Gestational length and preterm delivery

Gestational length, in completed weeks, is either calculated using fetal biometric ultrasonography measurements (first or early second trimester) or by clinical data such as the date of the last menstrual period (LMP), the date of embryo transfer or by a combination of these. The mean, or average, gestational length in a particular population depends on the distribution of deliveries in each gestational week. This distribution is bell shaped with a longer left tail composed of PTDs. A shorter mean gestational length in a population can result from a shift of the whole distribution curve to the left, or it can result from a high proportion of preterm infants. In the first case, a slight shift to the left may inflate the PTD rate due to the relatively large number of deliveries that occur at 35−36 weeks of gestation. In the second case, in a population where PTD is frequent as shown by an exaggerated left tail, the effect on the mean gestational length will be moderate due to the much larger proportion of term deliveries.

Although disparities in the average gestational length is an important concern, PTD is of more practical importance, as a cause of infant mortality and morbidity (e.g., respiratory distress) and as a risk factor for a number of adverse health outcomes over the life course (e.g., neurodevelopmental disabilities and metabolic disorders) . The mechanism by which PTD occurs is not well understood, but it is considered multifactorial with diverse risk factors implicated including behavioural, psychosocial, sociodemographic and community factors . Both the causes and the severity of consequences of PTD are known to vary according to gestational age, and consequently PTD is often further categorized into moderate (32–36 weeks) and very preterm delivery (<32 weeks). For clinicians, gestational age categories are particularly useful to predict specific neonatal and infant prognosis. Gestational length also delineates thresholds for intervention in preterm labour or preterm premature rupture of membranes (PPROM), such as the administration of corticosteroids, tocolysis and referral to higher levels of care .

In addition to gestational age cut-offs, PTD can also be categorized by clinical subtypes, specifically according to whether the delivery is ‘spontaneous’ (preterm labour with intact membranes, which is poorly understood, or PPROM, often stemming from undiagnosed infection) or due to obstetric, or provider-initiated, intervention . The majority of PTDs occur spontaneously; however, up to 25–35% of PTDs are provider-initiated due to maternal (i.e., preeclampsia) or fetal complications (i.e., fetal distress), where the pregnancy is ended by caesarean section (CS) or induction of labour. Despite this distinction, both types of PTD share common pathways as the causes of provider-initiated PTD may be identical to those that lead to spontaneous PTD if no actions were taken .

Gestational length according to race/ethnicity

Whereas numerous studies report on PTD rates, studies reporting racial or ethnic differences in average gestational length are few. Typically, the mother’s race/ethnicity has been examined, using broad categories such as ‘Black’, ‘Hispanic’, ‘Asian’ and ‘non-Hispanic White’. Several studies have reported a shorter average gestational length in Black women, such as in an American study that found a difference of 5 days compared with White women. Furthermore, the results showed that the most common gestational week of delivery at term was the 39th week among Black women and the 40th week among White women . However, this pattern is not uniformly observed. Among Somali-born migrant women to Scandinavia, the average gestational length, as well as the PTD rate, was found to be comparable to the native-born, mostly White, population . Among nulliparous women progressing to spontaneous labour in the United Kingdom (UK), Black (of both African and Caribbean descent) and Asian (Indian, Pakistani and Bangladeshi descent) women had a shorter mean duration of pregnancy (39 weeks) compared with women of European origin (40 weeks) . Among women living in eight South American countries, women reporting Latin American ancestry had on average a 2–4 days shorter duration of pregnancy than among women reporting European ancestry, who may be the descendants of more recent migration waves .

Despite much debate, there is no consensus as to the clinical significance and potential implications of these disparities in the average duration of pregnancy. Future comparative studies should assure a standardized method of pregnancy dating as well as account for potential confounders. We will now discuss the more commonly investigated endpoint of PTD.

