|Home | About | Journals | Submit | Contact Us | Français|
We compared risk for several medical illnesses between immigrant and US-born older Mexican Americans to determine the relationship between functional health and years of US residency among immigrants.
Cross-sectional, multistage probability sample data for 3050 Mexican Americans aged 65 years or older from 5 US southwestern states were analyzed. Self-rated health, medical illnesses, and functional measures were examined in multivariate regression models that included nativity and years of US residency as key predictors.
Self-rated health and medical illnesses of immigrant and US-born groups did not differ significantly. Immigrants with longer US residency had significantly higher cognitive functioning scores and fewer problems with functional activities after adjustment for predisposing and medical need factors.
Among older Mexican Americans, immigrant health advantages over their US-born counterparts were not apparent. Immigrants had better health functioning with longer US residency that may derive from greater socioeconomic resources. Our findings suggest that the negative acculturation–health relationship found among younger immigrant adults may become a positive relationship in later life.
More than 30 years ago, Teller and Clyburn reported more favorable birth outcomes in Bexar County, Texas, for Spanish-surnamed residents than for non-Latino White and African American residents.1 Despite the disadvantaged economic and social position of many Latinos, additional reports appeared of unexpected favorable birth and longevity outcomes for Latinos (primarily Mexican Americans) compared with other ethnic groups.2–5 These findings were considered paradoxical, since they ran contrary to the negative socioeconomic health gradient documented in the public health literature.6,7 The Latino health paradox contends that despite experiencing disproportionate exposure to risk factors for excess morbidity and mortality associated with low socioeconomic position, Latinos, primarily Mexican American immigrants, generally have more favorable health outcomes than Mexican Americans born in the United States, most other minorities, and nonminorities.8 The longer these healthy immigrants resided in the United States and acculturated, the more likely they were to report deterioration in health status indicators compared with recent immigrants.9
The mental health researchers Rogler et al. referred to “acculturation as an exogenous force shaping the conditions for the rise to psychological distress.”10(p588) This phenomenon was later coined the “acculturative stress” hypothesis and widely used in mental health research of the Latino population. However, acculturative stress models have important limitations in health research because they confound the effects of cultural change on health behaviors with pathology. In addition, they are temporally unspecific and thus inconsistent with developmental models of health. We offer an alternative, the “acculturation–health” hypothesis, to emphasize that health outcomes are instead conditional on multiple life-course contingencies that vary in importance as determinants over one’s life span. The temporal relationships between acculturation and health can range from negative to positive during the lifespan of an individual. Although the negative effects of acculturation are commonly described in the literature, positive outcomes are observed as well and also require explanations.11–15 Acculturative stress assumes a cumulative, linear, dose–response association of acculturation with health. An acculturation–health hypothesis accounts for transactions between endogenous (individual-level) and exogenous (external) factors that affect health differentially over the life course. These factors are systematically influenced by acculturation across multiple domains of life activity.
The proposed acculturation–health model assumes that there are multiple points in the life course that are critical for improving health and lowering the risk of weathering effects seen in aging among minority groups.16 Carefully focused research could lead to timely and effective interventions that improve health outcomes across the life course. The assumption that a constant acculturation–health relationship is generally negative and leads to declines in health among Mexican Americans is limiting because it is overly deterministic and restricts opportunities for identifying determinants of long-range outcomes and life-course stages when they are most consequential. Although we can assume that living in a new society introduces behavioral, social, and environmental changes that may negatively influence health, there may be benefits as well, such as acquisition of new occupational skills, greater exposure to public health information, and use of preventive medicine. Nevertheless, it is unclear how these changes are expressed in the long run in the health of older Mexican Americans.
