Barriers to treatment for mental disorders in six countries of the Americas: A regional report from the World Mental Health Surveys

Abstract

Background:

Mental health treatment is scarce and little resources are invested in reducing the wide treatment gap that exists in the Americas. The regional barriers are unknown. We describe the barriers for not seeking treatment among those with mental and substance use disorders from six (four low- and middle-income and two high-income) countries from the Americas. Regional socio-demographic and clinical correlates are assessed.

Methods:

Respondents (n=4648) from seven World Mental Health surveys carried out in Argentina, Brazil, Colombia, Mexico, Peru, and the United States, who met diagnostic criteria for a 12-month mental disorder, measured with the Composite International Diagnostic Interview, and who did not access treatment, were asked about treatment need and, among those with need, structural and attitudinal barriers. Country-specific deviations from regional estimates were evaluated through logistic models.

Results:

In the Americas, 43% of those that did not access treatment did not perceive treatment need, while the rest reported structural and attitudinal barriers. Overall, 27% reported structural barriers, and 95% attitudinal barriers. The most frequent attitudinal barrier was to want to handle it on their own (69.4%). Being female and having higher severity of disorders were significant correlates of greater perceived structural and lower attitudinal barriers, with few country-specific variations.

Limitations:

Only six countries in the Americas are represented; the cross-sectional nature of the survey precludes any causal interpretation.

Conclusions:

Awareness of disorder or treatment need in various forms is one of the main barriers reported in the Americas and it specially affects persons with severe disorders.

1. Introduction

Differences in accessing mental health treatment between high- and low-and-middle-income countries (LMICs) extends to all types of treatment and levels of care. In inpatient care, LMICs have 20 times less beds per unit and 30 times less admissions; in outpatient care there are 40 times less patients being treated, and 15 times less staff at the outpatient levels (Lora et al., 2020). In addition, there may be differences in treatment barriers. Recently, a qualitative review on LMICs found that low perceived treatment need and low motivation were reported as the most common treatment barriers for substance use treatment, followed by stigma, not perceiving a problem, and structural barriers (e.g., cost, limited availability) (Sarkar et al., 2021). However, none of the studies included in the review were from LMICs in the Americas, and the inclusion criteria did not cover all mental disorders, only substance use.

In the Americas, mental and substance use disorders have been considered among the most important public health concerns. In this region, the burden related to mental health has been estimated between 10.5% and 19% of disability adjusted life years (DALYs) and 22% to 34% of years lived with disability (YLDs) (Kohn et al., 2018; Vigo, 2019). Despite this, only a small proportion of people with mental or substance use disorders receive adequate treatment for their condition (Alonso et al., 2018; Degenhardt et al., 2017; Thornicroft et al., 2017; Vigo et al., 2020).

We have previously estimated that, in the Americas, about one in four persons with any 12-month mental disorder had received any kind of treatment (Borges et al., 2020). This treatment gap is consistent with other regional studies that estimated a prevalence of 71.2% persons that required care but did not obtain it (Kohn et al., 2018). There is evidence of variation in treatment gap by disorder severity and by sub-region. Kohn et al. (2018) found that the treatment gap for severe mental disorders in high-income countries in North America (i.e., Canada and the United States) was 53.2%, whereas it reached 78.8% in Mesoamerica (i.e., Costa Rica, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, and Panama, all of them LMICs). At the national level, a striking treatment gap of 95% was reported for Guatemala (Kohn et al., 2018).

Perceived treatment need has an important role in treatment seeking and, potentially, in attitudinal barriers. A recent article with the general population in the United States (Moeller et al., 2020) found that low perception of treatment need is prevalent in 77% of those with substance use disorders. Moreover, the authors found that perceived treatment need is stable and even increases over time (up to 84%). If this pattern is found in LMICs in the Americas it would imply more complex challenges considering that internationally, treatment seeking latency ranges from 1 to 30 years (Wang et al., 2007b).

While it could be assumed that in LMICs the treatment gap for mental disorders might be mainly related to financial hardship and poverty, other factors may play a key role as well.

Personal and structural barriers for not seeking treatment in the region have been sporadically reported, including stigma and misconceptions about the natural history of the disease, among others (Saldivia et al., 2004; Santos Cruz et al., 2013). A previous study from the World Mental Health (WMH) Initiative has reported on the barriers to seeking help, classifying the participating countries according to income categories (Andrade et al., 2014). However, previous literature has not specifically focused on the Americas, and it has not considered potentially relevant determinants of help-seeking behaviors such as disease severity, stigma, and attitude toward treatment, among others.

The region of the Americas is full of contrasts, comprising some of the richest and poorest countries in the world with large variations in the organization of mental health services (Pan American Health Organization, 2013). To expand treatment coverage and to reduce the mental health treatment gap in the region, detailed comparable data and analyses are needed, but rarely available. Review of the current epidemiologic literature suggests that few reports from the region have considered the role of disease severity and its relationship to several types of barriers to treatment.

Therefore, the aim of this article is to describe the structural and attitudinal barriers for seeking treatment among those with 12-month mental and substance use disorders, based on WMH data from the countries of the region of the Americas. We also assessed their socio-demographic correlates and their association with disease severity, evaluating the heterogeneity across countries by testing how individual countries vary from the overall pooled estimates for the region.

2. Methods

2.1. Participants

Respondents were from seven WHO WMH surveys carried out in six countries in the Region of the Americas (two from Colombia): two surveys in countries classified at the time of data collection as LMICs (Colombia-national and Peru), three in upper-middle-income countries (Brazil, Colombia-Medellin and Mexico) and two in high-income countries (Argentina and the United States). The U.S. survey was based on a nationally representative household sample, three surveys (Argentina, Colombia-national and Mexico) were representative of urban areas, and the remaining three were representative of selected metropolitan areas (Sao Paulo, Brazil, Medellin, Colombia- and Metropolitan Lima, Huancayo, Iquitos, Arequipa, and Chiclayo, Peru). In all surveys, trained lay interviewers conducted face-to-face interviews with respondents aged at least 18 years old. Respondents were selected using multistage household probability samples. The total sample size was 35,645. The weighted average response rate across all countries was 79.8% (supplementary Table S1). The local human research committees of each survey gave their respective approval. Subsampling procedures to reduce respondent burden were used, as described elsewhere (Heeringa et al., 2008). Among the total sample, 6710 participants met criteria for a mental disorder in the last 12 months. In this report, we focused on the respondents that did not seek mental health services but met criteria for a mental disorder in the last 12 months (n= 4648), a group that is clearly in need of treatment. We focused only on past-year diagnosis to reduce memory bias in both psychiatric symptoms and reasons for not seeking treatment.

2.2. Instrument

The interviewers administered, face-to-face, the computer-assisted interview-version of the WMH Survey Initiative-Composite International Diagnostic Interview (CIDI) (Kessler and Ustun, 2004; Robins et al., 1988); this fully structured diagnostic interview yielded-DSM-IV diagnoses.

All measures that were used are embedded within the CIDI. Following WHO guidelines (see Pennell et al. (2008) for further details), the original English version of the CIDI was translated, back-translated, and harmonized in different languages for use in the WMH surveys. Classification accuracy of CIDI diagnoses compared with diagnoses from blinded clinical reappraisal interviews has been reported with an Area Under the Curve (AUC) of 0.76 for any disorder and ranging from 0.62 to 0.93 for individual disorders (Haro et al., 2006). Disorders that were considered in this report include the standard disorders measured in the CIDI. The CIDI instrument also includes the Sheehan Disability Scales (Sheehan et al., 1996) which evaluate how much the symptoms interfere with four different areas of life on a scale from 0 to 10. The Spanish Language version has good indicators of validity (discriminating between those with and without depression with an AUC of 0.81) and internal consistency (α = 0.83) (Luciano et al., 2010).

