- Email: [email protected]
Trends in the Mexican infant mortality paradox over the past two decades
Annals of Epidemiology 24 (2014) 831e836
Contents lists available at ScienceDirect
Annals of Epidemiology journal homepage: www.annalsofepidemiology.org
Original article
Trends in the Mexican infant mortality paradox over the past two decades Abdulrahman M. El-Sayed MD, DPhil a, b, *, Magdalena M. Paczkowski PhD a, Dana March PhD a, Sandro Galea MD, DrPH a a b
Department of Epidemiology, Columbia University, New York, NY College of Physicians and Surgeons, Columbia University, New York, NY
a r t i c l e i n f o
a b s t r a c t
Article history: Received 15 October 2013 Accepted 11 September 2014 Available online 18 September 2014
Purpose: Mexicans in the United States have lower rates of several important population health metrics than non-Hispanic whites, including infant mortality. This mortality advantage is particularly pronounced among infants born to foreign-born Mexican mothers. However, the literature to date has been relegated to point-in-time studies that preclude a dynamic understanding of ethnic and nativity differences in infant mortality among Mexicans and non-Hispanic whites. Methods: We assessed secular trends in the relation between Mexican ethnicity, maternal nativity, and infant mortality between 1989 and 2006 using a linked birthedeath data set from one US state. Results: Congruent to previous research, we found a significant mortality advantage among infants of Mexican relative to non-Hispanic white mothers between 1989 and 1991 after adjustment for baseline demographic differences (relative risk ¼ 0.78, 95% confidence interval, 0.62e0.98). However, because of an upward trend in infant mortality among infants of Mexican mothers, the risk of infant mortality was not significantly different from non-Hispanic white mothers in later periods. Conclusions: Our findings suggest that the “Mexican paradox” with respect to infant mortality is resolving. Changing sociocultural norms among Mexican mothers and changes in immigrant selection and immigration processes may explain these observations, suggesting directions for future research. Ó 2014 Elsevier Inc. All rights reserved.
Keywords: Mexican paradox Population health trends Infant mortality Ethnicity Disparities
Introduction There has been substantial interest in racial and ethnic differences in health outcomes, such as life expectancy, all-cause and cause-specific mortalities [1,2], and morbidity due to high-burden diseases [3], in the United States over the past two decades. Investigations into the mechanisms underlying these differences in health have demonstrated them to be, in part, driven by differences in access to social and economic resources, such as education, income, wealth, and health services [4,5] between minorities and their counterpartsda manifestation of structural, institutional, and individual marginalization among these groups [6e9]. Mexicans in the United States, as well as other Hispanic groups, generally have lower socioeconomic position relative to non-Hispanic whites, with higher poverty [10], lower income [11], and lower educational attainment compared with their non-Hispanic white counterparts [12]. Yet despite this, their health profile has been * Corresponding author. Department of Epidemiology, Columbia University, 722 W. 168th Street, Rm 505, New York, NY 10032. E-mail address: [email protected] (A.M. El-Sayed). http://dx.doi.org/10.1016/j.annepidem.2014.09.005 1047-2797/Ó 2014 Elsevier Inc. All rights reserved.
similar, if not better, than that of non-Hispanic whites along a number of important health metrics. This “paradox” has been observed in life expectancy and mortality [13e15], health-risk behaviors [16], preterm birth and low birth weight [17e19], and infant mortality [20]. Two principal explanations, among several, have been proposed to explain this paradox. One explanation is that the paradox may be a result of different social and cultural resources and mores among Mexicans relative to non-Hispanic whites. Cultural resources such as collectivism and reciprocal obligation, and dietary, exercise, and childcare norms may operate to mitigate the adverse health consequences of socioeconomic marginalization in this group [21,22]. Evidence supporting this explanation comes from the observation that the paradox is substantially stronger among immigrants, who are more likely to adhere to these cultural and social mores, and that the ethnic advantage dissipates with each subsequent US-born generation [20]. Observations suggesting that residence in ethnically dense neighborhoods may reinforce protective social norms, thereby improving health among immigrants, is also supportive of this explanation [23,24]. An alternative explanation may be that the apparent health advantage of immigrants in the United States is an artifact of
832
A.M. El-Sayed et al. / Annals of Epidemiology 24 (2014) 831e836
population selection for individuals who are healthier at baseline. This selection may occur via two mechanisms: first, immigration itself selects for a healthier population at baseline (coined the “healthy migrant” hypothesis), and second, those in poor health are more likely to return to Mexico, leaving only the healthiest Mexicans behind in the United States (coined the “salmon-bias” hypothesis) [25,26]. This selection argument is supported by the observation that other ethnic immigrant groups, such as Eastern Europeans and Arab Americans, who may not benefit from ostensibly protective Mexican cultural mores, also appear to have health metrics that outperform their socioeconomic profiles across diverse contexts, including the United States [27e29], Canada [30], and Europe [31e33]. Others have attempted to explain the paradox via misclassification of ethnicity in death certificates [26,34]. Although the literature about the “Mexican paradox” is well developed, research in this area has been largely confined to static cross-sectional analyses that have documented this paradox at single points in time. Yet as both cultural norms and immigration norms may change with time, a dynamic examination of the paradox may yield important insight into its etiology and future trajectory. Therefore, we explored secular trends in the relation between Mexican ethnicity and infant mortality over the past two decades, aiming both to illuminate the etiology of the paradox and to document its trajectory over time. Methods
them. Information about race and ethnicity was collected for all mothers. We created three categories of race/ethnicity, Mexican, other Hispanic, and non-Hispanic white to whom our sample was restricted. Our sample consisted of 1,753,746 non-Hispanic white, 70,678 Mexican, and 29,819 other Hispanic infants; this accounted for 74.3% of the original sample of all singleton births occurring between 1989 and 2006 in the State of Michigan. We defined infant mortality as death of the infant from the day of birth through the first year after birth, and we calculated the total number of deaths among all infants aged less than 1 year during the study period. Data were then grouped into six periods for analysis: 1989 to 1991, 1992 to 1994, 1995 to 1997, 1998 to 2000, 2001 to 2003, and 2004 to 2006. Other available information included data about maternal age, education, marital status, place of birth, parity, prenatal care, and payment source for labor and delivery. Age and education were analyzed in three categories each: less than 20, 20 and 35, and greater than 35, and less than high school diploma, high school diploma and/or some college, or college degree or greater, respectively. Marital status was dichotomized and analyzed as married or not at the time of parturition. Maternal place of birth was divided into two categories, foreign (including birth anywhere outside of the 48 continental states, D.C., Hawaii, or Alaska) or United States. The Kotelchuck Index was used to assess prenatal care and was analyzed dichotomously as adequate versus other [35]. Payment source for labor and delivery was analyzed as a dichotomous variable: private insurance versus Medicaid, self-pay, or other.
