) for analysis. Missing data were treated as missing, and statistical significance was set as p < 0.05. Demographic information such as firefighter current and past employment, age, gender, sleep, and mental health questionnaire scores were described using means, frequencies, percentages, and standard deviations. Overall prevalence estimates for each mental disorder were calculated using (where appropriate) the mean score and the established dichotomous cut-offs.
The prevalence rates and differences in positive screens for mental disorders between individuals who had screened positive for clinical insomnia and those who had not were calculated using several cross tabulations with χ 2 tests. For assessing differences between different types of firefighters with respect to the association between sleep disruption and positive screens for mental health disorders, moderator analyses were conducted. A series of logistic regression calculations were then performed to assess the magnitude of differences in mental disorders identified as significantly different between firefighters with and without a positive screen for clinical insomnia, while controlling for age, and gender. The Bonferroni correction was applied to control for multiple comparisons. Pearson correlation analyses were performed between all measures. Finally, six statistical mediation analyses were conducted using PROCESS [40] to assess the mediator effect of stress on the relationship between sleep disturbance and mental disorder symptoms (i.e., PTSD, PHQ9, GAD7, PDSS, SIPS, or AUDIT; Figure 1). Each statistical mediator models included mental disorder symptoms as the dependent variable, stress as the mediator variable, and sleep disturbance as the independent variable. of logistic regression calculations were then performed to assess the magnitude of differences in mental disorders identified as significantly different between firefighters with and without a positive screen for clinical insomnia, while controlling for age, and gender. The Bonferroni correction was applied to control for multiple comparisons. Pearson correlation analyses were performed between all measures. Finally, six statistical mediation analyses were conducted using PROCESS [40] to assess the mediator effect of stress on the relationship between sleep disturbance and mental disorder symptoms (i.e., PTSD, PHQ9, GAD7, PDSS, SIPS, or AUDIT; Figure 1). Each statistical mediator models included mental disorder symptoms as the dependent variable, stress as the mediator variable, and sleep disturbance as the independent variable.
The study was granted ethical clearance by the University of Regina Institutional Research Ethics Board (File #2016-107). Prior to access to the survey, individuals indicated their willingness to participate by clicking "I agree" at the end of an electronic study information letter.

Results
A total of 1217 firefighters participated in the survey. Most participants (n = 1014; 84%) identified as male, 1086 (89%) were career firefighters, and 141 (11.6%) identified their position as primarily administrative (e.g., Chief, Deputy Chief), 352 (28.9%) as leadership (e.g., chief officers and Captains in emergency response, as well as officers in support divisions such as training, prevention, mechanical, public education, and communications), 310 (25.5%) as a senior firefighter (e.g., First Class), 23 (1.9%) as junior (i.e., 5 to 7 years in career service), and 56 (4.6%) as 'other' (e.g., mixed role). A small proportion (n = 131, 10.8%) identified themselves as volunteers. See Table 1 for full demographic information. ANOVA produced no statistically significant relationships between positive mental health disorder screenings and the categories of firefighter, or between career and volunteer fighters and positive mental health disorder screenings; as such, the remaining results are for the entire sample of 1217 combined.  The study was granted ethical clearance by the University of Regina Institutional Research Ethics Board (File #2016-107). Prior to access to the survey, individuals indicated their willingness to participate by clicking "I agree" at the end of an electronic study information letter.

