If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Oregon Health & Science University, Department of Obstetrics and Gynecology, Portland, OregonOHSU-PSU School of Public Health, Portland, OregonInstituto Nacional de Salud Publica (INSP), Center for Population Health Research (CISP), Cuernavaca, MexicoHealth Research Consortium (CISIDAT), Cuernavaca, Mexico
Much reproductive health research on the Latina population overlooks heterogeneity by national origin, nativity, and age and also ignores how U.S.-based populations differ from those in “sending” nations. The purpose of this study is to describe a history of adolescent birth, age at first sex, and contraceptive use at first sex in the Mexican-origin population in both the United States and Mexico.
We developed a binational dataset merging two comparable nationally representative cross-sectional surveys in the United States and Mexico and used covariate balancing propensity scores to balance the age structure of our four samples: U.S.-born Latinas of Mexican origin, foreign-born Latinas of Mexican origin, U.S.-born non-Latina Whites, and Mexican women residing in Mexico. We used a negative binomial regression and calculated the predicted probability of experiencing at least one adolescent birth for each ethnicity/nativity group, stratified by 5-year age group. We also described age and contraceptive use at first sex.
Foreign-born Latinas of Mexican origin and Mexicans in Mexico had similar adjusted probabilities of reporting an adolescent birth (30.1% and 29.9%, respectively), which were higher than those of Mexican-Americans (26.2%) and U.S.-born non-Latina Whites (11.6%). History of an adolescent birth is declining across all four groups among younger ages. Differences do not appear to be driven by the timing of first sex but by contraceptive use, which is increasing among younger age groups.
Access to and use of effective contraception rather than timing of initiation of sexual activity is a key determinant of U.S. Latina and Mexican adolescent births.
The use of contraception, and not age at first sex, is a likely driver of adolescent birth among Latinas of Mexican origin on both sides of the Mexico-U.S. border. Research should examine the intersections of nativity, country of origin, and access to resources when studying reproductive health outcomes.
Supporting adolescents (aged 10–19 years) and young people to achieve their educational, economic, and health potential include supporting reproductive autonomy and providing resources to prevent unwanted pregnancies and births. Three-fourths of pregnancies among U.S. adolescents aged 15–19 years are reported as unintended, compared with 45% in the overall population aged 15–44 years [
]. Scholars argue that adolescent birth rates should be studied as a marker of social and economic inequality both within and between countries. Indeed, adolescent births are not a cause of poverty but rather a result of pervasive social and structural inequity—motherhood is a meaningful and valued role when opportunities for educational and economic mobility are constrained.
Adolescent birth rates in the United States are declining across racial/ethnic groups and by socioeconomic status [
]. Inequities in adolescent births by race/ethnicity or socioeconomic status in the United States mirror inequities in health more broadly in the United States, the result of social and structural inequity [
]. For example, Latinas experienced one of the steepest declines in adolescent births of any race/ethnic group; however, they are still having higher birthrates than non-Latina White adolescents (25.3 vs. 11.4 per 1,000 females aged 15–19 years) [
Adolescent births among Latinas are also of special interest given the large U.S. Latino population (we use the term “Latino” or “Latina” rather than “Hispanic” here, acknowledging that debate exists around these terms [
In addition to national origin and nativity, studying health inequities among the Latina population also requires a more nuanced understanding of the way in which both sending societies and contexts of reception in the United States may affect reproductive health. When examining immigrants in particular, it may be a mistake to ascribe outcomes only to the social environment in the United States or to “Mexican culture” [
]. Rather, examining Latinas of Mexican origin by nativity relative to other populations in the United States and Mexico country can further allow for a better assessment of the possible impacts of immigration on those who engage in it and, to some extent, their descendants [
Given sex and contraceptive use are proximal determinants of pregnancy, the purpose of this study is to describe a history of adolescent birth, age at first sex, and contraceptive use at first sex in four population groups: U.S.-born Latinas of Mexican origin, foreign-born Latinas of Mexican origin, U.S.-born non-Latina Whites, and Mexican women residing in Mexico. We construct a binational dataset and fully adjust for population age structure to isolate the relationship of nativity and ethnicity with key adolescent reproductive health behaviors and outcomes.