Data Analysis The data used for this study were collected by the Michigan Department Community of Health and consisted of all singleton births that occurred between January 1, 1989, and December 31, 2006, in the State of Michigan. Data about births were then crosslinked to all deaths before one completed year of age among
We calculated descriptive statistics, stratified by time interval, for the study sample. To assess trends in key measures over time, we used the CochraneArmitage test for trend. We used P values to assess significance (a ¼ 0.05, two tailed) for all trend tests. We then
Table 1 Demographic characteristics and infant mortality by study period among all singleton live births in Michigan, 1989 to 2006 Characteristics
Age (y) 19 20e34 35þ Education Less than high school High school, some college College or greater Married* Foreign born Race Non-Hispanic White Hispanic Hispanic race White Black Hispanic ethnicity Mexican Other Private source of payment Prenatal carey Inadequate Intermediate Adequate Adequate plus Parity Infant died before first year * y
Period 1989e1991 (N ¼ 316,900), n, %
1992e1994 (N ¼ 294,325), n, %
1995e1997 (N ¼ 285,706), n, %
1998e2000 (N ¼ 283,064), n, %
2001e2003 (N ¼ 278,787), n, %
2004e2006 (N ¼ 268,571), n, %
X2CAP
31,169 (9.8) 259,949 (82) 25,782 (8.1)
28,194 (9.6) 236,470 (80.3) 29,661 (10.1)
28,089 (9.8) 223,946 (78.4) 33,671 (11.8)
26,396 (9.3) 220,964 (78.1) 35,704 (12.6)
23,341 (8.4) 217,976 (78.2) 37,470 (13.4)
21,423 (8) 210,182 (78.3) 36,966 (13.8)
<.0001 <.0001 <.0001
49,643 209,096 58,161 271,992 13,641
44,094 187,263 62,968 233,215 14,340
41,136 173,905 70,665 215,995 16,788
41,998 164,937 76,129 210,815 16,726
42,249 156,714 79,824 203,297 27,282
40,590 146,253 81,728 188,935 28,145
(15.1) (54.5) (30.4) (70.3) (10.5)
<.0001 .0001 <.0001 <.0001 <.0001
(15.7) (66) (18.4) (85.8) (4.2)
(15) (63.6) (21.4) (79.2) (4.9)
(14.4) (60.9) (24.7) (75.6) (5.9)
(14.8) (58.3) (26.9) (74.5) (5.9)
(15.2) (56.2) (28.6) (72.9) (9.8)
305,318 (96.3) 11,582 (3.7)
282,724 (96.1) 11,601 (3.9)
272,235 (95.3) 13,471 (4.7)
275,260 (93.9) 17,734 (6.1)
258,796 (92.8) 19,991 (7.2)
246,619 (91.8) 21,952 (8.2)
<.0001 <.0001
11,396 (98.4) 186 (1.6)
11,462 (98.8) 139 (1.2)
13,330 (99.0) 141 (1.1)
16,559 (99.0) 167 (1.0)
19,794 (99.0) 197 (1.0)
21,729 (99.0) 223 (1.0)
<.0001 <.0001
6805 (58.75) 4777 (41.25 229,794 (72.5)
7337 (63.2) 4264 (36.8) 206,588 (70.2)
9519 (70.7) 3952 (29.3) 206,683 (72.3)
12,923 (77.3) 3803 (22.7) 211,412 (74.7)
16,233 (81.2) 3758 (18.8) 194,045 (69.6)
17,861 (81.4) 4091 (18.6) 172,551 (64.2)
<.0001 <.0001 <.0001
29,785 42,511 166,974 77,630 1.39 2000
23,234 34,151 153,747 83,193 1.41 1577
20,947 32,431 145,200 87,128 1.39 1358
22,106 30,983 139,831 90,144 1.37 1275
21,056 29,846 135,438 92,447 1.38 1304
19,517 32,245 136,190 80,619 1.40 1192
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001
(9.4) (13.4) (52.7) (24.5) (1.45) (0.6)
At time of parturition. Per the Kotelchuck adequacy of prenatal care index.
(7.9) (11.6) (52.2) (28.3) (1.48) (0.5)
(7.3) (11.4) (50.8) (30.5) (1.49) (0.5)
(7.8) (10.9) (49.4) (31.8) (1.48) (0.5)
(7.6) (10.7) (48.6) (33.2) (1.51) (0.5)
(7.3) (12) (50.7) (30) (1.53) (0.4)
A.M. El-Sayed et al. / Annals of Epidemiology 24 (2014) 831e836 Table 2 Rate of infant mortality per 1000 singleton live births among Mexican infants overall and stratified by maternal nativity as well as among non-Hispanic white infants in Michigan, 1989 to 2006 Period
Mexican
1989e1991 1992e1994 1995e1997 1998e2000 2001e2003 2004e2006
Non-Hispanic White
Foreign born
US born
Overall
5.5 3.4 3.5 3.8 4.8 4.9
6.4 5.4 4.8 4.5 4.7 4.4
6.8 6.5 5.2 5.0 5.7 5.5
6.3 5.3 4.7 4.5 4.6 4.3
calculated infant mortality rates, stratified by ethnicity and nativity, over time. Next, we used bivariate and multivariable Poisson regression models, adjusted for age, education, marital status, parity, prenatal care, and payment source for labor and delivery, to assess the relation between Mexican ethnicity and infant mortality relative to non-Hispanic whites in each time interval. Results Table 1 shows descriptive statistics over time among our sample. The incidence of infant mortality overall declined from 6.3 per 1000 live singleton births to 4.4 per 1000 throughout our study period. This mirrors a US-wide decline in infant mortality from 9.8 per 1000 live births in 1989 to 6.7 per 1000 live births in 2006 [36,37]. Both the proportion of foreign-born and Mexican mothers increased with time. Overall, the distribution of maternal age and the proportion of mothers achieving at least a college degree increased with time throughout our study period. The proportion of infants born to mothers who were married at parturition declined, as did the parity of mothers. All tests for trend were statistically significant (P < .01) across our study period. Table 2 and Figure 1 shows the incidence of infant mortality stratified by ethnicity among Mexican and non-Hispanic white infants. Infant mortality declined among infants of both ethnic groups with time. However, disparities in infant mortality increased: the absolute difference in infant mortality was 0.5 per 1000 singleton live births among infants born to Mexican relative to non-Hispanic white mothers in 1989 to 1991, increasing to 1.2 per 1000 singleton live births by the end of the study period. Moreover, between 1998e2000 and 2001e2003, the incidence of infant mortality among infants born to Mexican mothers increased slightly despite a steady decline among those born to non-Hispanic whites throughout the study period. The incidence of infant mortality stratified by ethnicity and maternal birthplace (among Mexican mothers) is shown in Table 2
833
and Figure 2. US-born Mexican mothers had the highest incidence of infant mortality in all time intervals although risk decreased throughout the study period among this group from 7.2 to 6.2 per 1000 singleton live births. At the beginning of the study period, risk for infant mortality among non-Hispanic whites was higher than among foreign-born Mexican mothers. However, after 2000, risk for infant mortality among foreign-born Mexican mothers overtook that of non-Hispanic white mothers. Risk for infant mortality among non-Hispanic white mothers decreased consistently throughout the study period from 6.3 to 4.4 per 1000 singleton live births, whereas risk among foreign-born Mexican mothers decreased between 1989-1991 and 1992-1994, and then increased steadily between 1992-1994 and the end of the study period. Figure 3 shows log relative risk for infant mortality among Mexican mothers relative to non-Hispanic white mothers as well as relative risk and 95% confidence intervals in text, from models adjusted for age, education, marital status, source of payment for labor and delivery, prenatal care, and parity. Risk for infant mortality among Mexicans increased relative to non-Hispanic whites. In 1989 to 1991, Mexican mothers had lower risk for infant mortality (relative risk ¼ 0.78, 95% confidence interval, 0.62e0.98) than their non-Hispanic white counterparts. However, by 2004 to 2006, relative risk among Mexican mothers relative to non-Hispanic whites had increased to 1.01 (95% confidence interval, 0.83e1.22).
