Mental health service use among Black adolescents in Ontario by sex and distress level: a cross-sectional study

Abstract

Background:

Race is a social construct reflecting broader systemic forces that can affect health, including mental health. We sought to ascertain whether patterns of mental health care service use are associated with race among adolescents in Ontario.

Methods:

We conducted a cross-sectional study using data from the 2015–2019 Ontario Student Drug Use and Health Survey. We assessed mental health care access for students in grades 7–12 younger than 20 years based on their responses about their care usage in the past 12 months. We used logistic and Poisson regression models to analyze differences in service utilization, with interaction terms for sex and mental distress (measured using the Kessler Psychological Distress Scale–6 Items).

Results:

Black male students with low distress were nearly twice as likely as White males to report initiating care (odds ratio [OR] 1.50, 95% confidence interval [CI] 1.09–2.06). However, when Black males’ distress worsened to moderate levels, they became less than half as likely to access care than their White peers (OR 0.41, 95% CI 0.20–0.84). Black females faced disparities at all distress levels, with the gap widening as distress increased (moderate distress OR 0.78, 95% CI 0.46–1.34; serious distress OR 0.60, 95% CI 0.40–0.89). Even after initiating care, Black females mostly had lower odds of access frequency than White females (low distress OR 0.78, 95% CI 0.66–0.92; moderate distress OR 1.00, 95% CI 0.84–1.19; serious distress OR 0.60, 95% CI 0.42–0.85).

Interpretation:

Black survey respondents with psychological distress were less likely to report using mental health services than their White peers, with Black female respondents being the least likely to access care. Policy and practice should seek to address systemic racism and a lack of culturally relevant care for Black adolescents with mental distress.

Race is a socially constructed concept that can deeply shape the human experience.¹ In North America, race is associated with physical traits and country of ancestry. Race can influence the community beliefs to which a person subscribes, attitudes toward mental health, and health care–seeking behaviours.^2–4 Racial appearance affects how people are treated, with an impact on health care experiences and comfort with accessing care.^5,6 Racial differences between patients and providers can further affect health care.⁷ Yet, little data exist on mental health service utilization among racialized groups in Canada. This dearth of data can obscure disparities in mental health access among racialized groups and hinder efforts to address structural racism in mental health services in Canada. Recent discussions highlight a mental health crisis in Canada’s Black communities, particularly concerning unmet needs.^8–11 Key recommendations include creating accessible, culturally competent mental health services for Black people in Canada and measuring and reporting care outcomes.^8,9 The second recommendation aligns with Ontario’s Anti-Racism Act, 2017, and its 3-year plan to collect race-based data to eliminate systemic racism and advance racial equity.¹² However, despite this directive, data on the mental health service use of Black people in Canada remain limited, making it difficult to fully understand and address disparities in care.

Adolescence is a crucial developmental stage and a critical period for onset of mental health problems.¹³ In Canada, Black adolescents disproportionately access services through crisis situations, such as justice system interactions or when intensive care is required, suggesting they are less likely to access mental health care until intense intervention is needed.^14,15 However, little further data exist on service use for Black adolescents, and this is even more limited when the intersection of sex is considered. The Ontario Student Drug Use and Health Survey (OSDUHS) is one of the few regularly collected data sets that allows for the assessment of demographic differences in adolescent mental health and service use. We leveraged OSDUHS data to provide much-needed empirical data on mental health service utilization patterns among Black adolescents in Ontario, addressing a substantial gap in the literature and advancing our understanding of racial disparities in mental health access.

In this study, we sought to examine whether Black adolescent middle- and high-school students in Ontario accessed mental health services differently from White students, and to investigate how sex and psychological distress affect the relationship between race and mental health service use. Specifically, we asked: Are there racial differences in accessing mental health care at least once in the past 12 months, and is this modified by sex and mental distress level? Among students who accessed services, are there racial differences in the frequency of utilization, and is this relationship modified by sex and mental distress level?

Methods

Study design and setting

We conducted a cross-sectional secondary data analysis using data from the OSDUHS.¹⁶

Ontario is home to more than half (52.4%) of Canada’s Black population, and robust data on adolescent mental health, including Black adolescents, are available through the OSDUHS.^16,17

Data source

The OSDUHS is a population-based survey of Ontario middle-and high-school students (grades 7–12), conducted every 2 years by the Centre for Addiction and Mental Health. The goals of the survey are to capture trends in drug use, mental health, physical health, gambling, and risk behaviours, as well as to identify risk and protective factors.¹⁶

The partnership of the OSDUHS with Ontario schools enables a robust data-collection strategy that minimizes selection bias by capturing a representative sample of students directly within educational settings.¹⁶ A comparable data source, the Canadian Community Health Survey, primarily targets the general population rather than adolescents and is likely to underrepresent Black respondents. The OSDUHS provides a more reliable and comprehensive data set for analyzing trends and disparities in mental health service utilization among Black and White adolescents in Ontario.

