Rates Of Preterm Birth And Low Birthweight: An Analysis Of Racial And Ethnic Populations

Abstract

Using National Center for Health Statistics data (2016–2020), we evaluated variation in low birthweight and prematurity among racial/ethnic subcategories (e.g., Hispanic Mexican). Disparities, as large as 2.3-fold among rates of low birthweight for subcategories within the multiple race category, underscore the need for granular data stratification and analysis by racial/ethnic subcategories to address root causes of inequities in infant outcomes.

Introduction

Preterm birth (defined as <37 weeks gestation)¹ and low birthweight (defined as <2,500 grams) are associated with adverse physical and neurocognitive outcomes throughout the life course, and prematurity contributes to 36% of infant mortality.² Reducing rates of adverse infant outcomes by addressing underlying root causes could therefore serve as a primary mechanism for reducing disparities in infant mortality and other health outcomes.^3,4

There is a major gap in understanding differences in rates of adverse infant outcomes within granular racial and ethnic groups, and such information is critical for addressing disparities. Using data from the National Center for Health Statistics (NCHS) birth certificate files (years 2016 through 2020), we evaluated differences in rates of low birthweight and preterm birth among racial/ethnic subgroups (e.g., Chinese) within broader, commonly utilized racial categories (e.g., Non-Hispanic Asian [“Asian”]). Overall, we found large variation between subcategories within broader categories, with rates of low birthweight varying as high as 2.3-fold (Exhibit 1) among the non-Hispanic Multiple Race (“Multiple Race”) category and preterm birth varying as high as 2.0-fold (Exhibit 2) among the Asian category.

Exhibit 1:

Variation in rates of low birthweight (<2,500 grams), by broad racial/ethnic category, United States, 2016–2020

Source: Author’s analysis of National Center for Health Statistics (NCHS) birth certificate data, years 2016 through 2020.

Notes: Rates are age- and year-adjusted. Ranges for such categories were constructed among five Hispanic subcategories, seven non-Hispanic Asian subcategories, 21 non-Hispanic Multiple Race subcategories, and four non-Hispanic Native Hawaiian or Other Pacific Islander subcategories. AIAN=American Indian or Alaska Native; NHOPI = Native Hawaiian or Other Pacific Islander.

Exhibit 2:

Variation in rates of preterm birth (<37 weeks gestation), by broad racial/ethnic category, United States, 2016–2020

Source: Author’s analysis of National Center for Health Statistics (NCHS) birth certificate data, years 2016 through 2020.

Notes: Rates are age- and year-adjusted. Ranges for such categories were constructed among five Hispanic subcategories, seven non-Hispanic Asian subcategories, 21 non-Hispanic Multiple Race subcategories, and four non-Hispanic Native Hawaiian or Other Pacific Islander subcategories. AIAN=American Indian or Alaska Native; NHOPI = Native Hawaiian or Other Pacific Islander.

Reducing disparities in adverse infant outcomes will require addressing long-term systematic inequities rooted in racism and social and economic oppression. In such efforts, it is important to consider how aggregating racial/ethnic groups (e.g., combining all Hispanic populations) may ultimately impact policy recommendations and resource allocation,⁵ which may fail to take into account the unique experiences and structural barriers faced by each group.

Study Data and Methods

The data for this study included NCHS birth certificate files from years 2016 through 2020. After excluding plural births (3.3%), births with imputed race (6.5%), births to foreign residents (0.2%), and births with missing gestational age, birthweight, or race/ethnicity (0.4%), we identified 17,055,800 singleton births. The NCHS data included aggregate racial/ethnic categories of non-Hispanic American Indian or Alaska Native (AIAN), Black, and White, as well as more granular subcategories within four broader categories. These subcategories included seven Asian subcategories, four Non-Hispanic Native Hawaiian/Other Pacific Islander [“NHOPI”] subcategories, five Hispanic subcategories based on nationality, and 26 Multiple Race subcategories. Five of the 26 Multiple Race subcategories (n=487 births) were excluded for having <20 low birthweight or preterm births, resulting in 21 Multiple Race subcategories used in our analysis.

In addition to evaluating specific racial/ethnic subcategories within broader categories, we evaluated differences in outcomes among infants born to US-born compared to foreign-born individuals within each category. Please note that the US-born group includes around 0.16% of births with missing or unknown nativity.