Discussion We studied over 1.7 million infants born between 1989 and 2006 in one US state. We found increases in infant mortality among foreign-born Mexican infants between 1992e1994 and 2004e2006 and among US-born Mexican infants between 1998-2000 and 2004-2006. The overall mortality advantage favoring Mexican infants we observed at the outset of our study period dissipated as a result of these increases during our study period. Our findings are consistent with prior observations demonstrating that infants of foreign-born Mexican mothers had lower risk of infant mortality than US-born counterparts [38] as well as studies demonstrating an ethnic advantage relative to nonHispanic whites [20]. However, this is the first study, of which we are aware, that has directly considered secular trends in the Mexican infant mortality paradox. There are two plausible explanations for the changes we observed. First, if Mexican cultural mores are protective against infant mortality, it is plausible that the mechanism through which they influence this outcome includes access to sociocultural resources like communalism and shared obligation that are reinforced via social norms within Mexican enclaves in which a large
Fig. 1. Risk of infant mortality per 1000 singleton live births to Mexican and non-Hispanic white mothers stratified by ethnicity in Michigan, 1989 to 2006.
834
A.M. El-Sayed et al. / Annals of Epidemiology 24 (2014) 831e836
Fig. 2. Risk of infant mortality per 1000 singleton live births to US-born non-Hispanic white mothers as well as Mexican mothers stratified by maternal nativity in Michigan, 1989 to 2006.
proportion of US Mexicans live. Ethnic enclaves are characterized by the cultural distinction they maintain from surrounding areas [39]. Residence within such enclaves may then serve to maintain and insulate protective social norms [40], which may ultimately mediate the mortality advantage among infants of foreign-born Mexican mothers that we observed. In support of this hypothesis, for example, Cagney et al. [23] demonstrated that respiratory morbidity among foreign-born Hispanics was inversely associated with the proportion of foreign-born Hispanics in their residential communities. The dynamics of Mexican enclaves are changing along with the US Mexican population more broadly as US-born Mexicans grow as a proportion of the US Mexican population overall [41]. Between 2000 and 2010, for example, births to native Mexicans outpaced immigration as the largest driver of Mexican population growth for the first time in US history [41]. US-born Mexicans are, by definition, more acculturated and less likely to adhere to traditional cultural norms and mores than their immigrant counterparts. It is plausible, then, that as the proportion of US-born, more acculturated Mexicans has increased in enclaves, social norms in these contexts have shifted away from the more traditional norms that may have, in part, mediated the mortality advantage among infants of foreign-born Mexican
mothers. This proportional increase in the US-born Mexican population in enclaves may then explain the dissipation of the mortality advantage among infants of foreign-born Mexican mothers observed here. There is evidence to indirectly support this potential explanation. For example, one study of birth weight, a leading determinant of infant mortality, among infants born to US- and foreign-born Mexican-American women found that residence in ethnic enclaves (i.e., enclaves of US-born Mexican Americans) was associated with lower birth weight among infants born to US-born women; controlling for residence in immigrant enclaves (i.e., enclaves of foreign-born Mexican Americans) strengthened the association. However, residence in immigrant enclaves was positively associated with birth weight, after controlling for residence in ethnic enclaves [42]. A second plausible explanation supports the “healthy migrant” hypothesis (i.e., the immigration process selects for a healthier population at baseline, and that the better baseline health of this population then explains the mortality advantage of foreign-born Mexicans [25,26]). The relative health of the immigrant population may vary with time. In fact, the healthy migrant hypothesis suggests that the propensity to immigrate is a function, directly or indirectly, of baseline health status. In that respect, it is also
Fig. 3. Adjusted* log relative risk of infant mortality among singleton live births to Mexican mothers relative to non-Hispanic White mothers in Michigan, 1989e2006. *Multivariable models adjusted for age, education, marital status, source of payment for labor and delivery, prenatal care, and parity. CI, confidence interval; RR, relative risk.
A.M. El-Sayed et al. / Annals of Epidemiology 24 (2014) 831e836
plausible that baseline health may also predict the time-toimmigration among immigrants, such that the healthiest may be more likely to immigrate before their less healthy counterparts. It would follow, then, that the baseline health of immigrant Mexicans should have declined with time as the healthiest Mexicans immigrated before their less healthy counterparts. Therefore, our observation of a deteriorating infant mortality advantage among infants of foreign-born Mexican mothers may be explained by a decline in the baseline health of immigrant Mexican mothers resulting from the dynamics of immigrant selection. Our findings do not favor either of these competing explanations regarding the etiology of the Mexican infant mortality paradox over the other. The first explanation supports the hypothesis that the paradox may result from protective social and cultural mores among immigrants, whereas the second supports the selection hypothesis. In that regard, it remains unclear as to which, if either, is a better explanation for the Mexican birth paradox more generally. Our study has several limitations. First, our covariate set was limited. In particular, data about maternal anthropometric characteristics and/or previous health complaints were unavailable. However, these factors are more likely to mediate than to confound the relations of interest here, precluding adjustment in our multivariable models. Second, we are limited by the validity of birth certificate data in vital registry files. Although it has been demonstrated that demographic data are accurate [43e45], questions have been raised about the accuracy of ethnicity reports in these files [25,26]. However, trend analyses of this sort are beholden to public registry files for comprehensive data over long time horizons, and therefore, these limitations were inevitable. Despite these challenges, our findings have several important implications for future research regarding the health of ethnic minorities in the United States. Our findings suggest that the relative health advantage that immigrant Mexicans in the United States have enjoyed for some time may be dissipating. Explorations into the temporal dynamics of the Mexican birth paradox in other United States contexts are needed to compare with the findings we have presented here. Similar health “paradoxes” have been documented among other ethnic minority groups [29,46]. Investigating secular trends in the infant mortality outcomes of these groups would improve our understanding of generalizable trends in immigrant health across ethnic identities. Future work may fruitfully consider the influence of US-born Mexicans on health norms among foreign-born residents of Mexican enclaves and trends in the baseline health of immigrant Mexicans over time.
Acknowledgments The authors thank Glenn Copeland and Glenn Radford from the Michigan Department of Community Health for their health in acquiring the data.
References [1] El-Sayed AM, Tracy M, Scarborough P, Galea S. Ethnic inequalities in mortality: the case of Arab-Americans. PloS One 2011;6(12):e29185. [2] Sudano JJ, Baker DW. Explaining US racial/ethnic disparities in health declines and mortality in late middle age: The roles of socioeconomic status, health behaviors, and health insurance. Soc Sci Med 2006;62(4):909. [3] Cooper R, Cutler J, Desvigne-Nickens P, Fortmann SP, Friedman L, Havlik R, et al. Trends and disparities in coronary heart disease, stroke, and other cardiovascular diseases in the United States: findings of the national conference on cardiovascular disease prevention. Circulation 2000;102(25):3137e47. [4] Ashton CM, Haidet P, Paterniti DA, Collins TC, Gordon HS, O’Malley K, et al. Racial and ethnic disparities in the use of health services. J Gen Intern Med 2003;18(2):146e52.
