Mental Health and Wellness of Service Providers Working with People Experiencing Homelessness in Canada: A National Survey from the Second Wave of the COVID-19 Pandemic.

OBJECTIVE: This study examined the scope of common mental health problems and perceived impacts of the COVID-19 pandemic among direct service providers working with people experiencing homelessness in Canada.
METHOD: This cross-sectional study used an online survey that was disseminated to homeless service, supportive housing, and harm reduction organizations and networks. Data were collected on depression, anxiety, stress, post-traumatic stress, compassion satisfaction and fatigue, and substance use problems as well as pandemic-related changes in mental health and wellness. A total of 701 service providers completed the survey and were included in data analysis. Descriptive statistics were used to examine the primary research questions, with hierarchical multiple regression models also being fit to explore mental health and wellness differences by occupational service setting.
RESULTS: Most direct service providers (79.5%) working with people experiencing homelessness reported a decline in their mental health during the pandemic. There were high rates of common mental health problems within the sample that are largely consistent with those found among health-care workers during the pandemic. Occupational service settings were not associated with the severity of mental health problems, indicating pervasive issues across the workforce, though providers who were younger and spent more time in direct service roles were at greater risk.
CONCLUSIONS: The common mental health problems and negative impacts of the pandemic among service providers working with people experiencing homelessness highlight a highly vulnerable workforce that could benefit from improved access to supports. Given the similarities between our findings and other studies examining essential workforces, it is recommended that initiatives that provide accessible mental health care to the health-care workforce during the pandemic be expanded to include homeless and social service providers.

Introduction

Approximately 6,300 workers provide care and support to the 235,000 people in Canada who experience homelessness every year. The size of this workforce grows exponentially when including the adjunct social and community service sector, which may also work with people experiencing or at risk of homelessness. The health of service providers is instrumental to the delivery of quality and effective care, as burn out is associated with lower service satisfaction and poorer health outcomes for patients. The consequences of inadequate or poor service delivery are even greater for people experiencing homelessness, as these can lead to service disengagement and prolonged homelessness. Yet, direct service work is demanding and stressful, and homeless sector services often have limited resources and difficulties retaining staff. A pre-pandemic study showed that one-third of emergency shelter workers in Alberta reported post-traumatic stress symptoms, and one-quarter had elevations on an index of burn out. However, beyond this evidence, there is a dearth of research on the rates of common mental health problems among service providers working with people experiencing homelessness.

The COVID-19 pandemic represents a third formidable challenge faced by the homelessness sector in Canada, which was already struggling with the affordable housing and overdose crises. People experiencing homelessness were identified early on in the pandemic as a population at greater risk of negative outcomes from the virus and pandemic, assertions that are now empirically supported by a robust and growing evidence base. Efforts to reduce the spread of COVID-19 led to changes in how services were delivered to people experiencing homelessness (e.g., reduced emergency shelter capacity, provision of virtual supports), alongside the proliferation of sizable outdoor encampments in many cities. The complexities of this changing landscape have far-reaching implications, including to the mental health and well-being of direct service providers working in the homelessness and housing sectors. However, this essential workforce’s well-being throughout the pandemic has yet to be examined. This cross-sectional study used national, representative data to examine the scope of common mental health problems and perceived impacts of the pandemic among direct service providers working with people experiencing homelessness in Canada.

Methods

Participants and Recruitment

An environmental scan of homeless, supportive housing, and harm reduction sector service organizations and networks in each province and territory was conducted, with invitations to participate in the study being subsequently sent to over 300 identified agencies and groups. Individuals were eligible to participate in the study if they (a) were 18 years of age or older, (b) worked in Canada, (c) provided direct services to people experiencing homelessness, and (d) worked in homeless (including community-based health services specializing in care for people experiencing homelessness), supportive housing, or harm reduction services. A total of 948 individuals began the online survey. Of them, 701 (73.9%) completed the measures used in this study and were included in the analysis. A CAD $3 donation to a charity of participants’ choosing was provided to service providers who completed the survey. This study was reviewed and approved by the Centre for Addiction and Mental Health’s Research Ethics Board.