2.3. Measures

2.3.1. Disorders

As in our previous report about treatment for mental disorders in the Americas (Borges et al., 2020), we identified those respondents positive for at least one of the following WMH-CIDI mental and substance use disorders: (1) affective disorders: major depressive disorder (MDE), dysthymia and bipolar disorder (we used a broad definition that included bipolar I, II and subthreshold, defined as a history of recurrent sub-threshold hypomania in the presence of MDE, or a history of recurrent hypomania in the absence of recurrent MDE, or a history of recurrent sub-threshold hypomania in the absence of concurrent MDE (Kessler et al., 2006)); (2) anxiety disorders: panic disorder, agoraphobia, social phobia, specific phobia, adult separation anxiety disorder, generalized anxiety disorder and posttraumatic stress-disorder; (3) substance use disorders: alcohol and drug abuse and dependence and (4) behavioral disorders: attention deficit/hyperactive disorder, and intermittent explosive disorder.

2.3.2. Disorder severity

WMH-CIDI uses trans-diagnostic criteria to classify the clinical presentations of disorders as serious, moderate, or mild (Demyttenaere et al., 2004; Evans-Lacko et al., 2018). People who meet criteria for any 12-month DSM-IV disorder were classified as suffering a serious disorder if they had any of the following: (a) high level of impairment on the Sheehan Disability Scales (SDS); (b) a suicide attempt in the past 12 months; (c) substance dependence with a physiological dependence syndrome; or (d) 12-month bipolar I disorder (Endicott et al., 1976; Sheehan et al., 1996). Respondents not classified as having a serious disorder were classified as moderate if impairment was rated as at least moderate in any SDS domain or if the respondent had substance dependence without a physiological dependence syndrome. All other respondents were classified as mild.

2.3.3. Barriers and reasons for not using services

2.3.3.1. Perceived treatment need

Respondents with any 12-month mental or substance use disorder who reported no use of mental health services were asked whether there was a time in the past 12 months when they felt they might have needed to see a professional for problems with their emotions, nerves, or mental health.

As described elsewhere (Borges et al., 2020), treatment options for mental health services included psychiatrists, other mental health specialists (i.e., psychologists, counselors, psychotherapists, mental health nurses and social workers), general medical practitioners (i.e., family physicians, general practitioners and other medical doctors, nurses, occupational therapists, or other health care professionals); human services, (e.g., religious or spiritual advisors), and complementary-alternative medicine (e.g., self-help groups, any other healers).

Those who did not think they needed help or thought they needed help for less than four weeks were categorized as having “low perceived need”. Respondents who reported needing help for more than 4 weeks were categorized as “perceived need”, and then were asked about the potential structural and attitudinal barriers that may have led to not receiving the care they felt they needed.

2.3.3.2. Structural barriers

As in previous studies from the WMH surveys (Andrade et al., 2014), structural barriers were identified by respondents’ endorsement of the following statements: financial barriers: “My health insurance would not cover this type of treatment”, or “I was concerned about how much money it would cost”; availability: “I was unsure about where to go or who to see”, “I could not get an appointment”, or “I was not satisfied with available services”; transportation: “I had problems with things like transportation, childcare, or scheduling that would have made it hard to get to treatment”; inconvenient: “I thought it would take too much time or be inconvenient”. Those who reported at least one of those, were coded positive in a composite variable labeled “any structural barrier”.

2.3.3.3. Attitudinal barriers and reasons

Attitudinal barriers were identified by endorsement of the following statements: wanted to handle on onés own: “I wanted to handle the problem on my own”; perceived ineffectiveness: “I didn’t think treatment would work”, or “I received treatment before and it did not work”; stigma: “I was concerned about what others might think if they found out I was in treatment”, or “I was scared about being put into a hospital against my will”; thought it would get better: “I thought the problem would get better by itself”; problem was not severe: “The problem didn’t bother me very much”, or “The problem went away by itself, and I did not really need help”. Respondents who endorsed more than one attitudinal barrier/reason for not seeking help were coded positive for “any attitudinal barrier”.

2.3.4. Socio-demographic correlates

Socio-demographic variables were age (categorized as 18–34, 35–49, 50–64 and 65+ years), sex, and marital status (married/cohabitating, previously married, never married). Completed years of education and family income were grouped into four categories: low, low-average, high-average and high, based on country-specific distributions, as detailed elsewhere (Evans-Lacko et al., 2018). After some preliminary analysis, due to sample size limitations in the statistical models for attitudinal barriers, we further dichotomized age, education, and income, by collapsing the first two and the last two categories for these models.

2.4. Data analysis strategy

The data were weighted to adjust for differential probabilities of selection and nonresponse. Estimates of standard errors for proportions were obtained by the Taylor series-linearization method implemented in the SAS’ survey analysis procedures (SAS Institute Inc, 2017). Logistic regression-analysis (Hosmer and Lemeshow, 2000) was performed to study socio-demographic correlates. Estimates of standard errors of odds ratios from logistic regression coefficients were obtained, along with their corresponding 95% confidence intervals, adjusted for design effects. Statistical significance was evaluated with two-sided design-based tests with α=0.05. All models included controls for survey and mental disorders.

First, we estimated the proportion of individuals reporting low perceived treatment need or any structural and attitudinal barrier, by disorder severity and by country, as well as pooled across the region, among those respondents with mental disorders who did not seek treatment. Second, we estimated the proportion of respondents with each individual structural and attitudinal barrier by severity for each country and pooled across the region. Third, we examined between-country variation in associations of barriers with sociodemographic variables by including all predictor-by-survey interactions in logistic regression models using a dummy coding scheme that kept the product of all country-specific odds ratios (ORs) equal to 1. This method allowed us to detect significant between-country variation with respect to the overall effect, by evaluating the statistical significance of deviation of within-survey coefficients from the median value (Mortier et al., 2018). The reported survey-specific ORs show to what extent the survey-specific effect deviates from the overall effect. For example, if the reported OR for females (versus males) in the United States is 1.5, then it would be necessary to multiply it by the reported overall effect (e.g., OR = 1.2) to obtain the survey-specific effect in the United States (i.e., OR = 1.8). To study the attitudinal barriers further, we also modelled the most important reason in the Americas for not seeking treatment despite perceiving need: people with a 12-month disorder wanting to handle the problem on their own.

2.5. Statement of ethics

The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. All participants gave informed consent and the Ethical Review Board of the responsible research institutions in each country gave approval of the survey.

3. Results

3.1. Barriers for not seeking treatment

Among 4648 respondents with mental disorders but no service use in the past 12 months, low perceived treatment need was reported by 1998 respondents (43%). People with moderate or mild disorders reported low perceived treatment need significantly more frequently than those with a severe disorder (49.7% vs. 31.9%; p <0.001) (Table 1). This pattern was observed across countries, reaching a higher than twofold difference in perceived need by level of severity in Argentina (48.4% vs. 22.2%; p =0.002). The highest proportion of low perceived treatment need was found in Brazil among those with moderate/mild disorder severity (63.0%). However, also in that level of severity, close to half of respondents in Argentina (48.4%) and the United States (48.8%) reported low perceived treatment need. Among those with severe disorders, respondents in Medellin, Colombia (46.4%) and Brazil (41.0%) reported high prevalence of low perceived treatment need. Attitudinal barriers were endorsed by a higher proportion of respondents than structural barriers (for severe disorders: 63.9% vs. 26.6% respectively and for mild/moderate disorders: 48.2% vs. 11.4%, respectively). Both structural and attitudinal barriers were more likely to be reported by those with severe cases than mild to moderate cases.