Mexican Americans represent more than two-thirds of US Latinos and, perhaps accordingly, most previous studies have focused on the relationship between Mexican American acculturation (or various proxies of acculturation, such as language preference and years of US residency) and health.17 For younger Mexican Americans, most studies have reported negative relationships between acculturation and health11—for example, for birth outcomes18,19 and perinatal health behaviors,20,21 general health behaviors (e.g., nutrition and physical exercise),22 vascular disease risks (e.g., diabetes),23,24 and substance abuse and mental disorders.25,26 Several reports have shown positive associations between acculturation and higher use of preventive medical services, which may relate to the higher economic position of more acculturated and US-born Mexican Americans.27–30 Among older adults, the prevalence of depression and dementia are reportedly lower among acculturated Mexican Americans.12,13 Furthermore, there is evidence that greater social assimilation, as evidenced by English-language acquisition and more years of education, are associated with lower disability rates and higher self-esteem.14,15,31
These general findings have led scientists to hypothesize that immigrants are somehow healthier than US-born Mexican Americans and non-Latino Whites and that with longer US residency, acculturation erodes immigrants’ health.9,32 The negative aspects of the acculturation–health relationship dominate current thinking about Mexican American health, especially regarding acculturation-related changes in risk behaviors such as diet, exercise, and substance abuse.33 However, if the association between acculturation and health is conditional and predicated on various endogenous and exogenous life-course factors, recent work among older adults may better inform us about the long-term acculturation–health relationship.
Our first aim was to examine whether the “healthy immigrant” phenomenon persists among older Mexican Americans by comparing the rates of several medical conditions between US-born and immigrant groups. Second, we sought to determine if functional health was negatively associated with longer exposure to the host country within a representative sample of immigrant Mexican Americans. On the basis of previous work,12–14,31 we expected that the health status of US-born and immigrant older Mexican Americans would not differ. Third, we sought to examine if longer US residency would be associated with better functional health among older Mexican American immigrants. We expected that increased wealth, access to health-care and services would facilitate or mediate better health functioning.6,34,35 Finally, in examining the health of older Mexican Americans, we aimed to compare the acculturative stress hypothesis and the acculturation–health hypothesis for a better understanding of their utility for the study of health.
We used the public-use baseline data set of the Hispanic Established Populations for the Epidemiologic Studies of the Elderly (H-EPESE). The H-EPESE is a multistage, stratified, probability sample that was modeled after other EPESE studies of non-Latino Whites and African Americans. The H-EPESE methodology presented here was extracted from study documentation available through the Interuniversity Consortium for Political and Social Research.
H-EPESE participants were taken from the population of noninstitutionalized Mexican Americans aged 65 years or older at baseline (September 1993–June 1994) who were residing in the 5 US southwestern states of Arizona, California, Colorado, New Mexico, and Texas. Sampling weights, adjusted for unequal probability of selection, nonresponse, and poststratification, were calculated to ensure that sample-based estimates would reflect census estimates.
To address our first aim, we examined self-rated health and several chronic medical conditions. The self-rated health measure is widely accepted to be predictive of morbidity and mortality.28 The 5-point, Likert-type response was dichotomized (excellent to fair=0; poor=1) to identify respondents with the highest level of disability and to avoid response biases previously observed among Latinos.36 The health outcome variables of interest included self-reported physician diagnoses of diabetes, hypertension, heart disease, and stroke.
To address our second aim, we examined 2 functional health outcome measures. For neuropsychiatric function, we examined raw continuous total scores on the Folstein Mini-Mental State (MMS; range=0–30) examination and the Center for Epidemiologic Studies-Depression (CES-D; range=0–60) scale.37,38 To increase the reliability of the MMS information, we excluded respondents (n=18) with very low MMS total scores (<10 points), who were likely to be cognitively impaired.39 For general health self-management functioning, we examined data on respondents’ self-reported requirements of assistance in activities of daily living and instrumental activities of daily living.40 Activities of daily living are basic tasks (e.g., grooming and toileting), whereas instrumental activities of daily living are more complex (e.g., telephoning and shopping). These functional measures are widely used clinically and in research for evaluating older adults’ capacity for independent living.
To address our first aim, the key predictor variable was nativity (i.e., US born versus foreign born). With regard to our second aim, we restricted the sample to immigrants. The principal predictor variable was the self-reported number of years of US residency, which was used as a proxy measure for acculturation. In addition to the main predictor, we examined factors based on the Andersen model of health care utilization that could potentially explain any associations found between functional health and years in the United States (or possible mediators of the hypothesized relationships).34 For these Andersen model factors, we chose predisposing demographic factors (age and gender), enabling factors (education, home ownership as an indicator of wealth, and health insurance coverage), and health need factors (the aforementioned medical conditions).
