Fetal growth: a review of terms, concepts and issues relevant to obstetrics

Authors

  • C. Mayer,

    1. Department of Obstetrics and Gynaecology, University of British Columbia and the Children's and Women's Hospital of British Columbia, Vancouver, Canada
    Search for more papers by this author
  • K. S. Joseph

    1. Department of Obstetrics and Gynaecology, University of British Columbia and the Children's and Women's Hospital of British Columbia, Vancouver, Canada
    2. School of Population and Public Health, University of British Columbia, Vancouver, Canada
    Search for more papers by this author

Correspondence to: Dr C. Mayer, Room 2H30, Department of Obstetrics and Gynecology, Women's Hospital of British Columbia, 4500 Oak Street, Vancouver, British Columbia, Canada V6H 3 N1 (e-mail: cmayer@cw.bc.ca)

ABSTRACT

The perinatal literature includes several potentially confusing and controversial terms and concepts related to fetal size and growth. This article discusses fetal growth from an obstetric perspective and addresses various issues including the physiologic mechanisms that determine fetal growth trajectories, known risk factors for abnormal fetal growth, diagnostic and prognostic issues related to restricted and excessive growth and temporal trends in fetal growth. Also addressed are distinctions between fetal growth ‘standards’ and fetal growth ‘references’, and between fetal growth charts based on estimated fetal weight vs those based on birth weight. Other concepts discussed include the incidence of fetal growth restriction in pregnancy (does the frequency of fetal growth restriction increase or decrease with increasing gestation?), the obstetric implications of studies showing associations between fetal growth and adult chronic illnesses (such as coronary heart disease) and the need for customizing fetal growth standards.

INTRODUCTION

Fetal size and fetal growth trajectories are important indicators of fetal health. This article reviews fetal growth from an obstetric perspective and addresses various issues including the physiologic mechanisms that determine fetal growth trajectories, known risk factors for abnormal fetal growth, diagnostic and prognostic issues related to restricted and excessive growth and temporal trends in fetal growth.

The perinatal literature contains several potentially confusing terms and concepts related to fetal size and growth. These include distinctions between fetal growth ‘standards’ and fetal growth ‘references’, and between fetal growth charts based on estimated fetal weight vs those based on birth weight. There is also a lack of clarity around fundamental concepts such as the incidence of fetal growth restriction in pregnancy: does the frequency of fetal growth restriction increase or decrease as gestation advances (or is it a constant 3 or 10% for each gestational week)? The demonstrated associations between in-utero growth and adult chronic illnesses (such as coronary heart disease) are another potential source of confusion in obstetrics: should pregnancy interventions be predicated on improving such long-term outcomes? Finally, there is the seductive proposition regarding the need for customizing fetal growth standards. Should fetal size and growth be assessed in the context of fetal gender alone or should other physiologic parameters such as maternal height, weight, parity and ethnicity/race provide additional context? This review attempts to clarify these issues by providing a brief synthesis of the prevailing perspectives.

UTEROPLACENTAL FUNCTION AND FETAL GROWTH

Fetal growth is heavily modulated by placental function, with the placenta serving the critical respiratory, hepatic and renal functions of the fetus. Early placental problems can occur because incomplete trophoblast invasion results in a failure of remodeling of the myometrial arteries and reduced uteroplacental blood flow that is typically associated with pre-eclampsia and fetal growth restriction[1].

Both animal and human studies show that the ability of the uteroplacental unit to support the fetus diminishes steadily as gestation advances[2-7]. Whereas increases in umbilical artery diameter, blood-flow velocity and blood-flow volume with advancing gestation represent changes in response to the increasing demands of a growing fetus, these parameters in fact conceal a gradually declining ability of the uteroplacental system to meet fetal demands from midpregnancy onwards. For instance, studies in chronically instrumented fetal lambs show significant decreases in the arterial partial pressure of oxygen, base excess, oxygen saturation and oxygen content, and significant increases in the arterial partial pressure of carbon dioxide between 90 and 103 days' gestation[7]. In humans as well, the average uterine blood-flow volume per kg of estimated fetal weight declines from 993 mL/min/kg at 24 weeks to 360 mL/min/kg at 34 weeks and to 296 mL/min/kg at 38 weeks' gestation[5].