Table 1.

Barriers for not seeking treatment among all respondents with 12-month mental disorders who did not use services in that period, by severity of disorder. WMH Americas surveys.

		Low perceived need for treatment				Any structural barrier				Any attitudinal barrier
Survey	n	Severe	Moderate/Mild	χ21	p-value	Severe	Moderate/Mild	χ21	p-value	Severe	Moderate/Mild	χ21	p-value
		% (S.E.)	% (S.E.)			% (S.E.)	% (S.E.)			% (S.E.)	% (S.E.)
Argentina	379	22.2 (5.7)	48.4 (4.6)	9.8	0.002	22.4 (5.4)	9.0 (2.0)	4.7	0.031	70.1 (5.5)	51.4 (4.5)	5.5	0.02
Brazil	909	41.0 (3.9)	63.0 (2.7)	22.1	<0.001	25.0 (3.6)	9.7 (1.7)	16.9	<0.001	52.9 (3.7)	33.5 (2.6)	20.9	<0.001
Colombia	706	24.3 (5.0)	42.1 (3.2)	10.8	0.001	31.7 (6.6)	13.0 (1.7)	6.9	0.009	73.6 (5.0)	56.3 (3.2)	9.8	0.002
Medellin, Colombia	411	46.4 (6.7)	58.9 (3.8)	2.8	0.094	19.5 (4.2)	11.6 (2.6)	2.7	0.101	52.1 (6.7)	39.4 (3.8)	2.9	0.092
Mexico	550	25.8 (4.1)	41.9 (3.2)	11.7	<0.001	29.9 (4.5)	15.6 (1.9)	9.8	0.002	68.0 (5.1)	55.1 (3.2)	5.3	0.022
Peru	304	24.7 (5.3)	35.4 (3.0)	2.1	0.144	29.7 (7.5)	20.1 (2.1)	1.4	0.229	68.0 (5.1)	61.3 (3.0)	0.9	0.334
United States	1389	25.9 (3.3)	48.8 (1.8)	41.1	<0.001	28.6 (2.9)	9.1 (1.1)	50.8	<0.001	72.7 (3.1)	50.0 (1.8)	38.4	<0.001
Overall	4648	31.9 (1.9)	49.7 (1.1)	67.5	<0.001	26.6 (1.7)	11.4 (0.7)	70.5	<0.001	63.9 (1.9)	48.2 (1.1)	52.9	<0.001

n shown is the denominator n of all respondents with 12-month mental disorders who did not use services in that period in each country.

Abbreviations: SE, Standard Error

Analyses performed on part II sample

3.2. Specific treatment barriers by severity

3.2.1. Structural barriers

Among those who reported treatment need but did not access treatment, we further analyzed the structural and attitudinal barriers. Tables 2 and 3 show the proportion of respondents with perceived treatment need (n=2650) reporting each structural and attitudinal barrier. First, structural barriers were reported less frequently in HIC: Argentina (21.1%) and the United States (22.1%), and most frequently in LMIC: e.g., Brazil and Peru (32.5%) (Table 2). Overall, the most frequent structural barriers reported were financial (18.5%) and availability (17.7%), with a significantly higher proportion of severe cases (27.2% and 26.8% respectively) reporting those barriers as compared to moderate (18.3% and 17.0% respectively) and mild cases (11.4% and 10.7% respectively).

Table 2.

Structural barriers for not seeking treatment among the subgroup with 12-month mental disorders who perceived a need for mental health care but did not access any, according to level of severity and survey. WMH Americas surveys.

			Any severity		Severe		Moderate		Mild
Structural barrier	Survey	Total with need but no service (n)	n	% (S.E.)	n	% (S.E.)	n	% (S.E.)	n	% (S.E.)	χ22	p-value
Financial
	Argentina	234	32	13.1 (3.3)	16	23.9 (8.0)	8	9.5 (3.9)	8	6.0 (2.7)	3.9	0.145
	Brazil	431	103	25.0 (3.6)	52	32.1 (5.1)	39	24.5 (4.9)	12	13.6 (4.9)	14.3	<0.001
	Colombia	430	95	19.1 (2.4)	27	26.3 (6.6)	44	18.2 (3.8)	24	14.7 (3.7)	2	0.360
	Medellin, CO	196	33	19.7 (3.4)	16	27.1 (6.6)	13	20.9 (5.6)	4	--	5.1	0.079
	Mexico	352	77	18.8 (2.8)	26	23.5 (5.7)	36	23.6 (4.0)	15	10.9 (3.4)	7.5	0.025
	Peru	203	44	21.2 (2.9)	11	25.9 (6.0)	19	20.9 (4.2)	14	19.5 (5.9)	0.9	0.647
	United States	804	124	15.0 (1.7)	48	26.0 (4.2)	55	14.5 (2.4)	21	8.8 (2.4)	11.6	0.003
	Overall	2650	508	18.5 (1.1)	196	27.2 (2.3)	214	18.3 (1.5)	98	11.4 (1.5)	34.9	<0.001
Availability
	Argentina	234	31	11.7 (2.4)	11	14.4 (4.3)	13	11.9 (4.0)	7	8.4 (4.0)	1.8	0.404
	Brazil	431	101	22.7 (2.5)	55	30.5 (5.1)	32	19.4 (3.2)	14	14.8 (4.6)	5	0.083
	Colombia	430	91	20.3 (2.6)	27	30.8 (7.6)	39	18.4 (3.8)	25	14.9 (3.5)	2.6	0.275
	Medellin, CO	196	36	21.3 (4.1)	14	21.7 (5.9)	19	32.8 (7.5)	3	--	13.8	0.001
	Mexico	352	94	22.1 (2.8)	38	33.5 (6.5)	32	21.2 (4.1)	24	15.0 (4.0)	6.6	0.038
	Peru	203	40	18.8 (3.0)	12	28.1 (9.5)	16	19.8 (4.7)	12	13.7 (3.6)	2.5	0.291
	United States	804	107	12.6 (1.5)	42	24.2 (3.5)	43	11.3 (2.0)	22	6.8 (1.2)	17.9	<0.001
	Overall	2650	500	17.7 (1.0)	199	26.8 (2.2)	194	17.0 (1.4)	107	10.7 (1.1)	36.6	<0.001
Transportation
	Argentina	234	10	--	3	--	3	--	4	--	1.5	0.472
	Brazil	431	47	9.6 (1.7)	29	15.5 (2.9)	14	8.1 (3.4)	4	--	17.9	<0.001
	Colombia	430	37	6.5 (1.2)	12	11.3 (4.1)	15	4.7 (1.6)	10	5.4 (2.2)	2.3	0.320
	Medellin, CO	196	8	4.2 (1.6)	4	--	4	--	0	0	++	++
	Mexico	352	34	7.5 (1.6)	14	11.0 (2.8)	12	8.7 (2.8)	8	4.0 (1.7)	5.6	0.061
	Peru	203	16	5.8 (1.3)	5	10.2 (3.3)	7	5.3 (1.5)	4	4.5 (2.3)	2.7	0.260
	United States	804	52	5.6 (1.1)	27	13.4 (3.0)	19	4.9 (1.2)	6	1.5 (0.6)	15.6	<0.001
	Overall	2650	204	6.5 (0.6)	94	11.7 (1.4)	74	5.7 (0.9)	36	3.1 (0.6)	35.8	<0.001
Inconvenient
	Argentina	234	14	6.8 (2.3)	3	--	6	9.4 (4.7)	5	--	1.4	0.496
	Brazil	431	41	8.0 (1.5)	27	13.8 (3.7)	12	6.0 (2.3)	2	--	9.5	0.009
	Colombia	430	47	9.6 (1.9)	15	13.2 (3.9)	20	10.1 (3.6)	12	6.0 (2.4)	2.3	0.322
	Medellin, CO	196	11	5.8 (2.0)	7	10.1 (3.8)	4	--	0	0	++	++
	Mexico	352	47	13.0 (2.1)	18	19.7 (5.3)	21	15.7 (3.4)	8	5.7 (2.2)	9	0.012
	Peru	203	33	14.7 (3.2)	12	28.0 (8.2)	14	14.0 (4.3)	7	9.9 (4.2)	4.3	0.119
	United States	804	86	10.0 (1.3)	32	18.7 (3.2)	39	10.0 (1.6)	15	4.5 (1.3)	14.2	<0.001
	Overall	2650	279	9.8 (0.7)	114	15.0 (1.6)	116	9.9 (1.1)	49	5.0 (0.8)	28.8	<0.001
Any structural barrier
	Argentina	234	51	21.1 (3.7)	20	28.8 (6.4)	20	22.4 (6.6)	11	10.8 (4.1)	6	0.050
	Brazil	431	138	32.5 (4.0)	74	42.3 (5.6)	45	29.3 (5.4)	19	21.0 (5.5)	8.9	0.012
	Colombia	430	128	27.0 (2.7)	40	41.8 (7.7)	57	25.6 (4.5)	31	17.3 (3.9)	5.6	0.061
	Medellin, CO	196	56	31.0 (4.3)	24	36.3 (7.1)	25	39.5 (7.4)	7	--	10.5	0.006
	Mexico	352	129	30.4 (3.2)	46	40.2 (6.8)	50	32.2 (4.7)	33	21.9 (4.5)	5.5	0.064
	Peru	203	70	32.5 (2.4)	17	39.5 (9.0)	30	30.7 (4.8)	23	31.6 (5.4)	0.8	0.679
	United States	804	183	22.1 (2.2)	70	38.5 (3.5)	76	20.4 (2.9)	37	13.7 (2.4)	43.4	<0.001
	Overall	2650	755	27.0 (1.2)	291	39.1 (2.4)	303	26.2 (1.8)	161	17.6 (1.6)	54.2	<0.001

