- Email: [email protected]
Customized fetal growth standards: rationale and clinical application
Customized Fetal Growth Standards: Rationale and Clinical Application Jason Gardosi
Accurate assessment of fetal growth status requires the definition of an optimal standard, which represents the growth potential of the baby. Against this standard, individually ‘customized’ percentiles can be calculated. They improve the distinction between normal and abnormal, and help in our understanding and diagnosis of pathological fetal growth. This method can be used as a tool for epidemiological analysis as well as for prospective clinical monitoring. © 2004 Elsevier Inc. All rights reserved. etal growth restriction (FGR) can be defined as the failure of the fetus to reach its growth potential. This potential varies in individual babies and needs to be adjusted or ‘customized’ for each pregnancy. Customized assessment of birth weight or growth therefore means an evaluation against the optimal, individually adjusted standard.
F
Establishing the Standard for Growth The development of a customized standard requires consideration of the following factors: 1. Accurate dating. Ultrasound dating is much more accurate than menstrual dating. Both methods can have error, but the error is much wider for menstrual dates and the distribution is skewed.1,2 Because the distribution of menstrual dating error is positively skewed, many birth weight points at term appear at later gestations than they actually should be, leading to an artificial flattening of the growth curve and apparent increase in “post-term” births.2,3 In reality, fetal growth in utero in normal pregnancy continues without diminished velocity until birth,4,5 and even birth weight charts that are based on accurately dated pregnancies have a near-straight line at term.6 Dating error can also severely affect accuracy in the preterm gestation range. 2. The growth standard also needs to be individually adjusted for physiological factors known to affect birth weight and growth. Significant variables include maternal height, weight in early pregnancy, parity and ethnic group, as well as the sex of the baby.7,8 Paternal height also plays a role but this factor is relatively minor9 and is usually not adjusted for. The
coefficients for adjustment are derived by multivariate analysis of large, well-dated birth weight databases representing unselected maternity populations. There are an infinite number of combinations of these variables, and these can be calculated by computer to give an individually adjusted or “customized” optimal weight value for each pregnancy—the “Term Optimal Weight” (TOW) at 280 days. 3. The growth and birth weight standard need to be free from pathology. Multivariate analysis of the constitutional variables mentioned will allow for pathological factors that affect growth, eg, diabetes, antepartum haemorrhage, or smoking. These factors need to be identified but not used to adjust the predicted birth weight at term, so that the latter will represent a truly optimal weight, free from pathology. For example, in a pregnancy in which the mother smoked, the birth weight is likely to be reduced, and this is in fact a dose depended effect: the baby of a mother who smokes 1 to 9 cigarettes per day will at 40 weeks weigh on average 143.1 g less; if she smokes 10 to 19 cigarettes, the baby will weigh 200.4 g less; and 20⫹ cigarettes, 248.5g less than a baby of a nonsmoking mother. Although including the (negative) coefficient for smoking would be likely to improve the From the Perinatal Institute, Birmingham, UK. Address reprint requests to Jason Gardosi, MD, FRCOG, FRCSED, Director, Perinatal Institute, Crystal Court, Aston Cross, Birmingham, West Midlands, B6 5RQ UK; e-mail: [email protected]. © 2004 Elsevier Inc. All rights reserved. 0146-0005/04/2801-0004$30.00/0 doi:10.1053/j.semperi.2003.12.002
Seminars in Perinatology, Vol 28, No 1 (February), 2004: pp 33-40
33
34
Jason Gardosi
Figure 1. “Proportionality” fetal growth curve. The line represents an equation derived from an in utero weight curve, transformed into a % weight versus gestation curve for any predicted term (280 day) birth weight point.8 % weight ⫽ 299.1 ⫺ 31.85 GA ⫹ 1.094 GA2 ⫺ 0.01055 GA3.
prediction of birth weight, not adjusting for this factor will improve the identification of a weight deficit caused by growth restriction, if it did occur. 4. Once the TOW is identified, the in utero weight curve should be outlined to predict how the term weight is to be reached. We use a log polynomial equation for fetal weight versus gestational age, derived from Hadlock’s fetal weight standard based on a normal population,10 and converted the function into one of % term weight versus gestation, taking Hadlock’s own term weight as 100%. The resultant “proportionality curve” (Fig 1) represents a normal growth equation that is remarkably similar in populations from different countries and continents. Such a curve can be used to link into any TOW; the growth dynamics in a normal pregnancy ending with this predicted weight point are outlined by a “Gestation Related Optimal Weight” chart.8 As a consequence of using a fetal rather than a neonatal weight based curve, the negative skewness of birth weight distributions in the preterm period is also avoided. The skewed distribution exists because of the association between spontaneous preterm birth and fetal growth re-
striction. It is inappropriate to use a standard for preterm neonatal weight assessment that is derived from other preterm baby weights, as by definition these are abnormal. The normal range around the median curve is derived by using the coefficient of variation of birth weight at term ⫽ 11%.8 Thus, the 90th and 10th percentile lines represent a range of ⫾ 1.28 ⫻ 11% ⫽ 14% around the median or optimal weight. Using a percentage around the median has the advantage that the normal range becomes smaller as the median is reduced because of individual adjustment (customization) of the optimal weight. Thus, the normal range of fetal weight for a small mother is smaller than that for a large mother.
GROW Software Using accurate gestation dates and physiological variables known at the beginning of pregnancy (maternal height, weight, ethnic group, and parity), a term optimal weight can be calculated; through this point a fetal growth curve is drawn to predict the optimal weight at each gestation from 24 weeks. These elements are incorporated into a software program [GROW (Gestation Re-
Customized Fetal Growth Standards
35
Figure 2. (A and B) Two examples of customized fetal growth curves with GROW.exe version 5.11. (www.gestation. net). The charts can be used to calculate previous baby weights and ultrasound estimated fetal weight(s) in the current pregnancy. Serial fundal height measurements can also be plotted. The graphs are adjusted to predict the optimal curve for each pregnancy, based on the variables entered (maternal height and weight, parity, ethnic group). In the example, a baby born at 37.0 weeks weighing 2,500 g was within normal limits for Mother A (51st percentile) but FGR for Mother B (5th percentile) as the latter’s predicted optimal growth curve is steeper. The pregnancy details entered are shown on the top left, together with the (computer-) calculated body mass index (BMI). The horizontal axis shows the day and month of each gestation week, calculated by the software on the basis of the EDD.
lated Optimal Weight)] that is freely available for download from www.gestation.net (Perinatal Institute, Birmingham, UK). Figures 2A and B show examples of individually adjusted or “customized” fetal growth charts.