Analyses of complex sample survey data were completed with Stata software version 10.1 (StataCorp LP, College Station, Texas). All statistical estimates were weighted. Standard errors were also adjusted for sample design.41 We performed univariate analyses to calculate sample estimates describing nativity and health outcomes. With regard to our first aim, we then fitted multivariate logistic regression models to the data to investigate the relationships between the 4 health outcomes and nativity with age, gender, and years of education as covariates. Odds ratios expressing the relative influence of nativity on the odds of having the health outcomes were calculated on the basis of the multivariate logistic regression models, along with design-based 95% confidence intervals (CI) for the estimated odds ratios.
To address our second aim, a linear regression model assuming normally distributed errors was found to fit the MMS data after 18 observations with low values dropped (i.e., MMS total <10). Next, negative binomial regression was used to model the log-transformed counts of activities of daily living and instrumental activities of daily living because of evidence of overdispersion for these dependent variables. Covariates for these models included the Andersen model factors.
Because the enabling factors were expected to mediate associations between years of US residency and the outcomes of interest, Baron and Kenny’s steps were used to test for mediation.42 We first examined associations between years of US residency and the outcomes without controlling for the enabling factors. If significant associations were found, the association between US residency and each enabling factor was next examined. If these associations were also significant, we reassessed associations between years of US residency and the outcomes, controlling for the enabling factors; the magnitude of the decrease in the relationship of acculturation with the outcomes was determined. Complete mediation was evident if the association of acculturation with the outcome of interest was no longer significant after including enabling factors in the model and finding that those enabling factors had a significant association with the outcome. Partial mediation was evident if both acculturation and the enabling factors were significantly associated with the outcome, but with a smaller association.
Table 1 shows selected characteristics of the foreign-born respondents, the US-born respondents, and the total H-EPESE sample. Compared with the US-born group, foreign-born Mexican Americans were significantly older (t=5.19; P <.001), had significantly fewer years of education (t=−8.96; P<.001), were significantly less likely to own their home (design-based F1,208 =76.56; P <.001), and were significantly less likely to have health insurance coverage (design-based F1,208 =14.97; P <.001). In general, US-born Mexican Americans had higher levels of economic enabling factors than their foreign-born counterparts.
In the multivariate logistic regression models, foreign- and US-born Mexican Americans did not significantly differ in self-rated health (odds ratio [OR]=1.03; 95% CI=0.78, 1.35), diabetes (OR=1.14; 95% CI=0.90, 1.44), hypertension (OR=0.97; 95% CI=0.76, 1.24), heart attack (OR=1.31; 95% CI=0.93, 1.84), or stroke (OR=1.15; 95% CI=0.75, 1.76) after adjustment for the relationships of age, education, and gender. These results indicate that health did not significantly differ by nativity.
Functional outcomes were grouped into 2 domains: (1) neuropsychiatric functioning (MMS and CES-D) and (2) self-management functioning (activities of daily living and instrumental activities of daily living). Two multivariate models were used to estimate the associations between the outcomes of interest and years of US residency. In the first models, age and gender were included as predisposing factors along with health need factors. The second models were designed to determine whether any associations between the functional health outcomes and years of US residency were mediated by socioeconomic enabling factors (Tables 2 and and33).
Multivariate logistic regression results (not shown) indicated that longer US residency was associated with significantly higher education and higher odds of home ownership and of having health insurance coverage. Therefore, the next models tested for possible mediation between enabling factors (i.e., education, home ownership, and health insurance coverage), US residency, and the functional outcomes. In addition, path analyses (available on request from the authors) were conducted with MPLUS version 3.13 (MPLUS Statistical Software, Los Angeles, California) to account for complex survey design variables and to directly test the significance of total, direct, and indirect effects between our variables of interest.43 Although the path mediation results yielded the same conclusions, we chose to present the multivariate regression results for ease of interpretation.
The first multivariate linear regression model results indicated that with more years of US residency, immigrants had significantly higher MMS scores after adjustment for predisposing and health need factors. When socioeconomic enabling factors were included in model 2, the magnitude of the association between US residency and MMS scores decreased by 55.8%, although the association remained significant (Table 2). This suggests that socioeconomic enabling factors partially mediated the association between MMS scores and years of US residency. Excluding observations with low MMS scores did not substantively affect this association.