^--Percentage less than twice the SE or sample size < 30

⁺⁺χ² test was not computed due to an empty cell

Abbreviations: SE, Standard Error

Analyses performed on part II sample

Table 3.

Attitudinal barriers for not seeking treatment among the subgroup with 12-month mental disorders who perceived a need for mental health care but did not access any, according to level of severity and survey. WMH Americas surveys.

			Any severity		Severe		Moderate		Mild
Attitudinal barrier	Survey	Total with need but no service (n)	n	% (S.E.)	n	% (S.E.)	n	% (S.E.)	n	% (S.E.)	χ22	p-value
Wanted to handle own
	Argentina	234	153	60.9 (4.8)	27	38.7 (8.0)	72	70.0 (5.8)	54	73.0 (6.4)	10.4	0.006
	Brazil	431	293	66.1 (2.9)	116	65.5 (7.5)	105	67.0 (6.1)	72	65.7 (6.4)	0	0.984
	Colombia	430	298	71.7 (3.2)	61	62.5 (6.8)	130	73.1 (5.1)	107	76.8 (4.1)	3.7	0.161
	Medellin, CO	196	141	73.4 (4.0)	40	74.0 (6.8)	56	66.4 (6.7)	45	84.1 (6.9)	3.2	0.198
	Mexico	352	226	67.1 (3.4)	48	61.4 (7.2)	98	68.6 (4.6)	80	69.6 (5.7)	0.9	0.652
	Peru	203	126	65.6 (2.9)	21	52.8 (8.0)	54	64.9 (5.0)	51	72.1 (5.9)	5.1	0.080
	United States	804	581	73.2 (1.4)	111	62.7 (3.3)	251	73.9 (2.8)	219	79.0 (2.9)	13.1	0.002
	Overall	2650	1818	69.4 (1.1)	424	61.5 (2.7)	766	70.5 (1.9)	628	74.9 (1.9)	15.7	<0.001
Perceived ineffectiveness
	Argentina	234	54	24.4 (3.9)	19	30.2 (7.4)	19	25.8 (6.2)	16	16.3 (5.3)	3.2	0.205
	Brazil	431	77	16.4 (2.5)	41	20.3 (5.0)	25	14.3 (3.9)	11	13.1 (6.3)	1	0.610
	Colombia	430	85	19.6 (2.8)	26	34.4 (7.7)	34	14.6 (3.4)	25	15.5 (2.9)	5	0.085
	Medellin, CO	196	39	17.3 (3.0)	19	25.4 (5.9)	16	16.3 (4.8)	4	--	4.4	0.115
	Mexico	352	55	13.2 (1.8)	24	20.1 (4.5)	23	17.2 (3.4)	8	4.6 (1.8)	18.1	<0.001
	Peru	203	26	11.3 (2.0)	6	14.6 (5.9)	11	12.1 (2.8)	9	8.9 (3.6)	1.1	0.588
	United States	804	140	16.0 (1.3)	46	26.0 (4.4)	58	14.9 (1.7)	36	11.1 (2.3)	7.5	0.024
	Overall	2650	476	16.7 (0.9)	181	24.9 (2.4)	186	15.6 (1.3)	109	11.2 (1.4)	23.5	<0.001
Stigma
	Argentina	234	4	0.8 (0.4)	3	--	1	--	0	0	++	++
	Brazil	431	41	7.5 (1.1)	27	12.4 (2.5)	13	6.4 (1.7)	1	--	14.2	<0.001
	Colombia	430	59	12.9 (2.0)	23	24.7 (5.8)	24	10.6 (3.1)	12	6.8 (2.4)	7	0.031
	Medellin, CO	196	27	15.9 (4.1)	14	25.8 (8.3)	11	13.5 (5.1)	2	--	3.5	0.175
	Mexico	352	43	9.6 (1.4)	20	16.9 (4.2)	16	9.7 (2.1)	7	4.4 (1.9)	7.9	0.020
	Peru	203	22	10.4 (2.9)	8	19.5 (6.9)	8	11.0 (4.6)	6	5.7 (2.6)	3.8	0.150
	United States	804	77	8.9 (1.2)	38	21.3 (3.2)	28	7.2 (1.6)	11	3.1 (0.9)	23.5	<0.001
	Overall	2650	273	9.3 (0.7)	133	17.5 (1.7)	101	8.2 (1.0)	39	3.8 (0.7)	49.4	<0.001
Thougth would get better
	Argentina	234	35	13.5 (2.5)	7	10.1 (4.4)	17	17.3 (4.7)	11	12.0 (4.5)	1.4	0.507
	Brazil	431	70	15.0 (2.6)	40	20.7 (4.2)	22	14.5 (3.8)	8	6.3 (2.0)	8.8	0.013
	Colombia	430	77	17.5 (2.7)	27	29.3 (7.1)	33	16.0 (3.9)	17	10.4 (2.7)	5.7	0.059
	Medellin, CO	196	37	17.9 (3.4)	18	25.7 (6.2)	15	19.8 (6.5)	4	--	9.6	0.009
	Mexico	352	71	17.7 (2.3)	24	22.5 (5.0)	29	19.2 (3.5)	18	12.9 (3.9)	2.2	0.330
	Peru	203	33	14.7 (2.6)	9	22.2 (9.0)	12	12.4 (3.9)	12	14.3 (4.1)	1	0.610
	United States	804	103	11.6 (1.4)	43	23.1 (3.5)	42	10.3 (1.8)	18	5.8 (1.2)	20.7	<0.001
	Overall	2650	426	14.7 (0.9)	168	22.3 (2.0)	170	14.2 (1.3)	88	8.7 (1.0)	35.1	<0.001
Problem was not severe
	Argentina	234	33	15.8 (3.5)	14	24.5 (7.6)	11	10.6 (4.0)	8	13.4 (6.0)	2.3	0.319
	Brazil	431	64	14.8 (2.7)	27	15.6 (3.5)	26	14.9 (4.6)	11	13.1 (5.7)	0.2	0.918
	Colombia	430	93	24.4 (4.1)	27	25.5 (5.6)	41	29.6 (6.5)	25	15.3 (3.8)	6.5	0.040
	Medellin, CO	196	32	15.8 (3.1)	14	21.7 (6.1)	14	18.1 (5.1)	4	--	8.4	0.016
	Mexico	352	74	21.4 (2.9)	22	23.9 (4.6)	23	14.0 (3.3)	29	26.7 (5.8)	5.9	0.054
	Peru	203	57	28.3 (2.8)	14	37.0 (8.8)	27	25.9 (6.3)	16	27.6 (6.0)	1.1	0.570
	United States	804	138	16.6 (1.2)	51	27.1 (3.6)	57	15.9 (2.6)	30	10.9 (1.9)	14	0.001
	Overall	2650	491	18.9 (1.1)	169	23.3 (1.9)	199	18.7 (1.9)	123	15.3 (1.6)	10.7	0.005
Any attitudinal barrier
	Argentina	234	226	96.6 (1.2)	57	90.1 (3.4)	97	99.4 (0.6)	72	100	++	++
	Brazil	431	395	90.3 (2.0)	158	89.7 (2.9)	149	92.3 (3.0)	88	88.0 (4.5)	0.8	0.684
	Colombia	430	415	97.3 (0.7)	93	97.1 (0.7)	182	97.2 (1.3)	140	97.5 (1.4)	0.1	0.972
	Medellin, CO	196	188	96.3 (1.6)	58	97.2 (2.1)	77	93.5 (3.5)	53	99.8 (0.2)	4.4	0.110
	Mexico	352	324	93.9 (1.4)	82	91.6 (4.0)	133	92.0 (2.2)	109	97.2 (1.3)	4.2	0.121
	Peru	203	187	94.1 (1.1)	35	90.3 (4.5)	85	93.1 (2.2)	67	97.1 (1.5)	4	0.139
	United States	804	784	97.3 (0.6)	177	97.9 (1.1)	336	97.4 (1.1)	271	96.8 (1.3)	0.4	0.822
	Overall	2650	2519	95.3 (0.5)	660	93.9 (1.0)	1059	95.5 (0.8)	800	96.2 (0.9)	3.2	0.200

