Mental disorder symptoms among paramedics in a single Canadian site 1

There is growing recognition in research and policy of a mental health crisis among Canada’s paramedics but despite this, epidemiological surveillance of the problem is in its infancy. Just weeks before the emergence of the COVID-19 pandemic, we surveyed paramedics from a single, large, urban paramedic service in Ontario, Canada to assess for clinically significant symptoms of Post-Traumatic Stress Disorder (PTSD), depression, and anxiety and to identify potential risk factors for each. In total, we received 589 completed surveys (a 97% completion rate) and found that 11% met the screening criteria for PTSD, 15% for depression, and 15% for anxiety, with 1 in 4 active-duty paramedics meeting the criteria for any of the three as recently as February 2020. In adjusted analyses, the risk of a positive screen varied as a function of employment classification, gender, self-reported resilience, and previous experience as a member of the service’s peer support team. Our findings support the position that paramedics have high rates of current mental illness – a problem likely to have worsened since the onset of the COVID-19 pandemic. We echo the calls of researchers and policymakers for urgent action to support paramedic mental health in Canada.


Introduction
A 2012 systematic review and meta-analysis calculated the international pooled prevalence of Post-Traumatic Stress Disorder (PTSD) among rescue workers at 10% [1]. When stratified by occupation, ambulance workers were found to be at an increased risk for PTSD with a pooled prevalence of 14% [1]. In fact, the ambulance worker strata was used as the reference category in the authors' relative risk calculation modelling. The findings parallel a growing body of research internationally that points to high rates of mental illness among Emergency Medical Services (EMS) workers [2][3][4][5][6][7][8] both in general and comparatively within the first responder group.
Addressing a conspicuous absence of similar research in Canada, a recent national survey of public safety personnel found that 1 in 4 participating paramedics met the screening criteria for PTSD, 1 in 3 for depression, and 1 in 3 for an anxiety disorder, with nearly half screening positive for any one mental disorder [9] and higher rates being observed among female paramedics. Related research has also suggested that exposure to trauma [10][11][12][13][14][15][16]; chronic pain [17]; disturbed sleep [18]; suicidal ideation, planning, and attempts [19]; alcohol use disorders [9]; and a history of childhood abuse [20] are common among paramedics, again varying across demographic categories, including gender. The downstream consequences of this pattern of poor mental health are significant, and include lost time from work [21], family hardship [22], reduced quality of life [23][24][25][26], and suicidality [7,27], all of which can contribute to professional burnout [6,28], workplace incivility [29,30], attrition [31] and potentially compromise patient care [32]. The situation has been characterized in research and policy as a 'crisis in Canada' [33,34].
One response to the problem has been a growing interest in workplace resiliency training programs. Resilience is generally held to be the ability for individuals to 'bounce back' from adversity [35], and the resilience narrative suggests that: (1) individuals can cultivate skills that enhance their resilience; and (2) more resilient individuals are less likely to experience mental illness [36][37][38][39]. The result has been an increasing adoption of workplace resiliency training within the public safety professions through programs such as the Road to Mental Readiness (R2MR) developed by the Canadian Armed Forces [40], and its civilian analogue The Working Mind [41].
Evidence supporting the resilience hypothesis, however, has generally been underwhelming, with recent research showing only small effects on mental health literacy, stigma reduction, or the risk of mental illness [42][43][44].
Any response to the mental health crisis within the paramedic community requires a nuanced understanding of the epidemiology of work-related mental illness. Although the extant research in Canada paints a concerning picture, there are methodological limitations that cloud its interpretation. In generating the dataset, the research team cast an intentionally broad net, using a combination of social media advertising, and employer and paramedic association list servers to recruit participants. In total, approximately 600 paramedics from across Canada participated, but both the response rate and characteristics of the population from which the sample was drawn are unknown. This makes understanding the scope of the problem challenging, as results from selfselected samples can be difficult to interpret. It may be, for example, that paramedics who have experienced mental health challenges are more likely to volunteer for survey research on the topic. Or, conversely, the opposite may be true. In either case, however, the precision of the existing prevalence estimates -and by extension, our understanding of risk factors -may be called into question when the representativeness of study samples is uncertain. Therefore, our objective was to generate robust estimates of the prevalence of symptoms consistent with various mental disorders among paramedics in a single, large, urban paramedic service in Ontario, Canada. Given the growing interest in the resilience narrative in the population, we also sought to explore the relationship between self-reported resilience on the risk of a positive mental disorder screen. Our study draws on and replicates the work of Carleton and colleagues [9] while carefully controlling participant selection.

