Ethnic and Sex Differences in Suicide Rates Relative to Major Depression in the United States

The data were gathered in the Epidemiologic Catchment Area (ECA) study (1980–1984) (16) and the Hispanic Health and Nutrition Epidemiologic Survey (1982–1984) (3) and have been previously reported (3–5). Both surveys used the National Institute of Mental Health Diagnostic Interview Schedule (DIS) (17) to establish DSM-III diagnoses. We used the 1-year prevalence rates of major depression for blacks, whites, and Los Angeles Hispanics for persons aged 20 to 74 years from the ECA study, which was conducted in five sites in the United States. That study used a two-stage, stratified, clustered sample of respondents both in the community and in institutions. Only community rates are reported here. We also used depression rates for Puerto Ricans in mainland United States, Mexican Americans, and Cuban Americans from the Hispanic Health and Nutrition Epidemiologic Survey data, conducted by the National Center for Health Statistics. This study surveyed the three largest Hispanic groups in the geographic areas in which they were most concentrated. The Hispanic Health and Nutrition Epidemiologic Survey used a complex, multistage, stratified, clustered sample of respondents aged 6 months to 74 years (only subjects aged 20 to 74 years were included in this analysis).

The responses to the DIS were subjected to a computer algorithm that generated DSM-III diagnoses for major depression in the last year. Subjects were categorized as depressed if they gave affirmative responses to questions about the presence of current major depression (excluding bereavement) or responded that an episode of major depression had ended within the last year.

We used the U.S. suicide rates reported by Desenclos and Hahn (18), which are based on the Centers for Disease Control National Center for Health Statistics Mortality Detail Data Tapes 1986–1988. This time period is near the periods covered by the depression data but does not overlap them. Age-adjusted death rates per 100,000 person years (all ages) were calculated on the basis of the 1980 census and 1987 postcensus age- and sex-specific population estimates. Eighteen states and the District of Columbia report mortality by Hispanic origin. National population estimates for Hispanics, non-Hispanic whites, and non-Hispanic blacks were extrapolated from these geographic areas by taking into account the proportion of the U.S. total that these areas represented for non-Hispanic blacks and whites, Hispanics, and specific Hispanic subgroups.

The 1-year prevalence rates of major depression for each subgroup were standardized to the sex and age distribution of the 1987 midyear population of persons of Hispanic origin in the United States (19). We standardized the rates to the age and sex distribution of the 1987 midyear Hispanic population so that we could compare the 1-year prevalence rates of major depression across ethnic groups in relation to the suicide rates for these groups, which were based on 1987 U.S. population estimates. Standardization was done according to the methods of Breslow and Day (20). Weights were calculated in accordance with the methods of Pothoff et al. (21). The rates and proportions were calculated to provide estimates as though the ECA study and Hispanic Health and Nutrition Epidemiologic Survey had the same age and sex distributions as the 1987 midyear Hispanic population. Standard errors were derived as described by Fleiss (22). All analyses were done by using the frequency procedure (PROC FREQ) in SAS version 6.09 (SAS Institute, Cary, N.C.) (23). We compared 1-year depression rates across ethnic groups by using an overall chi-square test (24). Then we compared the rate for whites with the rate for each of the other ethnic groups, using the chi-square test. We also compared differences between the rates of depression in women and men in the five ethnic groups by using the Breslow-Day test for homogeneity of ratios (20).

The annual suicide rates (reported per 100,000 persons) were divided by 1,000 to represent suicides per 100 persons in each group. This rate was then divided by the annual prevalence rate of depression per 100 persons in each group. This provides an estimate of the number of suicides per year per case of major depression per year in each group.

To determine whether effects of sociodemographic variables on differences in depression rates among ethnic groups might account for differences in suicide, we calculated the rates of major depression for subjects with three sociodemographic characteristics considered to be risk factors for suicide: low income, not working for pay, and disrupted marital status. A logistic regression analysis (25) was conducted to examine the effects of these three sociodemographic variables on the depression rate for each ethnic group.

Although Mexican Americans had a modestly lower 1-year prevalence of depression than whites, Cuban Americans, blacks, and whites had similar, although not identical, rates. In contrast, the 1-year prevalence of major depression for Puerto Ricans in the United States was about double that for whites. Potter et al. (5) suggested that since the rates of major depression in Puerto Rico have been shown to be the same as in the general U.S. population (27), perhaps the migratory process selectively involves Puerto Ricans with higher levels of psychological distress. Alternatively, Vera et al. (28) reported that Puerto Ricans in the New York area, most of whom are impoverished, have the same depression rate as impoverished Puerto Ricans on the island. This is supported by our finding that Puerto Ricans with incomes of less than $5,000 per year had significantly higher depression rates than their counterparts. We also found that Puerto Ricans as a whole had a higher depression rate than whites. It has been suggested that Puerto Ricans in New York have a different response style, leading them to report more symptoms (29). It is unclear why island Puerto Ricans as a whole would not show this tendency when asked about psychiatric symptoms (27). Furthermore, that impoverished Puerto Ricans both in New York and in Puerto Rico would show similar rates of major depression (28) that differ from the rate for the general population of Puerto Rico (27) suggests that differences in depression rates between New York Puerto Ricans and whites or other Hispanic groups are related to issues other than national origin or identification, possibly socioeconomic.

Studies from a number of countries have indicated mood disorders in approximately 60% of suicide victims: Finland, 59% (10); Hungary, 58% (30); Australia, 55% (31); England, 70% (32); Sweden, 59% (33). These studies have been conducted in largely white populations. However, the apparent stability of the percentage of suicide victims with mood disorders before death is striking, suggesting a close relationship between depression rates and suicide rates. Consideration of ethnicity and gender demonstrates a more complex picture.