Data Collection

Data were collected using an online survey that consisted of both standardized measures and instruments developed or adapted for this study. The survey was created in REDCap Electronic Data Capture and hosted at the Centre for Addiction and Mental Health. The survey was available for 2.5 months during the second wave of the COVID-19 pandemic in Canada, from November 12, 2020, to January 31, 2021. During this period of time, the average number of daily cases and deaths due to COVID-19 across the country were 6,200.70 and 114.96, respectively. The survey was available only in English, as a number of measures had not been validated in other languages including French. Common mental health problems were measured using the Depression Anxiety and Stress Scales (DASS-21), Professional Quality of Life Scale (ProQOL), Abbreviated PTSD Checklist for Civilians (PCL-6), the CAGE Adapted to Include Drugs (CAGE-AID), and items examining the effects of the COVID-19 pandemic.

The DASS-21 is a set of 3 self-report scales measuring the emotional states of depression, anxiety, and stress. A subscale score is computed for each emotional state ranging from 0 to 42, with higher scores reflective of more severe symptomatology. The measure has adequate convergent and discriminant validity and good internal consistency in both clinical and community samples.

The ProQOL is a 30-item measure of the positive and negative effects of helping others who experience suffering and trauma. Despite widespread use of the ProQOL in workplace mental health research, its psychometric properties are not well-established. Due to this measurement limitation, the tool was scored in 2 ways. The first was consistent with the measure’s original procedures to enable contextual comparisons between our findings and those of other studies examining the health and well-being of health professionals during the pandemic and the homeless service workforce. This standard scoring approach involved computation of subscale scores for compassion satisfaction, burn out, and compassion fatigue that ranged from 10 to 50; higher scores on the Compassion Satisfaction subscale reflect greater pleasure derived from work, whereas higher scores on the other 2 subscales are indicative of more negative occupational impacts. Internal consistency for the subscales in our sample ranged from adequate to good (Cronbach αs = .77 to .89). Two additional scores—(1) Compassion Satisfaction and (2) Compassion Fatigue—were computed following procedures based on Rasch analysis, which demonstrated satisfactory content validity (ProQOL-21). The Compassion Satisfaction subscale consisted of 10 items that produced a score from 10 to 36, whereas Compassion Fatigue consisted of 11 items and yielded a score from 10 to 46; higher scores are reflective of greater satisfaction and fatigue, respectively.

The PCL-6 is a 6-item screening tool for post-traumatic stress symptoms in the past month. A total score was computed that ranged from 6 to 30, with higher scores suggestive of problems due to post-traumatic stress. The full scale has well-established reliability and validity for use with civilians. The PCL-6 has been shown to account for 94.3% of the variance of the full scale, and a cut-off of 14 has excellent sensitivity and adequate specificity.

The CAGE-AID was used to screen for problems due to substance use. The 4 yes–no items were summed to create a total score ranging from 0 to 4. Two or more positive responses indicate a positive screen; this approach has adequate sensitivity and good specificity.

The effects of the COVID-19 pandemic on the mental health and wellness of service providers were assessed using items developed for this study, with content considerations informed by other COVID-19 surveys. Questions asked about contraction of COVID-19, provision of homeschooling, provision of care to non-child dependents (e.g., elderly parents), receipt of the Canada Emergency Response Benefit (CERB), financial problems, changes in work hours, access to personal protective equipment in the workplace, support from work colleagues, changes in work effectiveness, changes in substance use, and changes in mental health and wellness. All items measuring perceived changes in mental health and substance use during the COVID-19 pandemic used a 5-point Likert-type scale.

Demographic and occupational information was also gathered on gender, age, ethnicity, relationship status, level of education, occupational service setting, length of time in current role and service sector, hours and term of work, lived experience of homelessness and behavioural health problems, and amount of weekly direct contact with service users. Several items on access to health care drawn from previous protocols were also included.

Data Analysis

Descriptive statistics were used to examine the rates of common mental health problems and effects of the COVID-19 pandemic within the sample. Hierarchal multiple linear regression models were used to explore the extent to which common mental health problems were associated with providers’ occupational service settings. Predictor variables were entered into the regression models in 2 blocks. The first block consisted of individual characteristics and occupational roles (gender, age, relationship status, and percentage of work involving direct service contact) as proximal factors of common mental health problems. The second block included occupational service settings (homeless services, harm reduction programs, supportive housing, and community-based health services) as more distal, contextual predictor variables. χ2 tests, Mann–Whitney U tests, and independent samples t tests were used to examine any differences between survey completers and non-completers. All analyses were conducted using SPSS Version 25.

