Associations between subjective social status and DSM-IV mental disorders: Results from the World Mental Health Surveys

Abstract

Importance

The inverse social gradient in mental disorders is a well-established research finding with important implications for causal models and policy. This research has used traditional objective social status (OSS) measures such as education, income and occupation. Recently, subjective social status (SSS) measurement has been advocated to capture perception of relative social status, but to date there are no studies of associations between SSS and mental disorders.

Objective

To estimate associations of SSS with DSM-IV mental disorders in multiple countries and to investigate whether the associations persist after comprehensive adjustment of OSS.

Design; Setting; Participants

Face-to-face cross-sectional household surveys of community-dwelling adults in 18 countries in Asia, South Pacific, the Americas, Europe, the Middle East (n= 56,085). SSS was assessed with a self-anchoring scale reflecting respondent evaluations of their place in the social hierarchies of their countries in terms of income, education and occupation. Scores on the 1–10 SSS scale were categorised into four categories: low (scores 1–3); low-mid (scores 4 and 5); high-mid (scores 6 and 7); high (scores 8–10). OSS was assessed with a wide range of fine-grained objective indicators of income, education and occupation.

Main Outcome Measures

The Composite International Diagnostic Interview assessed 12-month prevalence of 16 DSM-IV mood, anxiety and impulse control disorders.

Results

Graded, inverse associations were found between SSS and all 16 mental disorders. Gross odds-ratios (lowest versus highest SSS categories) in the range 1.8–9.0 were attenuated but remained significant for all 16 disorders (ORs: 1.4–4.9) after adjusting for OSS indicators. The pattern of inverse association between SSS and mental disorders was significant in 14/18 individual countries, and in low, middle and high income country groups, but was significantly stronger in higher versus lower income countries.

Conclusions

Significant inverse associations between SSS and numerous DSM-IV mental disorders exist across a wide range of countries even after comprehensive adjustment for OSS. Although unclear whether these associations are due to social selection, social causation, or both, these results document clearly that research relying exclusively on standard OSS measures underestimates the steepness of the social gradient in mental disorders.

INTRODUCTION

Decades of research have established that socioeconomic status (SES) is inversely associated with many mental disorders.^1–4 Most of this research has used traditional indicators of SES such as education, income and occupation, referred to herein as measures of objective social status (OSS). However, a recent development in the research on the associations between SES and health has been the evaluation of subjective social status (SSS). Most studies have found that SSS is associated with physical health and psychological distress even after controlling for OSS,^5–14 a finding that has been explained by the idea that SSS captures subjective judgement of relative social position.^{7, 14} Relative social position has become a topic of great interest based on striking findings from the income inequality and physical health literature such as that African American men with a four-fold higher income than Costa Rican men nonetheless have a nine year shorter life expectancy.¹⁵ This shorter life expectancy has been attributed in part to the psychosocial effects of relative deprivation and status anxiety caused by the lower relative social position of African Americans.^{15, 16} More recently, greater income inequality among wealthy countries has also been associated with higher prevalence of mental disorders.¹⁷

Although use of SSS measures in mental health research has been advocated,¹⁸ prior studies have typically used measures of psychological distress such as the SF-36¹⁹ or GHQ,²⁰ and research on SSS and individual mental disorders has not yet been carried out. Examining a range of mental disorders is important as much of the past research on social stratification and mental health has measured depression as the outcome, but concepts of relative deprivation and status insecurity imply a wide range of emotional responses including anger, frustration, hostility and anxiety.¹⁸ The present study uses data from 20 of the WHO World Mental Health (WMH) Surveys to examine associations of SSS with 16 DSM-IV disorders, with the aim of determining whether these associations persist after controlling for multiple fine-grained measures of OSS. As prior research has suggested that SSS-health associations vary by culture^{10, 21} we also estimate SSS-mental disorder associations in individual countries. Additionally, because the association between income inequality and mental disorders has only been found in wealthy countries,¹⁷ we examine associations in countries grouped by income level and test whether associations vary across high, medium and low income countries.

