Depression in the United States and Japan: Gender, marital status, and SES patterns

Abstract

A number of investigators have claimed that higher depression scores and higher rates of depressive disorder are found worldwide in women, unmarried persons, and people of low socioeconomic status (SES). A closer look, however, indicates that patterns for Asian countries are less consistent than claimed. As a case in point, using comparable data from the National Family Research of Japan ‘98 survey ($N=6985$) and the National Survey of Families and Households in the US ($N=8111$), we examine the distributions of depressive symptoms by gender, marital status, and SES, with a short form of the CES-D Scale. Bivariate and multivariate analyses show that depressive symptoms are higher in women, unmarried persons, and those with lower family incomes in both countries, but there is no association between education and depression in Japan while symptoms are inversely related to education in the US. We argue that the lack of relationship between education and depression in Japan is not an artifact of measurement but a product of Japan's distinctive stratification processes relating to occupation. Cross-national variations around “general” patterns are important because they offer clues to more specific cultural and structural factors involved in the social etiology of mental disorder.

Introduction

According to reviewers, most cross-cultural studies show that women, unmarried persons, and those of lower education, income, or occupational prestige have higher rates of depressive disorder and/or depressive symptoms than men, the married, and those in more advantaged socioeconomic status (SES) positions (e.g., Kohn, Dohrenwend, & Mirotznik, 1998; Nolen-Hoeksema, 1990; Weissman et al., 1996). Although no one has claimed that these patterns are universal, reviewers usually imply that they are true in most countries. For example, Rosenfield (1989, p. 77) claims that higher depression rates in women exist “across cultures, over time, in different age groups, in rural as well as urban areas, and in treated as well as untreated populations.” Kohn and his colleagues (1998, p. 275) state that distributions of psychiatric disorders by gender and SES are well-established inside and outside the US, and, by implication, applicable worldwide. (See also Kessler, 2002, p. 61.) A closer look at the epidemiological literature indicates that such conclusions are premature, particularly with respect to Asian nations. This inference is underscored again in this paper when we compare gender, marital status, and socioeconomic differences in depressive symptoms in the US to those in Japan—which has rarely been included in cross-national comparisons.

In recent years, information on gender, marital status, and SES differences in depression in the US has been obtained from two types of studies (Kohn et al., 1998), those employing community screening scales such as the Center for Epidemiological Studies-Depression Scale (CES-D) (Radloff, 1977) and those using standardized interviews to assess psychiatric disorders that are described in the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association (1980), American Psychiatric Association (1987), American Psychiatric Association (1994)). In these latter types of research, symptoms defining the most common or serious disorders in the diagnostic manuals are converted into structured sets of questions that can be asked by trained lay interviewers in face-to-face surveys with probability samples of adults. Computer algorithms convert respondents’ answers into diagnoses which are more reliable than those based on psychiatrists’ judgments.

The most widely cited epidemiological studies that use standardized diagnostic interviews in the United States are the Epidemiological Catchment Area studies (ECA), which were carried out in five metropolitan areas in the US (Weissman, Bruce, Leaf, Florio, & Holzer, 1991); the National Comorbidity Survey (NCS), based on a national probability sample (Kessler & Zhao, 1999); and the National Comorbidity Survey Replication (NCS-R), again based on a nationally representative sample (Kessler et al., 2003). All three obtained similar findings for the distributions of depression by gender and marital status. Women had higher lifetime and/or 12-month prevalence rates of major depressive disorder than men (ECA, NCS, NCS-R) and higher lifetime and 1-year rates of dysthymia (a depressed mood that has endured for at least 2 years) compared to men (NCS). Formerly married persons had significantly greater lifetime and/or 1-year rates of depression than the married in the ECA (Weissman et al., 1991), the NCS (Kessler & Zhao, 1999) and the NCS-R (Kessler et al., 2003).

Distributions of major depression by SES in the three investigations were less consistent, however—inversely related only for some of the indicators commonly used for SES (e.g., education, income, employment status). In the ECA, individuals’ education, occupation, and income were unrelated to rates of major depression, but unemployed persons and those on welfare were more likely to have had a major depression during the past 12 months compared to their more advantaged counterparts (Robins et al., 1984; Weissman et al., 1991). In the NCS, both income and education varied inversely with the 12-month prevalence of major depression (Kessler et al., 1994). In the NCS-R, major depressive disorder was negatively related to education but not income; however, individuals who were not employed and those living in or near poverty had higher rates of major depressive disorders (Kessler et al., 2003). Thus, distributions of depression by measures of SES in the ECA, NCS, and NCS-R projects are less consistent than those by gender and marital status, although one still can discern an overall trend toward an inverse association.