ABNORMAL FETAL GROWTH

Fetal growth abnormalities are commonly diagnosed using criteria such as low birth weight, macrosomia, small-for-gestational age (SGA) and large-for-gestational age (LGA). Labeling fetuses as SGA or LGA based on normative values from a fetal growth standard is analogous to diagnosing malnutrition in children using a weight-for-age chart. Pediatric weight-for-age percentiles have been developed through the longitudinal follow-up of normal children with serial measurements obtained at regular intervals. Under this formulation, children who fall below the 3rd percentile or above the 97th percentile of weight for age are labeled malnourished. The theoretical basis for using the 3rd and 97th percentile cut-offs from such a weight-for-age standard is similar to the rationale of statistical inference using a P-value cut-off of 0.05 for rejecting the null hypothesis (i.e. a 2.5% error rate that is two tailed). It is also analogous to using the mean weight ± 2 SD of a standard population as cut-offs for abnormal weight-for-age, since approximately 5% of subjects will fall outside this range. Although by definition 3% of normal children will fall below the 3rd percentile weight-for-age cut-off (false positives), the probability that a child with undernutrition will fall below this cut-off is higher than 3% (with the magnitude of this latter probability a function of the severity of the pathologic process causing the under nutrition)[8, 9]. One important feature of this method of identifying abnormal growth is the use of normal children in the creation of the standard. This issue is of particular relevance because many fetal growth references are based on fetuses from normal and abnormal pregnancies, without sufficient acknowledgment of the implications for normative interpretation using percentiles (see below).

Small-for-gestational age

A fetus is labeled SGA if its size (e.g. estimated fetal weight, estimated abdominal circumference) falls below some cut-off percentile of size for gestational age (e.g. the 3rd percentile). Although the SGA label implies fetal growth restriction, such fetuses will include some normal yet constitutionally small fetuses by definition; the ratio of constitutionally small fetuses to fetuses that are small because of an abnormal, pathologic process will depend on the prevalence of such illnesses in the population. For instance, among normal pregnancies, 3% of fetuses will fall below the 3rd percentile of a fetal growth standard and all of these will be constitutionally small, whereas among women with pre-eclampsia substantially more than 3% of fetuses will fall below the 3rd percentile, and a substantial fraction of such fetuses will be pathologically small (i.e. growth restricted).

Fetal growth restriction

Fetuses whose growth potential has been compromised are referred to as growth restricted. Just as all SGA fetuses may not be growth restricted (some may be constitutionally small), growth-restricted fetuses may or may not be SGA. For instance, a fetus whose growth trajectory falls sequentially from a stable 60th percentile of estimated fetal weight to the 50th, 40th and 30th percentiles of estimated fetal weight for gestational age over several weeks could be labeled growth-restricted though not SGA[10]. Similarly, a fetus whose growth is substantially compromised owing to maternal smoking could be labeled growth-restricted even if the estimated fetal weight for gestational age is above the 10th percentile.

Large-for-gestational age and excessive fetal growth

A fetus is labeled LGA if its size (e.g. estimated fetal weight, estimated abdominal circumference) is above the 90th or 97th percentile (or other cut-off) of size for gestational age. As with SGA fetuses and growth-restricted fetuses, LGA fetuses will include normal (i.e. constitutionally large) fetuses by definition, and fetuses with excessive fetal growth do not have to be LGA. The ratio of constitutionally large fetuses to fetuses that are large because of a pathologic process (such as maternal diabetes mellitus) will depend on the prevalence of such pregnancy complications in the population.

