|Home | About | Journals | Submit | Contact Us | Français|
G. Borges collected data in Mexico, analyzed the data, and wrote the initial draft and the final version of the article. J. Breslau participated in planning and data analyses, wrote drafts, and reviewed the final version of the article. M. Su performed statistical coding and analyses and reviewed the final version of the article. M. Miller wrote drafts and reviewed the final version of the article. M. E. Medina-Mora collected data in Mexico and reviewed the final version of the article. S. Aguilar-Gaxiola reviewed the final version of the article. All authors helped to conceptualize the study.
We examined migration to the United States as a risk factor for suicidal behavior among people of Mexican origin.
We pooled data from 2 nationally representative surveys in the United States (2001–2003; n=1284) and Mexico (2001–2002; n=5782). We used discrete time survival models to account for time-varying and time-invariant characteristics, including psychiatric disorders.
Risk for suicidal ideation was higher among Mexicans with a family member in the United States (odds ratio [OR]=1.50; 95% confidence interval [CI]=1.06, 2.11), Mexican-born immigrants who arrived in the United States at 12 years or younger (OR=1.84; 95% CI=1.09, 3.09), and US-born Mexican Americans (OR=1.56; 95% CI=1.03, 2.38) than among Mexicans with neither a history of migration to the United States nor a family member currently living there. Risk for suicide attempts was also higher among Mexicans with a family member in the United States (OR=1.68; 95% CI=1.13, 2.52) and US-born Mexican Americans (OR=1.97; 95% CI=1.06, 3.65). Selection bias caused by differential migration or differential return migration of persons at higher risk of suicidal ideation or attempt did not account for these findings.
Public health efforts should focus on the impact of Mexico–US migration on family members of migrants and on US-born Mexican Americans.
Transnational migration shapes the lives of people of Mexican origin on both sides of the Mexico–US border. The 11 million Mexican-born individuals in the United States comprise approximately 10% of the total Mexican population and approximately one third of the total foreign-born population in the United States.1 The economic and social influence of this migrant population is multiplied through the impacts that their absence,2 return migration,3 and remittances4,5 have on communities in Mexico and through their participation in American society and the growth of the US-born Mexican American population.6 To study the mental health consequences of migration in this population, data from both sides of the border are needed but have rarely been available.
Suicide-related outcomes, which include suicidal ideation, having a specific plan to commit suicide, and making a suicide attempt, are particular concerns for the immigrant population for several reasons. First, evidence from studies conducted in the United States suggests that US-born Mexican Americans have a much higher risk for psychiatric disorders than do Mexican-born immigrants.7,8 Because psychiatric disorders are risk factors for suicidality,9 we expect a similar increase in suicidality to be associated with US birth among Mexican Americans in the United States. Second, there is evidence that return migrants and family members of migrants are more likely than are others in the Mexican population to use alcohol and illicit drugs and to have a substance abuse or dependence disorder.10 Because substance use and substance use disorders are risk factors for suicidality,11 migration may be associated with higher risk for suicidality in the general Mexican population.
Third, social stressors related to migration, both for migrants themselves and for the families they leave behind, may also increase the risk for suicidality. Migrants in the United States experience a range of stressors associated with acculturation, including social isolation and discrimination.12–14 In1 study of women in Mexico whose husbands were working in the United States, the women reported significant stressors associated with the loss of support, with adjustment to new obligations, and with the potential for family disintegration.15 A study in Albania found that the absence of a family member through labor migration was associated with higher rates of acute coronary syndrome.16 Despite the economic benefits of migration, these associated social stressors may lead to increased risk for suicidality among migrants and their families.
Research from the United States suggests that among Mexican Americans, those born in the United States are at higher risk for suicide-related outcomes and completed suicide than are those born in Mexico.17–21 In addition, the younger the age at immigration—which is an indicator of socialization in the United States—the higher the risk of mental disorders22 and of suicide-related outcomes.18 To date, however, no study has been able to simultaneously address the broader pattern of the association between migration and suicidality among family members who remain in Mexico, migrants who return to Mexico, migrants who remain in the United States, and US-born Mexican Americans.