Community mental health studies conducted in the US that employ depression screening scales, especially the widely used CES-D Scale, mirror the distributions found in the diagnostic interview surveys. Community studies consistently report higher symptoms of depression in women than men (Gove & Tudor, 1973; Kohn et al., 1998; Link & Dohrenwend, 1980; Mirowsky & Ross, 2003; Turner & Lloyd, 1999). Never-married and formerly-married individuals exhibit greater depression scores compared to the married (Mirowsky & Ross, 2003; Turner & Lloyd, 1999). And, in contrast to the findings of diagnostic interview studies, significant inverse relationships between scores on depression scales and education, income, and occupational status are repeatedly found (Kessler, 1982; Kohn et al., 1998; Link & Dohrenwend, 1980; Mirowsky & Ross, 2003; Turner & Lloyd, 1999). This is in part because depression scores are continuous measures (in contrast to dichotomous variables for diagnoses) and therefore are more likely to discern group differences (Mirowsky & Ross, 2003).

Unequal distributions of affective disorders and depressive symptoms by gender, marital status, and SES have also been established in many other nations (for comprehensive reviews see Dohrenwend et al., 1980; Kohn et al., 1998; Nolen-Hoeksema, 1990; WHO International Consortium in Psychiatric Epidemiology, 2000), leading researchers to assume that these patterns generalize across cultures and over time. However, a careful look at the range of countries in which studies have been conducted indicates that such conclusions are far too sweeping. Most have occurred in industrialized Western nations, including Canada, Great Britain, Australia, Germany, France, Spain, the Netherlands, Israel, Poland, Hungary, Sweden, and Norway, among many others, mostly European. Studies of Asian populations are exceedingly rare. Our search of English-language psychological, sociological, psychiatric, and public health journals uncovered only a handful of investigations in Asian countries from 1980 to the present that were based on representative samples. The majority of these were performed in Taiwan and South Korea as part of a 10-nation comparative epidemiological project by the Cross-National Collaborative Group, which utilized a standardized diagnostic interview based on DSM-III criteria, to facilitate comparisons across societies (Weissman et al., 1996). The ten countries were the United States, Canada, Puerto Rico, France, West Germany, Italy, New Zealand, Lebanon, Taiwan, and South Korea. A new multi-nation epidemiological effort that employs an interview based on DSM-III-R diagnostic criteria is in progress by the International Consortium of Psychiatric Epidemiology (ICPE) (Andrade et al., 2003). This study includes the US, Canada, the Czech Republic, Germany, the Netherlands, Turkey, Brazil, Chile, Mexico, and Japan. Studies that use depressive symptom scales with random samples of Asian adults are even rarer in English-language journals, although work by Lin and his colleagues in China is a notable exception (e.g., Lin & Lai, 1995).

When we examine the results of those few epidemiological and community mental health studies that have been conducted in Asian nations, we find fairly consistent distributions of depression by gender but surprisingly sparse consideration of marital status or SES differences.

The Cross-National Collaborative Group found higher lifetime and/or 12-month rates of major depressive disorder for women compared to men across all ten nations, including Taiwan and South Korea (Lepine, 2001), urban women in Seoul (Lee et al., 1990a), rural women in South Korea (Lee et al., 1990b), and women in Taiwan (Hwu, Chang, Yeh, Chang, & Yeh, 1966). Occurrences of major depressive and dysthymic disorders were also greater for women than men on Kangwha Island, South Korea (Lee, 1991) and in Hong Kong (Chen et al., 1993). Also, Lin and Lai (1995) reported that females had significantly higher mean CES-D depression scores than males in the city of Tianjin in the People's Republic of China.

In Japan, in contrast to other epidemiological studies, no significant gender differences in the prevalence of major depression or dysthymia were found (Kawakami, Shimizu, Haratani, Iwata, & Kitamura, 2004), although it is notable that the odds ratio for the 12-month prevalence of major depressive episodes in Japan was 2.5 women to 1.0 men, higher than in any other nation in the ICPE study (Andrade et al., 2003). Unlike earlier Cross-National Collaborative studies in which households were randomly sampled, Japanese participants in the ICPE survey were drawn from one prefecture's voter list, with a response rate of 57%, and diagnostic questions were self-administered rather than by interview. These design limitations may have affected results.

In general, then, as in Western nations, affective disorders and depressive symptoms seem more frequent in women than men in Asian countries, although studies are limited in number, findings are not perfectly consistent across nations, and research has been conducted in only a handful of Asian cultures and locations.

