Abstract
Objectives. We examined the association between race/ethnicity and all-cause mortality risk in US adults and whether this association differs by nativity status.
Methods. We used Cox proportional hazards regression to estimate all-cause mortality rates in 1997 through 2004 National Health Interview Survey respondents, relating the risk for Hispanic subgroup, non-Hispanic Black, and other non-Hispanic to non-Hispanic White adults before and after controlling for selected characteristics stratified by age and gender.
Results. We observed a Hispanic mortality advantage over non-Hispanic Whites among women that depended on nativity status: US-born Mexican Americans aged 25 to 44 years had a 90% (95% confidence interval [CI] = 0.03, 0.31) lower death rate; island- or foreign-born Cubans and other Hispanics aged 45 to 64 years were more than two times less likely to die than were their non-Hispanic White counterparts. Island- or foreign-born Puerto Rican and US-born Mexican American women aged 65 years and older exhibited at least a 25% lower rate of dying than did their non-Hispanics White counterparts.
Conclusions: The “Hispanic paradox” may not be a static process and may change with this population growth and its increasing diversity over time.
Despite Hispanics’ lower socioeconomic position (i.e., lower education and income) and lower rate of health insurance coverage, they exhibit lower all-cause mortality rates than do non-Hispanics. This “Hispanic paradox” for adult mortality outcomes has been researched over the past 2 decades.1–12 Recent data for deaths in the United States show that the all-cause age-adjusted death rate for Hispanics (546.1/100 000) was much lower than were those for non-Hispanic Blacks (978.6) and Whites (763.3).13 Although this finding has also been reported by several studies,2–7,12,14 the mortality advantage for Hispanic subgroups is less well characterized. Results from studies examining the mortality advantage among Hispanic subgroups have been mixed, with some reporting lower rates2,3,5,6,10,12,14 and others reporting similar or higher rates for Hispanic subgroups compared with non-Hispanic Whites.5,6,12,14 A notable observation is that several of these studies simultaneously examined only Mexicans, Puerto Ricans, and Cubans.2,3,5,12
Studies have also shown that compared with non-Hispanic Whites, the mortality advantage seems to be greater for foreign-born Hispanics than for their US-born counterparts.7,11,12,14–17 This advantage has been attributed to healthy in-migrant and unhealthy out-migrant selection effects,3,7,18 referred to as the “salmon bias hypothesis.”19 However, few studies examining the association between nativity status and mortality have focused on Hispanic subgroups, with 1 study focusing on Mexican Americans and other Hispanics14 and another on Mexican Americans, Puerto Ricans, Cubans, and other Hispanics.7 These studies suggested that the mortality advantage over non-Hispanic Whites may be specific to foreign-born Mexican Americans and other Hispanics.7,14 Therefore, given the continuous growth of the Hispanic population and its increasing heterogeneity regarding country of origin,20,21 studies focusing on Hispanic subgroups to examine this paradox are imperative.
To contribute to the limited literature on the Hispanic paradox on all-cause mortality risk among Hispanic subgroups according to nativity status, we used data from the National Health Interview Survey (NHIS) for the years 1997 through 2004 linked to the National Death Index (NDI) mortality files to examine the association between race/ethnicity and all-cause mortality risk for US adults aged 25 years and older and whether this association differs by nativity status. For these analyses, the Hispanic category was defined using national origin or ancestry to specified subgroups as follows: Puerto Rican, Mexican, Mexican American, Cuban, Central and South American, and other Hispanic.
METHODS
We used public use data from the NHIS files for the years 1997 through 2004 and the NHIS-NDI–linked mortality files obtained from the National Center for Health Statistics (NCHS) Web site.22,23 NHIS, a national survey using a 3-stage stratified cluster probability sampling design, conducts annual face-to-face household interviews of the US noninstitutionalized civilian population. A complete description of the survey description, plan, and design has been provided elsewhere.24,25 The NHIS-NDI–linked mortality public use files link NHIS survey participants for the years 1986–2004 to death certificate records from the NDI with follow-up through December 31, 2006.23 To link NHIS survey participants with NDI death certificate records, the NCHS used a probabilistic matching algorithm derived from social security number; first name, middle initial, last name, or surname; month, day, and year of birth; gender; father's surname; state of birth; race; state of residence; and marital status.26
Our outcome, all-cause mortality, was coded using the underlying cause of death as alive or dead at the end of the follow-up period. We constructed duration of follow-up according to the NCHS guidelines,27 using the NHIS interview year and year of death. In brief, we assigned all respondents one half year of follow-up for the year of their NHIS interview (including those who died the same year of the interview), then a full year of follow-up for each subsequent year until the year before death for those who died or until the end of 2006 for those assumed alive. For those who died after the year of their interview, another half year of follow-up was added during the year of their death.
The main independent variable was race/ethnicity, which we defined using several variables obtained from the public use NHIS files: ethnicity, national origin or ancestry, and race. We determined ethnicity from various questions depending on the year of the NHIS: For years 1997 and 1998, the NHIS asked the following question: “Do any of these groups represent (person)’s national origin or ancestry?” with Hispanic as a yes or no choice. For participants who answered yes to the ethnicity question, a follow-up question queried which group(s) represented the participant's national origin or ancestry: Puerto Rican, Mexican-Mexicano, Mexican American (including Chicano), Cuban or Cuban American, other Latin American nonspecific type, other Spanish, Hispanic or Spanish nonspecific type, Hispanic or Spanish type refused, Hispanic or Spanish type not ascertained, Hispanic or Spanish type don't know, and not Hispanic or Spanish origin. From 1999 to 2004, NHIS changed the question on ethnicity to “Do you consider yourself Hispanic or Latino?” with a choice of yes or no. The second question on national origin or ancestry remained the same but added the following choices: Dominican (Republic) and Central and South American.
We determined race from the following question: “What race do you consider yourself to be?” NHIS interviewers asked all survey participants this question, with the choices further categorized as White, Black or African American, Asian, American Indian and Alaska Native, Native Hawaiian and Pacific Islander, other, and multiple races. Using the aforementioned questions, we created the following categories to define race/ethnicity: Puerto Rican, Mexican, Mexican American, Cuban, Central and South American, and other Hispanic; and non-Hispanic Black, non-Hispanic White, and other non-Hispanic. Consistent with previous studies,28,29 Mexicans and Mexican Americans were kept as separate groups in our analyses because their self-identification as 2 distinct groups may represent issues associated with acculturation and assimilation to US society through ethnic identity, country of birth, and Mexican culture ownership or attachment, regardless of whether they were born in the United States or Mexico.