Results

The demographic and occupational characteristics of survey completers and non-completers were similar, with no significant differences being found between the 2 groups in age, gender, level of education, household income, and length of time working in one’s current job and the sector. Of the 701 direct service providers who completed the online survey, most participants were white women with college diplomas or bachelor degrees who worked full-time in their jobs with people experiencing homelessness (see Table 1). The mean age of participants was 38.74 years (SD = 12.42). Overall, the sample characteristics were similar to 2016 Census data on the composition of the homeless and social service workforces, with the exception of education (a higher proportion of participants had university degrees in our sample). Ontario and Eastern Canada were slightly more represented in our sample proportional to the size of the workforces in those provinces, whereas Quebec, the Prairies, and British Columbia were more underrepresented.

Table 1.

Sample Characteristics.

Characteristic	n	Percentage
Region
West (BC)	108	15.4
Prairies (AB, SK, MB)	95	13.6
North (YK, NT, NU)	7	1.0
Central (ON, QC)	416	59.3
East (NB, NS, NL, PE)	73	10.4
Gender
Female	561	80.0
Male	114	16.3
Transgender/nonbinary	21	3.0
Ethnicitya
White	614	87.6
Indigenous	54	7.7
Black (Africa, Canada/North America, and Caribbean region)	37	5.3
Asian (East, South, and South East)	33	4.7
Middle Eastern	12	1.7
Latin American	12	1.7
Indian-Caribbean	3	0.4
Other	33	4.7
Education
High school or less	52	7.4
College diploma	224	32.0
Bachelor’s degree	338	48.2
Graduate school degree	86	12.3
Occupational service setting
Homeless serviceb	280	39.9
Supportive housingc	197	28.1
Community health serviced	102	14.6
Harm reduction programe	75	10.7
Otherf	47	6.7
Work term
Full-time	569	81.2
Part-time	131	18.7
Lived experience of mental health or substance use problems	403	57.5
Lived experience of homelessness	108	15.4

Note. N = 701. BC = British Columbia; AB = Alberta; SK = Saskatchewan. MB = Manitoba; YK = Yukon; NT = Northwest Territories; NU = Nunavut; ON = Ontario; QC = Quebec; NB = New Brunswick; NS = Nova Scotia; NL = Newfoundland and Labrador; PE = Prince Edward Island.

a Exceeds 100% due to mixed heritages and multiple self-identified ethnicities.

b For example, emergency shelter, drop-in program, soup kitchen and meal program, street outreach, and respite program/warming centre.

c That is, permanent or transitional housing program with supports.

d For example, intensive case management, assertive community treatment, community health centre, and inner-city health team.

e For example, supervised consumption and overdose prevention site, needle exchange/safer inhalation service, opioid agonist therapy clinic, naloxone training/provision program, and managed alcohol or opioid program.

f For example, income support and employment programs, youth outreach and diversion, and victim prevention and protection services.

Rates of common mental health problems are presented in Table 2. Occupational service settings (homeless service, supportive housing, harm reduction, or community health service) were not significantly associated with common mental health problems in the hierarchal multiple linear regression models after accounting for gender, age, relationship status, and amount of direct service work, with the exception of a small negative association between supportive housing and compassion fatigue (see Tables 3 and 4). More time spent in direct contact with service users and younger age were significantly associated with greater problems across many of the common mental health domains, with small effect sizes.

Table 2.

Mental Health and Wellness of Direct Service Providers during the COVID-19 Pandemic.