835
[5] Betancourt JR, Maina AW. The institute of medicine report “unequal treatment”: implications for academic health centers. Mt Sinai J Med 2004; 71(5):314. [6] Williams DR, Collins C. Racial residential segregation: a fundamental cause of racial disparities in health. Public Health Rep 2001;116(5):404. [7] Williams DR. Racial/ethnic variations in women’s health: the social embeddedness of health. Am J Public Health 2002;92(4):588e97. [8] Williams DR, Neighbors HW, Jackson JS. Racial/ethnic discrimination and health: findings from community studies. Am J Public Health 2003;93(2):200e8. [9] Nazroo JY, Williams DR. The social determination of ethnic/racial inequalities in health. Social determinants of health. 2nd ed. Oxford, United Kingdom: Oxford University Press; 2006. [10] Lopez MH, Cohn D. Hispanic poverty rate highest in new supplemental measure. Washington, DC: Pew Hispanic Center; 2011. [11] U.S. Census Bureau. Historical income tables: income inequality. Available at: http://www.census.gov/hhes/www/income/data/historical/inequality/. Accessed December 2012. [12] Lopez MH, Center PH. Latinos and education: explaining the attainment gap. Washington, DC: Pew Hispanic Center; 2009. [13] Minino AM, Murphy SL. Death in the United States, 2010. NCHS Data Brief 2012;(99):1e8. [14] Rosenwaike I. Mortality differentials among persons born in Cuba, Mexico, and Puerto Rico residing in the United States, 1979-81. Am J Public Health 1987;77(5):603e6. [15] Elo IT, Turra CM, Kestenbaum B, Ferguson BR. Mortality among elderly Hispanics in the United States: past evidence and new results. Demography 2004;41(1):109e28. [16] Abraido-Lanza AF, Chao MT, Florez KR. Do healthy behaviors decline with greater acculturation? Implications for the Latino mortality paradox. Soc Sci Med 2005;61(6):1243e55. [17] Fuentes-Afflick E, Lurie P. Low birth weight and Latino ethnicity: examining the epidemiologic paradox. Arch Pediatr Adolesc Med 1997;151(7):665. [18] Hessol NA, Fuentes-Afflick E. The perinatal advantage of Mexican-origin Latina women. Ann Epidemiol 2000;10(8):516. [19] Chung JH, Boscardin WJ, Garite TJ, Lagrew DC, Porto M. Ethnic differences in birth weight by gestational age: at least a partial explanation for the Hispanic epidemiologic paradox? Obstet Gynecol 2003;189(4):1058e62. [20] Hummer RA, Powers DA, Pullum SG, Gossman GL, Frisbie WP. Paradox found (again): infant mortality among the Mexican-origin population in the United States. Demography 2007;44(3):441e57. [21] Abraido-Lanza AF. The Latino mortality paradox: a test of the “salmon bias” and healthy migrant hypotheses. Am J Public Health 1999;89(10):1543. [22] Barney C, Bulatao RA, Anderson NB, editors. Critical perspectives on racial and ethnic differences in health in late life. Washington, DC: National Academies Press; 2004. [23] Cagney KA, Browning CR, Wallace DM. The Latino paradox in neighborhood context: the case of asthma and other respiratory conditions. Am J Public Health 2007;97(5):919e26. [24] El-Sayed AM, Galea S. Community context, acculturation and low-birthweight risk among Arab Americans: evidence from the Arab-American birth-outcomes study. J Epidemiol Community Health 2010;64(2):155e60. [25] Palloni A, Morenoff JD. Interpreting the paradoxical in the Hispanic paradox. Ann N Y Acad Sci 2001;954(1):140e74. [26] Palloni A, Arias E. Paradox lost: explaining the Hispanic adult mortality advantage. Demography 2004;41(3):385e415. [27] Janevic T, Savitz D, Janevic M. Maternal education and adverse birth outcomes among immigrant women to the United States from eastern Europe: a test of the healthy migrant hypothesis. Soc Sci Med 2011;73(3):429e35. [28] El-Sayed A, Galea S. Explaining the low risk of preterm birth among Arab Americans in the United States: an analysis of 617451 births. Pediatrics 2009;123(3):e438. [29] Finkton DW, El-Sayed AM, Galea S. Infant mortality among Arab-Americans: findings from the Arab-American birth outcomes study. Matern Child Health J 2013;17(4):732e45. [30] Urquia M, Frank J, Moineddin R, Glazier R. Immigrants’ duration of residence and adverse birth outcomes: a population-based study. BJOG: An Int J Obstetrics Gynaecol 2010;117(5):591e601. [31] Khlat M, Darmon N. Is there a Mediterranean migrants mortality paradox in Europe? Int J Epidemiol 2003;32(6):1115e8. [32] Stirbu I, Kunst AE, Vlems FA, Visser O, Bos V, Deville W, et al. Cancer mortality rates among first and second generation migrants in the Netherlands: convergence toward the rates of the native Dutch population. Int J Cancer 2006;119(11):2665e72. [33] Uitenbroek DG, Verhoeff AP. Life expectancy and mortality differences between migrant groups living in Amsterdam, the Netherlands. Soc Sci Med 2002;54(9):1379e88. [34] Lariscy JT. Differential record linkage by Hispanic ethnicity and age in linked mortality studies: implications for the epidemiologic paradox. J Aging Health 2011;23(8):1263e84. [35] Kotelchuck M. An evaluation of the Kessner Adequacy of Prenatal Care Index and a proposed adequacy of prenatal care utilization index. Am J Public Health 1994;84(9):1414e20. [36] Centers for Disease Control and Prevention (CDC). Infant mortalityeUnited States, 1990. MMWR Morb Mortal Wkly Rep 1993;42(9):161e5.
836
A.M. El-Sayed et al. / Annals of Epidemiology 24 (2014) 831e836
[37] MacDorman MF, Mathews TJ. Infant deathseUnited States, 2000-2007. MMWR Surveill Summ 2011;60(1):49e51. [38] Hummer RA, Biegler M, De Turk PB, Forbes D, Frisbie WP, Hong Y, et al. Race/ ethnicity, nativity, and infant mortality in the United States. Social Forces 1999;77(3):1083e117. [39] Logan JR, Alba RD, Zhang W. Immigrant enclaves and ethnic enclaves in New York and Los Angeles. Am Sociol Rev 2002;67(2):299e322. [40] Portes A, Rumbaut RG. Immigrant America: a portrait. Berkeley, CA: University of California Press; 2006. [41] Pew Research Center. The Mexican-American boom: births overtake immigration. Available at: http://www.pewhispanic.org/2011/07/14/the-mexicanamerican-boom-brbirths-overtake-immigration/. Updated 2012. Accessed December 12, 2012.
[42] Osypuk TL, Bates LM, Acevedo-Garcia D. Another Mexican birthweight paradox? The role of residential enclaves and neighborhood poverty in the birthweight of Mexican-origin infants. Soc Sci Med 2010;70(4):550e60. [43] Buescher PA. The quality of the new birth certificate data: a validation study in North Carolina. Am J Public Health 1993;83(8):1163. [44] Northam S. The reliability and validity of birth certificates. J Obstet Gynecol neonatal Nurs 2006;35(1):3. [45] Reichman NE, Hade EM. Validation of birth certificate data. A study of women in New Jersey’s HealthStart program. Ann Epidemiol 2001; 11(3):186. [46] Gould JB, Madan A, Qin C, Chavez G. Perinatal outcomes in two dissimilar immigrant populations in the United States: a dual epidemiologic paradox. Pediatrics 2003;111:e676e82.