We included variables considered potential confounders from the public use NHIS files in these analyses. These variables included gender and US region of residence (Northeast, Midwest, South, or West) in the analysis as collected by NHIS. We categorized place of birth (hereinafter referred to as nativity status) into US-born for people who reported being “born in one of the 50 US states or the District of Columbia” and island or foreign-born for those who reported being “born in Puerto Rico, Guam, or other outlying territories of the United States” or “not born in the United States or a US territory.” Island- or foreign-born respondents were further asked how long they had been in the United States (in years), with categories ranging from less than 1 year to 15 years or more. For analytical purposes, we created a variable combining nativity status and length of stay in the United States and coded it as island- or foreign-born with less than 10 years in the United States, island or foreign-born with 10 years or more in the United States, and US born. We included age in the analysis as a continuous and a categorical (25–44, 45–64, and ≥ 65 years) variable.
Marital status was specified as married (married with spouse in the household, married with spouse not in household, and married with spouse in household unknown), divorced (divorced and separated), widowed, and single (never married). We did not include unknown marital status in our analysis. We categorized education as completing less than high school, being a high school graduate or receiving a general equivalency diploma, completing some college, and being a college graduate or higher. NHIS collected family income by asking each participant to select his or her family total annual income from 12 categories (ranging from $0 to > $75 000 and a refusal category) and adjusted it to 1997 income using the inflation calculator developed by the consumer price index from the US Bureau of Labor Statistics.30 Specifically, we adjusted the incomes in NHIS 1997–2004 to the equivalent dollar amount in NHIS 1997, with income categorized as less than $20 000, $20 000 to $34 999, $35 000 to $54 999, and $55 000 or greater. We kept records with missing values in the analysis in a separate income category.
We limited the analysis to NHIS-NDI–linked mortality file records of respondents aged 25 years and older for survey years 1997 to 2004 (n = 231 269). For analytic purposes, we excluded individuals ineligible for follow-up through December 31, 2006 (n = 14 352) and without information on education (n = 1401), marital status (n = 477), nativity status (n = 77), and mortality (n = 170), yielding a final sample of 214 792 records, including 16 633 deaths over 1 255 492 person-years (median = 5.8 years, range = 0.5–9.5).
We have presented selected characteristics of the population for Hispanic and non-Hispanic subgroups. We calculated death rates by dividing the number of deaths by the total number of person-years during the follow-up period. Because of heterogeneity of mortality risk by age and gender (P ≤ .001 for interaction among age, gender, and race/ethnicity), we presented analyses stratified by these variables.
We used Cox proportional hazards regression to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) relating all-cause mortality risk for Hispanic subgroup, non-Hispanic Black, and other non-Hispanic to non-Hispanic White adults before and after controlling for selected characteristics stratified by age and gender. To determine whether the strength of the association observed between race/ethnicity and all-cause mortality differed by nativity status in each age and gender group, we tested an interaction term in the fully adjusted models.
We conducted all data management procedures with SAS 9.2 (SAS Institute, Cary, NC).31 We conducted statistical analyses with SUDAAN 10.0.1 (RTI International, Research Triangle Park, NC)32 because of its ability to take into account the complex sampling design used in NHIS. In addition and as recommended by NCHS,33 we concatenated all 8 years of data into 1 file. We adjusted the survey weight variable provided by the NHIS-NDI–linked mortality files by dividing for the number of survey years (8) to obtain the average US civilian noninstitutionalized population.33 Sample sizes presented in Table 1 were unweighted, but all other estimates (proportions, standard errors, rates, and HRs with their 95% CIs) were weighted.
TABLE 1—
Distribution of Selected Characteristics for US Adults Aged 25 Years and Older According to Ethnicity: National Health Interview Survey, United States, 1997–2004
| Hispanic |
Non-Hispanic a |