Evidence for Customized Assessment The method sets an individually adjusted standard for fetal growth that allows better distinction between normal and abnormal. This can be
36
Jason Gardosi
Figure 3. Association between SGA and adverse perinatal outcome in 308,184 Swedish births 1992-1995.14 Outcomes: Stillbirths, Neonatal Deaths, and Low Apgars (⬍ 4 at 5 minutes). Comparison between definition of SGA as lowest 10% of births by customized percentile (SGA cust) and the lowest 10% by Swedish population based percentile (SGA pop), arranged in 3 categories: 1: SGA by both methods; 2: SGA by customized percentile only; and 3: SGA by population percentile only. Odds ratios and 95% Confidence Intervals are shown.
applied postnatally in the assessment of birth weight, and prospectively to assess fetal growth during the antenatal period. Birth Weight When assessing small for gestation age (SGA), it is clear that a large proportion of the population is misclassified if an unadjusted standard is used. In a heterogeneous population, for example, differences in birth weight between ethnic groups can be substantial. Individually adjusted birth weight percentiles are better correlated with Apgar scores7 and neonatal morphometry indices.11,12 They also better reflect adverse pregnancy events, even across geographical boundaries. SGA defined by
a customized standard derived from an English population is better correlated with operative deliveries for fetal distress and admission to neonatal intensive care in a Dutch population, than the local Dutch population standard.13 Analysis of a large Swedish dataset showed that SGA defined by a customized birth weight percentile was more closely associated with stillbirths, neonatal deaths, or low Apgar scores (⬍4) than the unadjusted population percentile.14 In fact, babies considered small by the general Swedish population standard but not by the customized standard did not have a larger risk of stillbirth, neonatal death, or low Apgar scores than the average-for-gestational age group (Fig 3). This study confirmed that small-normal babies are
37
Customized Fetal Growth Standards
not at greater risk than normal size babies. Furthermore, while SGA and FGR are not synonymous, “customized” SGA can be taken to represent FGR and used as a diagnostic category in retrospective cohort studies where birth weight, gestation, and physiological characteristics of mother and pregnancy are recorded. Intrauterine Weight Ultrasound derived fetal weight curves reproduce differences between physiological or constitutional characteristics in low-risk15 as well as high-risk16 populations. The use of fetal weight instead of individual scan biometry parameters allows adjustment of normal intrauterine growth limits on the basis of coefficients derived from large birth weight databases. Studies of individual scan parameters such as biparietal diameter, head circumference, abdominal circumference, and femur length are rarely large enough to be able to study variation in subgroups of populations. Instead, the variables can be determined from larger, population-based birth weight databases, and then applied to intrauterine growth curves. In addition, ultrasound databases usually represent a referred, more high-risk population. Third, the accuracy of individual measurements cannot be checked as there is no gold standard; whereas estimated fetal weight can be assessed against birth weight and has a random error of between 7% and 10 %.17-19 To check the accuracy of methods and formulae in a database, adjustment needs to be made for the gestational age difference between date of ultrasound measurement and date of birth.20 Normal or “optimal” in this model is defined retrospectively, from a cohort of pregnancies with normal outcome. Another model, proposed by Deter et al,21 uses ultrasound scans to project and predict normal growth for each fetus. Such individual adjustment has not been found to add accuracy.22 This is possibly because forward projection has to contend with ultrasound error, which can be ⫾10% for each measurement, and may vary from one measurement to the next. The error is then magnified by the forward projection. Furthermore, a projection to predict “normal” on the basis of scans in early pregnancy has to assume that the fetus is not already affected by abnormal growth. Otherwise
the predicted, projected range will not be optimal but pathological. Customized limits reduce false-positive diagnoses of FGR in a normal population.23 Furthermore, adjustment for the variables used in the model have been shown to improve the association between the slope of the growth curve and outcome variables such as fetal distress leading to operative delivery, or admission to the neonatal intensive care unit.24
Which Cut-off Limit Should be used? There is often debate about which limit should be used to denote small– or large for gestational age. This, however, depends on what the purpose for the cut-off is. Wide limits— eg, ⫾ 2 SD (standard deviation) or 3rd and 97th percentiles—will improve the strength of the association, but will reduce the number of cases detected. This is illustrated in Table 1, based on the analysis of the large Swedish database.14 Whereas the odds ratios are higher the lower the cut-off, the etiological fraction–ie, the proportion of stillbirths associated with SGA–is substantially higher when the 10th percentile limit is used. Receiver-operator studies showed the 8th percentile to be the best point for antenatal prediction of adverse outcome.25 Thus for antental use, the 10th percentile is a suitable cutoff limit to use with customized percentiles.
Clinical Application Improved differentiation between normal and abnormal growth can be of value in many aspects of perinatal audit and management.
Table 1. Rate of Stillbirths in an Unselected Swedish Population (n ⫽ 308,184), for Cut-off Limits for SGA at 10th, 5th, and 2.5th Centiles Stillbirths Cut-off
n
n
Rate (n/1000)
AOR
PAR
2.5% 5% 10%
8,018 14,186 30,818
196 246 323
24.4 17.3 10.5
11.2 8.3 5.3
19.6 23.8 28.9
Abbreviations: AOR, adjusted odds ratio; PAR, Population attributable risk or “etiological fraction”.