Longer US residency was associated with fewer depressive symptoms (lower CES-D scores) in models adjusted for demographic and medical conditions (not shown); however, these associations did not reach statistically significant levels.
For immigrants, negative binomial regression results indicated that fewer activities of daily living problems were associated with longer US residency. Because the association was not significant, however, no further analyses were conducted for activities of daily living problems. Overall, few older Mexican Americans reported activities of daily living problems.
For instrumental activities of daily living, longer US residency was associated with significantly fewer problems with higher-order self-management behaviors in model 1. In model 2, the association between instrumental activities of daily living and years of US residency was markedly reduced (36.9%) and became only marginally significant when enabling factors were included (Table 3). These results suggest that the enabling factors might act as mediators of the relationship between instrumental activities of daily living and years of US residency.
Two main findings emerged from this study. Consistent with previous work,32 immigrant and nonimmigrant older Mexican American adults reported comparable self-rated general health and similar histories of medical problems. Second, we found that among older Mexican American immigrants, longer US residency was associated with markedly better functional health. Furthermore, wealth and greater access to health care partially explained the relationship between acculturation and health among older Mexican Americans. Together, these findings indicate that among Mexican Americans, any health advantages that immigrants may have held over nonimmigrants in younger adulthood are no longer apparent in later years.
Among younger Mexican Americans, most studies have shown that immigrants have health advantages over their US-born Mexican American counterparts.44–47 Over time, the gap narrows between healthy immigrants and their US-born Mexican American counterparts.47 As Antecol and Bedard found among younger adults, immigrant health advantages completely converged with nonimmigrants and non-Latino Whites after 15 years in the United States.9 Among older US-born Mexican Americans, we found no health advantages for immigrants, which is consistent with another report.31 Among immigrants, however, we found that longer US residency was positively associated with improved functional health. Convergence between immigrant and nonimmigrant groups may have occurred in middle adulthood and a negative linear trajectory for immigrants apparently was not sustained into older adulthood, perhaps abetted by differential selection effects in surviving nativity groups. Indeed, based on this and previous work, the broader question remaining regarding the acculturative stress hypothesis is whether a continuously negative relationship between acculturation and health is sustainable at all. Together, our 2 sets of findings may tell us about 2 acculturation-health trajectories at distinct periods of life, yet using only the 2 time points does little to inform us about the intervening years.
Our findings suggest that the relationship between acculturation and health has contingencies and may be nonlinear. As Rogler et al. speculated regarding the relationship between acculturation and mental health, the association may be curvilinear.10 During the early and middle adulthood of immigrants, the acculturation–health relationship has been shown to have a negative trajectory; however, the lower limits of this downward trend toward worsening health status have not been well-studied. Our findings suggest that the long-range course is positive among older Mexican Americans. In the early postimmigration years, negative effects of acculturation on health (e.g., stress from discrimination and economic barriers, life in unhealthy neighborhoods, and unhealthy behavioral changes) may outweigh any observable positive aspects (e.g., availability of preventative medical services).13,14 It may be that the observed negative relationship between acculturation and health changes over time to the positive position we have reported here. Economic advantages accumulated over time may partially explain or mediate the positive acculturation–health relationship we observed among older adults.
The findings reported here are unique and contrast with most previous reports, which have shown acculturation-related declines in the health of immigrant Mexican Americans.11,48 An important difference between this and previous studies was our focus on the health and functional status of older Mexican Americans, the majority of whom immigrated in early adulthood and had resided in the United States for more than 40 years. Our findings are consistent with previous work that showed disability rates between immigrants and nonimmigrants and higher prevalence estimates of depression and dementia among older Mexican Americans with less acculturation.12,13,49
With increasing age, the likelihood of chronic illness and disability increases, as does the need for health and social services.50,51 Although disability rates among older adults have declined recently, sharp gradients in disability by socioeconomic status continue.52–54 Mexican Americans are not only disadvantaged in terms of education, income, and wealth but are also the least likely of all ethnic minorities to have health insurance coverage over the lifespan.29,30,55 Although nearly all US adults aged 65 years or older have some form of health insurance coverage, 14.6% of older Mexican American immigrants—more than 1 in 6—do not.56 Our findings suggest that older Mexican immigrants who have accumulated more economic assets may be better able to adjust to changing health and health care needs than their counterparts with fewer years in the United States. Because of their longer stay in the host country, these wealthier immigrants may also have other assets (e.g., better knowledge and access to resources) available to them as long-term US residents.