|||||||||
| Characteristics | Puerto Rican (n = 3472) | Mexican (n = 11 164) | Mexican American (n = 8217) | Cuban (n = 2187) | Central and South American (n = 3519) | Other Hispanic (n = 4807) | Black (n = 28 811) | White (n = 145 597) | Other (n = 7018) | Total (n = 214 792) |
| Age, y, mean (SE) | 45.2 (0.34) | 40.7 (0.21) | 44.4 (0.27) | 52.3 (0.74) | 42.1 (0.31) | 44.2 (0.58) | 46.1 (0.16) | 49.9 (0.08) | 44.4 (0.28) | 48.6 (0.07) |
| 25–44 | 54.1 (1.11) | 69.5 (0.75) | 57.8 (0.78) | 39.9 (1.58) | 64.4 (1.09) | 58.7 (1.70) | 53.0 (0.46) | 43.0 (0.22) | 56.2 (0.87) | 46.3 (0.20) |
| 45–64 | 33.8 (1.00) | 23.7 (0.63) | 30.0 (0.61) | 32.8 (1.23) | 28.3 (0.89) | 30.0 (1.05) | 33.4 (0.37) | 36.3 (0.18) | 33.3 (0.71) | 35.0 (0.16) |
| ≥ 65 | 12.1 (0.74) | 6.8 (0.37) | 12.2 (0.52) | 27.3 (1.99) | 7.4 (0.63) | 11.3 (1.03) | 13.8 (0.37) | 20.7 (0.20) | 10.5 (0.60) | 18.5 (0.17) |
| Men, proportion (SE) | 46.2 (1.15) | 53.1 (0.58) | 48.4 (0.69) | 50.6 (1.20) | 47.6 (0.95) | 46.8 (0.90) | 44.0 (0.35) | 47.8 (0.15) | 49.5 (0.74) | 47.6 (0.13) |
| Island- or foreign-born,b proportion (SE) | 56.6 (1.19) | 86.2 (0.048) | 16.8 (0.81) | 84.9 (0.87) | 92.7 (0.58) | 51.8 (3.20) | 9.1 (0.44) | 4.5 (0.10) | 69.5 (1.49) | 13.5 (0.18) |
| Years in United States, proportion (SE) | ||||||||||
| < 10 | 7.4 (0.59) | 26.5 (0.79) | 2.1 (0.23) | 18.9 (1.98) | 29.1 (1.22) | 13.1 (1.06) | 2.5 (0.19) | 1.0 (0.05) | 22.5 (0.87) | 3.6 (0.09) |
| ≥ 10 | 48.4 (1.11) | 57.7 (0.79) | 14.2 (0.72) | 65.4 (1.92) | 62.2 (1.21) | 38.1 (2.40) | 6.5 (0.32) | 3.5 (0.07 | 46.1 (1.10) | 9.7 (0.14) |
| Marital status, proportion (SE) | ||||||||||
| Married | 60.9 (0.98) | 78.0 (0.44) | 69.2 (0.63) | 69.4 (1.49) | 72.8 (0.83) | 66.5 (0.98) | 49.9 (0.52) | 71.9 (0.22) | 74.0 (0.63) | 69.5 (0.20) |
| Divorced | 5.9 (0.48) | 3.23 (0.18) | 4.9 (0.24) | 8.4 (1.13) | 3.2 (0.34) | 4.93 (0.55) | 8.4 (0.25) | 7.98 (0.10) | 4.6 (0.31) | 7.5 (0.09) |
| Widow | 16.8 (0.73) | 8.5 (0.27) | 13.5 (0.44) | 13.0 (1.03) | 11.0 (0.63) | 15.2 (0.55) | 19.1 (0.30) | 10.9 (0.11) | 7.9 (0.33) | 11.8 (0.10) |
| Single | 16.4 (0.82) | 10.3 (0.39) | 12.5 (0.42) | 9.3 (0.60) | 13.0 (0.55) | 13.4 (0.68) | 23.3 (0.39) | 9.2 (0.13) | 13.5 (0.49) | 11.2 (0.13) |
| Region of residence, proportion (SE) | ||||||||||
| Northeast | 58.4 (1.86) | 2.3 (0.34) | 0.62 (0.12) | 12.9 (1.35) | 25.9 (1.37) | 28.8 (2.06) | 16.1 (0.54) | 20.6 (0.31) | 18.0 (0.91) | 19.4 (0.25) |
| Midwest | 8.9 (0.92) | 11.4 (1.07) | 7.6 (0.54) | 2.4 (0.38) | 5.5 (0.73) | 5.40 (0.54) | 18.8 (0.76) | 28.5 (0.37) | 15.2 (1.00) | 24.9 (0.31) |
| South | 24.8 (1.68) | 29.8 (1.56) | 42.9 (2.26) | 78.9 (1.65) | 40.0 (1.58) | 28.7 (2.08) | 57.9 (1.05) | 34.3 (0.42) | 22.5 (1.21) | 36.6 (0.36) |
| West | 7.9 (0.76) | 56.5 (1.58) | 48.9 (2.06) | 5.8 (0.66) | 28.6 (1.44) | 37.1 (3.82) | 7.2 (0.32) | 16.7 (0.31) | 44.3 (1.46) | 19.1 (0.27) |
| Education, proportion (SE) | ||||||||||
| < high school | 33.4 (1.17) | 66.4 (0.73) | 32.6 (0.83) | 34.4 (1.77) | 33.1 (1.11) | 30.6 (0.93) | 23.4 (0.54) | 12.9 (0.21) | 14.3 (0.72) | 17.2 (0.19) |
| High school or GED | 29.7 (0.98) | 16.6 (0.43) | 29.9 (0.60) | 23.8 (1.18) | 24.0 (0.87) | 27.7 (0.83) | 31.4 (0.39) | 31.2 (0.24) | 19.5 (0.55) | 30.0 (0.20) |
| Some college | 16.0 (0.82) | 8.1 (0.34) | 18.8 (0.63) | 14.1 (1.08) | 16.1 (0.77) | 17.7 (0.73) | 20.0 (0.45) | 17.9 (0.13) | 13.6 (0.45) | 17.6 (0.12) |
| ≥ college graduate | 20.8 (1.03) | 8.9 (0.40) | 18.7 (0.66) | 27.7 (1.28) | 26.8 (1.05) | 24.0 (0.88) | 25.3 (0.50) | 38.1 (0.32) | 52.6 (0.94) | 35.2 (0.27) |
| Family income, proportion (SE) | ||||||||||
| < $20 000 | 29.0 (1.01) | 34.1 (0.85) | 22.7 (0.71) | 25.0 (1.28) | 23.6 (0.92) | 26.1 (1.12) | 29.0 (0.62) | 14.9 (0.19) | 17.2 (0.71) | 17.8 (0.18) |
| $20 000–$34 999 | 18.2 (0.77) | 23.9 (0.63) | 20.0 (0.60) | 18.7 (1.05) | 22.9 (0.80) | 21.9 (0.74) | 19.1 (0.28) | 16.8 (0.15) | 15.6 (0.49) | 17.5 (0.13) |
| $35 000–$54 999 | 13.2 (0.71) | 11.3 (0.43) | 16.3 (0.53) | 12.5 (0.88) | 14.3 (0.90) | 17.0 (0.69) | 14.0 (0.28) | 16.6 (0.13) | 13.2 (0.44) | 15.9 (0.12) |
| ≥ $55 000 | 18.7 (1.03) | 8.8 (0.44) | 21.1 (0.79) | 22.1 (1.37) | 19.0 (1.03) | 17.7 (0.91) | 16.5 (0.57) | 31.0 (0.30) | 35.5 (0.85) | 28.1 (0.25) |
| No. of deathsc | 183 | 354 | 388 | 186 | 51 | 240 | 2411 | 12 581 | 239 | 16 633 |
| All-cause death rate/100 000 person-y | 759.6 | 514.6 | 729.9 | 1343.7 | 258.8 | 662.8 | 1221.7 | 1226.8 | 499.0 | 1139.4 |
Note. GED = general equivalency diploma.
All P values for t-test and χ2 statistics comparing ethnic subgroups and groups were < .001.
Island or foreign born for Hispanics included island-born Puerto Ricans.
The number of deaths is unweighted.