METHODS

Samples and Procedures

This study used data from 20 surveys in 18 countries (Table 1). A stratified multi-stage clustered area probability sampling strategy was used to select adult respondents. Most of the surveys were based on nationally representative household (or population register) samples while Colombia, Mexico and Shenzhen were based on nationally representative household samples in urbanized areas. The weighted average response rate across all surveys included this paper was 75.2% (Table 1). The surveys are shown in Table 1 grouped by World Bank country income categories that are henceforth referred to as low, middle and high income country groups.

Table 1.

World mental health sample characteristics by world bank income categories.

Country	Survey	Sample characteristics	Field dates	Age range	Sample Size		Response Rate
					Part 1	Part 2
Low - lower middle income countries
Colombia	NSMH	All urban areas of the country (approximately 73% of the total national population)	2003	18–65	4426	2381	87.7
Colombia - Medellin	MMHHS	Medellin metropolitan area	2011–2	18–65	3261	1673	97.2
PRC Shen Zhen	Shenzhen	Shenzhen metropolitan area. Included temporary residents as well as household residents.	2006–7	18+	7132	2475	80.0
Iraq	IMHS	Nationally representative.	2006–7	18+	4332	4332	95.2
PRC Beijing/Shanghai	B-WMH/S- WMH	Beijing and Shanghai metropolitan areas.	2002–3	18+	5201	1628	74.7
Nigeria	NSMHW	21 of the 36 states in the country, representing 57% of the national population. The surveys were conducted in Yoruba, Igbo, Hausa and Efik languages.	2002–3	18+	6752	2143	79.3
Ukraine	CMDPSD	Nationally representative.	2002	18+	4724	1719	78.3
Upper- middle income countries
Mexico	M-NCS	All urban areas of the country (approximately 75% of the total national population).	2001–2	18–65	5782	2362	76.6
Peru	EMSMP	Nationally representative.	2004–5	18–65	3930	1801	90.2
Brazil	São Paulo Megacity	São Paulo metropolitan area.	2005–7	18+	5037	2942	81.3
Bulgaria	NSHS	Nationally representative.	2003–7	18+	5318	2233	72.0
South Africa	SASH	Nationally representative.	2003–4	18+	4315	4315	87.1
Lebanon	LEBANON	Nationally representative.	2002–3	18+	2857	1031	70.0
High- income countries
Japan	WMHJ 2002–2006	Eleven metropolitan areas.	2002–6	20+	4129	1682	55.1
New Zealand	NZMHS	Nationally representative.	2003–4	18+	12790	7312	73.3
Northern Ireland	NISHS	Nationally representative.	2004–7	18+	4340	1986	68.4
Portugal	NMHS	Nationally representative.	2008–9	18+	3849	2060	57.3
Israel	NHS	Nationally representative.	2002–4	21+	4859	4859	72.6
United States	NCS-R	Nationally representative.	2002–3	18+	9282	5692	70.9
Spain - Murcia	PEGASUS- Murcia	Murcia region	2010–2	18+	2621	1459	67.4
Total					104937	56085
Weighted average response rate (%)							75.2

The central WMH staff trained bilingual supervisors in each country. The WHO translation protocol was used to translate instruments and training materials. Some surveys were carried out in bilingual form while others were carried out exclusively in the country’s official language. Translation, back-translation, and harmonization of the WMH interview used standardized procedures that are discussed elsewhere.²² In most countries, internal subsampling was used to reduce respondent burden and average interview time by dividing the interview into two parts. All respondents completed Part 1 that included the core diagnostic assessment of most mental disorders. All Part 1 respondents who met lifetime criteria for any mental disorder and a probability sample of respondents without mental disorders were administered Part 2 (at the same interview sitting) that assessed the remaining mental disorders and collected a range of other information. Part 2 respondents were weighted by the inverse of their probability of selection for Part 2 of the interview to adjust for differential sampling, resulting in an unbiased sample. The analyses in this study are based on the Part 2 subsample (n=56,085).