The OSDUHS used a multistage, probability-proportionate-to-size systemic sampling design to select publicly funded English and French schools (public and Catholic) across Ontario’s 4 main regions: the Greater Toronto Area, and Northern, Western, and Eastern Ontario. Oversampling occurred in 10 designated public health units to provide more precise local data for those regions. Schools were eligible if they offered grades 7–12 and were excluded if they had fewer than 30 students in grades 7–8, fewer than 80 in grades 9–12, or lacked all 4 secondary grades.¹⁶

Within participating schools, researchers randomly selected eligible grades. Classes were excluded if they were special education, English as a second language, or had fewer than 4 students. Students were eligible if they could read English or French and provided signed consent. Those who were homeschooled, institutionalized, or unable to understand the survey language were excluded.¹⁶

Data collection for the 2015, 2017, and 2019 OSDUHS surveys took place between November and June in the 2014–2015, 2016–2017, and 2018–2019 academic years, respectively. The OSDUHS obtained school board and principal approvals. Within each school, a grade-stratified random selection of classes was made, ensuring equal probability for participation. In elementary and middle schools, 1–2 classes from each of grades 7 and 8 were selected, and in secondary schools, at least 1 class from each of grades 9–12 was chosen.¹⁶

For the survey, the OSDUHS used a paper-and-pencil questionnaire with 4 split-ballot forms to reduce respondent burden, varying in content by age group. Four versions of the OSDUHS questionnaire were used: Form A and Form B for both elementary (Grades 7–8) and secondary schools. To reduce burden and fit within a class period, the survey used a split-ballot design, with different sets of questions on each form. Forms A and B were distributed alternately to classrooms (i.e., A, B, A, B). Only Form A included questions on mental or emotional health, so only responses from Form A were included in this analysis. Elementary school forms were shorter and excluded certain topics, such as gender identity, sexual orientation, and illicit drug use.¹⁶ Participation was voluntary and anonymous, with active parental consent required for students younger than 18 years.¹⁶

Trained Institute for Social Research field staff administered the surveys during regular class periods, ensuring privacy and confidentiality. Completed questionnaires were collected and sent to the Institute for Social Research for data entry, with a 5% verification sample to ensure accuracy.¹⁶

In 2015, 349 schools were invited to participate, with 220 schools from 43 school boards agreeing, resulting in a participation rate of 63%. In 2017, 353 schools were invited and 214 schools from 52 school boards participated, resulting in a 61% participation rate. For 2019, 526 schools were invited, and 264 schools from 47 school boards participated, resulting in a participation rate of 50%. The primary reasons cited by nonparticipating schools were scheduling conflicts, with many indicating they were too busy or had already committed to other research projects. Participation was also affected by restrictions from certain school boards, which limited the number of times schools could be contacted for invitations.¹⁶ Survey completion rates among students were 59% in 2015, 61% in 2017, and 59% in 2019. The primary reasons for noncompletion were student absences and unreturned consent forms.¹⁶

Inclusion and exclusion criteria

We included in the analysis survey respondents from years 2015, 2017, and 2019, who identified as Black or White.¹⁶ We did not include data from 2020, given shifts in mental health experiences and behaviours related to the COVID-19 pandemic, how the pandemic disproportionately amplified pre-existing inequities and stressors for Black people in North America, and the uniquely acute mental health toll experienced by Black people in North America after the murder of George Floyd in 2020 and resulting racial unrest.^18,19 These events were important, do not reflect baseline conditions, and warrant independent investigation.

We excluded respondents who did not identify as Black or White, or identified as multiple races, from the analysis. We excluded multiple races as it is not possible to know which racial identity an individual most identified with, was raised around, or was perceived to be. Respondents with missing data on any variables of interest during analysis were excluded through listwise deletion. Given the low percentage of missing data (Table 1; 0.88% for White respondents and 1.8% for Black respondents regarding access to health care in the last year), the removal of cases with missing data is considered acceptable.

Table 1:

Descriptive characteristics of study population^*

Characteristic	All students		White students		Black students
	No. in unweighted sample	No. (%)† in weighted sample	No. in unweighted sample	No. (%)† in weighted sample	No. in unweighted sample	No. (%)† in weighted sample
Sex	n = 19 373	n = 19 853	n = 11 079	n = 10 500	n = 1289	n = 1761
Male	8591	10 231 (51.5)	4978	5484 (52.2)	564	894 (50.8)
Female	10 782	9623 (48.5)	6101	5016 (47.8)	725	867 (49.2)
Age, yr	n = 19 373	n = 19 849	n = 11 079	n = 10 500	n = 1289	n = 1761
11 or 12	2466	1914 (9.6)	1320	902 (8.6)	156	143 (8.1)
13	3314	2566 (12.9)	1889	1365 (13.0)	226	217 (12.3)
14	3355	3114 (15.7)	1909	1643 (15.6)	235	295 (16.7)
15	3395	3368 (17.0)	1997	1834 (17.5)	197	289 (16.4)
16	3089	3394 (17.1)	1805	1839 (17.5)	199	247 (14.0)
17	2789	3871 (19.5)	1656	2125 (20.2)	186	363 (20.6)
18	893	1478 (7.4)	482	753 (7.2)	75	158 (9.0)
19	68	144 (0.7)	21	39 (0.4)	15	50 (2.8)
Missing	4	–	0	–	0	–
Region	n = 19 373	n = 19 853	n = 11 079	n = 10 500	n = 1289	n = 1761
Greater Toronto Area	5935	7857 (39.6)	1987	2390 (22.8)	693	1214 (68.9)
Northern Ontario	2294	1232 (6.2)	1748	924 (8.8)	31	21 (1.2)
Western Ontario	5549	5832 (29.4)	3675	3889 (37.0)	232	226 (12.8)
Eastern Ontario	5595	4932 (24.8)	3669	3297 (31.4)	333	300 (17.0)
Self-rated mental or emotional health	n = 19 373	n = 19 662	n = 11 079	n = 10 420	n = 1289	n = 1719
Excellent	3895	4121 (21.0)	2183	2095 (20.1)	346	508 (29.5)
Very good	6058	6183 (31.4)	3575	3412 (32.7)	375	503 (29.3)
Good	5005	5193 (26.4)	2787	2659 (25.5)	309	385 (22.4)
Fair	2904	2834 (14.4)	1676	1513 (14.5)	179	243 (14.1)
Poor	1310	1331 (6.8)	769	741 (7.1)	49	80 (4.7)
Missing	201	–	89	–	31	–
K6 category
All sexes	n = 19 373	n = 19 100	n = 11 079	n = 10 178	n = 1289	n = 1689
Low	11 359	11 578 (60.6)	6779	6461 (63.5)	764	1029 (61.0)
Moderate	4018	4152 (21.7)	2141	2012 (19.8)	284	396 (23.5)
Serious	3287	3369 (17.6)	1826	1705 (16.8)	174	263 (15.6)
Missing	709	–	333	–	67	–
Males	n = 8591	n = 9765	n = 4978	n = 5302	n = 564	n = 848
Low	6041	7091 (72.6)	3657	4001 (75.5)	399	638 (75.2)
Moderate	1427	1733 (17.7)	738	830 (15.7)	92	135 (15.9)
Serious	746	941 (9.6)	407	470 (8.9)	36	75 (8.8)
Missing	377	–	176	–	37	–
Females	n = 10 782	n = 9335	n = 6101	n = 4876	n = 725	n = 840
Low	5318	4487 (48.1)	3122	2460 (50.4)	365	391 (46.6)
Moderate	2591	2419 (25.9)	1403	1182 (24.2)	192	261 (31.0)
Serious	2541	2429 (26.0)	1419	1235 (25.3)	138	188 (22.4)
Missing	332	–	157	–	30	–
In the last year, how many times did they see a doctor, nurse, or counsellor about their mental or emotional health?	n = 19 373	n = 19 605	n = 11 079	n = 10 426	n = 1289	n = 1724
Did not see a doctor, nurse, or counsellor	14 340	14 838 (75.7)	8160	7818 (75.0)	935	1283 (74.4)
Once	1523	1425 (7.3)	887	769 (7.4)	98	127 (7.3)
2 or 3 times	1474	1530 (7.8)	821	789 (7.6)	133	182 (10.5)
4 or 5 times	666	657 (3.4)	382	329 (3.2)	45	61 (3.6)
6 or 7 times	311	320 (1.6)	188	191 (1.8)	13	14 (0.8)
8 or 9 times	189	213 (1.1)	119	123 (1.2)	10	33 (1.9)
10 or more times	640	622 (3.2)	430	407 (3.9)	24	25 (1.5)
Missing	230	–	92	–	31	–
Length of time in Canada, yr	n = 19 373	n = 19 785	n = 11 079	n = 10 472	n = 1289	n = 1753
All of their life	16 066	15 808 (79.9)	10 436	9829 (93.9)	803	1010 (57.6)
≤ 2	531	665 (3.4)	48	50 (0.5)	97	158 (9.0)
3–5	648	792 (4.0)	83	92 (0.9)	87	113 (6.4)
6–10	985	1219 (6.2)	198	200 (1.9)	164	271 (15.4)
≥ 11	1063	1301 (6.6)	280	301 (2.9)	127	201 (11.5)
Missing	80	–	34	–	11	–

Note: K6 = Kessler Psychological Distress Scale–6 items, OSDUHS = Ontario Student Drug Use and Health Survey.

^*Survey weights provided by OSDUHS were applied to the data. The unweighted sample sizes are 19 373 for all adolescents, 11 079 for White, and 1289 for Black participants. We used post-stratification weights to adjust the data to better represent the broader student population. For example, before we adjusted for sex (before post-stratification weights), the 2019 survey was 45% male and 55% female, which was adjusted to 52% male and 48% female. Weighted sample sizes vary by characteristic.