RESULTS
Hispanic subgroups were more likely to be younger than were non-Hispanic groups, with the exception of Cubans (Table 1). More than one half of Hispanics were island- or foreign-born, with the majority reporting being in the United States for 10 years or more. The only exception to this was Mexican Americans. At least half of Hispanics and non-Hispanics reported being married. In general, Hispanic subgroups were more likely to live in the South and West regions of the United States, with the exception of Puerto Ricans, who tended to live in the Northeast. Non-Hispanic Blacks were more likely to live in the South. When compared with non-Hispanic Whites, Hispanic subgroups and non-Hispanic Blacks were more likely to have less than a high school education and more likely to earn an annual income of less than $20 000. It is worth noting that other non-Hispanics were more educated and reported higher income than did non-Hispanic Whites. The overall all-cause mortality rate was 1139.4 per 100 000, with Hispanics (640.0/100 000) exhibiting a lower rate than did non-Hispanics (1194.8/100 000; data not shown). However, there was great variability among Hispanics, with Central and South Americans having the lowest death rates (258.8/100 000) and Cubans having the highest (1343.7/100 000). All-cause mortality rates were similar for non-Hispanic Blacks and Whites (1221.7 vs 1226.8/100 000) but much lower for other non-Hispanics (499.0/100 000).
Mortality rates for Hispanic men were not different from those for non-Hispanic White men, regardless of subgroup and age group (Table 2). We observed this pattern before and after adjustment for age, nativity status and length in the United States, marital status, region of residence, education, and family income. However, for Hispanic women, Mexican Americans (HR = 0.13; 95% CI = 0.05, 0.35) and Central and South Americans (HR = 0.36; 95% CI = 0.14, 0.93) aged 25 to 44 years were less likely to die than were non-Hispanic White women. These estimates remained nearly identical after adjustment. Similarly, Cuban women aged 45 to 64 years exhibited at least a 38% lower death rate than did non-Hispanic Whites regardless of adjustment. For women aged 65 years and older, Puerto Ricans (HR = 0.49; 95% CI = 0.36, 0.68), Mexicans (HR = 0.68; 95% CI = 0.54, 0.85), Mexican Americans (HR = 0.68; 95% CI = 0.53, 0.87), and Central and South Americans (HR = 0.48; 95% CI = 0.27, 0.87) had lower unadjusted death rates than did non-Hispanic Whites, yet we observed significant associations only for Puerto Ricans (HR = 0.61; 95% CI = 0.44, 0.86) and Mexican Americans (HR = 0.75; 95% CI = 0.58, 0.98) after adjustment. For non-Hispanics, Black men younger than 65 years and women aged 45 to 64 years had higher death rates compared with their White counterparts before and after adjustment for selected characteristics. The estimates for the adjusted analyses were 1.33 (95% CI = 1.02, 1.73), 1.41 (95% CI = 1.22, 1.63), and 1.35 (95% CI = 1.15, 1.60), respectively. Other non-Hispanics were as likely to die as were non-Hispanic Whites with the exception of men aged 65 years and older (HR = 0.68; 95% CI = 0.51, 0.91).
TABLE 2—
All-Cause Mortality by Race/Ethnicity Among US Adults Aged 25 Years and Older: National Health Interview Survey, United States, 1997–2004
| Aged 25–44 Years |
Aged 45–64 Years |
Aged ≥ 65 Years |
||||
| Race/Ethnicity | CHR (95% CI) | AHR (95% CI) | CHR (95% CI) | AHR (95% CI) | CHR (95% CI) | AHR (95% CI) |
| Men | ||||||
| Puerto Rican | 1.06 (0.55, 2.05) | 1.01 (0.49, 2.07) | 1.19 (0.72, 1.97) | 1.26 (0.74, 2.14) | 0.93 (0.64, 1.35) | 1.25 (0.85, 1.85) |
| Mexican | 1.11 (0.80, 1.53) | 1.25 (0.75, 2.09) | 0.88 (0.63, 1.22) | 1.00 (0.68, 1.47) | 0.76 (0.57, 1.02) | 0.85 (0.62, 1.15) |
| Mexican American | 1.47 (0.87, 2.49) | 1.37 (0.80, 2.32) | 1.05 (0.75, 1.48) | 0.97 (0.68, 1.38) | 0.87 (0.70, 1.07) | 0.88 (0.71, 1.08) |
| Cuban | 1.12 (0.48, 2.61) | 1.43 (0.55, 3.71) | 0.96 (0.54, 1.70) | 1.03 (0.55, 1.91) | 0.96 (0.79, 1.18) | 1.03 (0.80, 1.31) |
| Central and South American | 0.39 (0.15, 1.02) | 0.56 (0.20, 1.58) | 0.24 (0.08, 0.68) | 0.37 (0.12, 1.11) | 0.60 (0.32, 1.14) | 0.87 (0.46, 1.64) |
| Other Hispanic | 0.73 (0.44, 1.21) | 0.85 (0.50, 1.45) | 0.78 (0.51, 1.19) | 0.81 (0.55, 1.21) | 0.78 (0.58, 1.04) | 0.80 (0.59, 1.07) |
| Non-Hispanic Black | 1.77 (1.37, 2.28) | 1.33 (1.02, 1.73) | 1.91 (1.66, 2.20) | 1.41 (1.22, 1.63) | 1.16 (1.05, 1.28) | 1.04 (0.95, 1.15) |
| Non-Hispanic other | 0.87 (0.50, 1.51) | 1.30 (0.72, 2.35) | 0.58 (0.41, 0.83) | 0.88 (0.61, 1.26) | 0.53 (0.40, 0.70) | 0.68 (0.51, 0.91) |
| Non-Hispanic White (Ref) | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Women | ||||||
| Puerto Rican | 1.48 (0.83, 2.63) | 1.21 (0.67, 2.19) | 1.25 (0.87, 1.79) | 1.09 (0.73, 1.63) | 0.49 (0.36, 0.68) | 0.61 (0.44, 0.86) |
| Mexican | 1.46 (1.03, 2.06) | 1.27 (0.76, 2.11) | 1.15 (0.75, 1.75) | 1.02 (0.64, 1.64) | 0.68 (0.54, 0.85) | 0.79 (0.63, 1.00) |
| Mexican American | 0.13 (0.05, 0.35) | 0.11 (0.04, 0.30) | 0.99 (0.68, 1.46) | 0.76 (0.52, 1.11) | 0.68 (0.53, 0.87) | 0.75 (0.58, 0.98) |
| Cuban | 0.90 (0.18, 4.43) | 0.73 (0.14, 3.85) | 0.62 (0.44, 0.87) | 0.58 (0.39, 0.87) | 0.85 (0.70, 1.03) | 1.08 (0.86, 1.36) |
| Central and South American | 0.36 (0.14, 0.93) | 0.34 (0.12, 0.93) | 0.50 (0.19, 1.34) | 0.65 (0.24, 1.79) | 0.48 (0.27, 0.87) | 0.64 (0.35, 1.18) |
| Other Hispanic | 0.87 (0.43, 1.77) | 0.75 (0.36, 1.56) | 0.86 (0.56, 1.33) | 0.78 (0.49, 1.24) | 0.95 (0.65, 1.40) | 1.17 (0.80, 1.71) |
| Non-Hispanic Black | 1.76 (1.36, 2.29) | 1.17 (0.89, 1.54) | 1.85 (1.58, 2.17) | 1.35 (1.15, 1.60) | 1.08 (0.99, 1.16) | 1.00 (0.92, 1.09) |
| Non-Hispanic other | 1.01 (0.56, 1.83) | 1.23 (0.68, 2.24) | 0.59 (0.37, 0.94) | 0.79 (0.49, 1.27) | 0.63 (0.48, 0.82) | 0.76 (0.57, 1.01) |
| Non-Hispanic White (Ref) | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
Note. AHR = adjusted hazard ratio; CHR = crude hazard ratio; CI = confidence interval. Unadjusted association is between race/ethnicity and all-cause mortality (CHR). Hazard ratios are adjusted for age, nativity status and length of stay in the United States, marital status, region, education, and family income (AHR).