Contents lists available at ScienceDirect
Annals of Epidemiology journal homepage: www.annalsofepidemiology.org
Original article
Trends in the Mexican infant mortality paradox over the past two decades Abdulrahman M. El-Sayed MD, DPhil a, b, *, Magdalena M. Paczkowski PhD a, Dana March PhD a, Sandro Galea MD, DrPH a a b
Department of Epidemiology, Columbia University, New York, NY College of Physicians and Surgeons, Columbia University, New York, NY
a r t i c l e i n f o
a b s t r a c t
Article history: Received 15 October 2013 Accepted 11 September 2014 Available online 18 September 2014
Purpose: Mexicans in the United States have lower rates of several important population health metrics than non-Hispanic whites, including infant mortality. This mortality advantage is particularly pronounced among infants born to foreign-born Mexican mothers. However, the literature to date has been relegated to point-in-time studies that preclude a dynamic understanding of ethnic and nativity differences in infant mortality among Mexicans and non-Hispanic whites. Methods: We assessed secular trends in the relation between Mexican ethnicity, maternal nativity, and infant mortality between 1989 and 2006 using a linked birthedeath data set from one US state. Results: Congruent to previous research, we found a significant mortality advantage among infants of Mexican relative to non-Hispanic white mothers between 1989 and 1991 after adjustment for baseline demographic differences (relative risk ¼ 0.78, 95% confidence interval, 0.62e0.98). However, because of an upward trend in infant mortality among infants of Mexican mothers, the risk of infant mortality was not significantly different from non-Hispanic white mothers in later periods. Conclusions: Our findings suggest that the “Mexican paradox” with respect to infant mortality is resolving. Changing sociocultural norms among Mexican mothers and changes in immigrant selection and immigration processes may explain these observations, suggesting directions for future research. Ó 2014 Elsevier Inc. All rights reserved.
Keywords: Mexican paradox Population health trends Infant mortality Ethnicity Disparities
Introduction There has been substantial interest in racial and ethnic differences in health outcomes, such as life expectancy, all-cause and cause-specific mortalities [1,2], and morbidity due to high-burden diseases [3], in the United States over the past two decades. Investigations into the mechanisms underlying these differences in health have demonstrated them to be, in part, driven by differences in access to social and economic resources, such as education, income, wealth, and health services [4,5] between minorities and their counterpartsda manifestation of structural, institutional, and individual marginalization among these groups [6e9]. Mexicans in the United States, as well as other Hispanic groups, generally have lower socioeconomic position relative to non-Hispanic whites, with higher poverty [10], lower income [11], and lower educational attainment compared with their non-Hispanic white counterparts [12]. Yet despite this, their health profile has been * Corresponding author. Department of Epidemiology, Columbia University, 722 W. 168th Street, Rm 505, New York, NY 10032. E-mail address: [email protected] (A.M. El-Sayed). http://dx.doi.org/10.1016/j.annepidem.2014.09.005 1047-2797/Ó 2014 Elsevier Inc. All rights reserved.
similar, if not better, than that of non-Hispanic whites along a number of important health metrics. This “paradox” has been observed in life expectancy and mortality [13e15], health-risk behaviors [16], preterm birth and low birth weight [17e19], and infant mortality [20]. Two principal explanations, among several, have been proposed to explain this paradox. One explanation is that the paradox may be a result of different social and cultural resources and mores among Mexicans relative to non-Hispanic whites. Cultural resources such as collectivism and reciprocal obligation, and dietary, exercise, and childcare norms may operate to mitigate the adverse health consequences of socioeconomic marginalization in this group [21,22]. Evidence supporting this explanation comes from the observation that the paradox is substantially stronger among immigrants, who are more likely to adhere to these cultural and social mores, and that the ethnic advantage dissipates with each subsequent US-born generation [20]. Observations suggesting that residence in ethnically dense neighborhoods may reinforce protective social norms, thereby improving health among immigrants, is also supportive of this explanation [23,24]. An alternative explanation may be that the apparent health advantage of immigrants in the United States is an artifact of
832
A.M. El-Sayed et al. / Annals of Epidemiology 24 (2014) 831e836
population selection for individuals who are healthier at baseline. This selection may occur via two mechanisms: first, immigration itself selects for a healthier population at baseline (coined the “healthy migrant” hypothesis), and second, those in poor health are more likely to return to Mexico, leaving only the healthiest Mexicans behind in the United States (coined the “salmon-bias” hypothesis) [25,26]. This selection argument is supported by the observation that other ethnic immigrant groups, such as Eastern Europeans and Arab Americans, who may not benefit from ostensibly protective Mexican cultural mores, also appear to have health metrics that outperform their socioeconomic profiles across diverse contexts, including the United States [27e29], Canada [30], and Europe [31e33]. Others have attempted to explain the paradox via misclassification of ethnicity in death certificates [26,34]. Although the literature about the “Mexican paradox” is well developed, research in this area has been largely confined to static cross-sectional analyses that have documented this paradox at single points in time. Yet as both cultural norms and immigration norms may change with time, a dynamic examination of the paradox may yield important insight into its etiology and future trajectory. Therefore, we explored secular trends in the relation between Mexican ethnicity and infant mortality over the past two decades, aiming both to illuminate the etiology of the paradox and to document its trajectory over time. Methods
them. Information about race and ethnicity was collected for all mothers. We created three categories of race/ethnicity, Mexican, other Hispanic, and non-Hispanic white to whom our sample was restricted. Our sample consisted of 1,753,746 non-Hispanic white, 70,678 Mexican, and 29,819 other Hispanic infants; this accounted for 74.3% of the original sample of all singleton births occurring between 1989 and 2006 in the State of Michigan. We defined infant mortality as death of the infant from the day of birth through the first year after birth, and we calculated the total number of deaths among all infants aged less than 1 year during the study period. Data were then grouped into six periods for analysis: 1989 to 1991, 1992 to 1994, 1995 to 1997, 1998 to 2000, 2001 to 2003, and 2004 to 2006. Other available information included data about maternal age, education, marital status, place of birth, parity, prenatal care, and payment source for labor and delivery. Age and education were analyzed in three categories each: less than 20, 20 and 35, and greater than 35, and less than high school diploma, high school diploma and/or some college, or college degree or greater, respectively. Marital status was dichotomized and analyzed as married or not at the time of parturition. Maternal place of birth was divided into two categories, foreign (including birth anywhere outside of the 48 continental states, D.C., Hawaii, or Alaska) or United States. The Kotelchuck Index was used to assess prenatal care and was analyzed dichotomously as adequate versus other [35]. Payment source for labor and delivery was analyzed as a dichotomous variable: private insurance versus Medicaid, self-pay, or other.