Low birth weight and macrosomia

Low birth weight (< 2500 g) and macrosomia (birth weight ≥ 4000 or ≥ 4500 g) are indices of size with substantial utility for various reasons including ease of measurement and strong correlation with adverse perinatal outcomes. The 10th percentile value at 37 weeks' gestation of current fetal growth references approximates the low birth weight cut-off (2452 g for females and 2552 g for males in the Canadian reference[11] and 2484 g for females and 2596 g for males in the United States reference[12]). The 90th and 97th percentiles of current fetal growth references loosely approximate the macrosomia cut-offs. The limitation of the low birth weight index arises because discounting gestational age makes the index a heterogeneous entity that includes both fetuses that are preterm and those that are SGA.

Morphologic heterogeneity in growth restriction and excessive growth

Growth-restricted fetuses have been traditionally categorized as symmetrically or asymmetrically growth restricted because of differences in both etiology and prognosis[13]. Symmetrical growth restriction is believed to occur as a result of a global insult such as aneuploidy or a viral infection early in pregnancy, while asymmetrical growth restriction – with a brain-sparing effect – is thought to be the result of complications later in pregnancy (e.g. pre-eclampsia). These findings have been challenged, however, with studies showing asymmetrical growth associated with aneuploidy[14], symmetrical growth associated with pre-eclampsia[15] and evidence of morbidity despite brain sparing in asymmetrical growth-restricted fetuses[16].

Asymmetric increases in abdominal circumference after 32 weeks' gestation characterize excessive fetal growth seen in diabetic pregnancies. Fetuses of diabetic mothers have an excess deposition of fat and more muscle growth in the interscapular areas and abdomen[17]. These morphologic differences are responsible for substantially higher rates of shoulder dystocia among macrosomic fetuses of diabetic pregnancies compared to macrosomic fetuses of non-diabetic pregnancies[18].

RISK FACTORS FOR ABNORMAL FETAL GROWTH

Numerous risk factors for fetal growth restriction and excessive fetal growth have been identified and can be classified into maternal, fetal and placental factors[19, 20].

Maternal factors

Low prepregnancy weight and poor weight gain in pregnancy are associated with higher rates of fetal growth restriction, though poor weight gain may be merely a marker of a pregnancy complication rather than the cause of growth restriction. Multifetal pregnancies can lead to mild or severe fetal growth restriction. Various maternal illnesses – including hypertensive disorders of pregnancy, diabetes mellitus, autoimmune diseases (such as anti-phospholipid syndrome and systemic lupus erythematosus), chronic maternal illnesses and infections – increase the risk for fetal growth restriction. Chronic hypertension, gestational hypertension and pre-eclampsia are the most common of these maternal conditions and typically increase the risk of fetal growth restriction by three- to four-fold[21-23]. Commonly abused drugs including tobacco, alcohol and cocaine increase the risk of growth restriction; women who smoke in pregnancy have a two- to three-fold higher risk of fetal growth restriction[24]. Therapeutic drug exposure during pregnancy to anticonvulsants such as diphenylhydantoin, anticoagulants such as heparin and folic acid antagonists such as methotrexate can also lead to fetal growth restriction.

Maternal diabetes mellitus, including Type I and Type II diabetes and gestational diabetes, is associated with excessive fetal growth. Maternal hyperglycemia due to diabetes leads to increased secretion of insulin by the fetus and muscle growth, deposition of excess fat and organomegaly in the fetus. Higher maternal age and obesity are other risk factors for excessive fetal growth.

Fetal factors

Aneuploidy and various other genetic syndromes are associated with fetal growth restriction. Down syndrome, trisomies 13 and 18 and Turner's syndrome are associated with higher rates of growth restriction. Gene deletions and duplications are also responsible for syndromes involving growth restriction. Genetic factors also underlie excessive fetal growth. For instance, fetal hyperinsulinemia associated with Beckwith–Wiedemann syndrome leads to increased fetal growth.

Placental factors

Various placental abnormalities, such as placenta previa, velamentous cord insertion and vasa previa, are associated with higher rates of fetal growth restriction. Uterine abnormalities that affect placental structure or function may also lead to compromised fetal growth. Abnormal vascular connections in the placenta are responsible for severe fetal growth restriction and growth discordance associated with multifetal pregnancies.