The association between race/ethnicity and all-cause mortality differed with age group and nativity status for men (P < .001) and women (P < .001; Table 3). Hispanic men were as likely to die as non-Hispanic White men, regardless of subgroup, age group, and nativity status. For Hispanic women, US-born Mexican Americans aged 25 to 44 years had a 90% lower rate of dying, whereas island- or foreign-born Cubans and other Hispanics aged 45 to 64 years were at least 58% less likely to die than were their non-Hispanic White counterparts. Moreover, US-born Mexican American and island- or foreign-born Puerto Rican women aged 65 years and older exhibited at least a 25% lower rate of dying compared with their non-Hispanic White counterparts. US-born non-Hispanic Black men younger than 65 years had at least a 33% higher death rate, and island- or foreign-born other non-Hispanic men older than 45 years were at least 50% less likely to die than were their White counterparts. For non-Hispanic women, US-born Blacks aged 45–64 years had a 39% higher rate of dying, whereas island- or foreign-born other non-Hispanics aged 65 years and older were 40% less likely to die than were their White counterparts.
TABLE 3—
All-Cause Mortality for Gender and Race/Ethnicity Among US Adults Aged 25 Years or Older According to Nativity Status: National Health Interview Survey, United States, 1997–2004
| Aged 25–44 Years |
Aged 45–64 Years |
Aged ≥ 65 Years |
||||
| Race/Ethnicity | US-Born, AHR (95% CI) | Island- or Foreign-Born, AHR (95% CI) | US-Born, AHR (95% CI) | Island- or Foreign-Born, AHR (95% CI) | US-Born, AHR (95% CI) | Island- or Foreign-Born, AHR (95% CI) |
| Men | ||||||
| Puerto Rican | 1.04 (0.45, 2.42) | 0.94 (0.26, 3.35) | 0.68 (0.28, 1.68) | 1.96 (0.92, 4.17) | 1.67 (0.65, 4.25) | 1.23 (0.80, 1.89) |
| Mexican | 1.19 (0.49, 2.91) | 1.07 (0.43, 2.65) | 0.83 (0.49, 1.42) | 1.24 (0.68, 2.28) | 0.64 (0.36, 1.13) | 0.96 (0.63, 1.47) |
| Mexican American | 1.43 (0.82, 2.51) | 0.96 (0.26, 3.51) | 0.87 (0.60, 1.27) | 1.80 (0.70, 4.61) | 0.88 (0.70, 1.10) | 0.89 (0.49, 1.64) |
| Cuban | 2.96 (0.89, 9.87) | 0.82 (0.22, 3.12) | 0.00 (0.00, 0.01)a | 1.29 (0.66, 2.51) | 0.88 (0.39, 1.98) | 0.96 (0.68, 1.34) |
| Central and South American | 0.00 (0.00, 0.00)a | 0.50 (0.15, 1.63) | 0.00 (0.00, 0.01)a | 0.35 (0.12, 1.09) | 4.65 (0.68, 31.8) | 0.67 (0.34, 1.35) |
| Other Hispanic | 0.92 (0.48, 1.80) | 0.76 (0.25, 2.26) | 0.98 (0.64, 1.48) | 0.69 (0.28, 1.68) | 0.95 (0.68, 1.34) | 0.52 (0.27, 1.01) |
| Non-Hispanic Black | 1.33 (1.01, 1.75) | 0.97 (0.32, 2.88) | 1.41 (1.22, 1.64) | 0.95 (0.42, 2.15) | 1.04 (0.94, 1.15) | 1.00 (0.61, 1.65) |
| Non-Hispanic other | 1.55 (0.80, 3.03) | 0.74 (0.19, 2.84) | 1.15 (0.74, 1.77) | 0.49 (0.24, 0.99) | 0.86 (0.64, 1.17) | 0.45 (0.25, 0.82) |
| Non-Hispanic White (Ref) | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Women | ||||||
| Puerto Rican | 0.91 (0.39, 2.15) | 2.03 (0.71, 5.80) | 0.47 (0.15, 1.48) | 1.33 (0.78, 2.29) | 0.61 (0.18, 2.05) | 0.64 (0.44, 0.95) |
| Mexican | 1.04 (0.35, 3.14) | 1.13 (0.43, 3.00) | 0.63 (0.28, 1.39) | 0.70 (0.38, 1.30) | 0.69 (0.44, 1.10) | 0.89 (0.63, 1.26) |
| Mexican American | 0.10 (0.03, 0.31) | 0.16 (0.02, 1.37) | 0.79 (0.53, 1.18) | 0.52 (0.17, 1.61) | 0.74 (0.55, 0.98) | 0.81 (0.42, 1.53) |
| Cuban | 0.00 (0.00, 0.00)a | 0.89 (0.14, 5.64) | 0.00 (0.00, 0.00)a | 0.42 (0.23, 0.75) | 1.15 (0.40, 3.29) | 1.14 (0.83, 1.57) |
| Central and South American | 0.00 (0.00, 0.00)a | 0.34 (0.11, 1.09) | 0.00 (0.00, 0.01)a | 0.54 (0.20, 1.51) | 0.30 (0.03, 2.64) | 0.67 (0.34, 1.33) |
| Other Hispanic | 0.88 (0.32, 2.39) | 0.80 (0.25, 2.51) | 1.10 (0.67, 1.82) | 0.40 (0.17, 0.94) | 1.44 (0.94, 2.21) | 0.77 (0.45, 1.34) |
| Non-Hispanic Black | 1.21 (0.89, 1.63) | 0.37 (0.09, 1.55) | 1.39 (1.18, 1.64) | 0.70 (0.34, 1.43) | 1.01 (0.93, 1.10) | 0.70 (0.34, 1.41) |
| Non-Hispanic other | 1.12 (0.51, 2.43) | 0.98 (0.36, 2.69) | 1.01 (0.61, 1.69) | 0.42 (0.16, 1.09) | 0.93 (0.68, 1.27) | 0.60 (0.36, 0.98) |
| Non-Hispanic White (Ref) | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
Note. AHR = adjusted hazard ratio; CI = confidence interval. AHRs adjusted for age, marital status, region, education, and family income.