Data Analysis The data used for this study were collected by the Michigan Department Community of Health and consisted of all singleton births that occurred between January 1, 1989, and December 31, 2006, in the State of Michigan. Data about births were then crosslinked to all deaths before one completed year of age among
We calculated descriptive statistics, stratified by time interval, for the study sample. To assess trends in key measures over time, we used the CochraneArmitage test for trend. We used P values to assess significance (a ¼ 0.05, two tailed) for all trend tests. We then
Table 1 Demographic characteristics and infant mortality by study period among all singleton live births in Michigan, 1989 to 2006 Characteristics
Age (y) 19 20e34 35þ Education Less than high school High school, some college College or greater Married* Foreign born Race Non-Hispanic White Hispanic Hispanic race White Black Hispanic ethnicity Mexican Other Private source of payment Prenatal carey Inadequate Intermediate Adequate Adequate plus Parity Infant died before first year * y
Period 1989e1991 (N ¼ 316,900), n, %
1992e1994 (N ¼ 294,325), n, %
1995e1997 (N ¼ 285,706), n, %
1998e2000 (N ¼ 283,064), n, %
2001e2003 (N ¼ 278,787), n, %
2004e2006 (N ¼ 268,571), n, %
X2CAP
31,169 (9.8) 259,949 (82) 25,782 (8.1)
28,194 (9.6) 236,470 (80.3) 29,661 (10.1)
28,089 (9.8) 223,946 (78.4) 33,671 (11.8)
26,396 (9.3) 220,964 (78.1) 35,704 (12.6)
23,341 (8.4) 217,976 (78.2) 37,470 (13.4)
21,423 (8) 210,182 (78.3) 36,966 (13.8)
<.0001 <.0001 <.0001
49,643 209,096 58,161 271,992 13,641
44,094 187,263 62,968 233,215 14,340
41,136 173,905 70,665 215,995 16,788
41,998 164,937 76,129 210,815 16,726
42,249 156,714 79,824 203,297 27,282
40,590 146,253 81,728 188,935 28,145
(15.1) (54.5) (30.4) (70.3) (10.5)
<.0001 .0001 <.0001 <.0001 <.0001
(15.7) (66) (18.4) (85.8) (4.2)
(15) (63.6) (21.4) (79.2) (4.9)
(14.4) (60.9) (24.7) (75.6) (5.9)
(14.8) (58.3) (26.9) (74.5) (5.9)
(15.2) (56.2) (28.6) (72.9) (9.8)
305,318 (96.3) 11,582 (3.7)
282,724 (96.1) 11,601 (3.9)
272,235 (95.3) 13,471 (4.7)
275,260 (93.9) 17,734 (6.1)
258,796 (92.8) 19,991 (7.2)
246,619 (91.8) 21,952 (8.2)
<.0001 <.0001
11,396 (98.4) 186 (1.6)
11,462 (98.8) 139 (1.2)
13,330 (99.0) 141 (1.1)
16,559 (99.0) 167 (1.0)
19,794 (99.0) 197 (1.0)
21,729 (99.0) 223 (1.0)
<.0001 <.0001
6805 (58.75) 4777 (41.25 229,794 (72.5)
7337 (63.2) 4264 (36.8) 206,588 (70.2)
9519 (70.7) 3952 (29.3) 206,683 (72.3)
12,923 (77.3) 3803 (22.7) 211,412 (74.7)
16,233 (81.2) 3758 (18.8) 194,045 (69.6)
17,861 (81.4) 4091 (18.6) 172,551 (64.2)
<.0001 <.0001 <.0001
29,785 42,511 166,974 77,630 1.39 2000
23,234 34,151 153,747 83,193 1.41 1577
20,947 32,431 145,200 87,128 1.39 1358
22,106 30,983 139,831 90,144 1.37 1275
21,056 29,846 135,438 92,447 1.38 1304
19,517 32,245 136,190 80,619 1.40 1192
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001
(9.4) (13.4) (52.7) (24.5) (1.45) (0.6)
At time of parturition. Per the Kotelchuck adequacy of prenatal care index.
(7.9) (11.6) (52.2) (28.3) (1.48) (0.5)
(7.3) (11.4) (50.8) (30.5) (1.49) (0.5)
(7.8) (10.9) (49.4) (31.8) (1.48) (0.5)
(7.6) (10.7) (48.6) (33.2) (1.51) (0.5)
(7.3) (12) (50.7) (30) (1.53) (0.4)
A.M. El-Sayed et al. / Annals of Epidemiology 24 (2014) 831e836 Table 2 Rate of infant mortality per 1000 singleton live births among Mexican infants overall and stratified by maternal nativity as well as among non-Hispanic white infants in Michigan, 1989 to 2006 Period
Mexican
1989e1991 1992e1994 1995e1997 1998e2000 2001e2003 2004e2006
Non-Hispanic White
Foreign born
US born
Overall
5.5 3.4 3.5 3.8 4.8 4.9
6.4 5.4 4.8 4.5 4.7 4.4
6.8 6.5 5.2 5.0 5.7 5.5
6.3 5.3 4.7 4.5 4.6 4.3
calculated infant mortality rates, stratified by ethnicity and nativity, over time. Next, we used bivariate and multivariable Poisson regression models, adjusted for age, education, marital status, parity, prenatal care, and payment source for labor and delivery, to assess the relation between Mexican ethnicity and infant mortality relative to non-Hispanic whites in each time interval. Results Table 1 shows descriptive statistics over time among our sample. The incidence of infant mortality overall declined from 6.3 per 1000 live singleton births to 4.4 per 1000 throughout our study period. This mirrors a US-wide decline in infant mortality from 9.8 per 1000 live births in 1989 to 6.7 per 1000 live births in 2006 [36,37]. Both the proportion of foreign-born and Mexican mothers increased with time. Overall, the distribution of maternal age and the proportion of mothers achieving at least a college degree increased with time throughout our study period. The proportion of infants born to mothers who were married at parturition declined, as did the parity of mothers. All tests for trend were statistically significant (P < .01) across our study period. Table 2 and Figure 1 shows the incidence of infant mortality stratified by ethnicity among Mexican and non-Hispanic white infants. Infant mortality declined among infants of both ethnic groups with time. However, disparities in infant mortality increased: the absolute difference in infant mortality was 0.5 per 1000 singleton live births among infants born to Mexican relative to non-Hispanic white mothers in 1989 to 1991, increasing to 1.2 per 1000 singleton live births by the end of the study period. Moreover, between 1998e2000 and 2001e2003, the incidence of infant mortality among infants born to Mexican mothers increased slightly despite a steady decline among those born to non-Hispanic whites throughout the study period. The incidence of infant mortality stratified by ethnicity and maternal birthplace (among Mexican mothers) is shown in Table 2
833
and Figure 2. US-born Mexican mothers had the highest incidence of infant mortality in all time intervals although risk decreased throughout the study period among this group from 7.2 to 6.2 per 1000 singleton live births. At the beginning of the study period, risk for infant mortality among non-Hispanic whites was higher than among foreign-born Mexican mothers. However, after 2000, risk for infant mortality among foreign-born Mexican mothers overtook that of non-Hispanic white mothers. Risk for infant mortality among non-Hispanic white mothers decreased consistently throughout the study period from 6.3 to 4.4 per 1000 singleton live births, whereas risk among foreign-born Mexican mothers decreased between 1989-1991 and 1992-1994, and then increased steadily between 1992-1994 and the end of the study period. Figure 3 shows log relative risk for infant mortality among Mexican mothers relative to non-Hispanic white mothers as well as relative risk and 95% confidence intervals in text, from models adjusted for age, education, marital status, source of payment for labor and delivery, prenatal care, and parity. Risk for infant mortality among Mexicans increased relative to non-Hispanic whites. In 1989 to 1991, Mexican mothers had lower risk for infant mortality (relative risk ¼ 0.78, 95% confidence interval, 0.62e0.98) than their non-Hispanic white counterparts. However, by 2004 to 2006, relative risk among Mexican mothers relative to non-Hispanic whites had increased to 1.01 (95% confidence interval, 0.83e1.22).