^†Unless otherwise specified.

Study variables

Andersen and Newman Framework

We used the Andersen and Newman Framework of Health Services Utilization to identify covariates influencing mental health use (Figure 1).²⁰ The framework identifies predisposing factors (e.g., sociocultural influences), enabling factors (e.g., resources facilitating access), and need factors (e.g., health needs driving service use) that influence service use.²¹ These factors, in addition to the directed acyclic graph in Figure 2 — a causal diagram illustrating relationships between variables — guided the selection of variables. Predisposing and need factors were controlled for in the analysis, but enabling factors were not.

Figure 1:

Conceptual framework for health service utilization behaviour, modified from Andersen’s behavioural model.²¹ The framework identifies predisposing factors (e.g., sociocultural influences), enabling factors (resources facilitating access), and need factors (health needs driving service use) that influence health service use.

Figure 2:

Directed acyclic graph depicting relationship between race and having accessed mental health services, mediated through access to extended health care benefits that include mental health (extended health care means health insurance benefits that go beyond the coverage provided by the standard Ontario Health Insurance Plan). Note: SES = socioeconomic status.

Exposure and outcome

Race was the primary “exposure” variable and served as a proxy for the broader systemic forces of racialization and racism in Canada. In analyzing the influence of race on mental health outcomes, we delved into the broader implications of being Black in Canada. We determined the outcome, access to mental health care, based on respondents’ self-reported frequency of accessing care for their emotional or mental health in the last year. We treated access as a binary variable, either accessing care or not in the previous year, and a count variable to assess frequency among respondents who accessed care in the last year. Further descriptions of covariates can be found in Table 2.

Table 2:

Variables used in study

Covariate	Description	Survey question	Survey response options	Recoding
Race (exposure)	A proxy for the broader systemic forces of racialization and racism in Canadian society, and Ontario specifically	Which of the following best describes your background?	Common racial groups in Canada including “Black” and “White”	None
Age (covariate)	Affects mental health access through awareness and attitudes to identify and address mental health problems22	“How old are you?”	Elementary school 2015 and 2017: “10 years of age or younger” to “16 years or older” 2020: “11 years of age or younger” to “15 years or older” High school 2015 and 2017: “10 years of age or younger” to “20 years or older” 2020: “12 years of age or younger” to “20 years or older”	All responses indicating ages ≤ 12 years were recoded to “11 or 12 years of age”; we excluded respondents who indicated “20 years or older” as they are no longer adolescents
Years in Canada (covariate)	Length of time student lived in Canada. Mental health access differs based on immigration status and the recency of immigration23	“How long have you lived in Canada?”	Five response options ranging from “all of my life” to “2 years or less”	None
Month and year of survey completion (covariate)	Accounts for potential seasonal fluctuations in mental health trends across different years and month to month24,25	Provided by OSDUHS	Provided by OSDUHS	None
Access to mental health care (outcome)	Whether students accessed care in the last year and how frequently	“In the last 12 months, how often have you seen a doctor, nurse, or counsellor about your emotional or mental health?”	2015: Options ranged from “Did not see a doctor/nurse/counsellor” to “10 or more times” over the past year, with single number intervals ranging from once to 9 times 2017 and 2019: Options ranged from “Did not see a doctor/nurse/counsellor” to “12 or more times” over the past year, with specific intervals such as once, 2–3 times, 4–5 times, and so on	Reformatted so that highest frequency response was “10 or more times.” Intervals between once and 10 or more were adapted to the 2017 and 2019 ranges
Sex (covariate and effect measure modifier)	Biological attributes related to physical and physiologic features and is typically categorized as female or male.26 Sex, and gender, are both associated with mental health attitudes and seeking behaviours.22,27 From 2015 to 2019, the OSDUHS collected information on sex, which may reflect both biological and gender-related factors	2015: “Are you male or female?” 2017 and 2019: “Were you born male or female?”	“Male” “Female”	None
Psychological distress (covariate and effect measure modifier)	Kessler Psychological Distress Scale–6 items	Six questions designed to assess symptoms of psychological distress, covering feelings of nervousness, hopelessness, restlessness, depression, effortfulness, worthlessness, and their frequency over the past 30 days	Each item is rated on a 5-point Likert scale and totalled together, with higher scores indicating higher levels of distress	Scores below 8 were considered low distress; those falling between 8 and 12 inclusive were moderate; and scores of 13 or higher were labelled serious distress16

Note: OSDUHS = Ontario Student Drug Use and Health Survey.

