Association of acculturation and country of origin with self-reported hypertension and diabetes in a heterogeneous Hispanic population

Background Hispanics are the fasting growing population in the U.S. and disproportionately suffer from chronic diseases such as hypertension and diabetes. Little is known about the complex interplay between acculturation and chronic disease prevalence in the growing and increasingly diverse Hispanic population. We explored the association between diabetes and hypertension prevalence among distinct U.S. Hispanic subgroups by country of origin and by degree of acculturation. Methods We examined the adult participants in the 2001, 2003, 2005, and 2007 California Health Interview Survey (CHIS). Using weighted logistic regression stratified by nativity, we measured the association between country of origin and self-reported hypertension and diabetes adjusting for participants’ demographics, insurance status, socio-economic status and degree of acculturation measured by citizenship, English language proficiency and the number of years of residence in the U.S. Results There were 33,633 self-identified Hispanics (foreign-born: 19,988; U.S.-born: 13,645). After multivariable adjustment, we found significant heterogeneity in self-reported hypertension and diabetes prevalence among Hispanic subgroups. Increasing years of U.S. residence was associated with increased disease prevalence. Among all foreign-born subgroups, only Mexicans reported lower odds of hypertension after adjustment for socioeconomic and acculturation factors. Both U.S.-born and foreign-born Mexicans had higher rates of diabetes as compared to non-Hispanic whites. Conclusions We found significant heterogeneity among Hispanics in self-reported rates of hypertension and diabetes by acculturation and country of origin. Our findings highlight the importance of disaggregation of Hispanics by country of origin and acculturation factors whenever possible.

While disease prevalence has been shown to vary by Hispanic subgroups [11][12][13][14], few studies have examined the differential effect of acculturation across Hispanic subgroups defined by country of origin. Acculturation is defined as a multidimensional process through which foreigners adopt the customs of a host country [13]. While a standard acculturation metric is not fully agreed upon [15], language remains an important acculturation metric since more than three-quarters of the U.S. Hispanic population speak a language other than English at home [1]. Other important components of acculturation include: nativity (U.S.-born or foreign-born), years of residence in the United States and citizenship status [15]. Although Hispanics have been traditionally studied as a homogenous group [8], understanding determinants of chronic diseases among distinct subgroups remains salient in planning public health interventions.
Since the Hispanic population in California is two times greater than that of the remainder of the U.S. (36.2% versus 15.1%) [16], we examined the hypothesis that diabetes and hypertension prevalence among U.S. Hispanics varies by country of origin and by the degree of acculturation using the California Health Interview Survey (CHIS).

Methods
We examined the adult participants in the 2001,2003,2005, and 2007 CHIS, the largest health survey conducted in any state. CHIS is a population-based random-digit dial telephone survey of California's population conducted every other year since 2001. CHIS employs a multi-stage sample design and includes cellphone-only households. In addition, interviews are conducted in English, Spanish, and three Asian languages. Using CHIS allows sampling of non-Carribean Hispanics in larger numbers than what is possible through national datasets. A detailed description of the history, design, data collection, and available data for CHIS is available at the website [17].