^--Percentage less than twice the SE or sample size < 30

⁺⁺χ² test was not computed due to an empty cell, or a cell with 100% of cases

Abbreviations: SE, Standard Error

Analyses performed on part II sample

3.2.2. Attitudinal barriers

Regarding attitudinal barriers (Table 3), the vast majority of respondents (95.3% overall and above 90% in all surveys) reported at least one, with no overall differences by severity of disorder. Overall, wanting to handle the problem by themselves was the most common reason for not seeking treatment, ranging from 60.9% in Argentina up to 73.4% in Medellin, Colombia. Stigma was the least cited reason, ranging from 0.8% to 15.9%, also in Argentina and in Medellin, Colombia. Each individual attitudinal barrier was more common in severe cases, except for wanting to handle it on their own, where the opposite was observed.

3.3. Socio-demographic predictors by country

3.3.1. Socio-demographic predictors of structural barriers

Overall, being female, with high-average and low-average educational attainment, and increased severity of disorders were significant correlates of perceiving structural barriers (Table 4). Females were 70% more likely than males to report structural barriers than males. Compared to those with the highest level of education, respondents with low average and high average education had twice the odds of reporting any structural barrier. Similarly, compared to mild cases, moderate and severe cases reported 1.7 and 2.5 the odds, respectively, of structural barriers. Controlling for sociodemographic variables, structural barriers significantly varied by disorder severity (χ²=21.4; p<0.001). Few deviations from the mean were observed for individual countries. In Argentina, previously married respondents, compared to those currently married were more likely to experience any structural barrier, and those with lower income experienced significantly less structural barriers than the average of the region. In Medellin, Colombia those with moderate severity, compared to mild severity, had threefold the odds of reporting any structural barrier. Few other deviations from the overall associations were observed.

Table 4:

Socio-demographic predictors of any structural barrier among the subgroup with 12-month mental disorders who perceived a need for mental health care but did not access any, in the WMH Americas Surveys. Country effect vs. overall effect.