DIAGNOSIS OF ABNORMAL FETAL GROWTH

Abnormal fetal growth is typically an ultrasound diagnosis based on discrepancies between actual and expected biometric measurements for a given gestational age. Diagnosis is highly dependent on accurate pregnancy dating. The ultrasound examination leading to diagnosis is usually performed as part of a routine screening examination or because risk factors for abnormal growth are present.

There are no universally accepted criteria for the diagnosis of abnormal fetal growth. Commonly used definitions for growth restriction are based on either estimated fetal weight or abdominal circumference below a threshold percentile of size for gestational age (e.g. 10th, 5th or 3rd percentile). Multiple definitions also exist for macrosomia and these are also based on ultrasound-estimated fetal weight or abdominal circumference (> 90th or > 95thpercentile or ≥ 4000 g or ≥ 4500 g estimated fetal weight).

Abnormal biometry prompts clinical and ultrasound examinations that serve to identify fetal, placental or maternal causes of abnormal fetal growth. Such investigations have a limited ability to discriminate between the constitutionally small (or large) fetus and the fetus at risk for an adverse perinatal outcome. However, despite its limitations, biometry remains the current gold standard for the diagnosis of abnormal fetal growth, as none of the proposed novel alternatives has clearly demonstrated superiority.

Estimated fetal weight and abdominal circumference

There is no consensus on whether the diagnosis of abnormal growth should be based on estimated fetal weight, estimated abdominal circumference or both. Several different equations have been proposed for the estimation of fetal weight. Despite the inclusion of multiple parameters in these prediction formulae (including the biparietal diameter, occipitofrontal diameter, head circumference, femur length, abdominal diameter and abdominal circumference), errors in estimated fetal weight range from 10–15% on average, but can be as high as 25%[25-27]. More importantly, errors in estimated fetal weight tend to be largest when accurate estimation of fetal size and fetal growth is most clinically relevant i.e. when the estimated fetal weight is either less than 1500 g or greater than 4000 g. Proposed formulae specifically designed for fetuses less than 1500 g do not consistently improve weight estimation beyond the commonly used Hadlock formula[28]. Although correlation between estimated fetal weight and actual birth weight is better than that obtained based on abdominal circumference alone[29], abdominal circumference may better predict neonatal outcomes in infants with a birth weight of between 400 g and 1000 g[30]. Similarly, patterns of growth may also alter the accuracy of estimated fetal weight. Among severely growth-restricted preterm fetuses that are asymmetrically grown, fetal weight may be better estimated by formulae that do not include femur length as a component in the prediction equation[31].

Errors in estimated fetal weight may also occur, in part, owing to the fact that estimation based on biometric measurements assumes a uniform density of tissue. For example, the relative amount of adipose tissue (less dense than lean body mass), is greater in the macrosomic fetus and lower in the growth-restricted fetus. This can lead to overestimation of estimated fetal weight in the macrosomic fetus and underestimation in the growth-restricted fetus.

Estimation of fetal weight for the purpose of identifying fetuses at risk for adverse outcomes related to macrosomia yields disappointing results. The detection rate for macrosomia is approximately 29% on average, and the false-positive rate is substantial (12%)[32].This is true for the diagnosis of macrosomia based on sonographically estimated fetal weight and sonographically estimated abdominal circumference. The inability to accurately diagnose macrosomia is an important reason why elective induction of labor or elective Cesarean delivery for suspected macrosomia is deemed inappropriate[33, 34]. Nevertheless, some experts recommend considering elective Cesarean delivery when the estimated fetal weight exceeds 4500 g in diabetic women and 5000 g in non-diabetic women[33, 34].

Fetal ponderal index

A number of small studies have examined estimated fetal ponderal index (weight/length3; a measure of leanness) as a predictor of adverse neonatal outcomes in low- and high-risk women[35, 36]. Although some studies show a good correlation between estimated fetal ponderal index and neonatal ponderal index, there is no evidence to suggest that the fetal ponderal index estimate is superior to estimated fetal weight in identifying the fetus at risk for the perinatal complications of growth restriction. It is also uncertain if the fetal ponderal index adds value in discriminating between the constitutionally small fetus and the growth-restricted fetus[35, 36].