Effect measure modifiers

We assessed sex and mental distress for effect measure modification. Sex refers to biological attributes related to physical and physiologic features and is typically categorized as female or male.²⁶ We analyzed sex rather than gender for consistency across survey years, as gender was introduced only in 2017, before which respondents were asked only about sex (male or female). We recognize this excludes nonbinary adolescents, whose mental health experiences may differ. We measured mental distress using the Kessler Psychological Distress Scale–6 items (K6).^28,29 The K6 is a robust and validated assessment tool for general use, including among adolescents in Canada, that assesses levels of distress in the 4 weeks before survey completion.^20,30,31 We considered scores below 8 to denote low distress, those from 8 to 24 moderate-to-serious distress, and scores of 13 or higher serious psychological distress.¹⁶ These ratings resulted in 3 categorical K6 scores: low, moderate, and serious.

Data analysis

We performed data preparation and analyses using SAS Studio 3.81 on SAS 9.4 software.³² All analyses applied survey weights, strata, and cluster provided by OSDUHS to be representative of the student populations in grades 7–12 enrolled in Ontario publicly funded schools. Survey weights controlled for the probability of selection, stratification, and clustering. Post-stratification weights adjusted for overrepresentation of younger females and underrepresentation of older males.¹⁶

For pooled analysis, survey data from the years 2015, 2017, and 2019 were combined and weighted to ensure representativeness and adjust for the survey’s complex design. This increased the sample size and enhanced the statistical power of the analysis, allowing for more reliable estimates and meaningful subgroup comparisons, particularly for Black respondents, who may be underrepresented in individual survey cycles. Strata and clusters were unique for each year and were recoded to be analyzed as such.

We used survey procedures for means, frequencies, and regressions.³³ We used logistic regression (PROC SURVEYLOGISTIC) to analyze differences between Black and White respondents in accessing mental health services at least once within the past 12 months (yes/no). We used Poisson regression (PROC GENMOD with Link = Poisson) to analyze count differences in service utilization among respondents who accessed services within the previous 12 months.³⁴ All models incorporated survey weights, clustering, and post-stratification adjustments to ensure accurate and representative estimates. We adjusted models for age, sex, years in Canada, mental distress, and survey year and month.

We assessed effect measure modification in 2 ways: by visualizing interaction effects for visual evidence of interaction, and by adding sex and mental distress as interaction terms and testing for statistical significance. If the p value for the interaction term was less than 0.05, we considered the interaction to be statistically significant. To visualize interaction effects, we generated adjusted predicted values using marginal potential outcome models with model coefficients produced from respective models. Using these coefficients, we calculated predicted probabilities (from logistic models) or counts (from Poisson models) using the full data set and modelling each possible combination of race and sex, or race and K6 score, depending on the interaction under investigation. We chose marginal potential outcome models so that the covariate distribution would mimic the distribution in the population for each of the potential comparisons. We then averaged predicted values across all race–sex groups to obtain standardized estimates representing what the outcomes would be if all groups had identical distributions of measured confounders.

When we identified effect modification for sex, the models were further stratified by sex. This resulted in 3 models: Model 1 comprised the main effect and control variables; Model 2 the main effect, control variables, and interaction of race with mental distress and sex; and Model 3 the main effect, control variables, interaction of race with mental distress, stratified by sex.

The logistic and Poisson regression models did not automatically produce ratio estimates, so we derived odds ratios (logistic) and rate ratios (Poisson) from model estimates. For Model 2, we computed ratio estimates and confidence intervals to examine associations within specific sex and distress level categories. For Model 3, we obtained ratios for noninteraction variables by exponentiating model coefficients, and estimated ratios for interaction variables separately.

Ethics approval

The OSDUHS protocol was approved by the Centre for Addiction and Mental Health and York University Research Ethics Boards and school board research review committees. Informed consent was obtained from study participants, and data use was approved by the University of Toronto Research Ethics Board.

Results

A total of 36 003 students participated in the 2015, 2017, and 2019 cycles of the OSDUHS survey. Of these, 19 384 students completed survey version A, which included mental health questions. We excluded respondents who did not identify exclusively as Black or White, or were aged 20 years or older, from the analysis (7016). The final analytic sample included 12 368 respondents: 11 079 White and 1289 Black (Figure 3). After we applied sampling weights, the weighted sample included 10 500 White and 1761 Black respondents (Table 1).

Figure 3:

Participant flow chart showing the progression of selecting participants for the study. A total of 36 003 students participated in the Ontario Student Drug Use and Health Survey (OSDHUS) conducted in 2015, 2017, and 2019. Of these surveys, 19 384 responded to questions about student mental health (OSDHUS survey Form A). We excluded from the analysis those who did not identify as Black or White, identified as multiple races (*of which 550 were Black with another race), had missing data, or were aged 20 years or older (nonadolescents), resulting in a final sample size of 12 368 respondents. Of the 12 368 respondents, 11 079 respondents identified as White and 1289 identified as Black.

After weighting, the overall sample was 51.5% male and 48.5% female. The majority of Black (58.8%) and White (52.8%) respondents rated their mental health as “excellent” or “very good.” Most respondents had low distress (K6), with similar proportions for Black (61.0%) and White (63.5%) respondents. About 74% of Black and White respondents had not sought mental health services in the past year. Notably, Black respondents were more concentrated in the Greater Toronto Area (68.9% v. 22.8%) and had been in Canada for 10 years or less (30.9% v. 3.3%), compared to White respondents (Table 1).