Preterm delivery according to race/ethnicity

In terms of an association between PTD and race/ethnicity, a recent meta-analysis of 22 studies (out of 30 possible studies) conducted in the US, the Netherlands, UK, Italy and Brazil comparing Blacks with Whites found a doubling of the adjusted odds ratio (OR = 2.0, 95% confidence interval (CI): 1.8, 2.2) with all but two individual studies finding significantly elevated risks . Subgroup analyses by gestational age groupings suggested that the excess of PTD among Blacks compared with Whites is more marked in earlier gestational age groupings. The majority ( n = 24 out of 30) of these studies were conducted in the US, with only two studies separating Black immigrants from the native-born Black population, which found greater risks among native-born Blacks than Black immigrants when compared with Whites. Furthermore, of all studies, ultrasound for gestational age dating was used in one study, 10 studies used LMP only and the remainder did not report the dating method. Five European studies compared Black-immigrant populations (sub-Saharan African, Surinamese, Ghanian, Antillian) with Whites and one UK study and two Brazilian studies compared native Blacks with Whites. Only the British study used a combination of ultrasound and LMP for gestational dating, whereas three studies used LMP, and the remainder did not report their method. This meta-analysis also examined Asian (17 studies) and Hispanic (11 studies, 10 conducted in the US) ethnicity, but it did not find a significant association with PTD for either group.

In terms of racial disparities in clinical PTD subtypes, the risk of spontaneous preterm labour, premature rupture of membranes and provider-initiated PTD were all significantly higher among US Blacks compared with US Whites, with the greatest disparity for spontaneous labour, showing a four times greater risk of very PTD (<32 weeks) and a two times greater risk of moderately PTD (weeks 32–36). More research is needed to establish whether there is variation in clinical PTD subtypes for other groups.

To sum up, there are ethnic and racial disparities in average gestational length and PTD, with US-born Black women (but not foreign-born Black women) at the highest risk of shorter average gestational length compared with US-born White women. In several other countries, foreign-born Black women were found to be at the highest risk of PTD compared with Whites or native-born women. We will now examine possible explanations for these disparities.

Understanding racial/ethnic disparities: measurement bias in gestational length and preterm delivery

As we have seen, several methods are used to determine gestational length; however, the choice of method might not be without importance when comparing population groups. Naegele’s rule estimates the duration of pregnancy to be 280 days from the first day of the LMP. However, LMP has a number of limitations, which affect its accuracy. Firstly, women may not reliably recall their LMP, particularly in the presence of first trimester bleeding, which may instead be interpreted as the beginning of normal menses. Secondly, LMP assumes that ovulation occurs on the 14th day of a 28-day cycle; however, it is not uncommon for women to have irregular cycles and also to experience delayed ovulation. Uncorrected, misclassification of gestational age using LMP can be in either direction, and it can consequently influence both preterm and post-term delivery rates . Of the studies cited in the above-mentioned section examining disparities in gestational length and PTD, the majority of studies that reported the dating method were based on LMP or the best clinical estimate including LMP , and they failed to correct for the potential misclassification of gestational length.

Ultrasonography before 20 weeks of gestation is often considered the ‘gold standard’ for gestational dating as it is more accurate (±3–5 days) than any other prenatal or postnatal estimate of pregnancy dating, particularly if conducted earlier in pregnancy . Ultrasound assessment assumes that below a certain gestational age, differences in fetal size are related to gestational age, which can lead to underestimation of the gestational age of small but normal fetuses by approximately 1–2 days. However, this systematic error is considered minor compared with the large errors inherent in LMP described earlier. Newer studies often use the ‘best obstetric/clinical estimate’, which is thought to improve accuracy by combining ultrasound and clinical information.

There is a possibility of pregnancy dating bias contributing to the ethnic/racial disparities seen in gestational length and PTD. A recent American study comparing gestational age estimates based on LMP with first trimester ultrasound assessment found important discrepancies (average differences, in days) among women of young maternal age, lower education, non-Hispanic Black race/ethnicity and maternal obesity . For most characteristics, this study indicated on average, consistently longer gestational lengths when the LMP was used compared with ultrasound. A notable exception was by race/ethnicity; non-Hispanic Black women exhibited a gestational length of 0.5 days shorter on average when LMP was used as compared with ultrasound, whereas the gestational length among non-Hispanic White women was 1.2 days longer using LMP compared with ultrasound. The authors suggest that such measurement biases may exacerbate the racial disparity in PTD, and that this should be verified in a more representative sample. This suggestion is consistent with findings from a California-based study where compared with US White women, US Black women had an 80% increased risk of PTD when gestational length was calculated from LMP, but only a 50% increased risk when calculated from ultrasound measurements .