Additional weights were used to adjust for differential probabilities of selection within households, to adjust for non-response, and to match the samples to population sociodemographic distributions. Measures taken to ensure data accuracy, cross-national consistency and protection of respondents are described in detail elsewhere.^{22, 23} All respondents provided written informed consent and procedures for protecting respondents were approved and monitored for compliance by the Institutional Review Boards in each country.²²

Measures

Mental disorders

All surveys used the WMH survey version of the WHO Composite International Diagnostic Interview (CIDI 3.0 ²³), a fully structured interview, to assess lifetime history and 12 month prevalence of DSM-IV mental disorders. The disorders included in this paper include anxiety disorders (panic disorder, agoraphobia without panic, specific phobia, social phobia, post-traumatic stress disorder, generalized anxiety disorder, obsessive compulsive disorder); mood disorders (major depressive disorder/dysthymia, bipolar broad (I, II and subthreshold)); substance use disorders (alcohol abuse and dependence, drug abuse and dependence); and impulse control disorders (intermittent explosive disorder, bulimia nervosa and binge eating disorder).

Subjective social status (SSS)

This was measured with the MacArthur subjective social status scale which is the most widely used indicator of SSS with good reliability and validity.^{5, 8, 14, 24} Participants were given a drawing of a ladder with 10 rungs described as follows: “Think of this ladder as representing where people stand in (country of interview). At the top of the ladder are the people who are the best off - those who have the most money, the most education and the most respected jobs. At the bottom are the people who are the worst off - who have the least money, least education and the least respected jobs or no job. The higher up you are on the ladder, the closer you are to the people at the very top; the lower you are, the closer you are to the people at the very bottom. Please place a large ‘X’ on the rung where you think you stand at this time in your life, relative to the other people in (country of interview). What is the number to the right of the rung where you placed the ‘X’?”

Objective social status (OSS)

Education was assessed by self-report of the number of years of schooling completed. Three education variables were created for each respondent. These were: number of years of education; country relative education score (number of years of education divided by the weighted median education (in years) for the respondent’s country); and neighbourhood relative education score (number of years of education divided by the weighted median education (in years) for each ‘neighbourhood’ [primary sampling unit] in the respondent’s country).

Income was assessed by asking respondents to estimate their total family household income in the past 12 months from all sources, before tax or anything was taken out of it, with show cards providing multiple income brackets in the currency of their country from which they could select the appropriate response. Respondents were also asked about personal income, but household income was used in this analysis. Four income variables were created for each respondent. These were: income percentile; income adjusted for household size; country relative income score; and neighbourhood relative income score. These two latter scores were created in a manner analogous to the education scores.

Occupational type was based on respondent information about occupation at the time of interview and classified into one of 28 occupation types, or as not working at the time of interview. Occupational status was categorised as: working (weighted percent: 59.2), student (4.9%), homemaker (12.6%), retired (11.5%), other (11.9%).

Statistical analysis

Scores on the 1–10 SSS scale were categorised into four categories: low (scores 1–3); low-mid (scores 4 and 5); high-mid (scores 6 and 7); high (scores 8–10). The high group was the reference group in all the regression models. Those who did not answer the question (for all countries combined: 3.6% [range across all countries 0.3% to 10.1%]; or with missing data or with outlying scores greater than 10 (0.6% [range 0–2.8%] were excluded from analyses. Country-specific logistic regression models estimated the associations of SSS with the aggregated indicator of any 12-month mental disorder controlling for current age, age squared, gender and country, and then additionally adjusting for all the OSS variables. We tested whether there were differences in strength of SSS-mental disorder associations across low, middle and high income country groups by including cross-product terms for the interaction of SSS with dummy variables representing high income countries and middle income countries (low income countries as reference).

In all countries combined, logistic regression models estimated the associations of SSS with specific 12-month mental disorders, controlling for current age, age squared, gender and country, then for the OSS indicators (income percentile score, income adjusted for household size, years of education, occupational type, occupational status, neighbourhood relative income score, country relative income score, neighbourhood relative years of education, country relative years of education, plus squared versions of the income and education variables).

Gender moderation of associations was investigated but associations did not vary materially for men and women. Significant age moderation of associations was found whereby associations were strongest for the two middle aged groups. However, because the inverse SSS-mental disorder gradient was evident for all age groups, we show results for all ages combined, controlling for current age and including age-squared in the models to capture some of the non-linearities in the relationship between SSS, age and mental disorders.