Domain	M (SD)	Clinical Range/Screen Interpretationa
Depressive symptoms (DASS-21)	10.99 (9.96)	Normal/mild (≤13): 65.9%
		Moderate (14 to 20): 18.7%
		Severe (21 to 27): 6.8%
		Extremely severe (≥28): 8.6%
Anxiety symptoms (DASS-21)	8.47 (8.61)	Normal/mild (≤9): 64.9%
		Moderate (10 to 14): 14.3%
		Severe (15 to 19): 7.7%
		Extremely severe (≥20): 13.1%
Stress symptoms (DASS-21)	15.04 (9.78)	Normal/mild (≤18): 67.8%Moderate (19 to 25): 15.0%
		Severe (26 to 33): 12.1%
		Extremely severe (≥34): 5.1%
Post-traumatic stress symptoms (PCL-C)	13.15 (5.20)	294 (41.9%) has a positive screen (≥14)
Problematic substance use (CAGE-AID)b	0.91 (1.24)	173 (24.7%) has a positive screen (≥2)
Burn out symptoms (ProQOL)	24.45 (5.97)	Low (≤22): 40.1%
		Moderate (23 to 41): 59.8%
		Severe (≥42): 0.1%
Secondary traumatic stress (ProQOL)	22.36 (7.04)	Low (≤22): 55.3%
		Moderate (23 to 41): 42.9%
		Severe (≥42): 1.7%
Compassion satisfaction (ProQOL)	38.26 (6.01)	Low (≤22): 0.7%
		Moderate (23 to 41): 68.5%
		High (≥42): 30.8%
Compassion satisfaction (ProQOL-21)	24.63 (5.59)	None available
Compassion fatigue (ProQOL-21)	24.09 (7.92)	None available

Note. DASS = Depression Anxiety and Stress Scales; PCL-C = PTSD Checklist for Civilians; ProQOL = Professional Quality of Life Scale.

a Clinical range/screen interpretation sources: DASS-21, ProQOL, PCL-6, and CAGE-AID = The CAGE Adapted to Include Drugs .

b M (SD) excludes participants who denied using substances (n = 128).

Table 3.

Hierarchical Multiple Regression Models Predicting Common Mental Health Problems on the DASS-21 and CAGE-AID.

Predictor	Depressive Symptoms				Anxiety Symptoms				Stress Symptoms				Problematic Substance Use
	β	B	SE	P	β	B	SE	P	β	B	SE	P	β	B	SE	P
Female, transgender, or nonbinary gendera	−.01	−0.26	1.04	0.80	.01	0.25	0.88	0.77	.04	1.14	1.00	0.25	−.12	−.39	.12	<0.01
Age	−.10	−0.08	0.03	0.02	−.16	−0.11	0.03	<0.001	−.14	−0.11	0.03	<0.001	−.15	−.02	.00	<0.001
Married or partneredb	−.06	−1.28	0.79	0.11	−.01	−0.09	0.67	0.89	.01	0.24	0.76	.76	−.02	−.05	.09	0.63
Time in direct contactc
26% to 50%	.03	0.77	1.38	0.58	.04	0.77	1.18	0.51	.06	1.53	1.33	0.25	.13	.38	.16	0.02
51% to 75%	.06	1.40	1.28	0.27	.09	1.63	1.09	0.14	.10	2.07	1.23	0.09	.07	.19	.15	0.20
76% to 100%	.08	1.56	1.23	0.21	.14	2.45	1.05	0.02	.15	2.97	1.19	0.01	.03	.08	.15	0.59
Work setting
Emergency shelter	−.04	−0.90	1.68	0.59	−.04	−0.78	1.43	0.59	−.06	−1.17	1.62	0.47	−.03	−.08	.20	0.68
Supportive housing	−.10	−2.29	1.72	0.18	−.09	−1.76	1.46	0.23	−.14	−2.98	1.66	0.07	−.05	−.13	.20	0.51
Community-based health	−.02	−0.65	1.86	0.73	−.05	−1.26	1.58	0.43	.00	−0.05	1.79	0.98	.00	−.01	.22	0.95
Harm reduction	.05	1.56	1.94	0.42	.07	1.86	1.65	0.26	.04	1.35	1.87	0.47	.01	.04	.23	0.86

Note. DASS = Depression Anxiety and Stress Scales; CAGE-AID = The CAGE Adapted to Include Drugs.

a Reference category: Male gender.

b Reference category: Single, divorced, or widowed.

c Reference category: 1% to 25% of time spent in direct contact with service users.

The bold values signifies the column (p-values).

Table 4.

Hierarchical Multiple Regression Models Predicting Common Mental Health Problems on the PCL-6 and ProQOL-21.