38
Jason Gardosi
“Unexplained” Stillbirths and Fetal Growth Restriction The commonly used Wigglesworth method26 to classify perinatal deaths consistently results in two thirds of stillbirths being classified as “unexplained.” A diagnosis of unexplained is often considered to be synonymous with “unavoidable,” which is not conducive to understanding what went wrong, to counseling the woman and her family, or for developing strategies for prevention. Yet, studies using data from the UK Confidential Enquiries into Stillbirths and Deaths in Infancy suggest a strong link between smallness for gestational age and antepartum stillbirth.27 Using customized percentiles in retrospective data allows the diagnosis of “IUGR” to be made on the basis of weight at delivery, gestational age, and physiological pregnancy characteristics. To assess the gestational age of stillbirths, an average of 2 days is deducted from the gestational age recorded at delivery.27a Studies of large, routinely collected databases in Sweden show that stillborn fetuses have a high chance of having been growth restricted before demise.14 Additional evidence comes from analysis of stillbirths in Oslo, where 52% of “unexpected” antepartum fetal deaths, which had revealed no findings on postmortem, were FGR (⬍10th customized percentile).28 While FGR is not in itself a “cause,” it is a clinically relevant condition preceding stillbirth. The classification system for stillbirth needs to aid our understanding of the foregoing clinical condition to aid prevention. Therefore, we have developed a new classification that looks at the “relevant condition at death” (ReCoDe ) (Table 2). Using this method to classify 313 antepartum stillbirths in the West Midlands in 2001 showed that while 230 (73.5%) were “unexplained” by the conventional Wigglesworth method, 132 of these (57.4%) fell into ReCoDe category A8 (Fetal Growth Restriction). Using ReCoDe, only 14.1 % of cases had no relevant condition identified (category H1). It is apparent that a more clinically relevant classification system can shed light on the large number of “unexplained” stillbirths, and can allow a better understanding of where the priorities need to lie for instituting strategies for prevention.
Table 2. ReCoDe Classification of Stillbirths by Relevant Condition at Death (www.perinate.org/ pnm/recode.htm) This system seeks to identify the condition(s) which existed at the time of death in-utero. The classification is based on the following principles: 1. Stillbirths are distinct from neonatal deaths and warrant their own classification. 2. There is hence no need for a sub-classification according to gestation, as “prematurity” is not a relevant cause or condition for stillbirths. 3. There is no subclassification according to weight, but one related to fetal growth status, based on weight-for-gestation. 4. The classification emphasises what went wrong, not necessarily “why”. Hence, more than one category can be coded. 5. The hierarchy starts from conditions affecting the fetus and moves outwards, in simple anatomical categories (A-F) which are subdivided into pathophysiological conditions. 6. The primary condition should be the highest on the list that is applicable to a case.
A. Fetus
B. Umbilical Cord
C. Placenta
D. Amniotic fluid
E. Uterus
F. Mother
G. Trauma H. Unclassified
1. Lethal congenital anomaly 2. Infection 2.1 Chronic-eg, TORCH 2.2 Acute 3. Non-immune hydrops 4. Iso-immunisation 5. Fetomaternal haemorrhage 6. Twin-twin transfusion 7. Intrapartum asphyxia 8. Fetal growth restriction* 9. Other 1. Prolapse 2. Constricting loop or knot† 3. Velamentous insertion 4. Other 1. Abruptio 2. Praevia 3. Vasa Praevia 4. Placental infarction 5. Other placental insufficiency‡ 6. Other 1. Chorioamnionitis 2. Oligohydramnios† 3. Polyhydramnios† 4. Other 1. Rupture 2. Uterine anomalies 3. Other 1. Diabetes 2. Thyroid diseases 3. Essential Hypertension 4. Hypertensive diseases in pregnancy 5. Lupus/Antiphospholipid Syndrome 6. Cholestasis 7. Drug abuse 8. Other 1. External 2. Iatrogenic 1. No relevant condition identified 2. No information available
Defined as ⬍10th customised weight-for-gestation percentile (centile calculator is available at www.gestation.net/centile)
*
† If severe enough to be considered relevant ‡ Histological diagnosis
Customized Fetal Growth Standards
Antenatal Detection of Fetal Growth Problems There are few studies that looked at the antenatal detection of growth failure. During routine antenatal care, only about a quarter of babies that end up being SGA are suspected as such before birth.29 In low-risk pregnancies, the rate is even lower–about 15%.30,31 This suggests that the assignment of the label of “low-risk” at the beginning of pregnancy puts the fetus at higher risk of a diagnosis of SGA being missed! Serial fetal growth assessment is appropriate in high-risk pregnancy when there is a history of obstetric problems, previous growth restriction, or adverse perinatal outcome. Estimated fetal weight plotted on general population charts may result in missed diagnoses (false-negatives) or unnecessary additional scans (false-positives) because of a constitutional small baby being plotted below the lower limit on a chart for the general population. Customized charts reduce the number of false-positive diagnoses if there is evidence of FGR.23 An ultrasound “growth” scan at every antenatal visit for the general maternity population has been suggested.32 However, routine third trimester scans have not been shown to improve outcome,33 and it is also doubtful whether serial scanning in the whole maternity population would prove cost effective, affordable, or acceptable. An alternative that we are currently evaluating within the National Health Service is to screen for FGR with 2 to 3 weekly fundal height assessments by well-trained midwives and doctors. Fundal height measurements are currently not well taught, not serially plotted and often only recorded in a haphazard fashion, against the number of weeks’ gestation, under the (false) expectation that fundal height in centimeters should equal the week of gestation. In fact, as is the case with birthweight and fetal growth, fundal height also varies with constitutional factors.34 The customized charts (Fig 2) have a second axis for fundal height, which is based on a fundal height versus fetal weight equation (details on www.gestation.net). Serial measurement of fundal height in the community by well-trained midwives using standard measurement techniques, with plotting on customized growth charts, and clearly defined referral pathways, has significantly increased the
39
antenatal detection of small as well as large for gestational age babies.35 At the same time, there was a decrease in the number of referrals from community clinics to the hospital for ultrasound scans and further investigations for fetal growth. This was because serial assessment and plotting helped clinicians to be reassured when growth was proceeding normally, whereas before there were more unnecessary requests for ultrasound. The scan in the first instance assessed liquor volume and calculated an estimated fetal weight (EFW), which was plotted on the same customized chart. If the EFW is within normal “customized” limits, surveillance is continued with routine, serial fundal height measurements; if the EFW is below the tenth customized centile line, or if the rate of growth since a previous EFW measurement was slower than that predicted by the growth curves, further investigation is recommended. This consisted mainly of Doppler and, if the pregnancy continues, serial (ie, fortnightly) ultrasound biometry. The change in EFW over time is a useful predictor of perinatal outcome,36 and Doppler studies in highrisk pregnancy reduce perinatal mortality.37 The impact of fetal growth screening on antenatal care and perinatal outcome needs large experimental and observational studies. There is a significant element of training involved, hence case by case randomization is not considered feasible. Cluster randomization of units is an alternative but requires standard protocols and care pathways that are not always in place. The use of customized charts for the assessment of fundal height and fetal weight is now recommended by the Green Top Guidelines of the Royal College of Obstetricians and Gynaecologists,38 and we are currently evaluating their large scale introduction into clinical practice in the West Midlands.