Variables
Patient demographic characteristics included age, sex, selfreported race [Hispanic, non-Hispanic White (NHW)], federal poverty level (FPL) (0-99%, 100-199%, 200-299%, ≥ = 300%), insurance (private, Medi-CAL, Medicare, uninsured), and education level (less than high school, high school/GED/vocational high school completion, some college or higher education). Self-identified Hispanics were categorized by country of origin: Mexican, Salvadorean, Guatemalan, Central American, South American, Puerto Rican, Hispanic European, Hispanics having two or more countries of origin (2+ Hispanic), and Other Hispanic. We merged Salvadorean, Guatemalan and Central American into one category to account for smaller sample size. Other Hispanic included those who self-identified as this category and those who identified as Hispanic European. We did not include Puerto Ricans in our analyses due to small sample size.
Given the multiple metrics for measuring acculturation [15], we examined four variables as measures of acculturation based on their utilization in prior studies [13,18]. Self-reported English language ability was dichotomized from a four level categorical variable into not well (not at all, not well) and well (well, very well). Years of residence in the U.S. was categorized as a four level variable (≤1 yr-4 yrs, 5-9 yrs, 10-14 yrs, 15+ yrs). Nativity was defined as U.S.-or foreign-born.
Citizenship was defined as U.S. citizen, naturalized citizen, and no-citizenship status. Years of residence in the U.S. and citizenship questions were only asked of those who were foreign-born.
Self-reported hypertension and diabetes were ascertained by the questions: "Has a doctor ever told you that you have high blood pressure?" and "Other than during pregnancy, has a doctor ever told you that you have diabetes or sugar diabetes?" Body mass index (BMI) was calculated by CHIS based on self-reported participant's weight and height. Smoking habits were assessed with two questions: "Altogether have you smoked at least 100 or more cigarettes in your entire lifetime?" and "Do you now smoke cigarettes every day, some days or not at all?" Based on examination of distribution plots, selfreported intensity of utilization of clinical services in the past 12 months was defined into three categories: no visits, 1-11 visits, greater than 11 visits.

Statistical analyses
We used the chi-square test to compare the distribution of participants' demographic and clinical characteristics of self-reported hypertension and diabetes. Weighted logistic regression models were created for participants' self-reported hypertension and diabetes. All models were adjusted for participants' demographic characteristics, insurance status, socioeconomic level, clinical characteristics, and acculturation factors (years in the U.S., citizenship, and English language ability). NHW was used as a reference group. Models were stratified by nativity status because prior work has shown that significant differences exist between foreign-and U.S.-born Hispanics [11]. Because the prevalence of hypertension and diabetes increases with age, we decided a priori to conduct separate sensitivity analyses stratifying the aforementioned regression models by age greater than 45 years. Finally, separate interaction analyses were conducted between Hispanic country of origin subgroups and two acculturation variables (years of residence in the U.S. and English language ability).
Due to the complex sampling design of CHIS, SUDAAN 9.2 statistical software (Research Triangle Park, North Carolina) was used for all analyses. Special sample weights provided by CHIS were included in the analysis to adjust for oversampling, non-response bias and post-stratification population totals. All p values reported are for 2-tailed tests and a value of <0.05 was considered statistically significant. All p values were adjusted for multiple comparisons between subgroups using the Bonferroni method.
The current study was deemed exempt by the Massachusetts General Hospital Institutional Review Board because the data were de-identified by CHIS and are publicly available.

Results
There were 33,633 self-identified Hispanics and 126,448 NHWs. There were 13,645 U.S.-born Hispanics (Table 1) and 19,988 foreign-born Hispanics (Table 2). Total selfidentified Hispanic subtypes included: Mexican (n = 26,091), Central American (n = 3,092), South American (n = 921), 2 + Hispanic (n = 2,245), and Other Hispanic (n = 1,284). Overall, regardless of nativity status, Hispanics were slightly younger than NHWs. Since our hypertension and diabetes findings did not change in separate sensitivity analyses stratifying for age greater than 45 years, our data is not presented stratified by age.

U.S.-born Hispanics
South American U.S.-Born Hispanics reported the highest rates of education and income level as compared with other Hispanic subgroups (Table 1). Mexicans, Central Americans, and Other Hispanics reported the highest rates of hypertension and obesity. Mexicans, 2+ Hispanics and Other Hispanics reported the highest rates of smoking among U.S.-born Hispanics. U.S.-born Hispanics rarely reported English language barriers.
In multivariable adjusted models, Mexicans, Central and South Americans were less likely to report hypertension as compared with NHWs (Table 3). Similarly, Mexicans  All p values less than 0.05 compared to Non-Hispanic whites. *Years in US and citizenship status only asked to foreign-born adults.
and Other Hispanics were more likely to report diabetes while Central and South Americans were less likely to report diabetes as compared with NHWs.