Mental health care use over last year

Increased odds of accessing mental health care in the last year was associated with higher mental distress and being female (Table 3, Model 1). Sex modified the effect of race on access (p = 0.022, Figure 4A) and K6 distress categories (p = 0.012), with Black females reporting 0.71 times lower odds (95% confidence interval [CI] 0.56–0.91; Table 3, Model 2) of accessing care than White females, with no such racial difference among males (odds ratio [OR] 1.17, 95% CI 0.72–1.91; Table 3, Model 2).

Table 3:

Odds of accessing mental health care at least once during the last year (Model 1 and 2)^*

Characteristic	OR (95% CI)
	Model 1: Main effects of race, age, and sex†	Model 2: Main effect of age and interactions of race with age and K6 score‡
Race
Black	1.00 (0.76–1.32)	–
White (ref.)	NA	NA
Age
1-yr increase	0.93 (0.88–0.97)§	0.93 (0.88–0.97)§
yr (ref.)	NA	NA
Sex
Female	1.19 (1.04–1.36)§	–
Male (ref.)	NA	NA
K6 score
Moderate	2.32 (1.99–2.70)§	–
Serious	4.19 (3.61–4.87)§	–
Low (ref.)	NA	NA
Sex and race
Male Black	–	1.17 (0.72–1.91)
Male White (ref.)	NA	NA
Female Black	–	0.71 (0.56–0.91)§
Female White (ref.)	NA	NA
K6 score and race
Low Black	–	1.33 (1.05–1.69)§
Low White (ref.)	NA	NA
Moderate Black	–	0.68 (0.42–1.10)
Moderate White (ref.)	NA	NA
Serious Black	–	0.84 (0.51–1.39)
Serious White (ref.)	NA	NA

Note: CI = confidence interval, K6 = Kessler Psychological Distress Scale–6 items, NA = not applicable, OR = odds ratio, ref. = reference category.

^*Modelling the odds of accessing mental health care at least once in the 12 months preceding survey completion (binary outcome) using logistic regression.

^†Model 1: adjusted for age, race, sex, length of time in Canada, K6 category (a measure of distress), month and year of survey completion.

^‡Model 2: adjusted for Model 1 variables and interaction of race with sex and K6 category separately.

^§Values indicate statistical significance at p < 0.05.

Figure 4:

Interaction plots showing adjusted predicted (A) probabilities and (B) counts of accessing mental health care by race and sex among adolescent grade 7–12 students in Ontario. We generated adjusted predicted values using marginal potential outcome models with model coefficients produced from respective models. (A) Predicted probabilities of accessing any mental health care in the past year, based on logistic regression (Table 3, Model 2). (B) Predicted frequency of mental health care use among those who accessed care at least once in the past year, based on Poisson regression (Table 5, Model 2).

After stratification by sex (Table 4), effect measure modification by distress level was still evident among males (p = 0.004, Table 4 and Figure 5A). Males with low distress had 1.50 higher odds (95% CI 1.09–2.06; Table 4) of accessing services, but those with moderate distress had 0.41 lower odds (95% CI 0.20–0.84; Table 4). We detected a nonsignificant difference among Black male respondents reporting serious distress, compared with White male respondents reporting serious distress (OR 1.85, 95% CI 0.48–7.05; Table 4); however, the wide CI suggests low statistical power for detecting a difference. Among females, effect modification was also present (p = 0.012) and showed differences among respondents with serious distress, where Black females exhibited 0.60 lower odds (95% CI 0.40–0.90; Table 4) of accessing care compared with White female respondents with serious distress.

Table 4:

Odds of accessing mental health care at least once (Model 3)^*

Characteristic	Model 3: Main effect of age and interaction of race with K6 score† OR (95% CI)
	Males	Females
Age
1-yr increase	0.84 (0.78–0.91)‡	1.02 (0.96–1.07)
yr (ref.)	NA	NA
K6 score and race
Low Black	1.50 (1.09–2.06)‡	1.20 (0.84–1.73)
Low White (ref.)	NA	NA
Moderate Black	0.41 (0.20–0.84)‡	0.79 (0.46–1.35)
Moderate White (ref.)	NA	NA
Serious Black	1.85 (0.48–7.05)	0.60 (0.40–0.90)‡
Serious White (ref.)	NA	NA

Note: CI = confidence interval, K6 = Kessler Psychological Distress Scale–6 items, NA = not applicable, OR = odds ratio, ref. = reference category.

^*Modelling the odds of accessing mental health care at least once in the 12 months preceding survey completion (binary outcome) using logistic regression.

^†Model 3: stratified Model 2, from Table 3, by sex and removed sex as a covariate.

^‡Values indicate statistical significance at p < 0.05.