In addition, studies that focus exclusively on the United States have not been able to assess the contribution of selective migration and return migration on risk of suicide. This is a significant gap in the literature because the effect of migration on suicidality cannot be examined without data from both the sending and receiving country. In our study, we combine data from 2 national surveys, 1 from Mexico and 1 from the United States, that used the same fully structured interview. Retrospective reports by respondents in these surveys allowed us to identify the timing of immigration to the United States, psychiatric disorders, and suicidality across the entire Mexican-origin population in both countries and to address 2 possible sources of bias that may have affected prior estimates of the relationship between immigration and suicide-related outcomes.
We combined data on the Mexican population from the Mexican National Comorbidity Survey (MNCS)23 with data on the Mexican-origin population in the United States from the Collaborative Psychiatric Epidemiology Surveys (CPES). Each of these studies have been described in detail elsewhere.24,25 Briefly, both the MNCS and the CPES used the same fully structured diagnostic interview—the World Mental Health version of the Composite International Diagnostic Interview (WMH-CIDI)—to assess suicide outcomes, psychiatric disorders, and their correlates.26 All study procedures were explained to potential participants, and written informed consent was obtained in the respondent’s preferred language.
The MNCS was conducted as part of the World Health Organization’s World Mental Health Survey Initiative, a coordinated series of household surveys carried out in 28 countries around the world.27 The MNCS was based on a stratified, multistage area probability sample of household residents in Mexico aged 18 to 65 years who lived in communities with a population of at least 2500 people. Interviews were conducted from September 2001 through May 2002. The response rate was 76.6%, with 5826 respondents interviewed. Data for 44 respondents with missing information on key survey identification variables were omitted, leaving a final sample of 5782. All respondents were administered a part 1 interview that included the measures of suicidality, and a selected subsample of 2362 were asked supplemental questions on risk factors and mental disorders; after weighting, this subsample was representative of the Mexican urban population.
Data on the Mexican American population come from 2 component surveys of the CPES, the National Comorbidity Survey Replication (NCSR)28and the National Latino and Asian American Survey (NLAAS).29 The NCSR was based on a stratified multistage area probability sample of the English-speaking household population of the continental United States. The NLAAS was based on the same sampling frame as the NCSR, with special supplements to increase representation of the survey’s target ethnic groups24; both English-speaking and Spanish-speaking households were interviewed. The NCSR was conducted from 2001 through 2003 and had a 70.9% response rate; the NLAAS was conducted from 2002 through 2003 and had a 75.5% response rate for the Latino sample. The combined sample of Mexican Americans comprised 1442 respondents. Full data on nativity are available for a total of 1284 respondents, which includes the subsample of NCSR Mexican Americans who were asked supplemental questions on risk factors and the full NLAAS sample of Mexican Americans. We used integrated weights developed by CPES biostatisticians 30 that use information on the common sampling frame to properly adjust the actual NCSR sample to the US national population within racial/ethnic groups.
The WMH-CIDI contains a module that assesses 3 suicide-related outcomes: suicidal ideation (“Have you ever seriously thought about committing suicide?”), suicide plans (“Have you ever made a plan for committing suicide?”), and suicide attempts (“Have you ever attempted suicide?”). For the interview, these questions were printed in a self-administered booklet and referred to by letter because evidence suggests that participants’ reports of such potentially sensitive behaviors are higher in self-administered than in interviewer-administered surveys.31 Respondents with a history of any suicide-related outcome were asked the age at which they first experienced that outcome.
Respondents to the MNCS were asked a series of questions about migration to the United States, including motivation for migration and whether they had members of their immediate family living permanently or temporarily in the United States. This information was used to define 3 population categories: (1) labor migrants (those who had stayed in the United States for at least 3 months and gave work as their reason for visiting the United States), (2) relatives of migrants (those with immediate family members in the United States at the time the survey was administered), and (3) those with no migration background (those not in neither migration category).
In the NCSR and the NLAAS, all respondents were asked their country of birth, and those born outside of the United States were asked the age at which they first came to the United States. We divided immigrants into 2 groups: immigrants who arrived as children (at 12 years or younger) and immigrants who arrived as adolescents or adults (at 13 years or older). The decision to use this cutoff for age at migration was made based on previous research we had performed on immigration and risk for psychiatric and substance use disorders,22 where the statistical significance of the difference between early- and late-arriving immigrants was maximized with this cutoff.