As the WMH data are both clustered and weighted, the design-based Taylor series linearization method²⁵ implemented in version 11 of the SUDAAN software system was used to estimate standard errors and evaluate the statistical significance of coefficients.

RESULTS

SSS distributions

The distributions of the original 10-point scale and 4 derived SSS categories are shown in Table 2 by country income groupings. The scores are approximately normally distributed with all scores on the 10 point scale observed in each country income group. However, both the 10-point and 4-category distributions do differ significantly across country income groups (χ² = 25.1, p < 0.001; (χ² = 57.8, p < 0.001, respectively). The nature of the differences is clearest in the 4 category distribution, where it can be seen that the proportions scoring low (in the 1–3 range) are larger in the lower-middle income countries (18.3%) and upper-middle income countries (17.6%) than in the high income countries (10.3%).

Table 2.

Distribution of subjective social status, by country income-group.

SSS (Relative standings in terms of money, education and job)	Subjective social status (SSS) scores (%, n)
	1	2	3	4	5	6	7	8	9	10	Other4
Low-lower middle income countries1	3.4 (662)	5.0 (923)	9.1 (1635)	11.6 (1929)	24.4 (3944)	15.1 (2444)	13.1 (1967)	8.3 (1289)	2.8 (406)	2.8 (417)	4.3 (735)
Upper-middle income countries2	3.7 (623)	4.4 (713)	8.4 (1360)	13.4 (2008)	23.4 (3335)	15.1 (2167)	12.6 (1757)	8.9 (1182)	2.5 (311)	1.8 (293)	5.8 (935)
High-income countries3	2.0 (603)	2.4 (771)	5.6 (1679)	9.4 (2602)	24.6 (6307)	19.6 (4617)	18.0 (4272)	10.9 (2431)	3.0 (642)	1.7 (371)	2.9 (755)
All countries combined	2.9 (1888)	3.7 (2407)	7.4 (4674)	11.1 (6539)	24.2 (13586)	17.1 (9228)	15.2 (7996)	9.6 (4902)	2.8 (1359)	2.1 (1081)	4.1 (2425)
	Low (Score 1–3)			Low-mid (Score 4–5)		High-mid (Score 6–7)		High (Score 8–10)
Low-lower middle income countries1	18.3 (3220)			37.6 (5873)		29.5 (4411)		14.6 (2112)
Upper-middle income countries2	17.6 (2696)			39.0 (5343)		29.4 (3924)		14.0 (1786)
High-income countries3	10.3 (3053)			35.0 (8909)		38.6 (8889)		16.1 (3444)
All countries combined	14.5 (8969)			36.8 (20125)		33.6 (17224)		15.1 (7342)

¹Colombia, Colombia (Medellin), PRC Shen Zhen, PRC Beijing/Shanghai, Iraq, Nigeria, Ukraine

²Mexico, Peru, Brazil, Bulgaria, South Africa, Lebanon

³Japan, New Zealand, Northern Ireland, Portugal, Israel, United States, Spain (Murcia)

⁴Don’t know, Refused to answer, Missing and Outliers

Country-specific associations of SSS with any 12-month mental disorder

Table 3 shows associations of the three lower categories of SSS (relative to the highest category) with the aggregated indicator of any 12-month mental disorder in individual countries, in low, middle and high country income groups, and among all countries combined. For all countries combined, there is a graded inverse association of SSS with any mental disorder, with ORs from low to high-mid SSS categories: 2.5, 1.7, 1.3. This inverse gradient is evident in all countries except Japan and Nigeria, and is significant in 14/18 countries, and in 15/20 individual surveys.

Table 3.

Associations between subjective social status and any 12-month mental disorder, by country¹.