Foreign-born Hispanics
Foreign-born Hispanics reported a lower educational attainment level, as compared with their U.S.-born counterparts (  (Table 2). In multivariable adjusted models, foreign-born Mexicans had the lowest odds of reporting hypertension and Other Hispanics the highest odds compared with NHWs (Table 3). Non-citizens, those with less than five years of residence in the U.S. and those with 10-14 years of residence in the U.S. were less likely to report hypertension in adjusted models (Table 4). After adjustment for clinical and demographic co-variates, limited English proficiency was not associated with self-reported hypertension.

U.S.-born versus foreign-born Hispanics
As compared with foreign-born Hispanics, U.S.-born Hispanics had higher educational attainment and income levels (Tables 1 and 2). Table 3 shows the differential effects of nativity on self-reported rates of hypertension and diabetes, after adjusting for demographic and clinical co-variates. Both U.S.-born and foreign-born Mexicans had lower odds of self-reporting hypertension and higher odds of reporting diabetes as compared with NHWs. U.S.-born Other Hispanics had the highest odds of self-reporting diabetes while foreignborn Other Hispanics had the highest odds of selfreporting hypertension.

Discussion
Using a large state sample of Hispanics over several years, our study has several important findings on the association of acculturation and chronic disease prevalence.   First, self-reported prevalence of hypertension and diabetes varied by country of origin. Second, stratifying by nativity status demonstrated significant differences in disease reporting by country of origin. Third, among all foreign-born subgroups, only Mexicans reported lower odds of hypertension after adjustment for socioeconomic and acculturation factors. Fourth, acculturation, as measured by years of residence in the U.S. and citizenship status, was an important predictor of hypertension and diabetes. This suggests that acculturation may differentially impact Hispanic subgroups. Our findings highlight the importance of disaggregation of Hispanics by country of origin and acculturation factors whenever possible. Nativity had important differential effects. Compared with NHW, we found that U.S.-born Mexicans, Central Americans South Americans and foreign-born Mexicans had lower odds of reporting hypertension. The rate of hypertension increased after adjustment for socioeconomic status and acculturation factors in all groups except for South Americans. There was no difference in the reported odds of hypertension compared to NHWs in fully adjusted models for the other Hispanic subgroups. As people spend more time in the U.S., lifestyle changes and a consequent increase in BMI may significantly lead to a poorer cardiovascular profile [19][20][21]. Our study also found that U.S.born Hispanics had higher BMIs compared to foreign-born Hispanics, regardless of country of origin. However, changes in clinical risk factors might differ in each Hispanic subgroup. Prior studies have shown that despite similar cardiac risk factors, Hispanic subgroups have varying degrees of subclinical CVD, including higher coronary artery calcium (CAC) scores and inflammation as measured by Creactive protein [12,[22][23][24]. This may be attributed to the genetic diversity in Hispanic subgroups [25]. The low prevalence of hypertension among Mexicans, despite acculturation, might be explained by the geographic proximity to their native land, which may play a stronger role in preservation of family ties and traditions despite length of residency in the U.S. and English language acquisition [26].
The "Hispanic Paradox" refers to the epidemiological finding that foreign-born Hispanicslargely Mexicanoften fare better than their white counterparts on morbidity and mortality outcomes, despite lower levels of income, education, and worse health care access [27][28][29]. This difference has been attributed to a healthy migrant effect, healthier behaviors, and/or cultural traditions. The healthy migrant effect posits that healthier persons are more likely to migrate thus producing increased longevity and health in the emigrant population. The acculturation hypothesis states that Hispanic cultural orientation results in healthier behaviors that result in better health outcomes and is thus protective against the effects of lower socioeconomic status in the U.S. [29]. Prior studies of the healthy migrant effect have focused primarily on Mexicans with few studies focusing on Hispanic subgroups [30].
Similar to prior studies, we also found lower odds of reported hypertension among foreign-born Mexicans even after adjustment for acculturation [31]. However, we found no significant difference in the odds of reported hypertension compared with NHWs after adjustment for socioeconomic and acculturation factors among all other foreign-born Hispanic subgroups. Our findings suggest that acculturation may differentially impact Hispanic subgroups and highlights the importance of disaggregation of Hispanics by country of origin and acculturation factors.
In agreement with existing literature, we found that reported diabetes rates varied with country of origin [6,7,32]. In our study, Mexicans, regardless of nativity status, have higher odds of reporting diabetes compared to NHWs, even after adjusting for BMI. In contrast, U.S.-born Central and South Americans had lower odds of reporting diabetes compared to NHWs. In addition, U.S.-born Other Hispanic and foreignborn Central Americans had higher odds of reporting diabetes. These findings might be attributed to genetic and behavioral differences, which make different subgroups more susceptible to diabetes and the metabolic syndrome and not hypertension. We found an increasing trend in the odd ratios for reporting diabetes as the years of residence in the U.S. increased (Table 4). This is consistent with prior studies that have shown an inverse relationship between increasing acculturation and a less healthy diet and increased rates of obesity [5,8,21,30,33]. Our findings show a complex relationship between greater acculturation and diabetes, likely because of the distinct Hispanic subgroups and numerous acculturation metrics available in our study.
We found higher odds of reporting hypertension among foreign-born Other Hispanics and higher rates of diabetes among U.S.-born Other Hispanics compared with NHWs. These differences may be due to a higher socioeconomic and acculturation status for this subgroup. Compared to other Hispanic subgroups, the Other Hispanics had high rates of college education, higher incomes, private insurance, and higher unadjusted rates of hypertension and diabetes. Among the foreign-born Hispanics, Other Hispanics were the most acculturated with 82% living in the U.S. greater than fifteen years and most speaking English well. The selfidentification as "Other Hispanic" may reflect people who do not fit a particular pre-established Hispanic subgroup (i.e., European Hispanics or Brazilian) or people who identify as biracial or bi-ethnic (i.e., participant with only one Hispanic parent). The self-identity of Hispanics is complex and is influenced by the country of origin and nativity status [16]. Future studies should include bi-racial or bi-ethnic Hispanics as they constitute a poorly studied subgroup.