Soft-tissue markers

Several methods of ultrasound visualization of soft tissue have been proposed as part of the assessment of excessive fetal growth. However, using subcutaneous tissue assessment to identify excessive growth does not appear to be more effective in identifying macrosomia than does estimated fetal weight based on standard biometry[37, 38]. A qualitative improvement in size assessment is required (through the addition of other modalities of size assessment or a novel use of existing methods) in order to improve diagnosis.

Interval growth

Interval growth may be used to identify fetuses deviating from their growth trajectory and can assist in differentiating normal from abnormal growth if gestational age is uncertain. There is no consensus regarding what constitutes normal growth velocity at any given gestational age for each biometric parameter, but assessment of abdominal circumference interval growth is commonly used in the clinical setting. Measurement of interval growth is particularly susceptible to inaccuracies in biometric measurements that are caused by intra- and interobserver variability[39]. This is especially true when the time interval between examinations is short. False-positive rates are highest in the late third trimester, when growth velocity slows and examinations are commonly repeated at 2-week intervals or less[40].

Three-dimensional ultrasonography

Equations using fetal thigh volume or fractional limb volumes obtained using three-dimensional (3D) ultrasound have been proposed for the assessment of estimated fetal weight, although preliminary results are equivocal[32, 41]. If 3D estimates of fetal size prove more accurate than two-dimensional estimates, the added cost, time and expertise required for such assessment will have to be addressed before this technology can be deployed widely.

First- and second-trimester diagnosis

Fetuses with growth restriction due to placental insufficiency are at high risk for adverse perinatal outcomes. While abnormal umbilical artery blood flow is diagnostic of placental insufficiency in the SGA fetus, such a finding is a late manifestation of abnormal placental development. Several first- and second-trimester serum and ultrasound markers of placental function have been used in attempts to arrive at an earlier diagnosis of fetal growth restriction due to placental insufficiency. Numerous studies have evaluated various combinations of first-trimester markers of placental function. Candidate variables for the early diagnosis of fetal growth restriction have included biometry (including the discrepancy between measured and expected crown–rump length), interval growth between the first and second trimesters, maternal serum markers of placental function (e.g. β-human chorionic gonadotropin, pregnancy-associated plasma protein-A), placental volume and morphometry assessment and uterine artery Doppler waveform analysis[42-44]. Although such models are able to identify fetuses at increased risk for growth restriction with varying degrees of accuracy, none of these methods has proven to be more accurate or clinically useful than standard biometric methods of assessment[45]. Similarly, the accuracy of second-trimester ultrasound and other markers for diagnosing macrosomia remains low[46].

Umbilical artery Doppler studies

Umbilical artery Doppler velocimetry has become the clinical standard for identifying early-onset fetal growth restriction, i.e. at < 34 weeks' gestation[47-49]. Doppler ultrasound determination of umbilical artery blood flow reveals impedance in the fetoplacental circulation, with absent or reversed end-diastolic blood flow diagnostic of severe fetal growth restriction. The use of umbilical artery Doppler ultrasound has led to reductions in perinatal death related to complications of placental insufficiency and iatrogenic preterm delivery[50, 51]. However, umbilical artery Doppler is not reliable for the identification of late-onset growth restriction and associated complications. Unfortunately, late-onset fetal growth restriction is more prevalent than growth restriction of early onset, and most adverse outcomes attributable to late-onset growth restriction occur in fetuses with normal umbilical artery Doppler waveforms. Middle cerebral artery Doppler waveform analysis is emerging as a promising diagnostic tool for the diagnosis of late third-trimester growth restriction among fetuses with normal umbilical artery Doppler waveforms, but further studies are required to support its widespread use[52]. The routine use of umbilical artery Doppler studies has no demonstrated utility in the absence of growth restriction or associated hypertension and no role in the assessment of the fetus at risk for macrosomia.

PROGNOSIS (COMPLICATIONS ASSOCIATED WITH ABNORMAL FETAL GROWTH)

Abnormalities of fetal growth serve as indicators of pregnancy complications, and are associated with adverse perinatal outcome.