Country	Relative standing in terms of money, education and job (SSS)
	Low4		Low-mid5		High-mid6
	OR	(95% C.I)	OR	(95% C.I)	OR	(95% C.I)
Low - lower middle income (pooled)2	2.0*	(1.6–2.5)	1.5*	(1.2–1.8)	1.2	(1.0–1.5)
Colombia	1.6*	(1.1–2.4)	1.6*	(1.0–2.5)	1.2	(0.8–1.8)
PRC ShenZhen	2.1	(0.9–4.7)	1.4	(0.9–2.0)	1.5	(1.0–2.3)
Iraq	1.9*	(1.1–3.1)	1.1	(0.7–1.7)	0.8	(0.5–1.2)
PRC Beijing/ Shanghai	3.8*	(1.7–8.1)	2.7*	(1.4–5.1)	1.2	(0.6–2.4)
Nigeria	1.9	(0.8–4.4)	2.0*	(1.0–4.1)	1.8	(0.8–4.0)
Medellin	2.3*	(1.4–3.9)	1.4	(0.9–2.4)	1.3	(0.8–2.2)
Ukraine	1.4	(0.4–4.7)	1.1	(0.3–3.2)	0.9	(0.3–2.7)
Upper- middle income (pooled)2	2.0*	(1.6–2.5)	1.5*	(1.2–1.8)	1.3*	(1.1–1.5)
Mexico	2.1*	(1.3–3.4)	1.8*	(1.3–2.6)	1.1	(0.8–1.5)
Brazil	2.8*	(1.8–4.6)	1.7*	(1.2–2.5)	1.5*	(1.0–2.3)
Bulgaria	1.9	(0.8–4.5)	1.6	(0.7–3.6)	1.5	(0.6–3.7)
Lebanon	3.2*	(1.3–7.7)	2.0	(0.9–4.4)	1.7	(0.8–3.6)
Peru	2.1*	(1.3–3.4)	1.1	(0.7–1.7)	1.0	(0.6–1.5)
South Africa	1.4*	(1.0–2.0)	1.3	(0.9–1.8)	1.2	(0.8–1.6)
High- income (pooled)2	3.1*	(2.6–3.7)	1.9*	(1.7–2.2)	1.3*	(1.1–1.5)
Japan	2.7	(0.9–8.6)	1.2	(0.4–3.1)	1.7	(0.7–4.4)
New Zealand	3.2*	(2.5–4.1)	2.0*	(1.6–2.4)	1.2	(1.0–1.6)
Northern Ireland	5.2*	(2.6–10.5)	2.6*	(1.6–4.0)	1.3	(0.8–2.0)
Portugal	2.4*	(1.3–4.3)	2.1*	(1.2–3.7)	1.7	(0.9–3.1)
Israel	4.7*	(3.2–6.7)	2.1*	(1.5–3.0)	1.8*	(1.3–2.6)
US	2.6*	(1.8–3.7)	1.7*	(1.4–2.0)	1.1	(0.9–1.3)
Murcia	1.5	(0.6–3.6)	1.4	(0.7–2.8)	1.3	(0.7–2.1)
All countries combined2	2.5*	(2.2–2.8)	1.7*	(1.6–1.9)	1.3*	(1.2–1.4)
	Adjusted for OSS3
Low - lower middle income (pooled)	2.0*	(1.5–2.5)	1.5*	(1.2–1.8)	1.2	(1.0–1.5)
Upper- middle income (pooled)	2.0*	(1.6–2.5)	1.5*	(1.2–1.8)	1.3*	(1.1–1.5)
High- income (pooled)	2.8*	(2.4–3.4)	1.9*	(1.6–2.1)	1.3*	(1.1–1.5)
All countries combined	2.3*	(2.1–2.6)	1.7*	(1.5–1.8)	1.3*	(1.2–1.4)

^*Significant at the 0.05 level, two-tailed test.

¹Models control for current age, age-squared and gender using subjective social status (ref = high) to predict the odds of having any 12-month mental disorder.

²Models control for current age, age-squared, gender and country using subjective social status (ref = high) to predict the odds of having any 12-month mental disorder.