In addition to the characteristics mentioned earlier (young maternal age, lower education, non-Hispanic Black race/ethnicity and maternal obesity), other studies examining discrepancies in gestational age comparing LMP with ultrasound assessment at <20 weeks of gestation found that such differences (both positive and negative) were more common for women who entered prenatal care after the second month of pregnancy , were unmarried and had severe preeclampsia . It is possible that many of the above-named characteristics are differentially distributed between ethnic, racial or migrant groups, therefore affecting studies of ethnic disparities.

Understanding racial/ethnic disparities: potential explanatory factors

What other factors can explain the racial and ethnic differences in gestational length and PTD? The heightened PTD risk among Black women in the US is described as a persistent public health problem, and it has been studied for decades with little to no headway made in identifying important strategies for reducing inequities . Socio-economic conditions (e.g., education), maternal behaviours (e.g., smoking), access to prenatal care, stress, infections (e.g., bacterial vaginosis) as well as genetics have been implicated in the PTD disparity between Blacks and Whites . Differentials in education and access to first trimester prenatal care cannot explain disparities nor can smoking or drug use as non-Hispanic Black women have lower rates of these behaviours than non-Hispanic White women . Four more promising lines of investigation have been identified. Firstly, numerous interventions during pregnancy related to maternal health have been tested with little success, suggesting that preconception health may need to be targeted as mechanisms that lead to PTD may be underway before pregnancy . Secondly, infections, specifically bacteria associated with bacterial vaginosis, are more common among African American women, and it is unknown why this is the case . However, antibiotic interventions have not resulted in any benefit . Thirdly, elements of psychosocial stress have been identified as possible mediating factors as African American women are more likely to experience stressful life events and racism than non-Hispanic Whites before or during pregnancy; however, studies have shown conflicting results .

Finally, there have been attempts to normalize the racial divide in PTD, particularly because other explanations have been generally unsuccessful, by suggesting that there are inherent genetic differences among Black populations that explain shorter gestational length ; however, there has not been much success in the genetic line of investigation. Examining complex gene–environment interactions may add to our understanding of racial/ethnic disparities . Under the genetic hypothesis, one would expect that US-born Black women constitute an intermediate risk group between foreign-born Black and US-born White women due to intermarriage and genetic mixing over previous generations . However, meta-analyses consistently show that US-born Blacks have markedly higher risks than their foreign-born counterparts . A socio-historical hypothesis pointing to continuous exposure to socio-economic and structural discrimination, from slavery to the urban underclass, provides a more comprehensive framework to explain the adverse outcomes of African Americans in the US . It has to be noted that although most studies control for at least one socio-economic indicator, given the complexity of social position, residual confounding by unmeasured socio-economic factors cannot be ruled out .

In the following, we will turn towards studies that mainly examine aspects of migration, looking at PTD among migrant mothers/fathers versus the native or domestically born.

Gestational length according to race/ethnicity

Whereas numerous studies report on PTD rates, studies reporting racial or ethnic differences in average gestational length are few. Typically, the mother’s race/ethnicity has been examined, using broad categories such as ‘Black’, ‘Hispanic’, ‘Asian’ and ‘non-Hispanic White’. Several studies have reported a shorter average gestational length in Black women, such as in an American study that found a difference of 5 days compared with White women. Furthermore, the results showed that the most common gestational week of delivery at term was the 39th week among Black women and the 40th week among White women . However, this pattern is not uniformly observed. Among Somali-born migrant women to Scandinavia, the average gestational length, as well as the PTD rate, was found to be comparable to the native-born, mostly White, population . Among nulliparous women progressing to spontaneous labour in the United Kingdom (UK), Black (of both African and Caribbean descent) and Asian (Indian, Pakistani and Bangladeshi descent) women had a shorter mean duration of pregnancy (39 weeks) compared with women of European origin (40 weeks) . Among women living in eight South American countries, women reporting Latin American ancestry had on average a 2–4 days shorter duration of pregnancy than among women reporting European ancestry, who may be the descendants of more recent migration waves .