Figure 5:

Interaction plots showing adjusted predicted (A) probabilities and (B) counts of accessing mental health care by race and Kessler Psychological Distress Scale–6 items (K6), stratified by sex, among adolescent grade 7–12 students in Ontario. We generated adjusted predicted values using marginal potential outcome models with model coefficients produced from respective models. (A) Predicted probabilities of accessing any mental health care in the past year, based on logistic regression (Table 4). (B) Predicted frequency of mental health care use among those who accessed care at least once in the past year, based on Poisson regression (Table 6).

Frequency of mental health care use

We observed effect measure modification by race for K6 distress categories (p = 0.043) and showed a trend toward significance for sex (p = 0.103; Figure 4B). Black females exhibited a 23% lower likelihood (rate ratio 0.77, 95% CI 0.65–0.90) of accessing services as frequently as White females (Table 5, Model 2), and we found no such racial difference among males. Age and higher mental distress were also associated with increased frequency of access (Table 5).

Table 5:

Rate ratios for mental health care access frequency (Model 1 and 2)^*

Characteristic	Model 1: Main effects of race, age, and sex† Rate ratio (95% CI)	Model 2: Main effect of age and interactions of race with age and K6 score‡ Rate ratio (95% CI)
Race
Black	0.89 (0.81–0.97)§	–
White (ref.)	NA	NA
Age
1-yr increase	1.03 (1.01–1.05)§	1.03 (1.01–1.05)§
yr. (ref.)	NA	NA
Sex
Female	1.06 (0.98–1.14)	–
Male (ref.)	NA	NA
K6 score
Moderate	1.27 (1.15–1.40)§	–
Serious	1.58 (1.45–1.72)§
Low (ref.)	NA	NA
Sex and race
Male Black	–	1.09 (0.89–1.34)
Male White (ref.)	NA	NA
Female Black	–	§0.77 (0.65–0.90)
Female White (ref.)	NA	NA
K6 score and race
Low Black	–	0.87 (0.74–1.02)¶
Low White (ref.)	NA	NA
Moderate Black	–	1.08 (0.92–1.27)
Moderate White (ref.)	NA	NA
Serious Black	–	0.81 (0.67–0.98)§
Serious White (ref.)	NA	NA

Note: CI = confidence interval, K6 = Kessler Psychological Distress Scale–6 items, NA = not applicable, ref. = reference category.

^*Modelling the rate of mental health care usage frequency in the 12 months preceding survey completion (count outcome) using Poisson regression.

^†Model 1: adjusted for age, race, sex, length of time in Canada, K6 category (a measure of distress), month and year of survey completion.

^‡Model 2: adjusted for Model 1 variables and interaction of race with sex and K6 category separately.

^§Values indicate statistical significance at p < 0.05.

^¶Values indicate statistical significance at p < 0.10.

After we stratified by sex (Table 6), effect measure modification by distress level may be present among females (p = 0.1482; Table 6 and Figure 5B), but this was not statistically significant. Black females with low and serious distress accessed services less frequently than White females with the same distress levels. The largest difference was among females with serious distress, where Black females had a 40% lower likelihood (rate ratio 0.60, 95% CI 0.42–0.85) of accessing services as frequently as their White counterparts. Effect measure modification by distress level was not evident among males (p = 0.873), who had no difference in frequency of access by distress.

Table 6:

Rate ratios for mental health care access frequency (Model 3)^*

Characteristic	Model 3: Main effect of age and interaction of race with K6 score† Rate ratio (95% CI)
	Males	Females
Age
1 yr increase	1.02 (0.99–1.05)‡	1.03 (1.00–1.06)‡
yr (ref.)	NA	NA
K6 score and race
Low Black	0.99 (0.86–1.15)	0.77 (0.65–0.91)‡
Low White (ref.)	NA	NA
Moderate Black	1.08 (0.83–1.40)	1.00 (0.84–1.18)
Moderate White (ref.)	NA	NA
Serious Black	1.04 (0.70–1.55)	0.60 (0.42–0.85)‡
Serious White (ref.)	NA	NA

Note: CI = confidence interval, K6 = Kessler Psychological Distress Scale–6 items, Ref. = reference category.

^*Modelling the rate of mental health care usage frequency in the 12 months preceding survey completion (count outcome) using Poisson regression.

^†Model 3: Stratified Model 2, from Table 5, by sex.

^‡Values indicate statistical significance at p < 0.05.

Interpretation

Black females consistently accessed mental health care less than White females. Black males with low distress had higher odds of accessing care than their White counterparts, but as distress level increased to moderate or serious, Black respondents (both male and female) were less likely to utilize services than White peers. Among respondents who accessed care in the previous year, Black females accessed care less frequently, despite high mental distress levels. Sex and mental distress modified the impact of being Black on mental health service use (Figure 4 and Figure 5), indicating differences in how Black girls and boys are socialized and experience racism, resulting in distinct behaviours regarding mental health service use.^35–37 Notably, Black female respondents consistently had poorer outcomes than all other groups. Additionally, as distress levels increased for Black adolescents, access to mental health services paradoxically reduced, unlike for White peers.