The adverse birth outcomes of interest were rates of low birthweight (<2,500 grams) and preterm birth (<37 weeks gestation), following NCHS classifications.¹ To assess differences in adverse birth outcomes among subcategories within broader categories, we evaluated each subcategory’s adjusted rate of preterm birth and rate of low birthweight relative to the respective overall category rate. Rates were calculated using marginal effects from logistic regressions adjusted for maternal age and year fixed effects. Testing for the significance of differences between a subcategory and its overall category is complicated because many common statistical tests require that the two comparator groups be independent of one another. We therefore adopted the simpler and more conservative test of determining significance by non-overlapping 95% confidence intervals between subcategories and their respective broader categories. It is possible that two values may be statistically different even with overlapping confidence intervals, as such, this approach identifies a conservative, minimum number of subcategories that are statistically different from their corresponding overall categories. Rates and confidence intervals are provided in the online supplemental material (Appendix Exhibit A1) for each subcategory.⁶

This study has some limitations. Due to small sample sizes for some categories, we were unable to evaluate outcomes stratified by other important demographic or healthcare characteristics, such as age or payer, or outcomes over time. Subcategory sample sizes are provided in the supplemental material in the appendix.⁶ Rates among subgroups with fewer births will have larger confidence intervals and should be interpreted with caution. Secondly, the study evaluated all racial and ethnic subcategories available in the NCHS data. It is critical that future studies evaluate differences in adverse birth outcomes among other important subpopulations, such as individual American Indian tribal nations or populations of Black women by ancestral country, and large surveys should include granular classifications of race and ethnicity in their data collection.⁷ Given the large disparities in adverse infant outcomes among Black individuals, collection of information about country and/or region of origin (e.g., Sub-Saharan Africa) may additionally be important.

Analyses were conducted in Stata 16.1. The study was deemed non-human subjects research by the University of Arkansas for Medical Sciences Institutional Review Board.

Study Results

Exhibits 3 and 4 present rates of low birthweight and pre-term birth, respectively, showing detail for all racial/ethnic subgroups. Among the full sample, the adjusted rate of low birthweight was 6.3%, and the rate of preterm birth was 8.0% (data not shown). We found large differences in rates of low birthweight (Exhibit 3) and preterm birth (Exhibit 4) between specific subcategories within broader racial/ethnic categories.

Exhibit 3:

Rate of low birthweight birth (<2,500 grams), by specific racial/ethnic subcategory, United States, 2016–2020

Source: Author’s analysis of National Center for Health Statistics (NCHS) birth certificate data, years 2016 through 2020.

Notes: Rates are age- and year-adjusted. Error bars represent 95% confidence intervals for each category and subcategory. AIAN=American Indian or Alaska Native; NHOPI = Native Hawaiian or Other Pacific Islander.

Exhibit 4:

Rate of preterm birth (<37 weeks gestation), by specific racial/ethnic subcategory, United States, 2016–2020

Source: Author’s analysis of National Center for Health Statistics (NCHS) birth certificate data, years 2016 through 2020.

Notes: Rates are age- and year-adjusted. Error bars represent 95% confidence intervals for each category and subcategory. AIAN=American Indian or Alaska Native; NHOPI = Native Hawaiian or Other Pacific Islander.

For low birthweight (Exhibit 3), all five Hispanic subcategories and all seven Asian subcategories had statistically different rates than their respective category. Rates of low birthweight among Hispanic subcategories ranged from 5.4% (Cuban) to 7.7% (Puerto Rican), and rates among Asian subcategories ranged from 4.5% (Chinese) to 8.7% (Filipino; Asian Indian). Among the 21 Multiple Race subcategories, there were 14 subcategories with different rates of low birthweight than the overall rate among the Multiple Race category. The rate of low birthweight among the Black/AIAN subcategory (11.3%) was 135% higher than the Multiple Race subcategory with the lowest rate (AIAN/Asian/NHOPI/White at 4.8%) and 57% higher than the rate of low birthweight for the Multiple Race category overall (7.2%). Of the 4 NHOPI subcategories, two categories (Samoan at 4.9%; Other Pacific Islanders at 7.5%) had a statistically different rate than the overall NHOPI category (6.8%). Of note is that the non-Hispanic Black category had a higher rate (11.7%) than any of the measured subgroups in any other category.