Conclusion For epidemiological analysis as well as for prospective assessment, customized percentiles improve the distinction between normal and abnormal, and help in our understanding and diagnosis of abnormal fetal growth.
References 1. Mul T, Mongelli M, Gardosi J: A comparative analysis of second-trimester ultrasound dating formulae in preg-
40
2.
3.
4.
5.
6.
7. 8.
9. 10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Jason Gardosi
nancies conceived with artificial reproductive techniques. Ultrasound Obstet Gynecol 8:397-402, 1996 Mongelli M, Wilcox M, Gardosi J: Estimating the date of confinement: Ultrasonographic biometry versus certain menstrual dates. Am J Obstet Gynecol 174:278-281, 1996 Gardosi J, Vanner T, Francis A: Gestational age and induction of labour for prolonged pregnancy. Br J Obstet Gynaecol 104:792-797, 1997 Gallivan S, Robson SC, Chang TC, et al: An investigation of fetal growth using serial ultrasound data. Ultrasound Obstet Gynecol 3:109-114, 1993 Mongelli M, Gardosi J: Longitudinal study of fetal growth in subgroups of a low risk population. Ultrasound Obstet Gynecol 6:340-344, 1995 Wilcox M, Gardosi J, Mongelli M, et al: Birth weight from pregnancies dated by ultrasonography in a multicultural British population. BMJ 307:588-591, 1993 Gardosi J, Chang A, Kalyan B, et al: Customised antenatal growth charts. Lancet 339:283-287, 1992 Gardosi J, Mongelli M, Wilcox M, et al: An adjustable fetal weight standard. Ultrasound Obstet Gynecol 6:168174, 1995 Wilcox MA, Newton CS, Johnson IR: Paternal influences on birthweight. Acta Obstet Gynecol Scand 74:15-18, 1995 Hadlock FP, Harrist RB, Martinez-Poyer J: In-utero analysis of fetal growth: A sonographic weight standard. Radiology 181:129-133, 1991 Sanderson DA, Wilcox MA, Johnson IR: The individualized birth weight ratio: a new method of identifying intrauterine growth retardation. Br J Obstet Gynaecol 101:310-314, 1994 Owen P, Farrell T, Hardwick JCR, et al: Relationship between customised birthweight centiles and neonatal anthropometric features of growth restriction. Br J Obstet Gynaecol 109:658-662, 2002 de Jong CLD, Gardosi J, Dekker GA, et al: Application of a customised birthweight standard in the assessment of perinatal outcome in a high risk population. Br J Obstet Gynaecol 105:531-535, 1998 Clausson B, Gardosi J, Francis A, et al: Perinatal outcome in SGA births defined by customised versus population based birthweight standards. Br J Obstet Gynaecol 108: 830-834, 2001 Mongelli M, Gardosi J: Longitudinal study of fetal growth in subgroups of a low risk population. Ultrasound Obstet Gynecol 6:340-344, 1995 de Jong CLD, Gardosi J, Baldwin C, et al: Fetal weight gain in a serially scanned high-risk population. Ultrasound Obstet Gynecol 11:39-43, 1998 Persson PH, Weldner BM: Intrauterine weight curves obtained by ultrasound. Acta Obstet Gynecol Scand 65: 169-173, 1986 Hadlock FP, Harrist RB, Sharman RS, et al: Estimation of fetal weight with the use of head, body and femur measurements–A prospective study. Am J Obstet Gynecol 151:333-337, 1985 Dalsgaard L, Wiberg N, Dragsted N: Quality control of ultrasound weight estimation in a central hospital. Ugeskrift for Laeger 164:2280-2283, 2002 Mongelli M, Gardosi J: Gestation adjusted projection of estimated fetal weight. Acta Obstet Gynecol Scand 75: 28-31, 1996
21. Deter RL, Rossavik IK, Harrist RB, et al: Mathematic modelling of fetal growth: Development of individual growth curve standards. Obstet Gynecol 68:156-161, 1986 22. Shields LE, Huff RW, et al: Fetal growth: A comparison of growth curves with mathematical modeling. J Ultrasound Med 5:271-274, 1993 23. Mongelli M, Gardosi J: Reduction of false-positive diagnosis of fetal growth restriction by application of customized fetal growth standards. Obstet Gynecol 88:844-848, 1996 24. de Jong CLD, Francis A, van Geijn HP, et al: Fetal growth rate and adverse perinatal events. Ultrasound Obstet Gynecol 13:86-89, 1998 25. De Jong CLD, Francis A, van Geijn HP, et al: Customized fetal weight limits for antenatal detection of fetal growth restriction. Ultrasound Obstet Gynecol 15:36-40, 2000 26. Wigglesworth JS: Monitoring perinatal mortality—A pathophysiological approach. Lancet 27:684-687, 1980 27. Maternal and Child Health Consortium. CESDI 8th Annual Report: Confidential Enquiry of Stillbirths and Deaths in Infancy, 2001. www.cemach.org.uk 27a. Gardosi J, Mul T, Mongelli M, et al: Analysis of birthweight and gestational age in antepartum stillbirths. Br J Obstet Gynaecol 105:524-530, 1998. 