Discussion We studied over 1.7 million infants born between 1989 and 2006 in one US state. We found increases in infant mortality among foreign-born Mexican infants between 1992e1994 and 2004e2006 and among US-born Mexican infants between 1998-2000 and 2004-2006. The overall mortality advantage favoring Mexican infants we observed at the outset of our study period dissipated as a result of these increases during our study period. Our findings are consistent with prior observations demonstrating that infants of foreign-born Mexican mothers had lower risk of infant mortality than US-born counterparts [38] as well as studies demonstrating an ethnic advantage relative to nonHispanic whites [20]. However, this is the first study, of which we are aware, that has directly considered secular trends in the Mexican infant mortality paradox. There are two plausible explanations for the changes we observed. First, if Mexican cultural mores are protective against infant mortality, it is plausible that the mechanism through which they influence this outcome includes access to sociocultural resources like communalism and shared obligation that are reinforced via social norms within Mexican enclaves in which a large
Fig. 1. Risk of infant mortality per 1000 singleton live births to Mexican and non-Hispanic white mothers stratified by ethnicity in Michigan, 1989 to 2006.
834
A.M. El-Sayed et al. / Annals of Epidemiology 24 (2014) 831e836
Fig. 2. Risk of infant mortality per 1000 singleton live births to US-born non-Hispanic white mothers as well as Mexican mothers stratified by maternal nativity in Michigan, 1989 to 2006.
proportion of US Mexicans live. Ethnic enclaves are characterized by the cultural distinction they maintain from surrounding areas [39]. Residence within such enclaves may then serve to maintain and insulate protective social norms [40], which may ultimately mediate the mortality advantage among infants of foreign-born Mexican mothers that we observed. In support of this hypothesis, for example, Cagney et al. [23] demonstrated that respiratory morbidity among foreign-born Hispanics was inversely associated with the proportion of foreign-born Hispanics in their residential communities. The dynamics of Mexican enclaves are changing along with the US Mexican population more broadly as US-born Mexicans grow as a proportion of the US Mexican population overall [41]. Between 2000 and 2010, for example, births to native Mexicans outpaced immigration as the largest driver of Mexican population growth for the first time in US history [41]. US-born Mexicans are, by definition, more acculturated and less likely to adhere to traditional cultural norms and mores than their immigrant counterparts. It is plausible, then, that as the proportion of US-born, more acculturated Mexicans has increased in enclaves, social norms in these contexts have shifted away from the more traditional norms that may have, in part, mediated the mortality advantage among infants of foreign-born Mexican
mothers. This proportional increase in the US-born Mexican population in enclaves may then explain the dissipation of the mortality advantage among infants of foreign-born Mexican mothers observed here. There is evidence to indirectly support this potential explanation. For example, one study of birth weight, a leading determinant of infant mortality, among infants born to US- and foreign-born Mexican-American women found that residence in ethnic enclaves (i.e., enclaves of US-born Mexican Americans) was associated with lower birth weight among infants born to US-born women; controlling for residence in immigrant enclaves (i.e., enclaves of foreign-born Mexican Americans) strengthened the association. However, residence in immigrant enclaves was positively associated with birth weight, after controlling for residence in ethnic enclaves [42]. A second plausible explanation supports the “healthy migrant” hypothesis (i.e., the immigration process selects for a healthier population at baseline, and that the better baseline health of this population then explains the mortality advantage of foreign-born Mexicans [25,26]). The relative health of the immigrant population may vary with time. In fact, the healthy migrant hypothesis suggests that the propensity to immigrate is a function, directly or indirectly, of baseline health status. In that respect, it is also
Fig. 3. Adjusted* log relative risk of infant mortality among singleton live births to Mexican mothers relative to non-Hispanic White mothers in Michigan, 1989e2006. *Multivariable models adjusted for age, education, marital status, source of payment for labor and delivery, prenatal care, and parity. CI, confidence interval; RR, relative risk.
A.M. El-Sayed et al. / Annals of Epidemiology 24 (2014) 831e836
plausible that baseline health may also predict the time-toimmigration among immigrants, such that the healthiest may be more likely to immigrate before their less healthy counterparts. It would follow, then, that the baseline health of immigrant Mexicans should have declined with time as the healthiest Mexicans immigrated before their less healthy counterparts. Therefore, our observation of a deteriorating infant mortality advantage among infants of foreign-born Mexican mothers may be explained by a decline in the baseline health of immigrant Mexican mothers resulting from the dynamics of immigrant selection. Our findings do not favor either of these competing explanations regarding the etiology of the Mexican infant mortality paradox over the other. The first explanation supports the hypothesis that the paradox may result from protective social and cultural mores among immigrants, whereas the second supports the selection hypothesis. In that regard, it remains unclear as to which, if either, is a better explanation for the Mexican birth paradox more generally. Our study has several limitations. First, our covariate set was limited. In particular, data about maternal anthropometric characteristics and/or previous health complaints were unavailable. However, these factors are more likely to mediate than to confound the relations of interest here, precluding adjustment in our multivariable models. Second, we are limited by the validity of birth certificate data in vital registry files. Although it has been demonstrated that demographic data are accurate [43e45], questions have been raised about the accuracy of ethnicity reports in these files [25,26]. However, trend analyses of this sort are beholden to public registry files for comprehensive data over long time horizons, and therefore, these limitations were inevitable. Despite these challenges, our findings have several important implications for future research regarding the health of ethnic minorities in the United States. Our findings suggest that the relative health advantage that immigrant Mexicans in the United States have enjoyed for some time may be dissipating. Explorations into the temporal dynamics of the Mexican birth paradox in other United States contexts are needed to compare with the findings we have presented here. Similar health “paradoxes” have been documented among other ethnic minority groups [29,46]. Investigating secular trends in the infant mortality outcomes of these groups would improve our understanding of generalizable trends in immigrant health across ethnic identities. Future work may fruitfully consider the influence of US-born Mexicans on health norms among foreign-born residents of Mexican enclaves and trends in the baseline health of immigrant Mexicans over time.
Acknowledgments The authors thank Glenn Copeland and Glenn Radford from the Michigan Department of Community Health for their health in acquiring the data.
References [1] El-Sayed AM, Tracy M, Scarborough P, Galea S. Ethnic inequalities in mortality: the case of Arab-Americans. PloS One 2011;6(12):e29185. [2] Sudano JJ, Baker DW. Explaining US racial/ethnic disparities in health declines and mortality in late middle age: The roles of socioeconomic status, health behaviors, and health insurance. Soc Sci Med 2006;62(4):909. [3] Cooper R, Cutler J, Desvigne-Nickens P, Fortmann SP, Friedman L, Havlik R, et al. Trends and disparities in coronary heart disease, stroke, and other cardiovascular diseases in the United States: findings of the national conference on cardiovascular disease prevention. Circulation 2000;102(25):3137e47. [4] Ashton CM, Haidet P, Paterniti DA, Collins TC, Gordon HS, O’Malley K, et al. Racial and ethnic disparities in the use of health services. J Gen Intern Med 2003;18(2):146e52.