Overview
Our study took place pre-COVID-19 between September 2019 through February 2020. We distributed an in-person cross-sectional survey to a single, large, urban paramedic service in Ontario, Canada. In addition to a demographic questionnaire, our survey contained validated self-report measures for resilience, PTSD, anxiety, and depression. Our study received ethics approval from the Hamilton Integrated Research Ethics Board (HiREB protocol number 7595), and all respondents provided informed consent to participate.

Setting and Participants
We conducted our study in Peel Region, Ontario, Canada. Peel Regional Paramedic Services is the publicly funded sole provider of land ambulance and paramedic services to the municipalities of Brampton, Mississauga, and Caledon, with a total population of 1.3 million residents across a mixed suburban and rural geography of 1,200km 2 . At the time of recruitment, the service employed a total of 714 paramedics responding to approximately 130,000 emergency calls annually, making the service the second largest in the province by staffing and caseload.
Workplace resiliency training in the form of the R2MR program was launched in the service in 2017.
Paramedics in Peel Region are required to complete semi-annual continuing medical education (CME) activities. We distributed our survey during the fall 2019 / winter 2020 CME sessions.
Following a brief presentation by the principal investigator, consenting paramedics were given approximately 20 minutes at the beginning of the day to complete the survey. All participants were given a list of mental health resources available in the community and a $10 Tim Horton's gift card. This recruitment strategy has been used previously among paramedic services in Ontario [45] and was specifically chosen for its potential to generate large response rates.

Survey
Unlike in previous studies, in which participants could complete surveys over multiple sittings, our participants had only a limited amount of time during their CME to participate. This necessarily constrained the number, length, and complexity of questionnaires we could include in our survey. Each instrument is described below.

Demographics
Our demographic questionnaire was developed through consensus among the research team and was intended to gather data on criteria identified in the literature to be associated with an increased risk of mental illness among paramedics. This included age, gender, relationship status, education, provider classification (primary or advanced care), employment classification (full vs. part-time), years of experience, and current or previous participation on the service's peer support team. The peer support team was established in 2017 and was originally intended to provide non-clinical, empathetic support for home and family stressors.
We decided not to collect data on race or ethnicity. While the link between race and ethnicity and health outcomes is well-established [46,47], and other studies of paramedic mental health have included ethnicity as a potential predictor [9], the relative racial and ethnic homogeneity of our sample (an extrapolation based on similar research [9,[48][49][50]) would likely have left our study under-powered to detect such an effect.

Resilience
We evaluated self-reported resilience using the 5-item Brief Resilience Scale (BRS) [51]. The BRS asks respondents to rate their agreement with various statements that characterize how well they recover from adversity (i.e., "I tend to bounce back quickly after hard times"). Response options range from 1 ("Strongly Disagree") to 5 ("Strongly Agree"). After reverse coding three items, the scores are summed and divided by the number of items answered. The scale categorizes respondents into 'low' (<3), 'normal' (3.00-4.30) or 'high' (>4.31) levels of resilience [51] and has been used in previous studies of mental health among public safety personnel [44]. Based on the resilience narrative, we hypothesized that self-reported resilience scores would be inversely associated with mental health symptoms.