Estimates may be unreliable as a result of small sample size.
DISCUSSION
We observed the Hispanic paradox of lower mortality rates for Hispanic subgroups relative to non-Hispanic Whites for women only: Mexican Americans and Central and South Americans aged 25 to 44 years, Cubans aged 45 to 64 years, and Puerto Ricans and Mexican Americans aged 65 years and older. Moreover, we observed a lower mortality risk for some Hispanic women subgroups by nativity status: US-born Mexican Americans aged 25 to 44 years and 65 years and older, island- or foreign-born Cubans and other Hispanics aged 45 to 64 years, and island- or foreign-born Puerto Ricans aged 65 years and older. Among non-Hispanics, Black men younger than 65 years and women aged 45 to 64 years exhibited higher rates of dying than did their White counterparts. Furthermore, the increased mortality risk burden observed among non-Hispanic Black men and women seems to be carried by those born in the United States.
Although previous studies have consistently shown that Hispanics, despite their low socioeconomic position and low access to care,2–7,10,12,14 exhibit lower or similar death rates for all-cause mortality than do non-Hispanic Whites, the evidence for Hispanic subgroups is mixed. Specifically, these studies found that compared with non-Hispanic Whites, some Hispanic subgroups had lower death rates,2,3,5,6,10,12,14 whereas other Hispanic subgroups had similar or higher death rates.5,6,12,14 For example, using data from the National Longitudinal Study, some studies have reported lower death rates for Mexicans, Cubans, and other Hispanics aged 65 years and older,2,3 with 1 study controlling for age, education, and household income.3 Moreover, studies using NHIS-NDI–linked mortality data also reported lower death rates for Mexican Americans5,6,12,14 and other Hispanics5,6,12 aged 65 years and older.
Most recently, studies using NHIS-NDI–linked mortality data have suggested that the Hispanic mortality advantage may further depend on age and gender.6,12 For example, Bond Huie et al.,6 using data from the 1986–1994 NHIS-NDI–linked files, found that younger Mexican Americans exhibited higher mortality risk, whereas older Mexican Americans and other Hispanics exhibited lower risk compared with non-Hispanic Whites after controlling for selected individual and contextual characteristics. Moreover, Borrell and Crawford,12 using 1990–2000 NHIS-NDI–linked mortality data, found that compared with non-Hispanic Whites, Mexican American men aged 25 to 44 years had a higher risk of dying, whereas Mexican American men aged 65 years and older had lower death rates after controlling for age, nativity status and length of stay in the United States, marital status, region, education, and income. Among women, Puerto Ricans younger than 65 years of age and Mexican Americans aged 25 to 44 years had higher mortality rates than did non-Hispanic Whites. Mexican American and other Hispanic women aged 65 years and older exhibited lower death rates than did non-Hispanic White women.
Our findings, although consistent with past studies showing that Hispanics have lower mortality rates compared with non-Hispanic Whites, suggest that the paradox was true only for Hispanic women. Specifically, Mexican Americans and Central and South Americans aged 25 to 44 years, Cubans aged 45 to 64 years, and Puerto Ricans and Mexican Americans aged 65 years and older had lower mortality risk than did their non-Hispanic White counterparts. Finally, and consistent with Markides and Coriel's Hispanic paradox seminal work,1 our findings indicate that Hispanic subgroups, in general, exhibit better mortality outcomes than do non-Hispanic Blacks, with the exception of men aged 65 years and older, among whom Hispanic subgroups and non-Hispanic Blacks had similar outcomes relative to non-Hispanic Whites.
Previous studies showing a Hispanic mortality advantage relative to non-Hispanic Whites suggest that this advantage, in general, seems to be greater among foreign-born Hispanics,7,11,14–17 especially older Hispanics.7,12,14 However, few studies have presented these results for Hispanic subgroups.7,14 Hummer et al.,14 using NHIS-NDI–linked mortality files for 1989–1994 for Mexican Americans and other Hispanics, found a protective effect for foreign-born Mexican American and other Hispanic adults aged 45 years and older as well as US-born Mexican Americans aged 45 to 64 years relative to US-born non-Hispanic Whites. Moreover, Palloni and Arias,7 using data from NHIS-NDI–linked files for 1986–1994, found that foreign-born Mexican American and other Hispanic adults aged 35 years and older had lower death rates than did non-Hispanic Whites regardless of gender, although stronger associations were observed in men. They did not observe this advantage for foreign-born Puerto Ricans or Cubans.7
Unlike previous studies, our study found a mortality advantage for Hispanic women only. More specifically, island- or foreign-born Cuban and other Hispanic women aged 45 to 64 years and island- or foreign-born Puerto Rican women aged 65 years or older were less likely to die than were their non-Hispanic White counterparts. We also observed a protective effect for US-born Mexican American women aged 25 to 44 years and 65 years or older.
Although the inconsistency of our findings with previous studies7,14 might be explained in the context of an analytic approach, years used, and possible changes of characteristics of immigrants to and in the United States and the Hispanic population subgroups used during analyses, this discrepancy in findings suggests a need for the continuing examination of the Hispanic paradox on mortality risk across age, gender, subgroup, and nativity status. Specifically, these findings suggest that the paradox may not be a static phenomenon and may change with Hispanic population growth and increasing diversity over time.