Results
A total of 1217 firefighters participated in the survey. Most participants (n = 1014; 84%) identified as male, 1086 (89%) were career firefighters, and 141 (11.6%) identified their position as primarily administrative (e.g., Chief, Deputy Chief), 352 (28.9%) as leadership (e.g., chief officers and Captains in emergency response, as well as officers in support divisions such as training, prevention, mechanical, public education, and communications), 310 (25.5%) as a senior firefighter (e.g., First Class), 23 (1.9%) as junior (i.e., 5 to 7 years in career service), and 56 (4.6%) as 'other' (e.g., mixed role). A small proportion (n = 131, 10.8%) identified themselves as volunteers. See Table 1 for full demographic information. ANOVA produced no statistically significant relationships between positive mental health disorder screenings and the categories of firefighter, or between career and volunteer fighters and positive mental health disorder screenings; as such, the remaining results are for the entire sample of 1217 combined. Table 2 presents the results of screening for sleep disturbances on the ISI, the perception of overall sleep quality, and the number of hours a night participants sleep when working and when not working. Most respondents (69.2%) reported overall quality of sleep as 'fair', 'poor', or 'very poor', 21.3% screened positive for clinical insomnia, and on average reported waking up feeling rested 2.7 out of 7 days a week. Table 3 compares the individuals with and without a positive screen for clinical insomnia, on the various screening measures of mental health disorders. The results indicate statistically significant differences between firefighters screening positive for clinical insomnia and those without a positive screen for clinical insomnia, on all mental health disorders except AUDIT. Table 4 provides odds ratios of screening positive for a mental health disorder for individuals who screened positive for clinical insomnia, controlling for age and gender. Firefighters who screen positive for clinical insomnia have statistically significantly higher odds of screening positive for PTSD, GAD, PD, SAD, and MDD (OR between 4.98 to 8.53; Table 4), a clinically important finding. Moderator analyses produced no statistically significant differences between different types of firefighters on the association between sleep disruption and positive screens for all mental health disorders. The current results suggest different firefighter categories experience similar impacts of positive clinical insomnia on positive screens for mental health disorders.   4 7.15 (4.38, 11.69) *** >0.05 PHQ-9 5 7.91 (5.27, 11.87) *** >0.05 PDSS 6 6.88 (3.81, 12.42) *** >0.05 SIPS 7 6.32 (3.83, 10.41) *** >0.05 AUDIT (risky drinking) 8 1 All correlations between measures were statistically significant, and in the expected direction (Table 5).   Table 6 presents statistical mediation analyses for all models, including total effects, direct effects, and indirect effects of sleep disturbance on mental disorder symptoms with mediation of stress. The results indicated a significant direct effect between sleep disturbance and mental disorder symptoms. Indirect effect of sleep disturbance on PTSD through stress was statistically significant. 14 PTSD = posttraumatic stress disorder; DASS = depression, anxiety, and stress symptoms (DASS-21); GAD-7 = generalized anxiety disorder Scale; PHQ-9 = patient health questionnaire; PDSS = panic disorder symptoms severity scale; SIPS = social interaction phobia scale; AUDIT = alcohol use disorders identification test; CI: confidence interval; ** p ≤ 0.01, *** p ≤ 0.001.