We conducted a secondary analysis leveraging two nationally representative demographic surveys: the National Survey of Family Growth (NSFG) (dataset) [
] in Mexico. We combined multiple cross-sections from each survey, covering approximately the same time period: 2013–2015, 2015–2017, and 2017–2019 waves of NSFG and 2014 and 2018 waves of ENADID. We verified that data distributions of all variables were similar between survey waves prior to combining. For NSFG data, we used the final poststratified, fully adjusted case weights released by the NSFG for combining 2013–2019 survey files. ENADID does not release case weights designed for combining survey waves, so we created a scaling factor based on the sample sizes of our study population (female respondents aged 15–44 years) [
] to adjust the 2014 and 2018 survey weights to be representative of the approximate population of Mexico in 2016, the midpoint of the two surveys. We then combined data from both surveys, which produced estimates that were identical to those produced by analyses stratified by survey while allowing us to simultaneously model outcomes for all ethnicity/nativity groups.
Our sample included female survey respondents aged 15–44 years who were identified as belonging to one of four ethnicity/nativity groups: Mexicans residing in Mexico (“Mexicans”), foreign-born Latinas of Mexican origin residing in the United States (“foreign-born”), U.S.-born Mexican origin (“Mexican-Americans”), and U.S.-born non-Latina White (“non-Latina White”). The first group used data from ENADID; the other three groups used data from NSFG. We included non-Latina White women in the analysis as a reference group because they experience lower rates of adolescent birth than the aggregate Latina population and are an important population, often examined in health disparities work to assess the degree and sources of ethnoracial disadvantage of people of color [
Our primary outcome was a history of adolescent birth, a retrospective outcome defined as a birth before age 20 years; we used a binary variable (yes/no) for descriptive analyses and a continuous variable (number of adolescent births) for regression modeling. While only women aged 15–44 years were surveyed in NSFG and ENADID, our retrospective outcome does capture births in early adolescence (aged less than 15 years) as long as respondents were old enough (15 years and more) to be surveyed. Our secondary outcomes were age at first sex and contraceptive method used at first sex. We categorized age at first sex as not applicable (i.e., had not yet had sex at the time of survey), age less than 15, 15–17, 18–19, and age 20 or more. We categorized method at first sex using the Center for Disease Control and Prevention's moderately and most effective framework: most effective methods (male or female sterilization, intrauterine device, and implant), moderately effective methods (short-acting hormonal methods and diaphragm), least effective methods (condoms, sponge, spermicide, emergency contraception, withdrawal, and fertility awareness), and no method [
Our primary independent variables were age at the time of survey, urban/rural status, and language spoken at home. Age was classified in 5-year age groups: 15–19, 20–24, 25–29, 30–34, 35–39, and 40–44 years. We considered a respondent to live in a rural area if their place of residence had a population less than 15,000, which was a cut point available in both surveys. We classified language based on the first mention of language spoken at home in the NSFG (English, Spanish, and other) and assigned all ENADID respondents as Spanish speakers. We also examined several additional demographic covariates at the time of survey: marital status, number of people living in the household, education, workforce status, and health insurance (Supplement provides details on all covariates).
Immigrants from Mexico are the youngest immigrant group in the United States [
], suggesting that population age structure may contribute to observed differences in reproductive behaviors and outcomes (Table 1). To reduce confounding by differences in age structure and urbanicity by ethnicity/nativity group and to isolate the relationship between our ethnicity/nativity groups and outcomes, we weighted our sample using covariate balancing propensity scores (CBPS) (Supplement) [
Table 1Sample characteristics among female respondents aged 15–44 years, National Survey of Family Growth (2013–2015, 2015–2017, and 2017–2019 waves) and National Survey of Demographic Dynamics (ENADID; 2014 and 2018 waves)
Survey and CBPS weights
Mexican in Mexico
Mexican in Mexico
Age (in years)
Place of residence
<15,000 people (Rural)
≥15,000 people (Urban/Suburban)
People in household
Currently in school
Completed standard education
Primary language at home
Labor force status
Current Insurance status
Estimates are weighted with complex survey weights (left) and complex survey weights incorporated into covariate balancing propensity scores (CBPS; right). All data are proportions.
We tabulated proportions of all independent variables by ethnicity/nativity group using survey weights alone and survey/CBPS weights to assess the effect of balancing; all other analyses used both survey/CBPS weights. We modeled a number of adolescent births using negative binomial regression, which we used to calculate the predicted probability of experiencing at least one adolescent birth adjusted for age group, urban/rural status, and language spoken at home for each ethnicity/nativity group, stratified by 5-year age group. Since respondents aged 15–19 years in our sample had not completed their adolescence and therefore had not been at risk of an adolescent birth for the same amount of time as respondents aged 20 years and older, we included an exposure variable in our models for time at risk of adolescent birth. We assigned 15-year-old respondents half a year at risk and added one additional year for each year of age, up to 4.5 years at-risk for 19-year-olds; all respondents aged more than 19 years were assigned 5 years at-risk. We aimed to produce population level estimates of a history of adolescent birth (a birth at any age less than 20 years), and therefore did not include age at first sex and method used at first sex in the model as proximate determinants of fertility [
] to be able to include all eligible respondents regardless of a sexual history.