Despite much debate, there is no consensus as to the clinical significance and potential implications of these disparities in the average duration of pregnancy. Future comparative studies should assure a standardized method of pregnancy dating as well as account for potential confounders. We will now discuss the more commonly investigated endpoint of PTD.

Preterm delivery according to race/ethnicity

In terms of an association between PTD and race/ethnicity, a recent meta-analysis of 22 studies (out of 30 possible studies) conducted in the US, the Netherlands, UK, Italy and Brazil comparing Blacks with Whites found a doubling of the adjusted odds ratio (OR = 2.0, 95% confidence interval (CI): 1.8, 2.2) with all but two individual studies finding significantly elevated risks . Subgroup analyses by gestational age groupings suggested that the excess of PTD among Blacks compared with Whites is more marked in earlier gestational age groupings. The majority ( n = 24 out of 30) of these studies were conducted in the US, with only two studies separating Black immigrants from the native-born Black population, which found greater risks among native-born Blacks than Black immigrants when compared with Whites. Furthermore, of all studies, ultrasound for gestational age dating was used in one study, 10 studies used LMP only and the remainder did not report the dating method. Five European studies compared Black-immigrant populations (sub-Saharan African, Surinamese, Ghanian, Antillian) with Whites and one UK study and two Brazilian studies compared native Blacks with Whites. Only the British study used a combination of ultrasound and LMP for gestational dating, whereas three studies used LMP, and the remainder did not report their method. This meta-analysis also examined Asian (17 studies) and Hispanic (11 studies, 10 conducted in the US) ethnicity, but it did not find a significant association with PTD for either group.

In terms of racial disparities in clinical PTD subtypes, the risk of spontaneous preterm labour, premature rupture of membranes and provider-initiated PTD were all significantly higher among US Blacks compared with US Whites, with the greatest disparity for spontaneous labour, showing a four times greater risk of very PTD (<32 weeks) and a two times greater risk of moderately PTD (weeks 32–36). More research is needed to establish whether there is variation in clinical PTD subtypes for other groups.

To sum up, there are ethnic and racial disparities in average gestational length and PTD, with US-born Black women (but not foreign-born Black women) at the highest risk of shorter average gestational length compared with US-born White women. In several other countries, foreign-born Black women were found to be at the highest risk of PTD compared with Whites or native-born women. We will now examine possible explanations for these disparities.

Understanding racial/ethnic disparities: measurement bias in gestational length and preterm delivery

As we have seen, several methods are used to determine gestational length; however, the choice of method might not be without importance when comparing population groups. Naegele’s rule estimates the duration of pregnancy to be 280 days from the first day of the LMP. However, LMP has a number of limitations, which affect its accuracy. Firstly, women may not reliably recall their LMP, particularly in the presence of first trimester bleeding, which may instead be interpreted as the beginning of normal menses. Secondly, LMP assumes that ovulation occurs on the 14th day of a 28-day cycle; however, it is not uncommon for women to have irregular cycles and also to experience delayed ovulation. Uncorrected, misclassification of gestational age using LMP can be in either direction, and it can consequently influence both preterm and post-term delivery rates . Of the studies cited in the above-mentioned section examining disparities in gestational length and PTD, the majority of studies that reported the dating method were based on LMP or the best clinical estimate including LMP , and they failed to correct for the potential misclassification of gestational length.

Ultrasonography before 20 weeks of gestation is often considered the ‘gold standard’ for gestational dating as it is more accurate (±3–5 days) than any other prenatal or postnatal estimate of pregnancy dating, particularly if conducted earlier in pregnancy . Ultrasound assessment assumes that below a certain gestational age, differences in fetal size are related to gestational age, which can lead to underestimation of the gestational age of small but normal fetuses by approximately 1–2 days. However, this systematic error is considered minor compared with the large errors inherent in LMP described earlier. Newer studies often use the ‘best obstetric/clinical estimate’, which is thought to improve accuracy by combining ultrasound and clinical information.