Variable		Overall (n = 2650)		Argentina (n = 234)		Brazil (n = 431)		Colombia (n = 430)		Medellin, Colombia (n = 196)		Mexico (n = 352)		Peru (n = 203)		United States (n = 804)
		aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)
Sex
	Female	1.7*	(1.2–2.3)	1.3	(0.6–3.0)	1.0	(0.6–1.7)	0.7	(0.4–1.2)	1.1	(0.5–2.7)	1.2	(0.7–2.4)	0.8	(0.3–2.1)	1.0	(0.7–1.5)
	Male	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ21 (p-value)	10.1* (0.001)		0.4 (0.509)		0.0 (0.956)		2.1 (0.150)		0.0 (0.832)		0.5 (0.497)		0.2 (0.685)		0.0 (0.945)
Age
	Age 18–34	1.0	(0.4–2.5)	0.4	(0.1–2.0)	0.7	(0.2–3.4)	0.3	(0.0–2.8)	1.6	(0.5–5.5)	0.4	(0.1–1.9)	0.7	(0.3–1.9)	25.4*	(3.4–188.6)
	Age 35–49	0.8	(0.3–1.9)	0.2*	(0.0–0.9)	0.9	(0.2–4.3)	0.2	(0.0–2.4)	1.1	(0.4–3.2)	0.6	(0.1–2.7)	1.4	(0.5–3.8)	30.9*	(4.6–208.3)
	Age 50–64	0.9	(0.3–2.2)	0.8	(0.2–3.3)	0.6	(0.1–3.3)	0.3	(0.0–3.2)	1.0	-	0.4	(0.1–2.0)	1.0	-	18.8*	(2.4–149.8)
	Age ≥65	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ22–3 (p-value)	2.9 (0.409)		6.2 (0.104)		1.5 (0.688)		1.9 (0.588)		0.8 (0.676)		2.6 (0.458)		4.3 (0.115)		13.8* (0.003)
Education
	Low	1.5	(0.9–2.5)	0.8	(0.3–2.8)	0.8	(0.3–2.3)	1.5	(0.6–4.0)	2.0	(0.3–15.9)	0.7	(0.3–1.6)	0.7	(0.2–3.0)	1.0	(0.5–2.1)
	Low average	1.7*	(1.1–2.7)	1.1	(0.3–4.0)	0.6	(0.2–1.6)	1.3	(0.5–3.5)	1.4	(0.4–5.0)	1.1	(0.5–2.7)	1.0	(0.2–4.0)	0.8	(0.4–1.5)
	High average	1.8*	(1.2–2.8)	0.7	(0.2–2.7)	1.3	(0.5–3.7)	0.9	(0.3–2.6)	1.5	(0.5–4.6)	1.1	(0.5–2.5)	0.8	(0.2–2.5)	0.8	(0.4–1.5)
	High	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ23 (p-value)	8.5* (0.036)		0.6 (0.889)		4.3 (0.230)		1.3 (0.723)		0.8 (0.856)		1.8 (0.622)		0.4 (0.948)		1.0 (0.791)
Marital status
	Married-cohabitating	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	Previously married	0.9	(0.6–1.2)	2.9*	(1.0–8.1)	1.1	(0.5–2.3)	0.8	(0.3–2.0)	0.8	(0.3–2.3)	1.0	(0.4–2.4)	0.6	(0.2–1.7)	0.9	(0.5–1.6)
	Never married	1.0	(0.7–1.3)	1.6	(0.6–3.8)	0.6	(0.3–1.3)	1.1	(0.6–2.1)	0.8	(0.4–1.8)	1.3	(0.6–2.6)	1.5	(0.6–3.7)	0.6	(0.4–1.1)
	χ22 (p-value)	0.7 (0.714)		4.3 (0.117)		2.4 (0.303)		0.5 (0.791)		0.3 (0.841)		0.5 (0.788)		2.0 (0.367)		2.7 (0.256)
Income
	Low	1.2	(0.8–1.7)	0.2*	(0.1–0.6)	1.2	(0.5–2.6)	0.9	(0.5–1.6)	2.4	(0.9–6.9)	1.2	(0.5–2.5)	2.2	(0.7–7.2)	0.7	(0.4–1.3)
	Low average	1.1	(0.7–1.6)	0.1*	(0.0–0.5)	1.7	(0.7–4.2)	1.2	(0.5–3.3)	1.9	(0.4–8.6)	1.3	(0.6–2.8)	2.2	(0.9–5.2)	0.7	(0.3–1.4)
	High average	0.9	(0.6–1.4)	0.2*	(0.1–0.8)	0.9	(0.4–2.0)	0.9	(0.4–2.0)	2.4	(0.6–9.4)	0.8	(0.4–1.8)	3.6	(0.9–14.1)	0.8	(0.4–1.5)
	High	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ23 (p-value)	2.8 (0.422)		11.9* (0.008)		2.0 (0.575)		0.7 (0.868)		3.0 (0.392)		1.4 (0.700)		4.4 (0.220)		1.4 (0.696)
Severity
	Severe	2.5*	(1.7–3.6)	1.8	(0.7–4.8)	0.9	(0.4–1.8)	1.0	(0.4–2.3)	1.7	(0.5–5.8)	0.7	(0.3–1.4)	0.5	(0.2–1.4)	1.0	(0.6–1.8)
	Moderate	1.7*	(1.2–2.4)	1.2	(0.4–3.3)	0.8	(0.4–1.7)	0.9	(0.4–1.8)	3.5*	(1.1–11.2)	0.7	(0.3–1.3)	0.6	(0.3–1.3)	0.8	(0.5–1.4)
	Mild	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ22 (p-value)	21.4* (<0.001)		1.7 (0.422)		0.4 (0.810)		0.2 (0.919)		6.5* (0.038)		1.6 (0.453)		2.2 (0.327)		1.3 (0.534)

Model Fit AIC = 1883.0

^aData are given as adjusted odds ratios (95% confidence interval) unless otherwise indicated

^*Significant at p = .05, 2-sided test

Reference categories are denoted as 1.0 -; Age groups 50–64 and 65+ were collapsed for Peru and Medellin due to cells with zero-counts.

The degrees of freedom for each chi-square test is based upon the number of groups available in each main category

Note: each row shows a separate logistic regression model with any structural barrier as the outcome variable, controlling for the other predictor variables (rows), survey, and all predictor-by-survey interaction dummies. The second column shows the overall adjusted predictor variable effect; the survey columns show to what extent the survey-specific adjusted predictor variable effect deviates from the overall adjusted predictor variable effect. For example, the survey-specific effect for females (versus males) in Mexico can be obtained by multiplying the aOR = 1.6 (the overall effect) by the aOR = 1.3 (the country-specific deviation), i.e., aOR = 2.1

Models include controls for groups of 12-Month DSM-IV / WMH CIDI disorders (any anxiety, any mood, any substance, and any externalized)

Intermittent explosive disorder was not assessed in Mexico and Medellin, were coded as zero; Imputed variables for alcohol and drug dependence were used for Colombia, Mexico, Peru and the U.S.; Lifetime ADHD was used in all countries, and was coded as zero for those with age > 45 in Colombia, Mexico, Peru and the U.S.

3.3.2. Socio-demographic predictors of attitudinal barriers

Overall, being female and having a moderate or severe disorder were associated to a lower proportion of any attitudinal barrier (Table 5). However, while in Argentina respondents with severe disorders were even less prone than the average to report this type of barrier, in the United States and Brazil, respondents were more likely to report it. Based on the high proportion of respondents that reported “wanting to handle it on their own”, we independently analyzed this attitudinal barrier (Table 6). Overall, compared to their counterparts, females and those with severe disorders were less likely to report this barrier. At the country level, respondents with severe disorders in Argentina, younger respondents in Brazil, and previously married respondents in Colombia were significantly less likely to report this barrier.

Table 5:

Socio-demographic predictors of any attitudinal barrier among the subgroup with 12-month mental disorders who perceived a need for mental health care but did not access any, in the WMH Americas Surveys. Country effect vs. overall effect.