The difference in the relationship between suicide rate and 1-year prevalence of major depression among ethnic groups in the United States is remarkable. Possible explanations include comorbid conditions associated with higher suicide rates, such as substance abuse, alcoholism, or borderline personality disorder, cultural and child-rearing environments, adequacy of antidepressant treatment, genetics, and neurobiology. For these groups, there are few data about comorbidity or other factors, such as history of physical abuse, that have significant influences on suicidal behavior in the context of major depression (14).

We found that Mexican Americans and Puerto Ricans of both genders and Cuban American females had lower suicide rates than expected given their depression rates. Black females also seemed to have a lower suicide rate given their depression rate, although the same was not true for black males. The finding that the suicide rate was almost as high for black males as for white males, proportional to their depression rates, was unexpected. Moreover, a study showing a lower suicide rate among blacks did not account for the rate of major depression (34). Explanations offered included the role of black churches, a bias in reporting, differences in social integration, and differences in emotional expressiveness (34, 35). Our results suggest that protective factors may be restricted to black females only. Further study is needed to identify these protective factors.

Our finding that, compared to non-Hispanic whites, Puerto Ricans and Mexican Americans of both genders and Cuban American females, but not Cuban American males, have a lower suicide rate relative to the rate of major depression is also unexplained. Cuban Americans, Puerto Ricans, Mexican Americans, blacks, and non-Hispanic whites are different ethnically. Cuban Americans are more frequently white, whereas Mexican Americans and Puerto Ricans tend to be mixed racially, with Puerto Ricans having a significant black contribution and Mexican Americans having a significant indigenous contribution. Moreover, cultural differences among these groups should not be underestimated. The same historical and geographic factors that contribute to differences in racial composition also contribute to cultural heterogeneity. In addition, reasons for immigration differ because of historical and political circumstances. For example, many Cuban Americans have immigrated to the United States for political rather than economic reasons. Differences in suicide rates relative to major depression warrant study so that factors that protect against suicide in major depression may be uncovered.

Reported suicide rates are lower in Cuba, Puerto Rico, and Mexico than in the United States (36). Furthermore, the suicide rate for Hispanics in the United States is almost one-half that for non-Hispanic whites (18, 37, 38). Although the relationship between suicide rate and major depression has not been addressed in these studies, Hoppe and Martin (37) offered the concepts of familism and fatalism as contributors to lower suicide rates among Hispanics. Familism, an emphasis on close relationships with extended kinship, may offer protection against stress. Fatalism, the expectation of adversity, may be an adaptive stance in the setting of chronic stress. Both may serve to lower suicide risk in the context of depression.

Variables linked to suicide include unemployment (15), marital isolation (14), and socioeconomic status (14). In our study, the differences in the rates of major depression among members of the five ethnic groups in those socioeconomic categories did not follow the pattern for suicide rates. In fact, the rate of major depression appeared to be influenced by low income only among Puerto Rican men and women. Yet Puerto Ricans were among the most protected from suicide given their high depression rate.

One of the study’s limitations is that the data about depression come from two surveys. Apart from specific differences in sampling techniques, the ECA study’s main purpose was to assess the prevalence of psychiatric disorders in five sites in the United States, by constructing samples similar to the U.S. population as a whole. The ECA study did not distinguish among Hispanic subgroups. The Hispanic Health and Nutrition Epidemiologic Survey, in contrast, sought information regarding health and nutrition, including but not restricted to psychiatric disorders. This survey targeted three Hispanic subgroups. However, between one-third and one-half of the respondents in the Hispanic Health and Nutrition Epidemiologic Survey did not receive the depression assessment. Alcohol use was the only other psychiatric variable that was assessed systematically. Also, this survey did not cover undocumented immigrants. In addition, both the ECA study and the Hispanic Health and Nutrition Epidemiologic Survey used the DIS administered by lay interviewers, the results of which have been shown to differ from results when clinicians interview the same subjects only a few weeks later (23).

Also, depression is only one factor leading to suicide. This study did not address substance abuse, schizophrenia, and psychosocial determinants. Also, there are no data with which to investigate factors such as familism or the role of religion in lowering suicide rates relative to depression across ethnic groups.

Another limitation is the assessment of suicide rates. Because only some states in the United States report mortality by Hispanic origin, the suicide rate for each Hispanic subgroup was extrapolated from the rates for states that do report whether or not the decedent was Hispanic. Furthermore, suicide rates may be obscured by accidents or violence. In addition, the suicide rates were from 1986 to 1988. The ECA data were collected between 1980 and 1984, and the Hispanic Health and Nutrition Epidemiologic Survey data were collected between 1982 and 1984. Thus, the 1-year prevalence rates of depression are from 3 to 7 years before the years for which suicide rates are reported. The depression rates for both samples were adjusted to the postcensus estimates for the population for 1987 to address this issue. However, considerations such as the economic environment, which changed over the course of the decade, and its possible differential effects on suicide rates and prevalence of depression could not be addressed. In addition, the suicide rates reported cover ages under 20 and over 74, which are not covered in the reports of depression prevalence.

A further limitation is the use of a general categorization for whites and blacks. Blacks and whites in the United States are of heterogeneous cultural backgrounds and countries of origin. Depression rates that appear similar for Hispanics when taken together, but not when Hispanics are considered as subgroups, might also vary if blacks or whites were to be considered as subgroups categorized by country of origin.

Mexican Americans and Puerto Ricans as a whole, and black and Cuban American females, have lower 1-year suicide rates relative to 1-year prevalence rates of major depression than do whites. Males in all ethnic groups had higher relative suicide rates than females in their respective ethnic groups. The robustness of the finding suggests the presence of as-yet undetermined biological, genetic, or cultural protective factors for suicide. The identification of these factors may guide suicide prevention efforts.