To further explore underlying mechanisms or determinants of observed adolescent births, we examined age at first sex and contraceptive method at first sex graphically by ethnicity/nativity and 5-year age group, excluding missing responses from ENADID respondents (1% of responses for both age and method used at first sex). We included all respondents in the multivariable model and when examining age at first sex but excluded respondents who had not yet had sex when examining a contraceptive method used at first sex to focus on a population with the potential to experience an adolescent birth. Finally, we restricted the denominator to those who experienced a history of adolescent birth and compared the percentage of women who used a contraceptive method at first sex by ethnicity/nativity group. We determined CBPS weights using R 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria) and the weight package (v0.10.2; Greifer, 2020) and conducted all other analyses using Stata 15.1 (Stata Corp, College Station, Texas). The Oregon Health and Science University Institutional Review Board approved this study as a minimal risk human subjects research.
Our final weighted analytic sample included 33.9 million non-Latina Whites, 4.3 million Mexican-Americans, 3.0 million foreign-born, and 28.4 million Mexicans from multiple waves of the NSFG spanning 2013–2019 and ENADID 2014 and 2018 (Table 1). Mexican-Americans skewed younger than the other groups with 27.7% aged 15–19 years compared to 15.0% non-Latina Whites, 6.2% foreign-born, and 19.0% of Mexicans. Non-Latina White and Mexican respondents were more likely to live in rural areas (21.9% and 36.5%, respectively) than either Mexican origin group in the United States (6.7% for Mexican-American and 9.0% for foreign-born). As a result of these skewed distributions, we applied CBPS weights, making age and place of residence (urban/rural) distributions similar across all ethnic/nativity groups with a sample of approximately 175,000 respondents for each ethnicity/nativity group (Table 1). After applying CBPS weights, foreign-born females were more often married/cohabiting (60.2%) compared to 54.7% of Mexican-Americans, 53.2% of non-Latina White, and 55.7% of Mexicans. About 21% of Mexican-Americans and 66% of foreign-born respondents spoke Spanish at home. The greatest proportion of uninsured respondents was observed in the foreign-born group (50.6%), compared to 9.8%, 15.8%, and 18.2% of non-Latina White, Mexican-Americans, and Mexicans, respectively (Supplemental Table 1 provides confidence intervals).
Foreign-born respondents had the highest unadjusted rates of history of adolescent birth (29.4%; Supplemental Table 2), followed by Mexican-Americans and Mexicans (27.6% and 25.9%, respectively); non-Latina Whites had the lowest rates at 12.2%. After accounting for time at risk of an adolescent birth and covariates (age group, urban/rural status, and primary language) in our multivariable model (Figure 1, left; Supplemental Table 3), the adjusted probabilities of at least one adolescent birth decreased slightly and remained lowest for non-Latina Whites (11.6%), followed by Mexican-Americans (26.2%). The adjusted proportion reporting an adolescent birth increased slightly for foreign-born (30.1%) and more for Mexicans (29.9%). Overall trends were similar to those by 5-year age group (Figure 1, right; Supplemental Table 3) and all four ethnicity/nativity groups saw similar trends by 5-year age groups; adjusted probabilities of an adolescent birth increased with increasing age up to 39 years; this indicates a downward trend in history of adolescent birth over time.
Differences in the probability of experiencing an adolescent birth do not appear to be driven by the timing of first sex. Foreign-born and Mexicans, the two groups with the highest probability of adolescent birth, were more likely to experience first sex after age 19 years (18.8% and 24.4%, respectively; Supplemental Table 2) than non-Latina Whites (12.9%), the group with the lowest proportion of adolescent births. When examined by 5-year age group, foreign-born and Mexicans again had the latest ages at first sex, although both of these groups showed a trend toward younger ages at first sex for younger age groups (Figure 2). Mexican-Americans were more similar to non-Latina Whites, who had the youngest age at first sex, which remained relatively constant over time.
