A few notes on birth weight

BW has been a popular metric in research studies. Its attractiveness stems from its precise and reliable measurement and its widespread availability in health information systems. Most studies have focused on LBW, defined as newborn weight <2500 g, which has been used to predict adult health and mortality . However, it is now increasingly recognized that LBW is not an optimal metric for either research studies or use in clinical practice. LBW is a single characteristic that may be the result of multiple causes affecting the uterine environment and mediated by short gestation (i.e., the length of gestation is the strongest correlate of BW), intrauterine growth restriction (IUGR), or both. The prevailing trend is to consider LBW as a complex syndrome with multiple etiologies rather than a single entity.

Birth-weight differences between immigration groups

There are two key findings regarding BW and immigration. First, comparisons between immigrants and the native-born individuals of industrialized countries give mixed results While in Canada and the UK immigrant mothers give birth to newborns who are generally lighter than those born to native-born mothers , the opposite had been documented in the US, France, and Belgium . Inconsistency in results may be due to multiple reasons, including different sociodemographic profiles and health behaviors , differing definitions of immigrant groups, comparison groups, and study settings . Therefore, interpreting the evidence is not straightforward.

The second main finding is that immigrant mothers in industrialized countries give birth to newborns generally heavier than those born to their same-country counterparts who did not migrate and remained in their home countries . This finding may be partly a reflection of the “healthy immigrant effect.” Selective migration may be affected by self-selection to emigrate, immigration admission, and integration policies of the receiving countries . For example, countries receiving primarily voluntary migrants are expected to have healthier immigrants than those admitting mainly refugees. Differences in the social environment and health-care system of the receiving countries are also thought to influence BWs post migration, although the evidence is scarce.

Are lower BWs bad? The literature has tended to interpret differences in BW and LBW between groups as pathological. However, significant BW differences may not necessarily have an impact on adverse outcomes. One classic example is that of infant sex. BWs of female newborns are on average about 200 g less than those of males, but their infant mortality rate is also lower. Besides, living at high altitudes leads to lower BWs but not to higher mortality rates In the same vein, in some countries where the proportion of LBW babies is high, such as Sri Lanka, their infant mortality is lower than expected . Should we therefore interpret differences in BW between immigrants and nonimmigrants as resulting from pathological processes or as adaptations of certain groups to differing environments? In other words, do lower BWs of immigrants mean that they are more affected by fetal growth restriction? Or, that babies of native-born parents in industrialized are more likely to be overweight? These questions have no easy answers, given the multiple determinants of BW. Part of the difficulty also lies in the limitations of BW as a meaningful study outcome. Both mean BW and LBW are jointly influenced by the duration of gestation and fetal growth, among other factors, making it problematic to understand the underlying processes behind the observed differences. While most research has traditionally focused on the risk of being small, less attention has been paid to large size at birth, particularly among immigrant groups .

Birth-weight differences between immigration groups

There are two key findings regarding BW and immigration. First, comparisons between immigrants and the native-born individuals of industrialized countries give mixed results While in Canada and the UK immigrant mothers give birth to newborns who are generally lighter than those born to native-born mothers , the opposite had been documented in the US, France, and Belgium . Inconsistency in results may be due to multiple reasons, including different sociodemographic profiles and health behaviors , differing definitions of immigrant groups, comparison groups, and study settings . Therefore, interpreting the evidence is not straightforward.

The second main finding is that immigrant mothers in industrialized countries give birth to newborns generally heavier than those born to their same-country counterparts who did not migrate and remained in their home countries . This finding may be partly a reflection of the “healthy immigrant effect.” Selective migration may be affected by self-selection to emigrate, immigration admission, and integration policies of the receiving countries . For example, countries receiving primarily voluntary migrants are expected to have healthier immigrants than those admitting mainly refugees. Differences in the social environment and health-care system of the receiving countries are also thought to influence BWs post migration, although the evidence is scarce.

Are lower BWs bad? The literature has tended to interpret differences in BW and LBW between groups as pathological. However, significant BW differences may not necessarily have an impact on adverse outcomes. One classic example is that of infant sex. BWs of female newborns are on average about 200 g less than those of males, but their infant mortality rate is also lower. Besides, living at high altitudes leads to lower BWs but not to higher mortality rates In the same vein, in some countries where the proportion of LBW babies is high, such as Sri Lanka, their infant mortality is lower than expected . Should we therefore interpret differences in BW between immigrants and nonimmigrants as resulting from pathological processes or as adaptations of certain groups to differing environments? In other words, do lower BWs of immigrants mean that they are more affected by fetal growth restriction? Or, that babies of native-born parents in industrialized are more likely to be overweight? These questions have no easy answers, given the multiple determinants of BW. Part of the difficulty also lies in the limitations of BW as a meaningful study outcome. Both mean BW and LBW are jointly influenced by the duration of gestation and fetal growth, among other factors, making it problematic to understand the underlying processes behind the observed differences. While most research has traditionally focused on the risk of being small, less attention has been paid to large size at birth, particularly among immigrant groups .