Short-term outcome

Fetal growth restriction is associated with substantially higher rates of perinatal morbidity and mortality. To a large extent the excess mortality and morbidity associated with fetal growth restriction arise from the cause of the growth restriction. Thus congenital anomalies may be the cause of fetal growth restriction and subsequent perinatal death. Chronic antepartum hypoxia due to maternal or placental causes is another mechanism that results in fetal growth restriction and perinatal mortality or serious neonatal morbidity. Among live births between 32 and 42 weeks' gestation, cerebral palsy rates are four- to six-fold higher among neonates with birth weight for gestational age below the 10th percentile of a fetal growth standard, as compared with neonates with a birth weight for gestational age between the 25th and 75th percentiles[53]. Similarly, cerebral palsy rates are about one-and-a-half to three times higher among live births with a birth weight for gestational age above the 97th percentile[53]. Other neonatal morbidity associated with growth-restricted fetuses includes polycythemia, hyperbilirubinemia and hypoglycemia, which may occur as a consequence of chronic hypoxia and depleted hepatic glycogen stores.

The short-term outcomes of excessive fetal growth include birth trauma, which arises mostly as a consequence of a mismatch between the size of the fetus and the maternal pelvis. Brachial plexus injury is the most common type of injury seen among macrosomic infants and is caused by lateral extension of the fetal neck during labor and delivery. Shoulder dystocia is responsible for most cases of obstetric brachial plexus injury, although a significant fraction of brachial plexus injury occurs in utero[54-56]. Other morbidity associated with excessive fetal growth includes polycythemia, hyperbilirunemia and hypoglycemia, which occurs secondary to fetal hyperinsulinemia and consequent hypoxia.

Long-term outcomes

Numerous longitudinal studies have shown associations between birth weight and blood pressure and birth weight and coronary heart disease[57-61]. Other related associations include those between low birth weight and hypertension and low birth weight and diabetes mellitus. The Barker hypothesis explains these associations as being a consequence of fetal programming in utero[57-59]. Should interventions to improve fetal growth be predicated on short-term outcomes or does the fetal programming literature justify interventions in order to prevent adult disease? It is noteworthy that associations between fetal growth indices and adult diseases are weak. For instance, systolic blood pressure in adult life is lower by 1–3 mmHg for a 1-kg increase in birth weight[62, 63]. Calculations show that increasing the mean population birth weight by as much as 300 g would result in a 6–7% reduction in coronary heart disease in a country with a birth-weight distribution such as that seen in Canada[63]. In fact, such large increases in fetal growth are neither attainable nor advisable given the potential negative consequences of excessive fetal growth, including pregnancy complications, maternal obesity and obesity in the offspring. It is also worth noting that women who give birth to low-birth-weight infants themselves have a substantially higher risk of coronary heart disease[64, 65]. These findings have led to explanations other than fetal programming for such long-term associations: women with a genetic cardiovascular susceptibility are at an increased risk of vascular complications, including pre-eclampsia and other hypertensive disorders, which makes them prone to having preterm/low-birth-weight infants who also inherit a susceptibility to heart disease. Clearly, the mechanisms underlying the associations between in-utero phenomena and adult disease remain speculative and the potential for intervention to address such long-term effects is unclear. Indeed, there are strong imperatives to intervene (or not) based on short-term outcomes, which usually trump long-term considerations.

RECENT TRENDS IN SGA AND LGA

There have been reductions in SGA live births and increases in LGA live births among term births over the last two decades[66, 67]. In the United States, rates of SGA live births decreased from 9.9% in 1992 to 8.2% in 2003, while rates of LGA infants first increased from 9.5% in 1992 to 10% in 2000 before decreasing to 9.3% in 2003. Macrosomia rates also decreased, from 2.2% in 1992 to 1.6% in 2003. Increases in maternal prepregnancy weight and body mass index, increases in gestational weight gain, and reductions in maternal smoking have acted to increase fetal growth, whereas increases in older maternal age, hypertension and diabetes have had the opposite effect. Some of the observed changes have occurred owing to improvements in the accuracy of gestational-age assessment[66]. A significant portion of the impact on population birth weight and gestational age has occurred because of earlier delivery at term following induction of labor or Cesarean delivery[67].