28. Froen JF, Gardosi J, Thurmann A, et al: Restricted Fetal Growth in Sudden Intrauterine Unexplained Death. Acta Obstet Gynecol Scand (in press) 29. Hepburn M, Rosenberg K: An audit of the detection and management of small-for-gestational age babies. Br J Obstet Gynaecol 93:212-216, 1986 30. Kean LH, Liu DT: Antenatal care as a screening tool for the detection of small for gestational age babies in the low risk population. J Obstet Gynaecol 16:77-82, 1996 31. Backe B, Nakling J: Effectiveness of antenatal care: A population based study. Br J Obstet Gynaecol 100:727732, 1993 32. Sim D, Beattie RB, Dornan JC: Evaluation of biophysical assessment in high risk pregnancy to assess ultrasound parameters suitable for screening in the low risk population. Ultrasound Obstet Gynecol 3:11-17, 1993 33. Jahn A, Razum O, Berle P: Routine screening for intrauterine growth retardation in Germany: Low sensitivity and questionable benefit for diagnosed cases. Acta Obstet Gynecol Scand 77:643-648, 1998 34. Mongelli M, Gardosi J: Symphysis-Fundus height and pregnancy characteristics in ultrasound-dated pregnancies. Obstet Gynecol 94:591-594, 1999 35. Gardosi J, Francis A: Controlled trial of fundal height measurement plotted on customised antenatal growth charts. Br J Obstet Gynaecol 106:309-317, 1999 36. Chang TC, Robson SC, Spencer JA, et al: Prediction of perinatal morbidity at term in small fetuses: Comparison of fetal growth and Doppler ultrasound. Br J Obstet Gynaecol 101:422-427, 1994 37. Alfirevic Z, Nielson JP: Doppler ultrasonography in high risk pregnancies: Systematic review with meta-analysis. Am J Obstet Gynecol 172:1379-1387, 1995 38. Royal College of Obstetricians and Gynaecologists. The investigation and management of the small-for-gestational age fetus. RCOG Green Top Guideline No.31, 2002. www.rcog.org.uk/resources/Public/Small_Gest_ Age_Fetus_No31.pdf
Accurate assessment of fetal growth status requires the definition of an optimal standard, which represents the growth potential of the baby. Against this standard, individually ‘customized’ percentiles can be calculated. They improve the distinction between normal and abnormal, and help in our understanding and diagnosis of pathological fetal growth. This method can be used as a tool for epidemiological analysis as well as for prospective clinical monitoring. © 2004 Elsevier Inc. All rights reserved. etal growth restriction (FGR) can be defined as the failure of the fetus to reach its growth potential. This potential varies in individual babies and needs to be adjusted or ‘customized’ for each pregnancy. Customized assessment of birth weight or growth therefore means an evaluation against the optimal, individually adjusted standard.
F
Establishing the Standard for Growth The development of a customized standard requires consideration of the following factors: 1. Accurate dating. Ultrasound dating is much more accurate than menstrual dating. Both methods can have error, but the error is much wider for menstrual dates and the distribution is skewed.1,2 Because the distribution of menstrual dating error is positively skewed, many birth weight points at term appear at later gestations than they actually should be, leading to an artificial flattening of the growth curve and apparent increase in “post-term” births.2,3 In reality, fetal growth in utero in normal pregnancy continues without diminished velocity until birth,4,5 and even birth weight charts that are based on accurately dated pregnancies have a near-straight line at term.6 Dating error can also severely affect accuracy in the preterm gestation range. 2. The growth standard also needs to be individually adjusted for physiological factors known to affect birth weight and growth. Significant variables include maternal height, weight in early pregnancy, parity and ethnic group, as well as the sex of the baby.7,8 Paternal height also plays a role but this factor is relatively minor9 and is usually not adjusted for. The
coefficients for adjustment are derived by multivariate analysis of large, well-dated birth weight databases representing unselected maternity populations. There are an infinite number of combinations of these variables, and these can be calculated by computer to give an individually adjusted or “customized” optimal weight value for each pregnancy—the “Term Optimal Weight” (TOW) at 280 days. 3. The growth and birth weight standard need to be free from pathology. Multivariate analysis of the constitutional variables mentioned will allow for pathological factors that affect growth, eg, diabetes, antepartum haemorrhage, or smoking. These factors need to be identified but not used to adjust the predicted birth weight at term, so that the latter will represent a truly optimal weight, free from pathology. For example, in a pregnancy in which the mother smoked, the birth weight is likely to be reduced, and this is in fact a dose depended effect: the baby of a mother who smokes 1 to 9 cigarettes per day will at 40 weeks weigh on average 143.1 g less; if she smokes 10 to 19 cigarettes, the baby will weigh 200.4 g less; and 20⫹ cigarettes, 248.5g less than a baby of a nonsmoking mother. Although including the (negative) coefficient for smoking would be likely to improve the From the Perinatal Institute, Birmingham, UK. Address reprint requests to Jason Gardosi, MD, FRCOG, FRCSED, Director, Perinatal Institute, Crystal Court, Aston Cross, Birmingham, West Midlands, B6 5RQ UK; e-mail: [email protected]. © 2004 Elsevier Inc. All rights reserved. 0146-0005/04/2801-0004$30.00/0 doi:10.1053/j.semperi.2003.12.002
Seminars in Perinatology, Vol 28, No 1 (February), 2004: pp 33-40
33
34
Jason Gardosi
Figure 1. “Proportionality” fetal growth curve. The line represents an equation derived from an in utero weight curve, transformed into a % weight versus gestation curve for any predicted term (280 day) birth weight point.8 % weight ⫽ 299.1 ⫺ 31.85 GA ⫹ 1.094 GA2 ⫺ 0.01055 GA3.