Self-Report Symptom Measures
The PTSD Checklist-5 (PCL-5; [52]) is a 20-item self-report measure that assesses four criteria specified in the Diagnostic and Statistical Manual (DSM) version 5 for diagnosing PTSD: intrusion, avoidance, alterations in cognition or mood, and arousal or reactivity. Symptoms are rated on a 5-point anchored scale from 0 ("Not at all") to 5 ("Extremely") within the past 30 days. Possible scores range from 0 to 80, with a summed score >31-33 providing a sensitivity of 88% and a specificity of 69% for probable PTSD [53] when compared to clinical interviews.
Consistent with similar studies in this population [9], we used a cut score of >32 to indicate a positive screen for PTSD. Typically, the diagnosis of (or screening for) PTSD is made in reference to an index trauma ('Criterion A'), evaluated using a separate questionnaire. However, given the increasing recognition in scholarship and policy that public safety personnel encounter multiple traumas as a result of their work, we omitted the Criterion A screen.
To evaluate for symptoms of depression, we used the 9-item Patient Health Questionnaire (PHQ-9) [54]. The PHQ-9 assesses the degree to which depressive symptoms (such as loss of interest, or difficulty concentrating) have affected the respondent over the past 14 days. Symptoms are rated on a 4-point anchored scale from 0 ("Not at All") to 3 ("Nearly Every Day"), with summed scores >9 corresponding to an 85% sensitivity and 82% specificity for probable depression when compared to clinical interviews [55].
Finally, to evaluate for symptoms of anxiety, we used the 7-item Generalized Anxiety Disorder (GAD-7) scale [56]. The GAD-7 assesses the degree to which symptoms of anxiety (such as feeling on edge) have affected the respondent over the last 14 days on a 4-point anchored scale from 0 ("Not at All") to 3 ("Nearly Every Day") with summed scores >9 corresponding to an 89% sensitivity and an 82% specificity for probable generalized anxiety disorder when compared to clinical interviews [56].

Analysis
We used descriptive statistics to characterize our data, including measures of central tendency, distribution (e.g., skewness and kurtosis), and dispersion for continuous variables and counts and percentages for categorical data. To explore group differences among our participants, we used one-way analysis of variance (ANOVA) and chi-square tests for continuous and categorical data, respectively. To evaluate the internal consistency of the self-report measures, we calculated Cronbach's alpha for each screening tool.
We used logistic regression modelling to explore the relationship between demographic characteristics and self-reported resilience on the risk of positive screens. We first constructed unadjusted, univariate logistic regression models to test the association between each demographic variable and our outcomes of interest. Given the exploratory nature of our study, we then entered all demographic variables into adjusted logistic regression models, making no effort to organize the variables hierarchically based on theoretical or statistical significance.
Where we made an effort at organizing our model parameters was in the exploration of interaction effects. Here, we constructed interaction terms based on group differences in demographic variables and their association with our outcomes of interest in our unadjusted models. Importantly, our primary interest was in the individual odds ratios of the covariates we included, rather than the predictive capacity of the models as a whole

Participation & Response Rate
We distributed a total of 607 surveys to paramedics attending the fall 2019 CME sessions, of which 600 completed surveys were returned. Of these, we excluded 11 for large portions of incomplete data, leaving a final sample of 589 surveys for analysis, corresponding to a response rate of 98.8% and a completion rate of 97%.
During the CME sessions, a total of 107 paramedics (15% of the total workforce) were on longterm leave ( Figure 1). Although we originally intended to distribute surveys to paramedics on leave via postal mail early in the new year, personnel within the service who could have facilitated this stage of recruitment were redeployed in response to the COVID-19 pandemic and we were unable to contact paramedics on leave. Five percent (n = 29) of our participants reported being a current or former member of the service's peer support team.
When stratified by gender, women were on average younger ( were also less likely to work full-time (OR 0.77, 95% CI 0.54-1.09) and more likely to be (or to have been) members of the peer support team (OR 1.25, 95% CI 0.59-2.65); however, neither difference reached the 5% significance threshold (p = 0.14 and 0.55, respectively).

Resilience
The internal consistency for the BRS in our survey was 0.85, consistent with previous investigations [44]. Across all participants, the average BRS score was 3.73 (95% Confidence Interval [CI] 3.68-3.79), corresponding to "Normal" levels of resilience [51]. A total of 63 (10.6%) of our participants met the criteria for "Low" levels of resilience. While mean BRS scores were higher among participants in a relationship (3.77 (SD 0.68) vs. 3.62 (SD 0.61), p=0.002) and advanced (compared to primary) paramedics (3.82 (SD 0.65) vs. 3.69 (SD 0.67), p= 0.02), we did not observe any significant differences in the proportions of participants meeting the threshold for "Low" resilience across demographic categories.

Unadjusted Models
In our unadjusted models (

Adjusted Models
Given the exploratory nature of our study, we included all demographic variables in our adjusted models. We also included interaction terms for gender*education (college versus university) and gender*employment status (part-time versus full-time) given the group differences we observed.
In our adjusted models, working full-time (OR 3.06, 95% CI 1.70-5.50, p<0.001) and having While "low" resilience persisted as a significant association when evaluating individual mental health outcomes, the associations with other demographic characteristics varied depending on the outcome being tested (Table 2b).