Several studies have also found better health outcomes for foreign-born non-Hispanic Blacks than their US-born counterparts.14–17,34–37 Hummer et al. found decreased odds of dying for foreign-born non-Hispanic Blacks aged 45 years and older compared with their non-Hispanic White counterparts.14 However, the study also revealed increased odds of dying for US-born Blacks regardless of age, with an apparent decrease in the odds of dying with age: those aged 18 to 44 years exhibited greater odds than did those aged 45 to 64 years and 65 years and older. Consistent with these findings, we found a high risk of mortality among US non-Hispanic Black men younger than 65 years and women aged 45 to 64 years compared with their non-Hispanic White counterparts. These findings taken together underscored the importance of examining nativity status whenever the data are available for all racial/ethnic groups.
Among the strengths of our study was the use of 8 years of a large national sample using the same design methodology, the availability of information on Hispanic subgroups to allow data disaggregation reflecting the diversity of the Hispanic population, in-person interviews yielding a high response rate, the use of bilingual interviewers to increase the participation of Spanish-speaking respondents, and the use of sample weights adjusted for oversampling and nonresponse. An important limitation was the use of the public use version of the NHIS-NDI–linked mortality files. The NHIS-NDI–linked mortality public use files have small perturbations for a small, selected number of records compared with the restricted use files. In addition, the public use files replace the exact follow-up time with an approximate follow-up time. However, a comparison of the public and restricted use files suggests that the results obtained from these 2 data sets were very similar for all-cause and selected cause-specific mortality.27 Thus, it is unlikely that our results were affected by our use of the NHIS-NDI public use files. Another limitation was the use of a missing category for income instead of the imputed income files provided by NHIS. SUDAAN does not support the use of imputed data for survival analysis.32 However, as sensitivity analyses, we repeated the analyses with and without the imputed files with logistic regression, and the results were nearly identical. Thus, it is unlikely that the use of the imputed files would have changed our results.
Additionally, the exclusion of NHIS participants as the result of missing data may have affected our results. However, we compared selected characteristics for participants included (n = 214 792) and those excluded (n = 2125) and found that the latter were older and more likely to be non-Hispanic Whites, non-Hispanic Blacks, or Mexicans and less likely to be married (all P values < .05). Thus, given the association of these characteristics with mortality risk, it is unlikely that our results were affected by this exclusion. For example, individuals who are older and non-Hispanic Blacks have a higher risk of death, whereas Mexican Americans and married individuals have a lower risk of death. Moreover, individuals who agreed to participate in NHIS may be different from those who refuse or chose not to participate. This refusal may have under- or overestimated our results toward the null depending on whether the decision to participate was founded on race/ethnicity. Finally, whereas previous studies3,7 have found inconsistent support for the salmon bias,19 out-migration may have affected our results. We found a protective effect only for island- or foreign-born Puerto Rican and US-born Mexican American women aged 65 years and older. However, Puerto Rican deaths are tabulated in US mortality statistics regardless of where they are reported, whereas US-born Mexican Americans are less likely to return home because of social networks and family ties in the United States. Thus, it is unlikely that our results were affected by unhealthy out-migration.
Our findings of a lower mortality risk for women only in selected Hispanic subgroups according to age groups and nativity status suggest that the Hispanic paradox may not be a static process and may depend on this population's changes over time. The latter was underscored by the mortality advantage findings observed in subgroups other than Mexican Americans, Cubans, and other Hispanics and for specific characteristics such as age, gender, and nativity status. Our findings also suggest that nativity status may help us understand health disparities not only among Hispanics but also in other racial/ethnic groups. Future studies should not only examine trends in mortality according to age, gender, race/ethnicity, and nativity status but also present mortality rates according to these characteristics to facilitate future comparison.
Acknowledgments
The National Institute of Dental and Craniofacial Research (grant R03DE017901 to L. N. B.) supported this work.
Human Participant Protection
Data were collected with the informed consent of the National Health Interview Survey respondents. The institutional review board of the National Center for Health Statistics approved our procedures. In addition, the institutional review board at Lehman College, CUNY approved the secondary data analyses.
References
- 1.Markides KS, Coreil J. The health of Hispanics in the southwestern United States: an epidemiologic paradox. Public Health Rep. 1986;101(3):253–265 [PMC free article] [PubMed] [Google Scholar]
- 2.Sorlie PD, Backlund E, Johnson NJ, Rogot E. Mortality by Hispanic status in the United States. JAMA. 1993;270(20):2464–2468 [PubMed] [Google Scholar]
- 3.Abraido-Lanza AF, Dohrenwend BP, Ng-Mak DS, Turner JB. The Latino mortality paradox: a test of the “salmon bias” and healthy migrant hypotheses. Am J Public Health. 1999;89(10):1543–1548 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Liao Y, Cooper RS, Cao G, et al. Mortality patterns among adult Hispanics: findings from the NHIS, 1986 to 1990. Am J Public Health. 1998;88(2):227–232 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hummer RA, Rogers RG, Amir HS, Forbes D, Frisbie WP. Adult mortality differentials among Hispanic subgroups and non-Hispanic Whites. Soc Sci Q. 2000;81(1):459–476 [PubMed] [Google Scholar]
- 6.Bond Huie SA, Hummer RA, Rogers RG. Individual and contextual risks of death among race and ethnic groups in the United States. J Health Soc Behav. 2002;43(3):359–381 [PubMed] [Google Scholar]