Black adolescent mental health must be discussed with consideration of the interaction between being Black, sex, and mental distress. Mental health professionals and researchers in Canada attribute barriers that Black adolescents experience in accessing care to lack of culturally competent care, inaccessible services, and cultural stigma due to racism.^15,38–40 Anti-Black racism and implicit biases often lead to misunderstandings and misdiagnoses, such as mistaking anxiety or depression for attention-deficit/hyperactivity disorder, resulting in ineffective or harmful treatments for Black adolescents.^38,40 These experiences foster mistrust in the health care system. Black adolescents and families, aware of potential judgment, discrimination, or criminalization of their symptoms, may avoid accessing care.¹⁵ Black adolescents expressing negative emotions are disproportionately seen as aggressive, often leading to crisis situations where they encounter law enforcement rather than receive proper mental health care.^15,41,42 This can further deter families from seeking help.¹⁵ At the core is a lack of culturally competent care. There is a shortage of mental health professionals who are equipped to meet the needs of Black adolescents, which fosters mistrust and reluctance to access services.¹⁵ Black adolescents also face longer wait times for mental health services, owing to financial barriers and a lack of mental health care providers in Black communities. ^15,38–40 Finally, Black adolescents may appraise their anxiety and depression as normal situational hardship due to the economic stress and systemic racism to which Black adolescents are disproportionally exposed.¹⁵

Black girls face unique discrimination and marginalization owing to the dual burden of racism and sexism, which negatively affects their mental well-being.^35–37 One critical issue is adultification, where Black girls are treated more like adults than children, leading to disproportionate discipline and overlooked mental health needs compared with their White peers, who are more readily seen as needing protection and care.^43–45 Cultural expectations to be “strong” and self-reliant can further discourage Black girls from seeking care.^46–49 The shortage of mental health providers who understand the unique cultural and sex- or gender-specific issues faced by Black girls results in mental health services that too often feel inaccessible or unhelpful.⁵⁰ Unconscious bias can perpetuate stereotypes that label Black girls as “resilient” or “less in need of help,” leading to under-diagnosis or inadequate treatment.^43,48

Our findings reveal a concerning trend for Black boys — they have no difficulty accessing care when distress is low, but as their distress intensifies, their likelihood of accessing care paradoxically decreases. This may stem from the criminalization of Black male distress, and Black males in general. Black boys are more likely to be perceived as older, less innocent, and more threatening than White peers.^45,51,52 When Black males exhibit signs of psychological distress, they are more likely to be met with disciplinary or punitive responses, sometimes with fatal consequences, rather than mental health support.^51–54 This may also explain why, once they overcome access barriers, their care frequency matches that of White peers.

Future research should examine longitudinal trends in Black adolescents’ mental health service use to track evolving disparities and their impact on help-seeking. Qualitative studies can deepen understanding of systemic barriers like distrust, discrimination, and family influence. Additionally, evaluating targeted interventions — such as increasing the number of Black mental health professionals, peer support, and school-based services — can inform policies to improve access and equity.

Limitations

The OSDUHS data provided a large, diverse sample, enhancing statistical power. However, Black people in Ontario comprise only a small proportion of the population (5.4% in 2021), and therefore Black students were a small sample of the OSDUHS student population (Table 1; 6.7%),¹⁷ which affected estimate precision by reducing statistical power to detect significant differences. The cross-sectional design captures data at one point in time, which limited our ability to assess changes or determine causality. This is especially important because the survey measures current mental and emotional health but asks about service use over the past 12 months, which makes it difficult to clearly link need with care received. We excluded 2098 multiracial respondents (including 550 who identified as Black) (Appendix 1, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.241733/tab-related-content) because the OSDUHS does not capture how multiracial respondents are perceived by others or how they primarily identify — factors that influence socialization and treatment. Although this decision allowed for clearer racial comparisons, it limits the generalizability of our findings and overlooks the complexities of multiracial identity, which future research should address.

The self-reporting of mental health care use introduces potential recall bias, social desirability bias, and variations in how Black and White respondents interpret mental health care services. Further qualitative differences in parents or students who consented to this study, selection bias, and biases inherent in self-reporting may exist owing to greater mistrust and disproportionately higher suspension and expulsion rates that Black students unfairly experience.⁵⁵ Consequently, results may not be generalizable to all Black adolescents in Ontario.

Finally, data were collected between 2015 and 2019, and findings may not fully reflect the current mental health landscape or patterns of service access.

Conclusion

Our study underscores the urgent need for policy and practice changes that address systemic racism and the lack of culturally relevant care affecting mental health use for Black adolescents. Black adolescents with psychological distress are less likely to use mental health services than their White peers, with Black girls being the least likely to access care. Intersectional strategies that tackle racism and the specific mental health challenges faced by Black students are needed. Enhancing mental health service utilization for Black adolescents demands culturally responsive and sex-specific adaptations to care.