There is a possibility of pregnancy dating bias contributing to the ethnic/racial disparities seen in gestational length and PTD. A recent American study comparing gestational age estimates based on LMP with first trimester ultrasound assessment found important discrepancies (average differences, in days) among women of young maternal age, lower education, non-Hispanic Black race/ethnicity and maternal obesity . For most characteristics, this study indicated on average, consistently longer gestational lengths when the LMP was used compared with ultrasound. A notable exception was by race/ethnicity; non-Hispanic Black women exhibited a gestational length of 0.5 days shorter on average when LMP was used as compared with ultrasound, whereas the gestational length among non-Hispanic White women was 1.2 days longer using LMP compared with ultrasound. The authors suggest that such measurement biases may exacerbate the racial disparity in PTD, and that this should be verified in a more representative sample. This suggestion is consistent with findings from a California-based study where compared with US White women, US Black women had an 80% increased risk of PTD when gestational length was calculated from LMP, but only a 50% increased risk when calculated from ultrasound measurements .

In addition to the characteristics mentioned earlier (young maternal age, lower education, non-Hispanic Black race/ethnicity and maternal obesity), other studies examining discrepancies in gestational age comparing LMP with ultrasound assessment at <20 weeks of gestation found that such differences (both positive and negative) were more common for women who entered prenatal care after the second month of pregnancy , were unmarried and had severe preeclampsia . It is possible that many of the above-named characteristics are differentially distributed between ethnic, racial or migrant groups, therefore affecting studies of ethnic disparities.

Understanding racial/ethnic disparities: potential explanatory factors

What other factors can explain the racial and ethnic differences in gestational length and PTD? The heightened PTD risk among Black women in the US is described as a persistent public health problem, and it has been studied for decades with little to no headway made in identifying important strategies for reducing inequities . Socio-economic conditions (e.g., education), maternal behaviours (e.g., smoking), access to prenatal care, stress, infections (e.g., bacterial vaginosis) as well as genetics have been implicated in the PTD disparity between Blacks and Whites . Differentials in education and access to first trimester prenatal care cannot explain disparities nor can smoking or drug use as non-Hispanic Black women have lower rates of these behaviours than non-Hispanic White women . Four more promising lines of investigation have been identified. Firstly, numerous interventions during pregnancy related to maternal health have been tested with little success, suggesting that preconception health may need to be targeted as mechanisms that lead to PTD may be underway before pregnancy . Secondly, infections, specifically bacteria associated with bacterial vaginosis, are more common among African American women, and it is unknown why this is the case . However, antibiotic interventions have not resulted in any benefit . Thirdly, elements of psychosocial stress have been identified as possible mediating factors as African American women are more likely to experience stressful life events and racism than non-Hispanic Whites before or during pregnancy; however, studies have shown conflicting results .

Finally, there have been attempts to normalize the racial divide in PTD, particularly because other explanations have been generally unsuccessful, by suggesting that there are inherent genetic differences among Black populations that explain shorter gestational length ; however, there has not been much success in the genetic line of investigation. Examining complex gene–environment interactions may add to our understanding of racial/ethnic disparities . Under the genetic hypothesis, one would expect that US-born Black women constitute an intermediate risk group between foreign-born Black and US-born White women due to intermarriage and genetic mixing over previous generations . However, meta-analyses consistently show that US-born Blacks have markedly higher risks than their foreign-born counterparts . A socio-historical hypothesis pointing to continuous exposure to socio-economic and structural discrimination, from slavery to the urban underclass, provides a more comprehensive framework to explain the adverse outcomes of African Americans in the US . It has to be noted that although most studies control for at least one socio-economic indicator, given the complexity of social position, residual confounding by unmeasured socio-economic factors cannot be ruled out .

In the following, we will turn towards studies that mainly examine aspects of migration, looking at PTD among migrant mothers/fathers versus the native or domestically born.