Among all respondents who have had sex (n = 585,507 with CBPS weighting), trends in the use of any contraceptive method at first sex (Figure 3) are increasing among younger age groups, mirroring the downward trends in a history of adolescent birth. Also mirroring patterns in history of adolescent birth by ethnicity and nativity/country, Mexican-Americans have patterns of contraceptive use and nonuse more similar to non-Latina White women than to Mexican born women. Overall, 32.0% of Mexican-Americans and 19.5% of non-Latina Whites did not use a contraceptive method at first sex, compared to 51.3% of foreign-born and 56.6% of Mexicans (Supplemental Table 4). Despite these similarities among Mexican-born women, there are important differences in method effectiveness; in some contrast to immigrants and, especially, both U.S.-born groups, Mexicans who used a method at first sex almost exclusively used a least effective method (Figure 3). Use of the most effective methods at first sex is rare in all groups but more common in the youngest age group.
Among females who have had an adolescent birth (n = 164,063 with CBPS weighting), the proportion who used a contraceptive method at first sex is the smallest among those who were foreign-born (28.2%) and Mexican (27.8%) compared to non-Latina Whites (65.5%) and Mexican-Americans (53.9%) (Figure 4).
Our study found that foreign-born Latinas and Mexicans had similar adjusted probabilities of reporting an adolescent birth (30.1% and 29.9%, respectively), which were higher than those of Mexican-Americans (26.2%) and non-Latina Whites (11.6%). All four ethnicity/nativity groups saw similar trends in the history of adolescent birth by 5-year age groups: adjusted probabilities of an adolescent birth increased with increasing age up to 39 years, indicating a downward trend in the history of adolescent birth over time in all groups. Differences by ethnicity/nativity group and age do not appear to be driven by the timing of first sex; in fact, the groups with the oldest age at first sex have higher probabilities of adolescent birth (foreign-born and Mexicans). Trends in the use of any contraceptive method at first sex are increasing among younger age groups, mirroring the downward trend in a history of adolescent birth. Among those who reported an adolescent birth, contraceptive use at first sex was greater among those who were foreign-born and Mexicans compared to Mexican-Americans and non-Latina Whites.
Examining key proximate determinants of fertility—sexual activity and contraceptive use—suggests that disparities between Mexican-American and non-Latina White women in the United States are not driven by sexual initiation but in spite of them, non-Latina Whites have earlier ages at initiation than all Mexican-origin groups. Instead, our analyses reveal that use of contraception at first sex is lower among the latter than the former, where Mexican-origin women also tend to use less-effective methods at first sex even if contraceptive use is becoming more common among all Mexican-origin groups on both sides of the border. Because contraceptive use at first sex is more common in younger age groups and younger age groups are having sex earlier, not later, it is likely contraceptive use rather than delayed sexual activity that helps explain declining probabilities of adolescent birth across ethnicity/nativity groups. Prior work, not disaggregated by ethnicity or nativity, showed that the proportion of adolescents in the United States who have ever had sex has remained stable over the past two decades [
]. Here, we have shown that these trends apply to the majority group of non-Latina Whites but also to U.S.-born and immigrant women of Mexican origin.
Despite these common patterns by cohort, our study found important differences in contraceptive use and type at first sex, particularly by nativity. Among respondents who had an adolescent birth in our study, contraceptive use at first sex was lower among those who were foreign-born and Mexicans compared to Mexican-Americans and non-Latina Whites. Together, our findings suggest again that use of contraception, and not age at first sex, is a main proximal determinant of adolescent pregnancy among heterogeneous groups of Latinas (U.S.-born, foreign-born, residing in Mexico). Furthermore, our findings also suggest that the effectiveness of contraception at first sex (and, likely, thereafter) could explain the differences in a history of adolescent birth we documented.
By incorporating nativity (as a marker of immigration), we show that respondents who were born in the United States (Mexican-American and non-Latina White) are more similar to each other on measures of adolescent birth, age at first sex, and, to a lesser extent, contraceptive use at first sex. Similarly, foreign-born and Mexicans are similar to each other. Historically, most research focused on reproductive health outcomes treats the Latino population as a monolith [
]. Our findings using nationally representative data on both sides of the border highlight the importance of examining nativity groups separately and including a country-of-origin comparison group in studies that evaluate Latina fertility.
Researchers have advanced cultural explanations of U.S. Latina and Mexican fertility, attributing earlier and/or higher Mexican-origin fertility to profamily values and traditions in Mexico [
]. Our results, showing similar patterns for both groups in adolescent birth and age at first sex do suggest immigrants might still be influenced by sexual and reproductive cultures prevalent in Mexico. However, at first sex, Mexican immigrant women in the United States are more likely to have used any and, especially, more effective contraception than women in Mexico. While this could be reflective of selectivity in immigration, for women immigrating to the United States before first sex, this pattern is likely driven by factors on the U.S. side. Thus, our findings that Mexican immigrant women use less and less-effective contraception at first sex than U.S.-born counterparts is particularly problematic and may not reflect Mexican “culture” but structural access to contraception in the United States.