ISSUES AND CONTROVERSIES IN THE FETAL GROWTH LITERATURE

Fetal growth standards vs fetal growth references

Pediatric weight-for-age standards, which have been developed based on the longitudinal follow-up of normal children, have served a singular, normative purpose. Fetal growth charts, on the other hand, have tended to be either normative or descriptive. The normative charts developed based on fetuses from normal pregnancies are referred to as fetal growth standards, while the descriptive fetal growth charts based on populations of fetuses/infants from normal and complicated pregnancies are called references[68].

Understanding the difference between fetal growth standards and fetal growth references is critical for percentile-based normative interpretation. The 3rd percentile of birth weight for gestational age obtained from a fetal growth standard is likely to be substantially higher than the 3rd percentile at the same gestational age obtained from a fetal growth reference, because the latter was constructed based on a population that included fetuses affected by various complications of pregnancy. Thus, fetal growth references typically use a less stringent percentile cut-off to identify abnormal growth (e.g. 5th or 10th percentile of birth weight-for-gestational age instead of the 3rd percentile).

Contemporary references for both Canada[11] and the United States[12] are based on all live births, including those with congenital anomalies and those whose mothers had pre-eclampsia, diabetes mellitus, etc. Other references[69] exclude some influences that are substantially associated with poor fetal growth (e.g. congenital anomalies, death before discharge) but not others. As mentioned, a more liberal birth weight-for-gestational age percentile cut-off (i.e. the 5th or the 10th percentile) based on these references is typically used for normative (prescriptive) purposes. On the other hand, use of the 10th percentile (estimated fetal weight for gestational-age cut-off) from a fetal growth standard based on normal pregnancies would represent a liberal definition of SGA (and result in a relatively larger number of false-positive cases). From an obstetric standpoint, the utility of charts based on birth weight (rather than estimated fetal weight) is questionable, since this index is unavailable for obstetric decision-making in the antenatal and intrapartum periods.

Patterns of growth restriction and excessive growth with increasing gestation

The assumption within fetal growth standards that the 3rd (or any other specific) percentile denotes an SGA cut-off at any gestational age appears to imply a constant rate of growth restriction across gestation. Assuming constancy of fetal growth restriction rates across pregnancy may be an untenable proposition, given the large changes in perinatal mortality with advancing gestation. In fact, the extended fetuses-at-risk formulation[70] suggests that rates of fetal growth restriction rise with increasing gestation in concert with gestational-age-specific increases in perinatal mortality rates (Figure 1). The fetuses-at-risk approach was first proposed over 20 years ago in connection with the gestational age-specific stillbirth rate, and under this formulation the stillbirth rate at any gestational age is calculated as the number of stillbirths at that gestational age expressed as a proportion of the number of fetuses at risk of stillbirth at that gestational age[71]. The pattern of increasing growth restriction with increasing gestation observed under the fetuses-at-risk model is congruent with rising perinatal mortality rates and consistent with the current physiologic understanding regarding the diminishing capacity of the uteroplacental system to meet fetal demands as gestation advances (see ‘Uteroplacental function and fetal growth’, above). However, the fetuses-at-risk approach and especially its extension to pregnancy-related phenomena other than stillbirth remain controversial.

Figure 1.

Gestational age-specific rates of small-for-gestational age (SGA) births and perinatal deaths among singletons and twins, Canada 1991–1997 (adapted from Joseph[70] with permission). image, Singleton SGA births; image, twin SGA births; image, singleton SGA deaths; image, twin SGA deaths.