Variable		Overall (n = 2650)		Argentina (n = 234)		Brazil (n = 431)		Colombia (n = 430)		Medellin, Colombia (n = 196)		Mexico (n = 352)		Peru (n = 203)		United States (n = 804)
		aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)
Sex
	Female	0.4*	(0.2–0.8)	2.2	(0.5–8.8)	1.0	(0.4–2.5)	0.9	(0.1–5.1)	-	-	0.5	(0.1–2.1)	0.8	(0.2–3.5)	1.4	(0.5–3.8)
	Male	1.0	-	1.0	-	1.0	-	1.0	-	-	-	1.0	-	1.0	-	1.0	-
	χ21 (p-value)	6.3* (0.012)		1.2 (0.283)		0.0 (0.966)		0.0 (0.869)		-		0.9 (0.347)		0.1 (0.724)		0.4 (0.507)
Age
	Age 18–49	1.0	(0.5–2.2)	0.7	(0.0–14.4)	1.1	(0.3–3.5)	3.7	(0.9–15.3)	1.4	(0.3–6.8)	0.5	(0.1–2.1)	-	-	0.5	(0.2–1.8)
	Age ≥50	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	-	-	1.0	-
	χ21 (p-value)	0.0 (0.929)		0.1 (0.795)		0.0 (0.874)		3.2 (0.075)		0.2 (0.692)		0.9 (0.343)		-		1.0 (0.307)
Education
	Low / Low average	0.6	(0.3–1.1)	1.0	(0.1–6.8)	1.5	(0.5–4.2)	1.6	(0.3–9.6)	0.1*	(0.0–0.4)	0.6	(0.2–1.8)	3.4	(0.7–15.5)	3.2*	(1.1–9.0)
	High average / High	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ21 (p-value)	2.7 (0.099)		0.0 (0.977)		0.5 (0.492)		0.3 (0.599)		8.4* (0.004)		0.9 (0.348)		2.4 (0.119)		4.8* (0.028)
Marital status
	Married-cohabitating	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	Previously married	1.1	(0.6–2.3)	0.4	(0.1–2.8)	0.9	(0.3–2.7)	0.5	(0.1–1.7)	1.0	-	1.6	(0.4–6.1)	2.7	(0.3–24.6)	1.2	(0.4–3.6)
	Never married	1.2	(0.6–2.2)	0.4	(0.1–3.1)	1.5	(0.3–6.7)	3.3	(0.7–15.1)	0.5	(0.1–3.0)	0.7	(0.2–2.6)	1.4	(0.4–4.3)	1.2	(0.3–4.1)
	χ21–2 (p-value)	0.4 (0.800)		5.3 (0.071)		0.3 (0.849)		5.7 (0.057)		0.7 (0.417)		1.0 (0.605)		1.1 (0.567)		0.2 (0.919)
Income
	Low / Low average	0.6	(0.3–1.2)	7.7	(0.8–72.0)	0.4	(0.1–1.5)	0.8	(0.1–5.2)	1.0	(0.1–8.7)	0.5	(0.1–1.4)	1.1	(0.2–5.3)	0.7	(0.2–3.0)
	High average / High	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ21 (p-value)	2.1 (0.152)		3.3 (0.071)		1.9 (0.168)		0.1 (0.822)		0.0 (0.979)		1.8 (0.181)		0.0 (0.879)		0.2 (0.678)
Severity
	Severe	0.3*	(0.1–0.6)	0.0*	(0.0–0.4)	4.3*	(1.5–12.3)	3.1	(0.8–12.1)	0.2	(0.0–2.7)	1.3	(0.3–5.4)	1.0	(0.2–5.0)	6.0*	(1.4–25.0)
	Moderate	0.4*	(0.2–0.8)	1.0	-	4.3*	(1.4–13.0)	2.0	(0.4–10.7)	0.0*	(0.0–0.4)	1.1	(0.3–3.4)	0.9	(0.3–3.2)	2.9	(0.9–9.9)
	Mild	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ21–2 (p-value)	12.5* (0.002)		8.1* (0.004)		9.1* (0.011)		2.7 (0.263)		10.7* (0.005)		0.2 (0.910)		0.0 (0.990)		6.5* (0.039)

Model Fit AIC = 642.4

^aData are given as adjusted odds ratios (95% confidence interval) unless otherwise indicated

^*Significant at p = .05, 2-sided test

Reference categories are denoted as 1.0 -; The Female and Previously Married categories for Medellin, the Moderate severity, and the Age category for Peru were excluded due to cells with zero-counts.

The degrees of freedom for each chi-square test is based upon the number of groups available in each main category

Note: each row shows a separate logistic regression model with any attitudinal barrier as the outcome variable, controlling for the other predictor variables (rows), survey, and all predictor-by-survey interaction dummies. The second column shows the overall adjusted predictor variable effect; the survey columns show to what extent the survey-specific adjusted predictor variable effect deviates from the overall adjusted predictor variable effect. Thus, the survey-specific effect in any given survey can be obtained by multiplying the overall aOR by the survey-specific aOR (i.e., the deviation from the overall effect)

Models include controls for groups of 12-Month DSM-IV / WMH CIDI disorders (any anxiety, any mood, any substance, and any externalized)

Intermittent explosive disorder was not assessed in Mexico and Medellin, were coded as zero; Imputed variables for alcohol and drug dependence were used for Colombia, Mexico, Peru and the U.S.; Lifetime ADHD was used in all countries, and was coded as zero for those with age > 45 in Colombia, Mexico, Peru and the U.S.

Table 6:

Socio-demographic predictors of wanting to handle it by their own (attitudinal barrier) among the subgroup with 12-month mental disorders who perceived a need for mental health care but did not access any, in the WMH Americas Surveys. Country effect vs. overall effect.

Variable		Overall (n = 2650)		Argentina (n = 234)		Brazil (n = 431)		Colombia (n = 430)		Medellin, Colombia (n = 196)		Mexico (n = 352)		Peru (n = 203)		United States (n = 804)
		aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)	aORa	(95% CI)
Sex
	Female	0.7*	(0.6–1.0)	0.9	(0.5–1.7)	0.8	(0.5–1.2)	1.3	(0.7–2.2)	1.0	(0.5–2.1)	1.0	(0.6–1.7)	1.1	(0.6–2.2)	1.0	(0.6–1.4)
	Male	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ21 (p-value)	5.1* (0.025)		0.1 (0.814)		1.1 (0.284)		0.6 (0.428)		0.0 (1.000)		0.0 (0.935)		0.1 (0.709)		0.0 (0.847)
Age
	Age 18–49	1.2	(0.9–1.8)	1.8	(0.7–4.4)	0.5*	(0.3–1.0)	0.9	(0.4–2.4)	1.8	(0.7–4.7)	0.7	(0.3–1.8)	1.1	(0.3–3.4)	0.8	(0.5–1.4)
	Age ≥50	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ21 (p-value)	1.3 (0.263)		1.7 (0.188)		4.3* (0.038)		0.0 (0.852)		1.7 (0.198)		0.5 (0.488)		0.0 (0.886)		0.6 (0.452)
Education
	Low / Low average	1.1	(0.8–1.5)	0.7	(0.3–1.8)	0.9	(0.5–1.7)	1.1	(0.6–2.1)	1.3	(0.5–3.0)	1.0	(0.6–1.9)	0.9	(0.5–1.8)	1.1	(0.7–1.7)
	High average / High	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ21 (p-value)	0.6 (0.444)		0.6 (0.448)		0.0 (0.836)		0.1 (0.711)		0.3 (0.587)		0.0 (0.934)		0.0 (0.868)		0.2 (0.689)
Marital status
	Married-cohabitating	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	Previously married	1.4	(0.9–2.1)	0.6	(0.2–1.7)	0.9	(0.4–1.7)	0.4*	(0.2–1.0)	1.2	(0.3–4.5)	2.1	(0.9–5.1)	1.7	(0.5–6.5)	0.9	(0.5–1.6)
	Never married	1.0	(0.8–1.4)	0.6	(0.3–1.4)	1.3	(0.6–2.8)	1.7	(0.8–3.3)	1.1	(0.5–2.6)	1.2	(0.7–2.1)	0.9	(0.4–1.7)	0.6	(0.4–1.0)
	χ22 (p-value)	2.7 (0.264)		1.8 (0.397)		0.7 (0.689)		7.4* (0.025)		0.2 (0.925)		2.9 (0.229)		0.9 (0.629)		3.3 (0.188)
Income
	Low / Low average	0.8	(0.7–1.1)	2.2	(0.9–5.1)	0.7	(0.4–1.2)	1.3	(0.8–2.2)	0.7	(0.3–1.6)	0.9	(0.5–1.5)	0.8	(0.4–1.5)	0.9	(0.6–1.4)
	High average / High	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ21 (p-value)	1.7 (0.187)		3.1 (0.077)		1.4 (0.232)		1.0 (0.311)		0.5 (0.459)		0.1 (0.707)		0.5 (0.489)		0.2 (0.663)
Severity
	Severe	0.5*	(0.3–0.7)	0.4*	(0.2–1.0)	2.1	(0.9–4.9)	0.9	(0.5–1.9)	1.1	(0.4–3.1)	1.3	(0.6–2.7)	0.8	(0.4–1.8)	1.0	(0.6–1.6)
	Moderate	0.8	(0.6–1.0)	1.3	(0.6–2.8)	1.5	(0.8–2.9)	1.1	(0.6–2.0)	0.4	(0.2–1.1)	1.3	(0.7–2.3)	0.9	(0.4–2.0)	1.0	(0.6–1.7)
	Mild	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-	1.0	-
	χ22 (p-value)	17.8* (<0.001)		6.2* (0.045)		3.6 (0.166)		0.1 (0.945)		6.0* (0.050)		1.0 (0.621)		0.2 (0.899)		0.1 (0.964)

Model Fit AIC = 2003.4

^aData are given as adjusted odds ratios (95% confidence interval) unless otherwise indicated

^*Significant at p = .05, 2-sided test

Reference categories are denoted as 1.0 −.