prediction of birth weight, not adjusting for this factor will improve the identification of a weight deficit caused by growth restriction, if it did occur. 4. Once the TOW is identified, the in utero weight curve should be outlined to predict how the term weight is to be reached. We use a log polynomial equation for fetal weight versus gestational age, derived from Hadlock’s fetal weight standard based on a normal population,10 and converted the function into one of % term weight versus gestation, taking Hadlock’s own term weight as 100%. The resultant “proportionality curve” (Fig 1) represents a normal growth equation that is remarkably similar in populations from different countries and continents. Such a curve can be used to link into any TOW; the growth dynamics in a normal pregnancy ending with this predicted weight point are outlined by a “Gestation Related Optimal Weight” chart.8 As a consequence of using a fetal rather than a neonatal weight based curve, the negative skewness of birth weight distributions in the preterm period is also avoided. The skewed distribution exists because of the association between spontaneous preterm birth and fetal growth re-
striction. It is inappropriate to use a standard for preterm neonatal weight assessment that is derived from other preterm baby weights, as by definition these are abnormal. The normal range around the median curve is derived by using the coefficient of variation of birth weight at term ⫽ 11%.8 Thus, the 90th and 10th percentile lines represent a range of ⫾ 1.28 ⫻ 11% ⫽ 14% around the median or optimal weight. Using a percentage around the median has the advantage that the normal range becomes smaller as the median is reduced because of individual adjustment (customization) of the optimal weight. Thus, the normal range of fetal weight for a small mother is smaller than that for a large mother.
GROW Software Using accurate gestation dates and physiological variables known at the beginning of pregnancy (maternal height, weight, ethnic group, and parity), a term optimal weight can be calculated; through this point a fetal growth curve is drawn to predict the optimal weight at each gestation from 24 weeks. These elements are incorporated into a software program [GROW (Gestation Re-
Customized Fetal Growth Standards
35
Figure 2. (A and B) Two examples of customized fetal growth curves with GROW.exe version 5.11. (www.gestation. net). The charts can be used to calculate previous baby weights and ultrasound estimated fetal weight(s) in the current pregnancy. Serial fundal height measurements can also be plotted. The graphs are adjusted to predict the optimal curve for each pregnancy, based on the variables entered (maternal height and weight, parity, ethnic group). In the example, a baby born at 37.0 weeks weighing 2,500 g was within normal limits for Mother A (51st percentile) but FGR for Mother B (5th percentile) as the latter’s predicted optimal growth curve is steeper. The pregnancy details entered are shown on the top left, together with the (computer-) calculated body mass index (BMI). The horizontal axis shows the day and month of each gestation week, calculated by the software on the basis of the EDD.
lated Optimal Weight)] that is freely available for download from www.gestation.net (Perinatal Institute, Birmingham, UK). Figures 2A and B show examples of individually adjusted or “customized” fetal growth charts.
Evidence for Customized Assessment The method sets an individually adjusted standard for fetal growth that allows better distinction between normal and abnormal. This can be
36
Jason Gardosi
Figure 3. Association between SGA and adverse perinatal outcome in 308,184 Swedish births 1992-1995.14 Outcomes: Stillbirths, Neonatal Deaths, and Low Apgars (⬍ 4 at 5 minutes). Comparison between definition of SGA as lowest 10% of births by customized percentile (SGA cust) and the lowest 10% by Swedish population based percentile (SGA pop), arranged in 3 categories: 1: SGA by both methods; 2: SGA by customized percentile only; and 3: SGA by population percentile only. Odds ratios and 95% Confidence Intervals are shown.
applied postnatally in the assessment of birth weight, and prospectively to assess fetal growth during the antenatal period. Birth Weight When assessing small for gestation age (SGA), it is clear that a large proportion of the population is misclassified if an unadjusted standard is used. In a heterogeneous population, for example, differences in birth weight between ethnic groups can be substantial. Individually adjusted birth weight percentiles are better correlated with Apgar scores7 and neonatal morphometry indices.11,12 They also better reflect adverse pregnancy events, even across geographical boundaries. SGA defined by
a customized standard derived from an English population is better correlated with operative deliveries for fetal distress and admission to neonatal intensive care in a Dutch population, than the local Dutch population standard.13 Analysis of a large Swedish dataset showed that SGA defined by a customized birth weight percentile was more closely associated with stillbirths, neonatal deaths, or low Apgar scores (⬍4) than the unadjusted population percentile.14 In fact, babies considered small by the general Swedish population standard but not by the customized standard did not have a larger risk of stillbirth, neonatal death, or low Apgar scores than the average-for-gestational age group (Fig 3). This study confirmed that small-normal babies are
37
Customized Fetal Growth Standards
not at greater risk than normal size babies. Furthermore, while SGA and FGR are not synonymous, “customized” SGA can be taken to represent FGR and used as a diagnostic category in retrospective cohort studies where birth weight, gestation, and physiological characteristics of mother and pregnancy are recorded. Intrauterine Weight Ultrasound derived fetal weight curves reproduce differences between physiological or constitutional characteristics in low-risk15 as well as high-risk16 populations. The use of fetal weight instead of individual scan biometry parameters allows adjustment of normal intrauterine growth limits on the basis of coefficients derived from large birth weight databases. Studies of individual scan parameters such as biparietal diameter, head circumference, abdominal circumference, and femur length are rarely large enough to be able to study variation in subgroups of populations. Instead, the variables can be determined from larger, population-based birth weight databases, and then applied to intrauterine growth curves. In addition, ultrasound databases usually represent a referred, more high-risk population. Third, the accuracy of individual measurements cannot be checked as there is no gold standard; whereas estimated fetal weight can be assessed against birth weight and has a random error of between 7% and 10 %.17-19 To check the accuracy of methods and formulae in a database, adjustment needs to be made for the gestational age difference between date of ultrasound measurement and date of birth.20 Normal or “optimal” in this model is defined retrospectively, from a cohort of pregnancies with normal outcome. Another model, proposed by Deter et al,21 uses ultrasound scans to project and predict normal growth for each fetus. Such individual adjustment has not been found to add accuracy.22 This is possibly because forward projection has to contend with ultrasound error, which can be ⫾10% for each measurement, and may vary from one measurement to the next. The error is then magnified by the forward projection. Furthermore, a projection to predict “normal” on the basis of scans in early pregnancy has to assume that the fetus is not already affected by abnormal growth. Otherwise
the predicted, projected range will not be optimal but pathological. Customized limits reduce false-positive diagnoses of FGR in a normal population.23 Furthermore, adjustment for the variables used in the model have been shown to improve the association between the slope of the growth curve and outcome variables such as fetal distress leading to operative delivery, or admission to the neonatal intensive care unit.24
Which Cut-off Limit Should be used? There is often debate about which limit should be used to denote small– or large for gestational age. This, however, depends on what the purpose for the cut-off is. Wide limits— eg, ⫾ 2 SD (standard deviation) or 3rd and 97th percentiles—will improve the strength of the association, but will reduce the number of cases detected. This is illustrated in Table 1, based on the analysis of the large Swedish database.14 Whereas the odds ratios are higher the lower the cut-off, the etiological fraction–ie, the proportion of stillbirths associated with SGA–is substantially higher when the 10th percentile limit is used. Receiver-operator studies showed the 8th percentile to be the best point for antenatal prediction of adverse outcome.25 Thus for antental use, the 10th percentile is a suitable cutoff limit to use with customized percentiles.