Discussion
The current results suggest a strong association between positive screens for mental health disorders and clinical insomnia. The current results evidenced high rates of sleep disturbance in all firefighter groups, but did not establish differences between different types of firefighters. Firefighters with insomnia also had higher frequencies and odds of screening positive for mental health disorders than those without. There were no statistical differences across firefighter classifications. Most support members (e.g., those involved in training and public education) work regular day hours without shift scheduling; however, to acquire a rank in a support division (e.g., training or fire prevention) requires having been an active emergency response firefighter at some point, either as a volunteer or during career service. Firefighters who respond to emergencies may continue to experience sleep disturbances and concomitant sequelae, well after they change roles.
The current results indicated 69.2% of participants reported having below adequate sleep quality, a finding consistent with that found among Chinese and Iranian (69.9%) [27] and U.S. firefighters (66%) [41], suggesting that this is an ongoing concern for firefighters globally. Disordered sleep increases a person's risk of suicide, symptoms of major depressive disorder, cardiovascular disease, diabetes, and decreases cognitive functioning [41][42][43]. Shift workers from diverse jobs working nights and rotating shifts have also reported lessened sleep quality and disordered sleep [44]. For shift workers, disordered sleep is explained by the irregularity of work hours and the disruption that these hours have on one's circadian rhythm [10,44]. Furthermore, similar results have been found among samples of American police officers, both on varied shift work (64.1%) and on day shifts (63.7%) [45], and slightly higher results were found in American paramedics (77%) [46]. The current results evidence poor sleep quality amongst different types of firefighters, regardless of the type of shifts being worked. This is perhaps due to the high-risk behaviors required from firefighters, the unpredictable emergencies they encounter, and their increased exposure to trauma [15]. Different factors may influence sleep in volunteer and career firefighters, but such supposition may require more qualitative investigations.
The current study also evidenced that 21.3% of firefighters had clinical insomnia. An international study on insomnia reported that, relative to health-related quality of life, insomnia is not simply a single-night symptom, but rather a severe condition, of which the impact on one's health-related quality of life can be compared to diabetes and major depressive disorder [47]. The current study also evidenced that firefighters with a positive screen for clinical insomnia were much more likely to screen positive for mental health disorders than firefighters that had a negative screen for insomnia. The firefighters with insomnia had higher positive screening frequencies for PTSD, MDD, GAD, SAD, PD, and risky alcohol use than firefighters without insomnia. The firefighters also had higher odds of screening positive for PTSD, GAD, PD, SAD, and MDD than their counterparts without insomnia. In addition to the GAD symptoms and lower quality of life that accompanies insomnia, Sing and Wong [48] found that it also has a mediating effect on the optimism associated with MDD. This implies that those with insomnia symptoms have less optimism towards their depression, which furthers their negative health path [48]. Firefighters are already predisposed to challenging mental disorder symptoms, due to the realities of their occupation [1,49], but insomnia serves as an exacerbating factor associated with decreased overall health optimism, which can further elevate risks to their health.
Researchers have documented insomnia as a comorbidity and a precursor to mental health disorders [20,50,51]. These findings in addition to those of the current study make a case for early intervention regarding sleep disturbances such as insomnia. A sleep health program which included three components-mandatory educational sessions, voluntary sleep disorders screening, and sleep disorders diagnosis and treatment for those who screened at risk for a sleep disorder-was implemented for U.S. firefighters [52]. Those that attended the educational sessions took 46% fewer days off due to disability and were 24% less likely to file an official injury report than the controls. An intervention program using behavior and imaging therapy found significant improvements in sleep, as well as symptoms of MDD and PTSD in firefighting participants [53]. Targeting disordered sleep in firefighters may result in fewer comorbidities and the prevention of numerous mental health disorders linked to insomnia (e.g., MDD, GAD, PTSD) [20,[52][53][54]. Career, volunteer, and paid on-call firefighters all have different vocational demands that differentially impact their work-life balance [55]; accordingly, each group may need tailored programming to support their health.
There may also be opportunities to develop organization-level modifications for work that lessens the alerting impact of fire calls on firefighters. Given it is adaptive and essential that firefighters learn to rest in a state of readiness, the longer-term impacts on their capacity to enjoy restful and restorative sleep are unclear. The expectation of having sleep interrupted by an alarm, and the need to urgently mobilize results in "light" sleep that is perceived as poor, from which one is easily woken [55]. A recent study compared cortisol levels for day and night alarms among firefighters, reporting that higher cortisol levels during night alarms may be due to a reduced ability for firefighters to predict when a call comes in, and a louder noise relative to the reduced background noise [56]. There is limited but emerging research that suggests adapting the alerting methods to a "ramp-up" tone; limited to station tasks, this response to a given call can reduce the impact on firefighter heart rate throughout their work days [57]. Such interventions may offer some preventive benefit, especially to firefighters with PPTE and whose elevated acoustic startle responses suggest vulnerability for posttraumatic stress [58]. Other studies have demonstrated that the "alarm stimulus signals the potential for exposure to trauma" [55], suggesting researchers should investigate how the physiological and psychological responses to alarms impact firefighters, their sleep, and their mental health.

Limitations of the Study
The sample size of the volunteer firefighters was relatively small, as was the number of women firefighters, which limited our ability to nuance sex or gender-based differences, limiting our confidence that these results apply to women and/or volunteer firefighters. The data were also derived from self-report surveys, which means personal and societal biases may have influenced the results. Due to these measures, participants were also required to remember and report their sleep and mental health data from their current state to the past year, which may have created recall biases. These limitations were offset by the use of screening tools with evidence of diagnostic discriminant validity. Another limitation is that we did not objectively assess the shift schedule these firefighters were on. Although firefighters generally operate within a shift-based schedule which influences their sleep, we know that these can differ throughout different regions across the country. Future research should investigate the influence of differing shift schedules on sleep and mental health disturbances. Additionally, as this study was cross sectional, the results should be interpreted with caution and no causal inferences should be made. In this study we were not able to assess the direction of the relationship between sleep and mental health disorders. Researchers have explained that the relationship between sleep disturbances and mental health may be bidirectional [59]. Future research should utilize a longitudinal approach in order to understand the directionality of this relationship for Canadian firefighters.

Conclusions
Many participants (21%) screened positive for clinically significant insomnia. Canadian firefighters report mental health disorders at a prevalence rate of 20.2% [26], which is approximately twice that of the general Canadian population. In the current study, participating firefighters who screened positive for insomnia were statistically significantly more likely to screen positive for PTSD, GAD, PD, SAD, and MDD (OR between 4.98 to 8.53) ( Table 4). Future researchers should explore the differences in the sleep and health between volunteer and professional firefighters. Insomnia is associated with several comorbidities and firefighters are already at increased risk, relative to the general population; as such, tailored interventions to improve the sleep of firefighters and other PSP should be developed and assessed by future researchers.