The degrees of freedom for each chi-square test is based upon the number of groups available in each main category

Note: each row shows a separate logistic regression model with handle own as the outcome variable, controlling for the other predictor variables (rows), survey, and all predictor-by-survey interaction dummies. The second column shows the overall adjusted predictor variable effect; the survey columns show to what extent the survey-specific adjusted predictor variable effect deviates from the overall adjusted predictor variable effect. Thus, the survey-specific effect in any given survey can be obtained by multiplying the overall aOR by the survey-specific aOR (i.e., the deviation from the overall effect)

Models include controls for groups of 12-Month DSM-IV / WMH CIDI disorders (any anxiety, any mood, any substance, and any externalized)

Intermittent explosive disorder was not assessed in Mexico and Medellin, were coded as zero; Imputed variables for alcohol and drug dependence were used for Colombia, Mexico, Peru and the U.S.; Lifetime ADHD was used in all countries, and was coded as zero for those with age > 45 in Colombia, Mexico, Peru and the U.S.

4. Discussion

While the WMH surveys is one of the largest transcultural projects ever done in psychiatric epidemiology, with large sample sizes, respondents with a 12-month mental disorder and no past 12-month service use are a reduced group of people. Because some of our data are sparce for some of the individual surveys included here, we focus our discussion on overall findings, with emphasis on trends of the data instead of plain “statistical significance”. Our findings indicate that 43% of people with a 12-month mental or substance use disorder did not seek services because of low perceived treatment need, while 57% perceived a need, but did not access services due to a combination of structural and attitudinal barriers. In light of the low levels of coverage for mental and substance use disorders globally and in the Americas (Borges et al., 2020; Wang et al., 2007a), these findings offer valuable insights for clinicians and decision-makers. Low perceived need may reflect two very different situations: first, it may reflect an inability to adequately gauge one’s own need for services that are unequivocally required, such as for severe major depressive and bipolar disorders; second, it may represent an informed personal decision in the context of milder clinical presentations. Indeed, research has shown that up to 20% of people with major depressive disorder recover on their own, so decisions to forego mental health services may be warranted both from a clinical and public health perspective (Boerema et al., 2017). Our data are consistent with these considerations: low perceived need was a reason not to seek services most frequently in people with milder disorders (50%) vs. people with severe disorders (32%). Given that people with severe disorders do require treatment (since our definition includes high levels of disability, substance use disorders, suicide attempts, or bipolar disorder type I during the past 12 months), it is concerning to find that a third of them did not perceive such need. This may signal the need to strengthen (a) public messaging efforts highlighting the importance of treating mental disorders, particularly those at the severe end of the spectrum, and the risks of untreated severe mental illness; and (b) screening practices for mental disorders at the primary care level to broadly deliver targeted messages for people with detectable severe syndromes. Also, it is noteworthy that net perceived treatment need in the Americas might be associated with immigration status, as suggested by previous studies from the United States (Breslau et al., 2017; Breslau et al., 2020; Villatoro et al., 2018), where more than half of its foreign-born population arrived from Latin America and the Caribbean (OECD et al., 2011).

The majority of people who did not access services despite having a 12-month disorder (57%) perceived treatment need but reported a variety of structural or attitudinal reasons for not receiving care. Throughout the region, attitudinal barriers were more frequently endorsed than structural barriers, and both types of barriers were reported more frequently by respondents with severe disorders (except for “wanting to handle the problem on their own”, which we discuss below). The most frequent structural barriers were financial concerns and availability of services, and they were highly prevalent for severely affected individuals, more than a fourth of whom cited them as reasons why they did not get the care they needed. Of note, structural barriers were the lowest in the two countries that qualified as high-income countries (HICs) at the time of the survey, Argentina (21.1%) and the United States (22.1%), whereas structural barriers in the LMICs in our sample were reported by a range of 27% to 32.5% of respondents. That is, in the LMICs close to a third of the persons that did not access treatment would be benefited by policy changes such as universal health care and increased availability of easy-to-access and low-cost treatment options.

Another important finding was that a great proportion of respondents reported they wanted to deal with their symptoms on their own. Interventions may need to be more focused on helping people help themselves. This has two main implications. First, public mental health interventions should include psychoeducation that may serve as toolkit to teach basic strategies to address mental health problems. A possible strategy could be self-guided internet-based interventions, which have been explored but not fully implemented in LMIC settings (Fu et al., 2020). Second, by offering self-help interventions for mild cases, outreach and outpatient services could focus on the moderate and severe cases.

A couple of caveats with respect to these analyses are in order. First, the distinction between HICs and LMICs may not be the best framework to explain the differences in prevalence of structural barriers described above. Indeed, the country that has the lowest level of “any structural barrier” was Argentina, which despite meeting criteria for HIC at the time of the survey, is a LMIC at the time of writing. It is more likely that the low structural barriers in Argentina were due to relatively high public health spending, and high level of mental health resources, especially psychologists and, to a lesser extent, psychiatrists (Organización Mundial de la Salud, 2011). Second, some of the “attitudinal barriers” we have identified are not necessarily barriers stricto sensu. The most prevalent attitudinal barrier was “I wanted to handle the problem on my own”, which together with “I thought the problem would get better by itself” presents a similar complexity to the one we described in our “low perceived need” category: as referenced above (Boerema et al., 2017), up to 20% of people with depression do get better on their own so a proportion of these respondents may be right in advancing these reasons not to seek services despite perceiving a need. In addition to these caveats, other limitations include that only six countries in the Americas are represented, and that not all surveys reflect the within-country regional heterogeneity. Our surveys mainly included urban populations, so it is possible that we have underestimated structural barriers that are likely to be greater in rural areas due to fewer services. While our data preclude testing this, it is noteworthy that Handley et al. (2014) also found greater perceived attitudinal than structural barriers to getting into treatment in a rural population of Australia. Having said that, there is no other source of primary mental health data in the region of comparable quality and methodological consistency. Another important limitation is that due to our focus on people that met 12-month disorder criteria, in some countries the sample size was too small, and some analyses were not able to be performed. Finally, the cross-sectional nature of the survey precludes any causal interpretation of the findings.

Despite the limitations, the implications of these findings for clinical and public health practice are relevant. The fact that women and people of lower educational levels had higher odds of perceiving structural barriers may signal discrimination toward women and unwarranted complexities in accessing mental health services by people unfamiliar with bureaucratic processes. Also, people with more severe disorders also have increased likelihood of running into structural barriers, which may indicate insufficiently facilitated access for the people with the highest level of need. And finally, our results seem to indicate that, in addition to country income level, relatively high financial and human resource investment in mental health does result in more adequate perception of need -awareness of the disease-, less structural barriers, and less stigma-related barriers.