Discussion
The goals of this study were to estimate the prevalence of symptoms consistent with three specific mental disorders potentially associated with public safety work, and to explore the relationship between these mental health outcomes and demographic variables and self-reported resilience. Because much of the extant research has relied social media or email list servers to recruit participants, the concern is that the possibility of response bias may produce results that over or underestimate the true prevalence. In that respect, our response rate of 98% is a strength of our investigation and our results findings are simultaneously encouraging and concerning.
Among our sample, 11% of our participants screened positive for PTSD, 15% for depression, and 15% for anxiety, with 25% screening positive for any one of the three mental health outcomes. Our estimates are lower than those reported among paramedics in a recent national study of public safety personnel in Canada [9]. This is encouraging because, while our findings are admittedly limited to a single site, it suggests that when participant selection is carefully controlled, the prevalence of mental disorder symptoms among paramedics may be lower than has been previously described in this population.
Nevertheless, our findings are concerning for two reasons. First, our study supports the position that the prevalence of current mental illness among paramedics is significantly higher than rates observed in the general population in Canada [9,10,13,19,60]. In total, 1 in 4 of the active-duty paramedics in our study site met the criteria for potentially diagnosable levels of any one of the three mental disorders we screened for -a problem that has likely only worsened since the emergence of the COVID-19 pandemic. What we don't know is the degree to which this proportion of participants have sought or are undergoing care for these symptoms. A number of studies speak to the stigmatization of mental illness within the public safety professions [61][62][63], and the reluctance of public safety personnel to seek out professional help [2,63,64]. It is unfortunately likely that many are not receiving care at all. Understanding barriers to accessing mental health care among paramedics is an important topic for future research.
Our second objective was to examine the associations between various demographic characteristics and the risk of screening positive for PTSD, depression, or anxiety. Previous research in the Canadian population would suggest that women are more likely than men to report current or past-year prevalence of any of the three, and that age, socioeconomic status, and education are also important predictors of mental health [65]. Similarly, recent findings point to differences in the risk of mental illness among paramedics when stratified by gender, age, education, and relationship status [9,19,60]. Our findings both align and contrast with this body of research. First, we did observe differences in risk attributable to gender. In our adjusted models, women were more likely to screen positive for depression and anxiety, but less likely for PTSD.
Help-seeking behavior has been shown to differ across genders [66]. This plays out particularly with depression, where, among men, hegemonic conceptualizations of masculinity and stigma conspire to limit reporting and diagnosis [66,67]. Among our sample, it may be that women are more willing to disclose symptoms and -for the PTSD outcome -more likely to be off work because of this gendered difference in help-seeking. In total, 61 members of the paramedic service were on leave due to illness or disability during the study, although we do not know the distribution of genders of the paramedics who were on leave. Further exploration of gendered differences in help-seeking and stigma in the context of public safety work is a topic worthy of further study.
Our findings also diverged from previous research among paramedics in the risk of mental illness when stratified by relationship status, education, and provider classification. Whereas previous studies have found higher rates of mental illness among advanced care paramedics [64] and protective effects of higher education [9,60] and being in a relationship [9,60], our findings did not bear this out. The only exception was where we found that participants with university level education were half as likely to screen positive for anxiety. Given that women in our sample were more likely to attend university than men and more likely to screen positive for depression or anxiety, we tested an interaction term between gender and education. While our point estimates for the term generally favored a protective effect (except for PTSD), our confidence intervals suggest that the effect is compatible with either an increased or decreased risk of adverse mental health outcomes. Where we did observe significant associations were in the relationships between self-reported resilience, employment classification, age, and experience as a member of the service's peer support team. Of the three, resilience and experience as a peer supporter warrant careful consideration.
Resilience is generally held to be the degree to which an individual can 'bounce back' from adversity [68]. The thought is that resiliency skills are teachable, and we have seen a growing trend of developing (and marketing) workplace resiliency training programs. When evaluated empirically, the effects of workplace resiliency training are modest [35,41,[68][69][70][71], with research suggesting only small gains in mental health, stigma-reduction, or attitudes toward help-seeking.
For example, a longitudinal study of Calgary police officers after completing the R2MR program did not indicate improvement in self-reported mental health symptoms or resilience at 6 or 12 months following the intervention [44]. As a whole, the topic of resiliency is not without controversy. Though cultivating resilience to adversity may be desirable, the narrative risks shifting the locus of control onto the individual, potentially removing the responsibility of employers to mitigate the risks posed by chronic workplace stressors or exposure to trauma.
Controversy aside, our findings consistently and strongly point to a relationship between selfreported resilience and the risk of mental illness. The topic warrants further study.
Peer support as a concept has been the subject of discussion in the public safety professions for many years, owing, in part to, critical incident stress management programs [72]. More broadly, peer support was popularized by the consumer-survivor movement of the 1970s in which patients eschewed from the (at the time) paternalistic medical models of psychiatric care in favor of seeking out the support of likeminded people with similar lived experiences [73]. Among public safety personnel, efficacy evidence of peer support teams is mixed [74][75][76], but the health of the peer supporters themselves has not (to our knowledge) been studied. In our sample, we observed that being or having been a member of the service's peer support team was associated with a 4-fold increase in the risk of screening positive for PTSD and a more than 3-fold increase in the risk of depression. Interpreting this relationship is difficult. In our site, peer supporters were recruited on the basis of having lived experience with adversity, including mental illness, potentially confounding the association. That said, recruitment for the program occurred two years before our study, and our self-report measures survey probe for symptoms present within the last 14-30 days. Although it is certainly possible that members of the peer support team screened positive at higher rates due to persistent symptoms of pre-existing mental illness, it is also possible that the exposure to vicarious trauma in providing empathetic support places peer supporters at an increased risk of poor mental health. Given the growing popularity of these programs and the dearth of both efficacy [75] and safety evidence, our findings emphasize the importance of studying peer support programs more closely, including the potential health consequences for peer supporters.