From birth weight to small for gestational age

As discussed earlier, the limitations of the traditional indicator of “prematurity” or LBW have become more apparent in the last few decades. It is now more widely accepted to conceptualize small newborn size as a result of IUGR, independent of gestational age. The measure of SGA represents an attempt to approximate IUGR and is based on a statistical definition by which babies are labeled as SGA if their BW falls below certain cutoff points at each gestational age. Commonly used cutoffs for SGA are the 3rd, 5th, or 10th percentiles (or <2 SD) of the BW distribution. A collection of the BW cutoff values across all weeks of gestation is summarized in a BW chart, also known as birth-weight curves. Most charts are under the category of BW “references”, which must be distinguished from the fetal growth “standards.” References describe the BW distribution of birth cohorts in a given place and time. They are usually population based and include a large number of healthy and unhealthy newborns. Standards are based on a prescriptive approach that excludes pregnancies affected by morbidities or environmental insults, with the intent of describing “normal” fetal growth and newborn size . Departures from the normal pattern are therefore interpreted as resulting from pathological perturbations. Moreover, various standards are not cross-sectional (i.e., measuring BW at the time of birth) but based on serial fetal ultrasound measurements to further describe normal growth patterns in utero. Fetal growth standards derive cutoff points from the estimated fetal weight in utero.

The incidence of SGA varies according to the references and standards. It is known that infants who are born preterm are usually smaller at a given week of gestation (e.g., 30 weeks) than those who remained in utero at that week and continued until term . This phenomenon is reflected in a higher proportion of SGA among preterm babies than among term babies when ultrasound-based standards are used. However, as BW references fix the proportion of SGA newborns at each gestational age (e.g., 10th percentile = 10% of the lightest babies), the SGA rate derived is the same among preterm than among term infants, thus underestimating IUGR among preterm babies.

A third type of BW charts is the customized standard that generates birth-weight percentiles adjusted for individual characteristics based on the Hadlock fetal growth equations . Customized standards have been introduced in clinical practices in parts of England and other settings . Studies show a stronger association between SGA thresholds based on the customized standards and perinatal mortality than with SGA thresholds derived from population-based references. However, this apparently better prediction of perinatal mortality by customized standards has been attributed to the inclusion of a larger proportion of preterm infants by using ultrasound standards rather than customization . Mikolajczyk and colleagues found that customizing SGA for maternal country of birth marginally improved prediction of adverse outcomes as compared to a non-customized ultrasound-based standard, but additional individual characteristics (e.g., infant sex, parity, maternal weight, and height) did not .

Population-based references have also been tailored to individual characteristics. While it is usual practice to stratify charts by infant sex, based on the observation that males are physiologically bigger than females, some have produced BW charts further stratified by multiplicity, parity, and race/ethnicity , with the intent to describe physiological variations in the BW distribution. These studies also imply that the use of a single reference or standard may not be appropriate for all newborns and that failure to consider specific BW distributions may result in over- or underestimation of abnormal newborn size, with the unintended consequences of misdiagnosis, both absence of or unwarranted intervention and parental anxiety.

Should minority-specific standards be used in clinical settings?

Several studies using a single reference or standard have found that foreign-born women generally have higher SGA rates than the majority locally born population of industrialized countries . For example, the SGA rate defined as a BW < 10th percentile on a single chart based on the Canadian-born population has been found to be about 20% among South Asian immigrants in Ontario, Canada . The interpretation of these disparities as pathological has been disputed. Do these higher rates of SGA indicate that South Asians are more likely to be growth restricted? Or, are these higher SGA rates an artifact of applying to immigrants’ BW curves created on a sample of heavier newborns?