Although our study used multiple measures of medical conditions and functional health to provide convergent findings, the MMS is susceptible to test biases. Our previous work with neuropsychological tests has shown that potential biases, such as age and education, were likely to have been limited by inclusion of these factors into our analyses.39,57 In addition, although the H-EPESE sample size allowed ample statistical power, one limitation was our reliance on years of US residency as a proxy measure of acculturation, which provides no information on variations in cultural and behavioral patterns. For example, for a variety of endogenous and exogenous reasons, many older Mexican Americans residing in the United States for decades may retain traditional language and other practices, whereas others may have experienced accelerated deculturation. Using multiple reliable and statistically independent acculturation measures would increase precision in understanding the relationship between acculturation and health.
The H-EPESE cohort may also be unique since many of the Mexican immigrants began entering the United States around 1925, through the years of the Braceros program at midcentury and into later decades. Although the demand for low-cost labor in the United States has remained steady, the attitude of tolerance toward Mexican Americans, particularly immigrants, has not, and their experiences may be similar and yet quite different from their contemporary counterparts. Additionally, it may be that the “acculturation intercept” at entry into the United States may be higher for modern immigrants through the new media source availability (e.g., Internet) and the oft-cited—but seldom well-defined—process of globalization. Finally, we were unable to determine whether the positive acculturation–health relationships we found could be explained by selection and return migration.4
We have interpreted our findings in the context of comparing the acculturative stress hypothesis and the acculturation–health hypotheses, and their ramifications as models for health research. The acculturation–health hypothesis has greater flexibility for understanding complex life-course acculturation effects. Acculturative stress is overdetermined (i.e., assumes a negative outcome) and cannot efficiently account for developmental changes in vulnerability that may cause positive outcomes in population health. We posit that multiple factors, including acculturation stress, affect the acculturation–health relationship but do not necessarily have consistent or deleterious effects over the life course.
The putative health advantages enjoyed by immigrants over US-born Mexican Americans are not apparent in later life. We found that longer US residency was associated with better functional health. Identifying characteristics associated with healthy immigrants early in life and characteristics of successful aging in older Latinos may ensure a healthy Latino population in the upcoming years. The results from this study suggest that socioeconomic advantages acquired through longer US residency may contribute to better functional health. There is clearly insufficient research on the aging population. Further research, especially longitudinal research designs that will increase our understanding of the health and aging of this large and growing segment of the US population, may help ensure the future health of the nation.
H. M. González is supported by National Institute of Mental Health grants MH 67726 and MH 84994. W. A. Vega and H. M. González are supported by the Robert Wood Johnson Foundation, Network for Multi-cultural Research on Health and Healthcare.
We express our sincere thanks to David V. Espino for his valuable, helpful, and supportive comments. We also thank Anne Dubois for her assistance in preparing the article for publication.
ContributorsH. M. González conceptualized the study and supervised all aspects of its implementation. M. Ceballos also conceptualized the study and contributed to the writing. W. Tarraf and B. T. West conducted all data analyses, interpreted the data, and helped draft the article. M. E. Bowen reviewed the article and assisted in the revisions. W. A. Vega reviewed all drafts of the article and guided its revisions. All authors worked on finalizing the data set, interpreting the data, and developing the narrative into this article.
Human Participant Protection
This existing data study was exempt from institutional review board evaluation.
Hector M. González, Institute of Gerontology, Wayne State University, Detroit, MI.
Miguel Ceballos, Department of Sociology and the Institute for Ethnic Studies, University of Nebraska, Lincoln.
Wassim Tarraf, Institute of Gerontology, Wayne State University, Detroit, MI.
Brady T. West, Center for Statistical Consultation and Research, University of Michigan, Ann Arbor.
Mary E. Bowen, Institute of Gerontology, Wayne State University, Detroit, MI.
William A. Vega, Department of Family Medicine, University of California, Los Angeles.