Predictor	Post-traumatic Stress				Compassion Satisfactiona				Compassion Fatigue
	β	B	SE	P	β	B	SE	P	β	B	SE	P
Female, transgender, or nonbinary genderb	−.02	−0.26	0.53	0.62	.10	1.55	0.58	0.01	.02	0.33	0.81	0.68
Age	−.14	−0.06	0.02	<0.001	.14	0.06	0.02	<0.001	−.08	−0.05	0.03	0.03
Married or partneredc	.00	−0.03	0.41	0.94	−.03	−0.34	0.44	0.45	.01	0.10	0.62	0.88
Time in direct contactd
26% to 50%	.08	1.08	0.71	0.13	−.06	−0.91	0.77	0.24	.16	3.16	1.08	<0.01
51% to 75%	.09	0.98	0.66	0.14	−.10	−1.29	0.72	0.07	.21	3.66	1.00	<0.001
76% to 100%	.19	2.01	0.64	<0.01	−.06	−0.64	0.69	0.36	.23	3.76	0.97	<0.001
Work setting
Emergency shelter	−.02	−0.22	0.87	0.80	.03	0.28	0.94	0.76	−.10	−1.57	1.32	0.23
Supportive housing	−.07	−0.83	0.89	0.35	.07	0.93	0.96	0.34	−.16	−2.84	1.35	0.04
Community-based health	−.03	−0.43	0.96	0.65	.04	0.55	1.04	0.60	−.01	−0.15	1.46	0.92
Harm reduction	.03	0.49	1.00	0.63	.01	0.26	1.09	0.82	.01	0.27	1.52	0.86

a Higher scores are indicative of greater compassion satisfaction (i.e., the directionality of the scale differs from the other common mental health problem indices).

b Reference category: Male gender.

c Reference category: Single, divorced, or widowed.

d Reference category: 1% to 25% of time spent in direct contact with service users.

The bold values signifies the column (p-values).

Participants reported health, social, and financial impacts in their lives due to the COVID-19 pandemic. Only 8 (1.1%) participants had tested positive for COVID-19 at any point since March 2020; however, a larger proportion (n = 69; 9.8%) believed that they had contracted COVID-19 but had not been tested for the virus. A total of 557 (79.5%) participants perceived that their mental health had declined; of them, 68.4% reported a slight decline, whereas 31.6% reported a substantial decline. Substance use increased for 274 (39.1%) participants, which was primarily alcohol (27.5%) and cannabis (20.8%). The majority of participants (n = 387; 55.2%) reported that they had been less able to access support from their social networks, but most (n = 535; 76.3%) also felt moderately or extremely supported by their co-workers throughout the COVID-19 pandemic. One hundred and fifty-nine (22.7%) participants reported that at least 1 individual that they served directly had died (any cause) during the COVID-19 pandemic.

Slightly more than half of participants (n = 360; 51.4%) had experienced financial problems of varying severity during the pandemic (slight: 22.4%; moderate: 19.4%; extreme: 9.6%). Yet, most service providers had either not experienced changes to their work hours (n = 364; 51.9%) or were working more hours (n = 211; 30.1%). Fifty-six (8.0%) and 62 (8.8%) participants reported major and minor work hour reductions, respectively. Only 104 (14.8%) participants had accessed CERB income support.

One hundred and thirty (18.5%) direct service providers identified an unmet need for treatment of mental health or substance use problems within the past year. Multiple reasons were often reported for not receiving behavioural health care, including not having time due to other commitments (54.6%), thinking the problem could be handled without treatment (54.6%), not having enough health insurance to afford treatment or counselling (40.0%), and not having any health insurance and being unable to afford care (33.8%).

Discussion

The findings provide a snapshot of the mental health and wellness of the workforce that serves people experiencing homelessness in Canada. A total of 79.5% of participants reported a decline in their mental health during the COVID-19 pandemic, suggesting worsening mental health in the workforce. As a survey conducted from November to December 2020 of approximately 18,000 health-care workers in Canada found that 70% reported worse mental health due to the pandemic, our study findings suggest that the pandemic is taking a similar, if not slightly greater, toll on the homeless service, supportive housing, and harm reduction workforces.