³Models control for current age, age-squared, gender, country, OSS (income percentile (1–100), income percentile adjusted for household size (1–100), current occupation, and education (number of years), country relative income score, neighbourhood relative income score, country relative education score, neighbourhood relative education score) and the squared version of all OSS variables except for current occupation using subjective social status (ref = high) to predict the odds of having any 12-month mental disorder

⁴low= scores 1–3 on the original 1–10 scale

⁵low-mid= scores 4 + 5 on the original 1–10 scale

⁶high-mid = scores 6 + 7 on the original 1–10 scale

The associations between SSS and any mental disorder for each pooled set of countries grouped by income level are stronger for the high income countries (ORs from low to high-mid SSS categories: 3.1, 1.9, 1.3) compared to middle income countries (ORs: 2.0, 1.5, 1.3) and low income countries (ORs: 2.0, 1.5 and 1.2). This country income group difference is statistically significant (χ²₆=22.2, p = 0.001). This interaction effect remained significant after inclusion of OSS in the models (χ²₆=16.1, p = 0.01); the results from models adjusting for OSS are presented in the lower part of the table.

Associations of SSS with individual mental disorders

In all countries combined, a graded, inverse pattern of association was found with all mental disorders unadjusted for OSS (Table 4, first three columns). ORs for the lowest SSS category relative to the highest range from 1.8 for intermittent explosive disorder and OCD to 9.0 for drug dependence, with the ORs for most disorders falling between 2.0 and 4.0.

Table 4.

Associations between subjective social status and 12-month mental disorders, all countries combined.

Type of disorder	Subjective social status (SSS)1			Subjective social status (SSS) (Adjusted for OSS – absolute status)2			Subjective social status (SSS) (Adjusted for OSS – relative status)3
	Low	Low-mid	High-mid	Low	Low-mid	High-mid	Low	Low-mid	High-mid
	OR (95% C.I)	OR (95% C.I)	OR (95% C.I)	OR (95% C.I)	OR (95% C.I)	OR (95% C.I)	OR (95% C.I)	OR (95% C.I)	OR (95% C.I)
Mood disorder
Major depressive episode / Dysthymia	3.2* (2.7–3.7)	1.9* (1.6–2.2)	1.3* (1.2–1.5)	3.0* (2.6–3.5)	1.9* (1.6–2.2)	1.3* (1.2–1.5)	3.0* (2.6–3.5)	1.9* (1.6–2.2)	1.4* (1.2–1.5)