Clinical Application Improved differentiation between normal and abnormal growth can be of value in many aspects of perinatal audit and management.
Table 1. Rate of Stillbirths in an Unselected Swedish Population (n ⫽ 308,184), for Cut-off Limits for SGA at 10th, 5th, and 2.5th Centiles Stillbirths Cut-off
n
n
Rate (n/1000)
AOR
PAR
2.5% 5% 10%
8,018 14,186 30,818
196 246 323
24.4 17.3 10.5
11.2 8.3 5.3
19.6 23.8 28.9
Abbreviations: AOR, adjusted odds ratio; PAR, Population attributable risk or “etiological fraction”.
38
Jason Gardosi
“Unexplained” Stillbirths and Fetal Growth Restriction The commonly used Wigglesworth method26 to classify perinatal deaths consistently results in two thirds of stillbirths being classified as “unexplained.” A diagnosis of unexplained is often considered to be synonymous with “unavoidable,” which is not conducive to understanding what went wrong, to counseling the woman and her family, or for developing strategies for prevention. Yet, studies using data from the UK Confidential Enquiries into Stillbirths and Deaths in Infancy suggest a strong link between smallness for gestational age and antepartum stillbirth.27 Using customized percentiles in retrospective data allows the diagnosis of “IUGR” to be made on the basis of weight at delivery, gestational age, and physiological pregnancy characteristics. To assess the gestational age of stillbirths, an average of 2 days is deducted from the gestational age recorded at delivery.27a Studies of large, routinely collected databases in Sweden show that stillborn fetuses have a high chance of having been growth restricted before demise.14 Additional evidence comes from analysis of stillbirths in Oslo, where 52% of “unexpected” antepartum fetal deaths, which had revealed no findings on postmortem, were FGR (⬍10th customized percentile).28 While FGR is not in itself a “cause,” it is a clinically relevant condition preceding stillbirth. The classification system for stillbirth needs to aid our understanding of the foregoing clinical condition to aid prevention. Therefore, we have developed a new classification that looks at the “relevant condition at death” (ReCoDe ) (Table 2). Using this method to classify 313 antepartum stillbirths in the West Midlands in 2001 showed that while 230 (73.5%) were “unexplained” by the conventional Wigglesworth method, 132 of these (57.4%) fell into ReCoDe category A8 (Fetal Growth Restriction). Using ReCoDe, only 14.1 % of cases had no relevant condition identified (category H1). It is apparent that a more clinically relevant classification system can shed light on the large number of “unexplained” stillbirths, and can allow a better understanding of where the priorities need to lie for instituting strategies for prevention.
Table 2. ReCoDe Classification of Stillbirths by Relevant Condition at Death (www.perinate.org/ pnm/recode.htm) This system seeks to identify the condition(s) which existed at the time of death in-utero. The classification is based on the following principles: 1. Stillbirths are distinct from neonatal deaths and warrant their own classification. 2. There is hence no need for a sub-classification according to gestation, as “prematurity” is not a relevant cause or condition for stillbirths. 3. There is no subclassification according to weight, but one related to fetal growth status, based on weight-for-gestation. 4. The classification emphasises what went wrong, not necessarily “why”. Hence, more than one category can be coded. 5. The hierarchy starts from conditions affecting the fetus and moves outwards, in simple anatomical categories (A-F) which are subdivided into pathophysiological conditions. 6. The primary condition should be the highest on the list that is applicable to a case.