Results from the standardized measures of common mental health problems revealed a similar, concerning narrative. Approximately one-third of participants reported moderate or more severe symptoms on indices of depression, anxiety, and stress. These rates were only slightly below those found in a small study of hospital-based critical care nurses in Western Canada during the COVID-19 pandemic. With regard to post-traumatic stress symptoms, 41.9% of direct service providers had a positive screen, which slightly exceeded that of a pre-pandemic study of emergency shelter workers in Alberta using the same measure (33% screened positive in that study). The finding may reflect increased exposure to stressful and traumatic events in the workplace wherein both the pandemic and worsening overdose crisis are possible contributory factors. Further, the rates of compassion satisfaction and burn out reported by direct service providers are consistent with earlier pre-pandemic research of homeless service workers in the United Kingdom; however, secondary traumatic stress symptoms were more severe among providers in our study. Our findings on compassion satisfaction, burn out, and secondary traumatic stress were also similar to those found in a small study of U.S. frontline health-care workers during the pandemic. As occupational service settings were very minimally associated with common mental health problems, the findings suggest pervasive issues across the workforce, though providers who are younger and spend more time in direct service roles may be at greater risk. Gender was minimally associated with common mental health problems, though additional differences may be obscured by male participants being older and less involved in direct contact with service users than female, transgender, and nonbinary providers. Overall, considering the found rates of common mental health problems and unmet treatment needs, the workforce serving the homeless population should be seen as one that is highly vulnerable and could benefit from improved access to mental health supports.

As community-based organizations serving the homeless population are often under-resourced, with many direct service providers also receiving low wages, worsening mental health of the workforce during the pandemic may threaten its sustainability. Given that insufficient time due to other commitments was among the most frequently reported reasons for not seeking needed behavioural health treatment, service providers could benefit from more flexibility in their workloads and hours. The provision of more wellness days and the expansion of relief staff rosters are expected to give service providers more time for help-seeking. Interventions are also needed to support community-based direct service providers working with people experiencing homelessness. Initiatives to provide accessible mental health care, including psychotherapy and psychiatric services, to frontline health-care workers during the pandemic could be aptly expanded to include those working in homeless and social services. Development of similar frontline wellness services in jurisdictions without such programs is also recommended. Governmental financial support to expand paid sick leave and enhance job security would also be beneficial for reducing health- and financial-related stress among the workforce.

The study findings also raise concern about the levels of grief and loss within the workforce. Given the high rates of overdose, suicide, and victimization among people experiencing homelessness, service providers who work with this population are regularly exposed to and grieve the deaths of people they support. Further, given the higher mortality rate associated with COVID-19 in the homeless population, the pandemic has likely increased service providers’ exposure to death and loss. People with lived experience of homelessness, mental health, and substance use problems who work in “peer” roles are especially vulnerable, as they are less likely to receive health benefits compared to other direct service staff. As more than 1 in 5 participants reported that they had served at least 1 individual who had died during the pandemic, increasing access to grief counselling through partnerships between mental health and social service systems is recommended. Organizations can also support direct service providers by engaging them in the development of “for staff, by staff” interventions for grief and loss.

This study had several limitations. First, due to the cross-sectional study design, the extent to which the COVID-19 pandemic contributed to the high rates of common mental health problems found in this study is unknown. Although there is cause for concern with regard to the mental health and wellness of the workforce, this may improve as Canada and rest of the world transition out of the pandemic. Nevertheless, this should be monitored closely and investigated further. Second, our study used convenience sampling, and it is unknown how many homeless service, supportive housing, and harm reduction agencies disseminated the survey invitation within their organizations. Further, direct service providers who were on leave from work for mental health reasons during the recruitment period or lost their jobs due to the pandemic are likely highly underrepresented in the sample. Because of this, it is possible that the found rates of common mental health problems and impacts of the COVID-19 pandemic on this workforce are underestimated. Third, direct service providers from Quebec were notably underrepresented in the sample, likely due to the survey only being available in English. As such, the findings may be less applicable to the homeless service, supportive housing, and harm reduction workforces in that province.

Conclusion and Future Directions

The study findings highlight the high rates of common mental health problems among direct service providers working with people experiencing homelessness in Canada, which most perceived to have worsened during the pandemic. As burn out and secondary traumatic stress can precipitate staff turnover in health and social services —a prevalent human resources issue in the homelessness sector —it is important these mental health problems be addressed. Deteriorating mental health among direct service providers in the homeless service, supportive housing, and harm reduction workforces may also increase risk of negative service delivery outcomes for people experiencing homelessness. Interventions to support the workforce’s mental health needs throughout and following the pandemic are needed. Further investigation to determine which groups are most at risk of mental health and wellness problems within this workforce is also recommended.