Behavioral or cultural approaches to examining immigration, fertility, and contraceptive use, such as acculturation [
], obscure the impact of contributing structural factors such as immigration and other social policies, including insurance and access to contraceptive services, that contribute to observed inequities in immigrant health [
]. In our study, the foreign-born group had the highest uninsurance rates (50.6%), compared to 9.8%, 15.8%, and 18.2% of non-Latina White, Mexican-American, and Mexicans, respectively. Furthermore, regardless of insurance status, immigrants and mixed-status families may be hesitant to access available services. For example, while the federal Title X program pays for contraceptive services for all low-income residents regardless of insurance or immigration status, mounting evidence shows that fear of immigration status and policies such as the “public charge” or government reporting of data to immigration authorities can influence access to health and social safety net programs regardless of eligibility (e.g., among Mexican-Americans) [
]. Future work should focus on the intersection of immigration history, insurance status, immigration policies, and reproductive health outcomes.
On the Mexican side of the border, access to contraception has long been part of national population policy and national strategy to prevent adolescent pregnancy. However, evidence has shown that access to the most effective methods of contraception, intrauterine devices and implants, is concentrated in the immediate postpartum setting. Thus, adolescents seeking to use the most effective contraception may need to experience a delivery to access these methods, which is a challenge to preventing undesired adolescent pregnancies and births.
While our binational covariate-balanced approach has strengths, this study has limitations. First, variables at the time of survey were not appropriate for examining retrospective outcomes and we were thus not able to incorporate additional factors known to be associated with current reproductive behaviors or outcomes (e.g., current educational level cannot be used to explain a history of adolescent birth). Second, we focus on a contraceptive method use at first sex and do not have a measure of patterns of contraceptive use during adolescence. Third, we have a much smaller sample in the NSFG compared to ENADID and small ethnicity/nativity subgroups. However, we leveraged several waves of NSFG to increase sample size and our samples are the largest population-based subgroup samples we are aware of. As a result, we only report point estimates, which suggest that U.S.-born Mexican origin respondents experience lower rates of adolescent birth than their Mexican-born counterparts. Fourth, we do not have data on immigrant generation or timing of immigration; analyses not considering the timing in which women migrate may confound the timing and mechanisms of important reproductive health outcomes [
]. This may result in an underestimation in our analysis, although this would only introduce bias if the Youth Risk Behavior Surveillance Survey–NSFG gap varied by race-ethnicity and/or nativity. Finally, we assess retrospective outcomes which are subject to recall bias, especially in older ages, and cannot be compared with rates. We capture a history of adolescent birth which has the advantage of capturing births that occurred when respondents were young (aged less than 15 years) adolescents but misses some early adolescent births if the adolescent is not yet aged 15 years and therefore not included in the NSFG or ENADID surveys.
Despite decreasing trends in history of an adolescent birth across age groups in all of our ethnicity/nativity groups, we also show persistent disparities among diverse groups of Latinas of Mexican origin (U.S.-born, Foreign-born, residing in Mexico) and compared with U.S.-born non-Latina White women. Together, our findings suggest that access to and use of effective forms of contraception rather than timing of initiation of sexual activity are likely key determinants of adolescent fertility among Mexican-origin women on both sides of the U.S.-Mexico border.
The authors would like to thank Emily Burney for her assistance with literature review.
Author Contributions: B.G.D., F.R., and E.B. conceived of the study design and analysis plan; E.B., E.F.R., and B.S.A. conducted the data management and analysis; B.G.D., E.B., and L.E.J. drafted the manuscript; B.G.D. secured funding; all authors contributed to interpretation of results and revision of the manuscript for intellectual content.
This study was supported by grant #1 PHEPA000004-01-00 from the Department of Health and Human Services , Office of Population Affairs (DHHS/OPA), Darney, PI. Dr. Darney's institution receives funding from Organon for projects in which she is the Principal Investigator.
As we write this editorial, both authors are living and working in states in the Southern United States where abortion trigger-laws have recently come into effect following the overturning of Roe v. Wade . It is impossible to enter a meaningful discussion on reproductive health in the United States without acknowledging that this is a particularly challenging time for birthing people. The shifting sociopolitical reproductive healthcare landscape creates precarity for people with uteruses, particularly Black and Brown communities.