The WMH-CIDI assesses a wide range of psychiatric disorders diagnosed according to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV),32 criteria. Disorders that were assessed in all the surveys were examined as potential risk factors for suicidality. The common diagnostic assessment in all surveys included measurement of mood disorders (major depressive disorder and dysthymia), anxiety disorders (panic disorder, agoraphobia without panic disorder, social phobia, generalized anxiety disorder, and posttraumatic stress disorder), impulse-control disorders (attention deficit hyperactivity disorder), and substance use disorders (alcohol abuse, drug abuse, alcohol abuse with dependence, and drug abuse with dependence). All respondents were asked the age at which they had first experienced any of those disorders. Analyses also used sociodemographic information collected in the WMH-CIDI on gender, age, educational attainment, age at first employment, and age at first marriage.
We used cross-tabulations to estimate the lifetime prevalence of suicidal ideation, plans, and attempts among Mexicans in the MNCS and Mexican Americans in the CPES. We used discrete-time survival analysis with time-varying covariates33 to study the associations between immigration and risk of suicidality, adjusting for sociodemographic variables and lifetime DSM-IV–diagnosed disorders. Discrete-time survival analysis was used instead of logistic regression to model the lifetime prevalence of suicide-related outcomes because it allowed us to characterize and account for the temporal ordering of the following: onset of mental disorders and suicide-related outcomes, time-varying sociodemographics (e.g., marital status), and most importantly, immigration status among CPES respondents. This, in turn, allowed us to compare the risk of the onset of suicide-related outcomes among Mexican-born immigrants in the CPES after their immigration to the United States. Differentiating person-years at risk for suicide-related outcomes by immigration history allowed us to test 2 main hypotheses: (1) after immigration, Mexican-born respondents in the CPES would be at greater risk for suicidality than would Mexicans who never immigrated (“immigration history”) and (2) after immigration, Mexican-born respondents in the CPES would be at lower risk for suicidality than would US-born Mexican Americans in the CPES (“nativity history”).
We also used differentiation of person-years at risk among the Mexican-born immigrants to examine 2 additional methodological questions about selection bias that have hampered prior studies of Hispanics in the United States34: (1) whether Mexican-born respondents in the CPES (i.e., those who eventually migrate) are at greater risk for suicide-related outcomes before immigration than are Mexicans who never immigrated (evidence of mental health–selective migration of persons at risk for suicidality) and (2) whether persons who return to Mexico after immigrating to the United States are at higher risk of suicidality than are Mexican-born respondents in the CPES who remain in the United States after immigration (evidence of selection bias among return migrants, the so-called “salmon bias” hypothesis).
We used a discrete-time approach instead of the more traditional continuous-time approach because the MNCS and the CPES recorded the time to event (age of onset of suicidality, immigration, and other time-dependent variables) at yearly intervals rather than on a continuous-time scale.35 Our use of survival analyses relied on retrospective age-at-onset reports to establish a temporal order between the predictors and the outcomes. This was done by treating the person-year as the unit of analysis and creating separate observational records for each year of a person’s life up to and including the year at first onset of the suicide-related outcomes being modeled. A dichotomous variable was created to distinguish the year of the outcome (coded 1) from years prior to the outcome (coded 0). Survival coefficients were converted to odds ratios (ORs) for ease of interpretation. We estimated standard errors and significance tests by the Taylor series method36 with SUDAAN version 8.01 (Research Triangle Institute, Research Triangle Park, NC) to adjust for the weighting and clustering of the data. We also report the 95%confidence intervals (CIs) of the ORs, which were adjusted for design effects on stratification and clustering and for unequal weighting of the observations.
The lifetime prevalence of suicidal ideation and attempts was higher in the CPES than in the MNCS (Table 1). Within the MNCS, the prevalence of all 3 suicide-related outcomes was higher among the relatives of migrants than among labor migrants or Mexicans with no migration background. Within the CPES, the prevalence of all 3 suicide-related outcomes was higher among the US born than among the Mexican born, and those who immigrated at younger ages showed a higher prevalence than did those who immigrated later.