Limitations
Our findings should be interpreted within the context of certain limitations. First, cross-sectional research does not lend itself to establishing causality. We acknowledge that we are assuming that the mental health outcomes we studied are attributable in some way to the participants' work as paramedics. This is increasingly supported by policy, however, given the increase in legislation in which a diagnosis of PTSD is presumed to be work related among public safety personnel to help facilitate access to treatment. Second, we acknowledge that self-report measures, while widely used, are a surrogate outcome and not diagnostic in and of themselves. Third, while our study site was carefully selected to be illustrative of a large, sophisticated, urban paramedic service, it is nevertheless a single site, and readers must exercise caution in generalizing our findings. Fourth, our approach to modelling and subsequent statistical power depended on the event rate observed in our study. We could estimate this beforehand, but the basis for our study was predicated on an assumption of overestimated prevalence, the degree of which was difficult to know at the outset. We attempted to account for this in the design of our survey to limit the number of predictors, but we acknowledge a risk of overfitting our models. We would evaluate this risk as low, given that we had between seven and nine events per covariate [77]. Finally, our study excluded 107 members of the service who were on leave during recruitment, making our sample vulnerable to a degree of selection bias, but even so, we would suggest that our findings have unique value in reflecting the mental health of the active-duty workforce.

Conclusion
Our findings are encouraging in that the rates of mental disorder symptoms that we observed in our study are lower than have been previously reported in Canada. Although admittedly limited to a single site, this suggests that the ways in which we gather this data may have important implications for its interpretation. At the same time, the rates we observed are notably higher than reported in the Canadian population at large. Fully 1 in 4 active-duty paramedics in our study had potentially diagnosable levels of symptoms consistent with either PTSD, depression, or anxiety, pointing to a mental health crisis within the profession that -with the emergence of the COVID-19 pandemic -has likely only worsened. We echo the growing calls within scholarship and policy for urgent action to support the mental health and well-being of public safety personnel in Canada.

Funding
This research was funded by a Canadian Institutes of Health Research (CIHR) catalyst grant (competition number 201809PPS). The article processing charge was covered by this grant.
We wish to express our gratitude and appreciation to Peel Regional Paramedic Services leadership team and the Ontario Public Services Employees Union (OPSEU) Local 277 for their support of this study, and to the members of the paramedic service for the important work they do in keeping our communities safe.

Author Contributions
Conceptualization

Informed Consent Statement
Informed consent was obtained from all subjects involved in this study.

Data Availability Statement
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy restrictions and data security procedures stipulated in the Research Ethics Board (REB) review of this project.