Strengths and limitations
A major strength of our study is its large sample size of Hispanics over a seven-year period that used the same sampling and data collection methodology in a state with one of the largest Hispanic populations. The survey questions allowed for a more detailed analysis of self-identified Hispanics by country of origin and acculturation factors than has been available in previous studies. We were able to assess acculturation in four dimensions (i.e., English language ability, years of residence in the U.S., citizenship, and nativity status). In addition, California's Hispanic demographic patterns allowed for a larger and more representative sample of Hispanics from Central and South America, groups traditionally underrepresented in other datasets. Importantly, our data allowed for statistical analyses that controlled for important confounders such as intensity of clinical services utilization.
Our study had several limitations. First, our data are cross-sectional and thus associations found are not proof of causality. Second, the self-reported nature of hypertension and diabetes may cause underestimation of disease rates, particularly because many Hispanics are uninsured and have less access to regular sources of healthcare. However, self-reported data for hypertension and diabetes have been shown to be highly correlated with physicians' records [34][35][36]. Similarly, self-reported English language ability, one of the metrics of acculturation, was self-reported, consistent with current U.S. census definitions. We lacked significant data on Caribbean Hispanics and Black Hispanics, which may represent a large segment of the Hispanic population in other parts of the U.S. Finally, Hispanics who self-identified as "2+ Hispanic" and "Other Hispanic" represent a heterogeneous group from multiple countries of origin. However, these groups represent primarily U.S.-born Hispanics. Inclusion of these categories is important since it is reflective of the complex demographic selfidentity of Hispanics that should be included in analyses of Hispanics.

Conclusion
In conclusion, we found significant differences in rates of self-reported hypertension and diabetes by Hispanic subgroups, nativity, and acculturation. Our findings underscore the importance of exploring within-group differences and heterogeneity among Hispanics in the U.S., differences that may account for the mixed and paradoxical associations between acculturation and various health indicators. Finally, better standardized data collection strategies that incorporate Hispanic subgroups and acculturation factors will allow for improved tracking of health status among Hispanics, which may better inform public health interventions.