In contrast to gender differences, few studies examine marital status differences in depression in non-Western nations, and findings differ. Across all ten nations in the Cross-National Collaborative Group project, including Taiwan and South Korea, separated and divorced individuals showed higher lifetime and/or 12-month prevalence of major depression compared to married persons (Weissman et al., 1996), and formerly married persons in Taiwan were more likely than married individuals to have had major depression (Hwu et al., 1966). However, on Kangwha Island, South Korea, married persons’ rates of major depression and dysthymia were elevated compared to the unmarried, a reversal in pattern (Lee, 1991). Finally, across the ten samples in the ICPE survey, unmarried persons were more likely to have had a major depressive episode only in four of the ten countries surveyed (all Western nations—Canada, Chile, the Netherlands, and US); there were no marital status differences in depressive episodes in Japan, the only Asian country examined in the ICPE (Andrade et al., 2003).

With respect to distributions of depression by SES in Asian countries, there is even more inconsistency. The lifetime and/or 12-month prevalence of major depressive disorder was elevated among individuals with low versus high levels of education in Taiwan (Hwu et al., 1966) and an inverse relationship between education and dysthymia was reported for Kangwha Island (Lee, 1991). However, income and education were unrelated to depressive episodes in the ICPE Japanese sample (Andrade et al., 2003). And in Tianjin, China, years of education, occupational prestige, and monthly earnings were unrelated to depressive symptom scores (Lin & Lai, 1995).

Studies of the distributions of affective disorder and/or depressive symptoms by gender, marital status, and SES have occurred primarily in three Asian nations (Taiwan, South Korea, and mainland China, with Japan constituting a very recent addition), have been few in number, have reported generally consistent distributions by gender, and have produced somewhat inconsistent patterns by marital status and SES. Given the relative paucity of available studies of Asian populations, the claim that women, the unmarried, and those of lower SES have higher rates of affective disorders and/or depressive symptoms both inside and outside the US is at best overstated. Although these patterns indeed appear to apply to some Western industrialized nations (particularly the US), considerably more evidence would be needed before one could confidently extrapolate them to most non-Western societies.

Consequently, as one of many still necessary steps in this direction, in this paper we compare distributions of depressive symptoms in nationally representative samples of Japanese and American adults. We take advantage of a first-ever, large-scale national survey conducted in Japan in 1998–1999, the National Family Research of Japan ‘98 (NFRJ98). This is the first nationally representative survey of an Asian nation that includes mental health measures and is open to researchers as public use data. The NFRJ98 includes multiple items from the CES-D. For comparison purposes, we employ the second wave of the National Survey of Families and Households (NSFH), conducted in 1994 in the US. This is ideal for comparison purposes as it too has a nationally representative sample, respondents were interviewed within the same time frame as the Japanese sample, and the survey includes CES-D items comparable to those used in the Japanese study. Neither study assessed affective disorders using DSM criteria, but as seen in US studies, gender, marital status, and SES distributions of symptoms mirror epidemiological patterns of major depression. As Mirowsky and Ross (2003) have observed, depression scales have the advantage of capturing greater variations in mental health than do dichotomous indicators of disorder, increasing the likelihood of discerning sociodemographic differences.

Epidemiological and community surveys establish patterns in the distributions of depression, and these patterns in turn serve as clues to the social etiology of depression. We believe it is useful to compare Japanese and American distributions of depression because Japan differs in a number of ways from the US in family, interpersonal, and business cultures (e.g., Benedict, 1946; Brinton, 1993; Hamabata, 1990; Markus & Kitayama, 1991; Nakane, 1973) as well as in family, social network, and workplace structures (Boling, 2000; Bramlett & Mosher, 2001; Tseng et al., 2001). At the same time, Japan is the only non-Western advanced industrialized democratic nation, and, much like the US, it is highly urbanized and economically and educationally advantaged (CIA, 2002; National Center for Education Statistics, 1999; Tseng et al., 2001). We expect to find similarities in the distributions of depressive symptoms in the two countries in part because they share these latter characteristics and in part because women, the unmarried, and people with lower education and income in both nations are more often exposed to social devaluation, discrimination, and major life stressors than their sociodemographic counterparts (e.g., Boling, 2000; Krause, Jay, & Liang, 1991; Turner & Lloyd, 1999; Wright, Baxter, & Birkelund, 1995), and these are conditions that raise the risk of depression. If we find that depressive symptoms are distributed differently in the two countries, however, this will point to more specific cultural and/or structural factors that differ between them rather than to broader discrepancies in industrialization, urbanization, GDP, educational levels, or quality of life, some or all of which distinguish South Korea, Taiwan, and the People's Republic of China, for example, from the US and Japan.