For preterm birth (Exhibit 4), four of the five Hispanic subcategories had statistically different rates than the overall Hispanic category, and all seven Asian subcategories had different rates than the overall Asian category, with infants born to Filipino individuals having 2.0 times the rate compared to infants born to Chinese individuals (10.0% vs 5.0%). Among the 21 Multiple Race subcategories, 8 subcategories had a rate of preterm birth that was statistically different than the overall Multiple Race category, with rates varying from 6.8% among infants born to Asian/White individuals to 12.0% among infants born to Black/AIAN individuals. Similar to low birthweight, infants born to Samoan individuals had lower rates of preterm birth than the overall NHOPI rate (8.7% vs 10.2%).

The same two Hispanic subcategories had higher rates for both outcomes (Puerto Rican and Other/Unknown Hispanic) and two of the same three subcategories had lower rates (Central/South American and Cuban). For both outcomes, Asian Indian, Filipino, and Other Asian subcategories had higher rates than the overall Asian category; additionally, Japanese, Korean, and Chinese subcategories had lower rates.

When assessing the rates of low birthweight (Exhibit 5) and preterm birth (Exhibit 6) among infants born to US versus foreign-born individuals, we found higher rates of adverse birth outcomes for US-born relative to foreign born individuals across six of the seven racial/ethnic categories.

Exhibit 5:

Rate of low birthweight (<2,500 grams), by racial/ethnic category and ancestry, United States, 2016–2020

Source: Author’s analysis of National Center for Health Statistics (NCHS) birth certificate data, years 2016 through 2020.

Notes: Rates are age- and year-adjusted. Error bars represent 95% confidence intervals for each category and subcategory. AIAN=American Indian or Alaska Native; NHOPI = Native Hawaiian or Other Pacific Islander.

Exhibit 6:

Rate of preterm birth (<37 weeks gestation), by racial/ethnic category and ancestry, United States, 2016–2020

Source: Author’s analysis of National Center for Health Statistics (NCHS) birth certificate data, years 2016 through 2020.

Notes: Rates are age- and year-adjusted. Error bars represent 95% confidence intervals for each category and subcategory. Dotted lines for each bar represent 95% confidence intervals. AIAN=American Indian or Alaska Native; NHOPI = Native Hawaiian or Other Pacific Islander.

Discussion

This study identified substantial differences in rates of adverse birth outcomes among granular racial/ethnic subcategories within Hispanic, Asian, Multiple Race, and NHOPI categories. The rate of low birthweight or preterm birth among specific subgroups were as much as 50% greater than the overall rate among the respective racial/ethnic category.

The findings from this study have multiple healthcare and public health policy implications. First, individuals of specific races/ethnicities within a given broader category may have different cultures, languages, and experiences related to personalized and institutionalized racism.^7,8 These important sociocultural differences, coupled with the large differences in rates of birth outcomes found in our study highlight the need for continued efforts among policy makers and public health officials to be mindful of aggregation of racial/ethnic groups in policy and resource allocation decisions and in programmatic outreach. Not only should resources potentially be focused on populations with the highest rates, but programs that avoid homogenization of a given racial/ethnic category may ultimately result in better identification and mitigation of root causes of the disparities in adverse infant outcomes. Second, researchers must consider how aggregation of race and ethnicity may mask important disparities. Researchers are often limited in their capacity to analyze specific subcategories due to sample size issues needed to obtain statistical significance, to avoid issues with statistical modeling, and to protect identification of an individual based on a rare clinical or demographic characteristic. Using methods to overcome such analytic challenges may be critical for research moving forward. A number of analytical approaches allow for improved reliability of estimates when using small cell sizes, such as Bayesian oversampling or bootstrapping methods to increase sample size.⁹ Analytic approaches can also be used to slightly modify the underlying data to protect individual-level information while still maintaining the same overall data distributions (e.g., smoothing techniques to mask outlier data points or creating synthetic data).¹⁰ Finally, analytic and surveillance efforts, including this study, are ultimately limited to the racial/ethnic categorization used during data collection. Public health and policy efforts should consider the value of more granular racial/ethnic surveillance data than the commonly utilized, broad categories. Given that many systematic inequities are historically rooted in state- and local-level policies, efforts at these levels may benefit from development of surveys with disaggregated racial/ethnic information specific to their communities for use in focused place-based initiatives.¹¹ Such efforts must simultaneously be careful to ensure that granular racial/ethnic data is used to reduce health inequities and structural barriers, rather than intensify inequities through racism or racial targeting.⁵ Partnerships with community leaders and members from the respective racial/ethnic groups will be critical for establishing guidelines and best practices for disaggregated reporting at national and local levels.