Bipolar disorder (broad)	2.9* (2.2–3.7)	1.4* (1.1–1.9)	1.1 (0.9–1.5)	2.4* (1.8–3.2)	1.3 (1.0–1.7)	1.1 (0.8–1.4)	2.4* (1.8–3.3)	1.3 (1.0–1.7)	1.1 (0.8–1.5)
Anxiety disorder
Panic disorder	3.9* (2.9–5.2)	1.9* (1.4–2.5)	1.3 (0.9–1.7)	2.9* (2.1–4.0)	1.6* (1.2–2.2)	1.2 (0.9–1.7)	2.9* (2.1–4.0)	1.6* (1.2–2.2)	1.2 (0.9–1.7)
Generalized anxiety disorder	3.5* (2.8–4.4)	2.2* (1.8–2.7)	1.4* (1.1–1.7)	3.1* (2.4–4.0)	2.1* (1.7–2.6)	1.4* (1.1–1.7)	3.2* (2.5–4.1)	2.1* (1.7–2.6)	1.4* (1.1–1.7)
Social phobia	2.7* (2.2–3.5)	1.7* (1.4–2.0)	1.2 (1.0–1.5)	2.6* (2.0–3.4)	1.7* (1.4–2.1)	1.2 (1.0–1.5)	2.6* (2.0–3.4)	1.7* (1.4–2.0)	1.2 (1.0–1.5)
Specific phobia	2.0* (1.7–2.3)	1.6* (1.4–1.8)	1.2* (1.0–1.4)	1.8* (1.5–2.1)	1.5* (1.3–1.8)	1.2* (1.0–1.4)	1.8* (1.5–2.2)	1.5* (1.3–1.8)	1.2* (1.0–1.4)
Agoraphobia without panic	2.5* (1.8–3.5)	1.6* (1.2–2.2)	1.1 (0.8–1.6)	2.1* (1.4–3.0)	1.4* (1.0–2.0)	1.1 (0.8–1.6)	2.1* (1.4–3.0)	1.4* (1.1–2.0)	1.1 (0.8–1.6)
Post-traumatic stress disorder	4.4* (3.3–5.9)	2.3* (1.8–3.0)	1.3 (1.0–1.8)	3.9* (2.9–5.3)	2.2* (1.7–2.9)	1.3 (1.0–1.8)	4.0* (2.9–5.4)	2.2* (1.7–2.9)	1.3 (1.0–1.8)
Obsessive compulsive disorder	1.8* (1.1–2.9)	1.3 (0.8–1.9)	1.1 (0.8–1.6)	1.9* (1.2–3.2)	1.3 (0.9–2.0)	1.1 (0.8–1.6)	2.0* (1.2–3.4)	1.4 (0.9–2.1)	1.2 (0.8–1.7)
Impulse-control disorder
Intermittent explosive disorder	1.8* (1.3–2.6)	1.7* (1.3–2.2)	1.2 (1.0–1.6)	1.9* (1.3–2.6)	1.8* (1.3–2.4)	1.3 (1.0–1.7)	1.9* (1.4–2.6)	1.8* (1.4–2.4)	1.3 (1.0–1.7)
Bulimia nervosa	3.0* (1.4–6.7)	1.6 (0.8–3.3)	1.1 (0.5–2.4)	2.8* (1.2–6.6)	1.7 (0.8–3.6)	1.3 (0.6–2.9)	3.2* (1.3–7.8)	1.9 (0.9–4.2)	1.4 (0.6–3.4)
Binge eating disorder	2.7* (1.7–4.5)	1.6* (1.0–2.6)	1.2 (0.7–1.9)	2.6* (1.5–4.4)	1.6 (1.0–2.7)	1.2 (0.7–2.0)	2.7* (1.6–4.6)	1.7* (1.0–2.7)	1.2 (0.8–2.0)
Substance-use disorder
Alcohol abuse	1.9* (1.4–2.5)	1.5* (1.2–2.0)	1.2 (0.9–1.5)	1.5* (1.1–2.1)	1.4* (1.0–1.8)	1.1 (0.8–1.5)	1.5* (1.1–2.1)	1.3* (1.0–1.7)	1.1 (0.8–1.4)
Alcohol dependence	3.2* (2.1–4.9)	2.0* (1.3–2.9)	1.4 (0.9–2.1)	2.1* (1.4–3.2)	1.5* (1.0–2.2)	1.3 (0.8–2.0)	2.0* (1.3–3.1)	1.5 (1.0–2.2)	1.2 (0.8–1.9)
Drug abuse	3.8* (2.4–6.0)	2.5* (1.6–3.8)	1.6* (1.0–2.5)	2.5* (1.5–4.0)	2.0* (1.3–3.2)	1.5 (1.0–2.4)	2.4* (1.5–3.9)	2.0* (1.3–3.1)	1.5 (1.0–2.4)
Drug dependence	9.0* (3.9–20.8)	4.2* (1.9–9.3)	2.4 (1.0–5.7)	5.8* (2.4–13.6)	3.4* (1.6–7.4)	2.2 (0.9–5.3)	5.7* (2.4–13.5)	3.4* (1.5–7.3)	2.2 (0.9–5.2)