We examined immigration history in relation to suicide-related outcomes across 6 Mexican-origin groups in Mexico and the United States, with controls for psychiatric disorders and sociodemographic characteristics (Table 2). People with no migration background (i.e., neither personal history of migration nor a family member working in the United States) were the reference group for each comparison. Compared with this group, risk of suicidal ideation was higher among people in the Mexican population with a family member in the United States, people in the United States who were born in Mexico and migrated to the United States prior to age 13 years, and US-born Mexican Americans. Risk of suicide attempt was higher among people in the Mexican population with a family member in the United States and US-born Mexican Americans. There were no differences in risk of suicide-related outcomes between the reference group, return migrants, and people interviewed in the United States who migrated at 13 years or older.
We also examined differences in risk of suicide-related outcomes across the 3 Mexican-origin groups in the United States (i.e., by nativity history). There were no differences in risk of suicide-related outcomes between Mexican-born respondents who immigrated to the United States prior to age 13 and US-born respondents. The risk of suicidal ideation was significantly lower among immigrants who arrived in the United States at age 13 or older than among those who arrived at earlier ages or those born in the United States.
Table 3 presents differences in risk of suicide-related outcomes in terms of selective migration (i.e., between nonmigrants and migrants prior to migration) and selective return migration, or “salmon bias” (i.e., between migrants who remain in the United States and those who return to Mexico). No evidence of either type of bias was found. With respect to selective migration to the United States, individuals who were born in Mexico and interviewed in the CPES were not at significantly lower risk of suicide-related outcomes during the years prior to migration to the United States than were Mexicans with no migration background. With respect to the “salmon bias” hypothesis, return migrants did not have greater risk of suicide-related outcomes than did migrants who remained in the United States, and migrants who arrived in the United States at 12 years or younger were at higher risk of suicide attempts than were return migrants (OR=2.78; 95%CI=1.02, 7.59).
To our knowledge, ours is the first study to examine suicide-related outcomes in a transnational migrant population using population-based data collected with the same survey instrument in both the migrant-sending and migrant-receiving countries. Suicide-related outcomes were associated with migration in both the sending and receiving countries, and these associations were not explained by basic sociodemographic differences, the prevalence of psychiatric disorders, or migrant selection bias. Compared with Mexicans with no personal history of migration and no family member in the United States, Mexicans with an immediate family member in the United States, Mexican-born immigrants to the United States who arrived in the United States prior to age13 years, and US-born Mexican Americans were at higher risk of suicidal ideation. Mexicans with an immediate family member in the United States and US-born Mexican Americans were also at higher risk of making a suicide attempt. These findings are consistent with previous reports that the prevalence of suicide-related outcomes is higher in the United States than in Mexico37 and provide for the first time a description of how risk for suicide-related outcomes is distributed with respect to migration back and forth across the Mexico–US border.