There is an incipient body of evidence comparing the predictive ability of adverse birth outcomes by a single chart versus a set of world region- or ethnic-specific references (i.e., one reference per each major immigrant or ethnic group) . These studies consistently reveal two main findings. First, SGA rates of some minority groups, such as South and East Asians, are substantially higher than those of the native born when a single reference is used but are similar when the world region- or ethnic-specific curves are used (e.g., SGA rates defined as BW < 10th percentile were, by definition, about 10% on South Asian- or Pakistani-specific curves but almost doubled when a curve based on the majority population was used) . Second, despite their higher rates of SGA based on a single chart, infants to South Asian mothers did not have a higher risk of adverse outcomes, such as perinatal or neonatal mortality and morbidity, when the world region- or ethnic-specific charts were used . This finding suggests that, in some cases, differences in absolute BW between immigrant groups may not be causally related to adverse outcomes in the same way as females do not have higher infant mortality than males, despite their lower BW. Recent evidence from the INTERGROWTH-21 multicountry study further suggests that differences in the absolute BW according to ethnic group or geographic origin may not necessarily reflect pathological differences. In this study, women from the Indian subsample had newborns about 400 g lighter than those born to European mothers , and yet all women meeting the inclusion criteria had healthy pregnancies, suggesting that the role of pathological influences on these differences has been reduced, although not eliminated, by the study design. Moreover, results from the INTERGROWTH-21 study further support the hypothesis that the higher rates of SGA observed among South Asian immigrants may be an artifact of applying cutoffs developed on a sample of heavier newborns composed of the majority locally born population. For example, in Ontario, the 10th percentile of BW among males born to Canadian-born mothers at 40 weeks gestation was 3.15 kg and 2.94 kg among male newborns of immigrants from South Asia . Interestingly, the 10th percentile observed for the pooled multicountry sample in the INTERGROWTH-21 study was 2.88 kg , much closer to that of the South Asian population in Ontario. According to the international standard, newborns of South Asian mothers seem to be doing well in Canada, while newborns of Canadian-born mothers seem to be exceeding growth expectations. Whether the heavier newborns of Canadian-born mothers reflect an optimization of fetal growth, a consequence of the obesity epidemic or a combination of the two processes is a matter for further research.

It has to be noted, however, that for some immigrant groups, such as refugees or undocumented migrants, having lower BW might indicate true IUGR due to adverse living conditions surrounding preconception and pregnancy .

To sum up, assessing SGA (and LGA) among immigrants using references based on the majority locally born population of industrialized countries is likely to misclassify SGA (and LGA) among some immigrant groups and misguide their care.

Then, should minority-specific references be used in clinical settings instead? To answer this question we must distinguish between the use of charts for research and for clinical practice. While research using minority-specific charts may help us understand the relationship between BW distributions of immigrant groups and their risk of adverse outcomes if compared against a meaningful benchmark, their application in clinical settings is problematic. First, assignment of newborns to immigrant or ethnic groups may not be accurate due to self-reporting. Second, challenges increase when one of the parents is foreign-born and the other is not. Third, if categories are based on ethnicity rather than on birthplace, things become even more complicated for second-generation mothers, who are not immigrants themselves but born to one or two immigrant ancestors. Fourth, extrapolation from one setting to another may not be appropriate, since the number and composition of minority groups may be dictated by diverse push and pull factors shaping local demographic profiles. Finally, adoption of minority-specific curves may contribute to the proliferation of newborn charts further hampering comparability.

Besides these limitations, owing to minority-specific charts, there are limitations common to population-based references that stem from the fact that they are contingent upon time and place, which may hamper their application in clinical settings. First, their generalizability to other populations may be limited due to their singularity. Second, BW charts may no longer represent the BW distribution of the population if profound demographic or behavioral changes occur over time. For example, changes in smoking or overweight rates or in the composition of immigrants originating from a given world region may affect the BW distributions and morbidity profiles, raising potential concerns about the need to update the BW charts to keep pace with the new situation. Third, the approach of generating local BW charts may lead to innumerable standards or references that may create unnecessary complexity and hamper comparability between populations . Fourth, and perhaps more important, locally based references may have the unintended consequence of “normalizing” otherwise abnormal distributions of BW. For example, if BW curves are created based on a population characterized by high obesity rates, both the SGA and the LGA cutoff points will be shifted upward and, therefore, the proportion of overweight babies will be underestimated and many overweight babies will be misclassified as normal. Such charts may be misleading , particularly if applied to healthy immigrant populations . Fifth, inclusion of unhealthy pregnancies in BW references is a major concern. Most newborn BW charts, including ethnic-specific references, have been constructed based on the entire population, including both healthy and unhealthy pregnancies, with differing exclusion criteria. Inclusion of varying and unknown number of unhealthy pregnancies may affect the BW distribution in unexpected ways, making it difficult to delimit normal size at birth.