^*Significant at the 0.05 level, two-tailed test.

¹Models control for gender, current age, age-squared and country using SSS categories as predictors (reference = high) to predict the odds of 12-month mental disorders in separate logistic regression models, for each mental disorder.

²Models control for gender, current age, age-squared, country, OSS (income percentile (1–100), income percentile adj for household size (1–100), current occupation, and education (number of years)) and the squared version of all OSS variables except for current occupation to predict the odds of 12-month mental disorders in separate logistic regression models, for each mental disorder.

³Models control for gender, current age, age-squared, country, OSS (income percentile (1–100), income percentile adj for household size (1–100), current occupation, and education (number of years), country relative income score, neighbourhood relative income score, country relative education score, neighbourhood relative education score) and the squared version of all OSS variables except for current occupation to predict the odds of 12-month mental disorders in separate logistic regression models, for each mental disorder.

Adjustment for OSS attenuated associations to a variable degree across disorders but most strongly for the substance use disorders and some of the anxiety disorders. Despite this attenuation SSS remained significantly associated with all disorders with most ORs remaining > 2.0. Odds ratios for the lowest SSS category relative to the highest range from 1.4 for alcohol abuse to 4.9 for drug dependence. Of individual disorders SSS was most strongly associated with drug dependence, but when considering associations between SSS and disorder groups, these were smallest in magnitude for any substance use disorder (OR of 1.6 for the lowest SSS category relative to the highest) and largest for any mood disorder (OR of 2.7 for the lowest SSS category relative to the highest).

DISCUSSION

In this general population sample from 18 countries, graded, inverse associations were found between SSS and all mental disorders, where SSS was measured as subjective perception of position in the country-specific hierarchy in terms of income, education and occupation. This pattern of association between SSS and mental disorders was evident in 18/20 surveys, significant in 15/20 surveys, and was significantly stronger in higher than lower income countries. Subjective social status remained associated with all mental disorders after adjustment for a large set of fine-grained objective social status (OSS) indicators.

Limitations of the study include the likelihood that sample selection biases (whereby those with the most severe mental disorders and the lowest socioeconomic status are less likely to be in the sample) may have restricted the range of measures and so attenuated the strength of associations. A further limitation is that our measures of OSS were restricted to education, income and occupation; inclusion of other measures of OSS such as wealth (assets), may have reduced the independent effects of SSS. Finally, the cross-sectional nature of the study prevents clarification of the temporal nature of the associations so that social causation and selection effects cannot be disentangled.

Within the context of these limitations, this study provides the first investigation of the relationship between SSS and diagnostic measures of a wide range of mental disorders. Prior research on SSS has only investigated depression, and this study shows that low SSS is associated with higher risk of all 16 mental disorders investigated, not just depression. Moreover, we found that associations between SSS and mental disorders persisted after more comprehensive adjustment for OSS than achieved in most prior studies. The major explanation advanced for why there are independent associations of SSS with health outcomes after controlling for OSS is that SSS measures subjective perception of relative social position.¹⁴ Perception of lower relative social position has been hypothesized to increase risk of mental ill-health through sense of relative deprivation and status insecurity, with associated feelings of shame, distrust, frustration and anxiety.^{7, 14, 17, 26, 27} Our findings of inverse associations between SSS and all mental disorders appear to offer considerable support for this hypothesis, although as noted above we cannot determine the relative contribution of social causation versus social selection processes.

The stronger association between SSS and mental disorders in the higher relative to lower income countries is interesting in light of recent research finding that greater income inequality was associated with higher prevalence of mental disorders in a group of high income countries.¹⁷ In this study, the stronger association between SSS and mental disorders in the higher income countries persisted even after adjustment for objective differences in absolute and relative household income, so this finding cannot be attributed to higher levels of income inequality in wealthier compared to less wealthy countries. Indeed, income inequality was highest in some of the middle income countries included in this study. If not greater income inequality, what then could explain this finding of a steeper SSS-mental disorder gradient in higher income countries? One contributing factor could be that advertising and media are more influential in higher income countries and this has the effect of making social inequalities more visible and encouraging social comparisons;¹⁷ this in turn could heighten status competition and status insecurity^{17, 27} leading to stronger associations between SSS and mental disorders in high income countries. Another possibility is that lower SSS may be more detrimental to mental health in higher income countries due to values that are more common in high income countries, where success is evaluated in terms of individual achievement and prestige.^{28, 29} In this regard it is interesting to note that Japan, considered a collectivist culture with a strong ethos of social relativism,²⁹ was the only high income country in this study with no clear SSS-mental disorder gradient (see also³⁰). However, most countries, including many others generally considered collectivist, did exhibit SSS-mental disorder associations.

In conclusion, this study found inverse, graded associations between SSS and each of 16 DSM-IV mental disorders that remained strong after adjustment for a large set of detailed OSS indicators. This pattern of association was evident in almost all countries but was significantly stronger in higher than lower income countries. Although interpretation of SSS-mental disorder associations is far from clear cut, the strength and consistency of these associations with mental disorders suggests that further research is warranted, using prospective designs that can help distinguish between social causation and selection processes. The study findings indicate that research into the social gradient in mental health that relies on the standard OSS measures of income, education and occupation will underestimate the steepness of the gradient.