- 7.Palloni A, Arias E. Paradox lost: explaining the Hispanic adult mortality advantage. Demography. 2004;41(3):385–415 [DOI] [PubMed] [Google Scholar]
- 8.Patel KV, Eschbach K, Ray LA, Markides KS. Evaluation of mortality data for older Mexican Americans: implications for the Hispanic paradox. Am J Epidemiol. 2004;159(7):707–715 [DOI] [PubMed] [Google Scholar]
- 9.Markides KS, Eschbach K. Aging, migration, and mortality: current status of research on the Hispanic paradox. J Gerontol B Psychol Sci Soc Sci. 2005;60(spec no 2):68–75 [DOI] [PubMed] [Google Scholar]
- 10.Turra CM, Goldman N. Socioeconomic differences in mortality among U.S. adults: insights into the Hispanic paradox. J Gerontol B Psychol Sci Soc Sci. 2007;62(3):S184–S192 [DOI] [PubMed] [Google Scholar]
- 11.Turra CM, Elo IT. The impact of salmon bias on the Hispanic mortality advantage: new evidence from social security data. Popul Res Policy Rev. 2008;27(5):515–530 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Borrell LN, Crawford ND. All-cause mortality among Hispanics in the United States: exploring heterogeneity by nativity status, country of origin, and race in the National Health Interview Survey–linked mortality files. Ann Epidemiol. 2009;19(5):336–343 [DOI] [PubMed] [Google Scholar]
- 13.Xu J, Kochanek KD, Murphy SL, Tejada-Vera B. Deaths: final data for 2007. Natl Vital Stat Rep. 2010;58(1):21. [PubMed] [Google Scholar]
- 14.Hummer RA, Rogers RG, Nam CB, LeClere FB. Race/ethnicity, nativity, and US adult mortality. Soc Sci Q. 1999;80:136–153 [Google Scholar]
- 15.Singh GK, Siahpush M. Ethnic-immigrant differentials in health behaviors, morbidity, and cause-specific mortality in the United States: an analysis of two national data bases. Hum Biol. 2002;74(1):83–109 [DOI] [PubMed] [Google Scholar]
- 16.Singh GK, Miller BA. Health, life expectancy, and mortality patterns among immigrant populations in the United States. Can J Public Health. 2004;95(3):I14–I21 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Singh GK, Hiatt RA. Trends and disparities in socioeconomic and behavioural characteristics, life expectancy, and cause-specific mortality of native-born and foreign-born populations in the United States, 1979–2003. Int J Epidemiol. 2006;35(4):903–919 [DOI] [PubMed] [Google Scholar]
- 18.Franzini L, Ribble JC, Keddie AM. Understanding the Hispanic paradox. Ethn Dis. 2001;11(3):496–518 [PubMed] [Google Scholar]
- 19.Pablos-Méndez A. Mortality among Hispanics. JAMA. 1994;271(16):1237. [PubMed] [Google Scholar]
- 20.Ramirez R, De la Cruz GP. The Hispanic Population in the United States: March 2002, Current Population Reports. Washington, DC: US Census Bureau; 2003 [Google Scholar]
- 21.US Department of Commerce, Economics, and Statistics Administration, US Census Bureau The American Community—Hispanics: 2004. American Community Survey Reports. ACS-03; 2007. Available at: http://www.census.gov/prod/2007pubs/acs-03.pdf. Accessed February 19, 2011 [Google Scholar]
- 22.National Center for Health Statistics, Centers for Disease Control and Prevention National Health Interview Survey. Available at: http://www.cdc.gov/nchs/nhis.htm. Accessed February 13, 2011
- 23.National Center for Health Statistics, Centers for Disease Control and Prevention NHIS Linked Mortality Public-Use Files. Available at: http://www.cdc.gov/nchs/data_access/data_linkage/mortality/nhis_linkage_public_use.htm. Accessed August 28, 2010
- 24.Botman SL, Moore TF, Moriarity CL, Parsons VL. Design and Estimation for the National Health Interview Survey, 1995–2004. Hyattsville, MD: National Center for Health Statistics; 2000 [Google Scholar]
- 25.National Center for Health Statistics National Health Interview Survey: research for the 1995–2004 redesign. Vital Health Stat 2. 1999;(126):1–119 [PubMed] [Google Scholar]
- 26.National Center for Health Statistics Office of Analysis and Epidemiology. The National Health Interview Survey (1986–2004) Linked Mortality Files: Mortality Follow-Up Through 2006: Matching Methodology; May 2009 Hyattsville, MD: Available at: http://www.cdc.gov/nchs/data/datalinkage/matching_methodology_nhis_final.pdf. Accessed August 28, 2010 [Google Scholar]
- 27.Data Linkage Team Comparative Analysis of the NHIS Public-Use and Restricted-Use Linked Mortality Files: 2010 Public-Use Data Release; March 2010. National Center for Health Statistics. Available at: http://www.cdc.gov/NCHS/data/datalinkage/nhis_mort_compare_2010_final.pdf. Accessed August 28, 2010
- 28.Borrell LN, Crawford ND. Disparities in self-reported hypertension in Hispanic subgroups, non-Hispanic black and non-Hispanic white adults: The National Health Interview Survey. Ann Epidemiol. 2008;18(10):803–812 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Borrell LN, Crawford ND, Dallo FJ, Baquero MC. Self-reported diabetes in Hispanic subgroup, non-Hispanic Black, and non-Hispanic White populations: National Health Interview Survey, 1997–2005. Public Health Rep. 2009;124(5):702–710 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.US Department of Labor, Bureau of Labor Statistics Consumer Price Indexes. CPI Inflation Calculator. Available at: http://www.bls.gov/bls/inflation.htm. Accessed July 24, 2007
- 31.SAS /STAT User's Guide, Version 9.2. Cary, NC: SAS Institute Inc; 2009 [Google Scholar]
- 32.Research Triangle Institute SUDAAN Language Manual, Release 10.0. Research Triangle Park, NC; 2008 [Google Scholar]
- 33.National Center for Health Statistics, Centers for Disease Control and Prevention NHIS Linked Mortality Files. Analytic Support-Analytic Guidelines. Available at: http://www.cdc.gov/nchs/data_access/data_linkage/mortality/nnhs_linkage.htm. Accessed January 22, 2011
- 34.Dey AN, Lucas JW. Physical and mental health characteristics of U.S.- and foreign-born adults: United States, 1998–2003. Adv Data. 2006; 369:1–19 [PubMed] [Google Scholar]
- 35.Hicks LS, Fairchild DG, Cook EF, Ayanian JZ. Association of region of residence and immigrant status with hypertension, renal failure, cardiovascular disease, and stroke, among African-American participants in the Third National Health and Nutrition Examination Survey (NHANES III). Ethn Dis. 2003;13(3):316–323 [PubMed] [Google Scholar]
- 36.Arthur CM, Katkin ES. Making a case for the examination of ethnicity of Blacks in United States Health Research. J Health Care Poor Underserved. 2006;17(1):25–36 [DOI] [PubMed] [Google Scholar]
- 37.Read JG, Emerson MO, Tarlov A. Implications of Black immigrant health for U.S. racial disparities in health. J Immigr Health. 2005;7(3):205–212 [DOI] [PubMed] [Google Scholar]