Customized standards vs population-based standards

The idea of customizing fetal growth standards has intuitive appeal. If small women tend to have smaller babies then incorporating maternal height, weight and ethnicity/race into the equation will help in the identification of babies who are small because of fetal growth restriction and not because of constitutional reasons[72-74]. Several studies have shown that customized fetal growth charts perform better than uncustomized charts in identifying infants at risk for adverse perinatal outcomes[75-77]. For instance, in a study using Swedish data, stillbirth rates among births classified as SGA and non-SGA were as follows: 2.0 per 1000 total births among births categorized as non-SGA by both the customized and uncustomized charts; 2.4 per 1000 total births among births categorized as SGA by the uncustomized chart and as non-SGA by the customized chart; 11.9 per 1000 total births among births categorized as non-SGA by the uncustomized chart and SGA by the customized chart; and 9.9 per 1000 total births among births categorized as SGA by both the customized and uncustomized charts[75]. Similar differentials between the four groups were seen for other outcomes such as neonatal death and 5-minute Apgar score < 4[75].

Customized fetal growth charts have been criticized because their improved performance appears to be a consequence of an artifact and not because of a real improvement in predictive ability[78-83]. The method used to construct customized fetal growth charts uses Hadlock's proportionality formula, which results in a substantially higher proportion of preterm infants being identified as SGA. If this artifactual identification of preterm infants is addressed, the apparent benefits of customization disappear. An indication of this is in fact evident even in the studies that appear to validate the customized charts. It is incongruous that the highest rates of stillbirth, neonatal death and live birth with a 5-minute Apgar score of < 4 are seen not in births categorized as SGA by both the customized and uncustomized charts (presumably the most growth-restricted category) but among the live births categorized as SGA by the customized chart and non-SGA by the uncustomized chart (presumably those with borderline growth restriction picked up by the customized chart and not by the uncustomized chart). The utility of customization appears to be limited because the factors used for customization are not powerful predictors of birth weight[79]. Although maternal height, weight and ethnicity/race are significant predictors of fetal size, they do not explain a large proportion of the variability in birth weight, and this limits their utility for customization.

These criticisms are refuted by proponents of customization who argue that the better identification of preterm SGA cases highlights the poor performance of uncustomized charts at preterm gestation[84]. Studies suggest that customized charts selectively identify SGA fetuses of mothers who smoke, have a high body mass index or other pathologic characteristics[85]. Other counterpoints include the finding that factors used for customization explain about 29% of the variability in birth weight for gestational age[86].

The contemporary literature is clearly divided on the merits of customized fetal growth standards[72-86]. Fortunately, contemporary clinical practice bases decisions regarding fetal health status on multiple indices of fetal health, including size, growth velocity, amniotic fluid volume and umbilical artery Doppler. It is interesting that recent developments in pediatrics have led to the creation of a new weight-for-age standard based on normal infants and children followed longitudinally, with subjects recruited from several different countries including Brazil, Ghana, India, Norway, Oman and the United States[87]. The logic of customizing fetal growth standards by race/ethnicity and abandoning customization at birth suggests some flaw in our conceptualization of fetal vs infant growth.

From a conceptual standpoint, it would appear that cut-off size values for diagnosing abnormal fetal growth should be based on studies that link estimated fetal size in normal pregnancy to rates of perinatal mortality and serious neonatal morbidity (or other longer-term outcomes). There is a need to move away from the percentile-based identification of fetal growth restriction (and excessive growth) to the identification of fetal growth restriction (and excessive growth) based on a clinically relevant outcome-based criterion[9]. For example, the estimated fetal size cut-off for growth restriction at any gestation should be the estimated size at and beyond which perinatal mortality and serious neonatal morbidity rates are significantly increased relative to optimal estimated size. Such cut-offs could become inputs, along with biochemical and other imaging parameters, in prognostic equations that serve to determine whether obstetric intervention through early delivery is appropriate or not.

CONCLUSION

The fetal growth literature includes some potentially confusing terms and concepts and an enormous body of sometimes conflicting evidence. The confusion derives in part from the relatively inaccessible nature of the fetus, which presents a challenge with regard to accurate measurement of size and growth. Future developments, including qualitatively improved methods for assessing fetal size, will bring coherence to this field through new concepts, better theories and more accurate evidence.

ACKNOWLEDGMENT

K. S. Joseph's work is supported in part by the Child and Family Research Institute.

Ancillary