835
[5] Betancourt JR, Maina AW. The institute of medicine report “unequal treatment”: implications for academic health centers. Mt Sinai J Med 2004; 71(5):314. [6] Williams DR, Collins C. Racial residential segregation: a fundamental cause of racial disparities in health. Public Health Rep 2001;116(5):404. [7] Williams DR. Racial/ethnic variations in women’s health: the social embeddedness of health. Am J Public Health 2002;92(4):588e97. [8] Williams DR, Neighbors HW, Jackson JS. Racial/ethnic discrimination and health: findings from community studies. Am J Public Health 2003;93(2):200e8. [9] Nazroo JY, Williams DR. The social determination of ethnic/racial inequalities in health. Social determinants of health. 2nd ed. Oxford, United Kingdom: Oxford University Press; 2006. [10] Lopez MH, Cohn D. Hispanic poverty rate highest in new supplemental measure. Washington, DC: Pew Hispanic Center; 2011. [11] U.S. Census Bureau. Historical income tables: income inequality. Available at: http://www.census.gov/hhes/www/income/data/historical/inequality/. Accessed December 2012. [12] Lopez MH, Center PH. Latinos and education: explaining the attainment gap. Washington, DC: Pew Hispanic Center; 2009. [13] Minino AM, Murphy SL. Death in the United States, 2010. NCHS Data Brief 2012;(99):1e8. [14] Rosenwaike I. Mortality differentials among persons born in Cuba, Mexico, and Puerto Rico residing in the United States, 1979-81. Am J Public Health 1987;77(5):603e6. [15] Elo IT, Turra CM, Kestenbaum B, Ferguson BR. Mortality among elderly Hispanics in the United States: past evidence and new results. Demography 2004;41(1):109e28. [16] Abraido-Lanza AF, Chao MT, Florez KR. Do healthy behaviors decline with greater acculturation? Implications for the Latino mortality paradox. Soc Sci Med 2005;61(6):1243e55. [17] Fuentes-Afflick E, Lurie P. Low birth weight and Latino ethnicity: examining the epidemiologic paradox. Arch Pediatr Adolesc Med 1997;151(7):665. [18] Hessol NA, Fuentes-Afflick E. The perinatal advantage of Mexican-origin Latina women. Ann Epidemiol 2000;10(8):516. [19] Chung JH, Boscardin WJ, Garite TJ, Lagrew DC, Porto M. Ethnic differences in birth weight by gestational age: at least a partial explanation for the Hispanic epidemiologic paradox? Obstet Gynecol 2003;189(4):1058e62. [20] Hummer RA, Powers DA, Pullum SG, Gossman GL, Frisbie WP. Paradox found (again): infant mortality among the Mexican-origin population in the United States. Demography 2007;44(3):441e57. [21] Abraido-Lanza AF. The Latino mortality paradox: a test of the “salmon bias” and healthy migrant hypotheses. Am J Public Health 1999;89(10):1543. [22] Barney C, Bulatao RA, Anderson NB, editors. Critical perspectives on racial and ethnic differences in health in late life. Washington, DC: National Academies Press; 2004. [23] Cagney KA, Browning CR, Wallace DM. The Latino paradox in neighborhood context: the case of asthma and other respiratory conditions. Am J Public Health 2007;97(5):919e26. [24] El-Sayed AM, Galea S. Community context, acculturation and low-birthweight risk among Arab Americans: evidence from the Arab-American birth-outcomes study. J Epidemiol Community Health 2010;64(2):155e60. [25] Palloni A, Morenoff JD. Interpreting the paradoxical in the Hispanic paradox. Ann N Y Acad Sci 2001;954(1):140e74. [26] Palloni A, Arias E. Paradox lost: explaining the Hispanic adult mortality advantage. Demography 2004;41(3):385e415. [27] Janevic T, Savitz D, Janevic M. Maternal education and adverse birth outcomes among immigrant women to the United States from eastern Europe: a test of the healthy migrant hypothesis. Soc Sci Med 2011;73(3):429e35. [28] El-Sayed A, Galea S. Explaining the low risk of preterm birth among Arab Americans in the United States: an analysis of 617451 births. Pediatrics 2009;123(3):e438. [29] Finkton DW, El-Sayed AM, Galea S. Infant mortality among Arab-Americans: findings from the Arab-American birth outcomes study. Matern Child Health J 2013;17(4):732e45. [30] Urquia M, Frank J, Moineddin R, Glazier R. Immigrants’ duration of residence and adverse birth outcomes: a population-based study. BJOG: An Int J Obstetrics Gynaecol 2010;117(5):591e601. [31] Khlat M, Darmon N. Is there a Mediterranean migrants mortality paradox in Europe? Int J Epidemiol 2003;32(6):1115e8. [32] Stirbu I, Kunst AE, Vlems FA, Visser O, Bos V, Deville W, et al. Cancer mortality rates among first and second generation migrants in the Netherlands: convergence toward the rates of the native Dutch population. Int J Cancer 2006;119(11):2665e72. [33] Uitenbroek DG, Verhoeff AP. Life expectancy and mortality differences between migrant groups living in Amsterdam, the Netherlands. Soc Sci Med 2002;54(9):1379e88. [34] Lariscy JT. Differential record linkage by Hispanic ethnicity and age in linked mortality studies: implications for the epidemiologic paradox. J Aging Health 2011;23(8):1263e84. [35] Kotelchuck M. An evaluation of the Kessner Adequacy of Prenatal Care Index and a proposed adequacy of prenatal care utilization index. Am J Public Health 1994;84(9):1414e20. [36] Centers for Disease Control and Prevention (CDC). Infant mortalityeUnited States, 1990. MMWR Morb Mortal Wkly Rep 1993;42(9):161e5.
836
A.M. El-Sayed et al. / Annals of Epidemiology 24 (2014) 831e836
[37] MacDorman MF, Mathews TJ. Infant deathseUnited States, 2000-2007. MMWR Surveill Summ 2011;60(1):49e51. [38] Hummer RA, Biegler M, De Turk PB, Forbes D, Frisbie WP, Hong Y, et al. Race/ ethnicity, nativity, and infant mortality in the United States. Social Forces 1999;77(3):1083e117. [39] Logan JR, Alba RD, Zhang W. Immigrant enclaves and ethnic enclaves in New York and Los Angeles. Am Sociol Rev 2002;67(2):299e322. [40] Portes A, Rumbaut RG. Immigrant America: a portrait. Berkeley, CA: University of California Press; 2006. [41] Pew Research Center. The Mexican-American boom: births overtake immigration. Available at: http://www.pewhispanic.org/2011/07/14/the-mexicanamerican-boom-brbirths-overtake-immigration/. Updated 2012. Accessed December 12, 2012.
[42] Osypuk TL, Bates LM, Acevedo-Garcia D. Another Mexican birthweight paradox? The role of residential enclaves and neighborhood poverty in the birthweight of Mexican-origin infants. Soc Sci Med 2010;70(4):550e60. [43] Buescher PA. The quality of the new birth certificate data: a validation study in North Carolina. Am J Public Health 1993;83(8):1163. [44] Northam S. The reliability and validity of birth certificates. J Obstet Gynecol neonatal Nurs 2006;35(1):3. [45] Reichman NE, Hade EM. Validation of birth certificate data. A study of women in New Jersey’s HealthStart program. Ann Epidemiol 2001; 11(3):186. [46] Gould JB, Madan A, Qin C, Chavez G. Perinatal outcomes in two dissimilar immigrant populations in the United States: a dual epidemiologic paradox. Pediatrics 2003;111:e676e82.