Our finding that the families of immigrants in Mexico were at greater risk of suicidal ideation and attempts than Mexican families without an immigrant background has not previously been reported, but it is consistent with prior reports of an increase in risk of alcohol and substance use disorders among families in the MNCS with an immigrant background. 10 This new finding is of importance because, according to the MNCS, approximately 48% of the Mexican population has relatives working in the United States. Immigration is known to disrupt family ties and may increase feelings of loneliness and insecurity among those left behind,2,15 thus imposing additional, unaccustomed, and stressful demands on family members. Our finding that the risk of suicide-related outcomes remained high among families in Mexico with an immediate family member working in the United States, even after we controlled for prior psychiatric disorders, suggests that suicide-related outcomes may be an important manifestation either of the stressors associated with disruption of family dynamics or of yet-to-be-identified aspects of having a family member residing in a foreign country. Little research on the health impact faced by the relatives of migrants is available, but it has been proposed that keeping in contact with the migrant relative and receiving remittances may buffer the negative effects of the migrant’s absence.16
To our knowledge, ours is also the first study to examine the impact of migration on suicidality in the United States by comparing post-migration risk among Mexican immigrants with that of the general population of Mexico. If the cause of increased risk for suicide-related outcomes is related to psychological distress associated with leaving the culture to which one is accustomed and integrating into a new and sometimes hostile environment in the receiving country, we would expect the increase in risk to be largest among older adolescents and adult migrants. Contrary to this expectation, adult migrants were not at higher risk of suicide-related outcomes than was the general Mexican population. Rather, an increase in risk of suicide-related outcomes occurred exclusively among migrants who arrived in the United States as children (i.e., prior to age 13 years). This finding is consistent with a new report on a large sample of Mexican American adolescents that found no differences in suicidality between immigrant and US-born adolescents.38 Taken together, these 2 findings suggest that stress because of acculturation may have less of an impact on suicidality than previously expected and that early socialization or cultural assimilation in the US milieu (in which baseline risk of suicide-related outcomes is also high19,37) may play a larger role than previously expected.18
A persistent question in the literature on Latin health and migration raised by studies conducted in the United States has been whether the lower level of risk observed among Mexican immigrants compared with US-born Mexican Americans can be explained by the selective migration of relatively healthy individuals or by the selective return migration of sick migrants to their country of origin.39 Our study, which we believe to be the first to directly examine this possibility in the context of suicidality, found that there were no differences in suicidality predating migration from Mexico to the United States and that return migrants were no more suicidal than were migrants currently in the United States. Thus, differences between foreign-born and US-born Mexican Americans cannot be attributed to selective migration or selective return.
Our findings must be evaluated in the context of several limitations. First, these household surveys excluded homeless and institutionalized people, populations known to have a high prevalence of suicidal behavior.40 Second, validity and reliability data on the measures of suicidal ideation, plans, and attempts were not obtained. Third, these analyses used data on retrospectively reported ages of onset that are subject to recall error, which probably means that the results we report were conservative. Retrospective data on the exact ages at immigration and return migration are lacking in the MNCS, which limited our ability to more adequately model the data from this survey.
Fourth, the Mexican sampling frame did not include the most rural parts of the country. To the extent that most Mexican immigrants to the United States come from areas not represented by our sample, our ability to control for premigration factors was limited. Fifth, despite using the same diagnostic interview, the 2 surveys differed in several ways, including language of interview and auspices of the survey. We cannot rule out the possibility that these methodological differences contributed to the observed differences in prevalence estimates in the CPES compared with the MNCS. Finally, the results are not generalizable to other immigrant groups that may have different baseline risk of suicidality, different histories of immigration, and a border situation different from that of Mexicans and Mexican Americans.
Mexican migration to the United States increased suicidal ideation and suicide attempts among the families of migrants in Mexico and among US-born citizens of Mexican origin. Given the large size of this migration, the public health consequences of this effect are likely to be profound for both countries. Further research on the migration patterns of suicidality that integrate data from sending and receiving countries are needed. Public health efforts should focus on providing buffers and psychosocial support to ameliorate the impact of Mexico–US migration on family members of migrants and on US-born Mexican Americans.
Support for survey data collection came from the National Institute of Psychiatry Ramon de la Fuente (grant INPRFM-DIES 4280) and Consejo Nacional de Ciencia y Tecnologia (grant CONACyT-G30544-H), with supplemental support from the Pan American Health Organization. Support also came from grants from Consejo Nacional de Ciencia y Tecnologia and Fundacion Mexico en Harvard to G. B. for a scholar affiliation at Harvard Injury Control Research Center, Harvard School of Public Health, and from a grant from the National Institute of Mental Health (MH K01-66057).
We thank the Inter-University Consortium for Social and Political Research for their work in making data from the National Comorbidity Survey Replication and the National Latino and Asian American Survey available for public use.
Human Participant Protection
The institutional review board of the National Institute of Psychiatry (Mexico City) approved this project, and the institutional review boards of the Cambridge Health Alliance, the University of Washington, and the University of Michigan approved all recruitment, consent, and interviewing procedures for the National Latino and Asian-American Survey. The recruitment, consent, and field procedures in the National Comorbidity Survey Replication were approved by the human subjects committees of both Harvard Medical School and the University of Michigan.