A. Fetus
B. Umbilical Cord
C. Placenta
D. Amniotic fluid
E. Uterus
F. Mother
G. Trauma H. Unclassified
1. Lethal congenital anomaly 2. Infection 2.1 Chronic-eg, TORCH 2.2 Acute 3. Non-immune hydrops 4. Iso-immunisation 5. Fetomaternal haemorrhage 6. Twin-twin transfusion 7. Intrapartum asphyxia 8. Fetal growth restriction* 9. Other 1. Prolapse 2. Constricting loop or knot† 3. Velamentous insertion 4. Other 1. Abruptio 2. Praevia 3. Vasa Praevia 4. Placental infarction 5. Other placental insufficiency‡ 6. Other 1. Chorioamnionitis 2. Oligohydramnios† 3. Polyhydramnios† 4. Other 1. Rupture 2. Uterine anomalies 3. Other 1. Diabetes 2. Thyroid diseases 3. Essential Hypertension 4. Hypertensive diseases in pregnancy 5. Lupus/Antiphospholipid Syndrome 6. Cholestasis 7. Drug abuse 8. Other 1. External 2. Iatrogenic 1. No relevant condition identified 2. No information available
Defined as ⬍10th customised weight-for-gestation percentile (centile calculator is available at www.gestation.net/centile)
*
† If severe enough to be considered relevant ‡ Histological diagnosis
Customized Fetal Growth Standards
Antenatal Detection of Fetal Growth Problems There are few studies that looked at the antenatal detection of growth failure. During routine antenatal care, only about a quarter of babies that end up being SGA are suspected as such before birth.29 In low-risk pregnancies, the rate is even lower–about 15%.30,31 This suggests that the assignment of the label of “low-risk” at the beginning of pregnancy puts the fetus at higher risk of a diagnosis of SGA being missed! Serial fetal growth assessment is appropriate in high-risk pregnancy when there is a history of obstetric problems, previous growth restriction, or adverse perinatal outcome. Estimated fetal weight plotted on general population charts may result in missed diagnoses (false-negatives) or unnecessary additional scans (false-positives) because of a constitutional small baby being plotted below the lower limit on a chart for the general population. Customized charts reduce the number of false-positive diagnoses if there is evidence of FGR.23 An ultrasound “growth” scan at every antenatal visit for the general maternity population has been suggested.32 However, routine third trimester scans have not been shown to improve outcome,33 and it is also doubtful whether serial scanning in the whole maternity population would prove cost effective, affordable, or acceptable. An alternative that we are currently evaluating within the National Health Service is to screen for FGR with 2 to 3 weekly fundal height assessments by well-trained midwives and doctors. Fundal height measurements are currently not well taught, not serially plotted and often only recorded in a haphazard fashion, against the number of weeks’ gestation, under the (false) expectation that fundal height in centimeters should equal the week of gestation. In fact, as is the case with birthweight and fetal growth, fundal height also varies with constitutional factors.34 The customized charts (Fig 2) have a second axis for fundal height, which is based on a fundal height versus fetal weight equation (details on www.gestation.net). Serial measurement of fundal height in the community by well-trained midwives using standard measurement techniques, with plotting on customized growth charts, and clearly defined referral pathways, has significantly increased the
39
antenatal detection of small as well as large for gestational age babies.35 At the same time, there was a decrease in the number of referrals from community clinics to the hospital for ultrasound scans and further investigations for fetal growth. This was because serial assessment and plotting helped clinicians to be reassured when growth was proceeding normally, whereas before there were more unnecessary requests for ultrasound. The scan in the first instance assessed liquor volume and calculated an estimated fetal weight (EFW), which was plotted on the same customized chart. If the EFW is within normal “customized” limits, surveillance is continued with routine, serial fundal height measurements; if the EFW is below the tenth customized centile line, or if the rate of growth since a previous EFW measurement was slower than that predicted by the growth curves, further investigation is recommended. This consisted mainly of Doppler and, if the pregnancy continues, serial (ie, fortnightly) ultrasound biometry. The change in EFW over time is a useful predictor of perinatal outcome,36 and Doppler studies in highrisk pregnancy reduce perinatal mortality.37 The impact of fetal growth screening on antenatal care and perinatal outcome needs large experimental and observational studies. There is a significant element of training involved, hence case by case randomization is not considered feasible. Cluster randomization of units is an alternative but requires standard protocols and care pathways that are not always in place. The use of customized charts for the assessment of fundal height and fetal weight is now recommended by the Green Top Guidelines of the Royal College of Obstetricians and Gynaecologists,38 and we are currently evaluating their large scale introduction into clinical practice in the West Midlands.
Conclusion For epidemiological analysis as well as for prospective assessment, customized percentiles improve the distinction between normal and abnormal, and help in our understanding and diagnosis of abnormal fetal growth.
References 1. Mul T, Mongelli M, Gardosi J: A comparative analysis of second-trimester ultrasound dating formulae in preg-
40
2.
3.
4.
5.
6.
7. 8.
9. 10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Jason Gardosi
nancies conceived with artificial reproductive techniques. Ultrasound Obstet Gynecol 8:397-402, 1996 Mongelli M, Wilcox M, Gardosi J: Estimating the date of confinement: Ultrasonographic biometry versus certain menstrual dates. Am J Obstet Gynecol 174:278-281, 1996 Gardosi J, Vanner T, Francis A: Gestational age and induction of labour for prolonged pregnancy. Br J Obstet Gynaecol 104:792-797, 1997 Gallivan S, Robson SC, Chang TC, et al: An investigation of fetal growth using serial ultrasound data. Ultrasound Obstet Gynecol 3:109-114, 1993 Mongelli M, Gardosi J: Longitudinal study of fetal growth in subgroups of a low risk population. Ultrasound Obstet Gynecol 6:340-344, 1995 Wilcox M, Gardosi J, Mongelli M, et al: Birth weight from pregnancies dated by ultrasonography in a multicultural British population. BMJ 307:588-591, 1993 Gardosi J, Chang A, Kalyan B, et al: Customised antenatal growth charts. Lancet 339:283-287, 1992 Gardosi J, Mongelli M, Wilcox M, et al: An adjustable fetal weight standard. Ultrasound Obstet Gynecol 6:168174, 1995 Wilcox MA, Newton CS, Johnson IR: Paternal influences on birthweight. Acta Obstet Gynecol Scand 74:15-18, 1995 Hadlock FP, Harrist RB, Martinez-Poyer J: In-utero analysis of fetal growth: A sonographic weight standard. Radiology 181:129-133, 1991 Sanderson DA, Wilcox MA, Johnson IR: The individualized birth weight ratio: a new method of identifying intrauterine growth retardation. Br J Obstet Gynaecol 101:310-314, 1994 Owen P, Farrell T, Hardwick JCR, et al: Relationship between customised birthweight centiles and neonatal anthropometric features of growth restriction. Br J Obstet Gynaecol 109:658-662, 2002 de Jong CLD, Gardosi J, Dekker GA, et al: Application of a customised birthweight standard in the assessment of perinatal outcome in a high risk population. Br J Obstet Gynaecol 105:531-535, 1998 Clausson B, Gardosi J, Francis A, et al: Perinatal outcome in SGA births defined by customised versus population based birthweight standards. Br J Obstet Gynaecol 108: 830-834, 2001 Mongelli M, Gardosi J: Longitudinal study of fetal growth in subgroups of a low risk population. Ultrasound Obstet Gynecol 6:340-344, 1995 de Jong CLD, Gardosi J, Baldwin C, et al: Fetal weight gain in a serially scanned high-risk population. Ultrasound Obstet Gynecol 11:39-43, 1998 Persson PH, Weldner BM: Intrauterine weight curves obtained by ultrasound. Acta Obstet Gynecol Scand 65: 169-173, 1986 Hadlock FP, Harrist RB, Sharman RS, et al: Estimation of fetal weight with the use of head, body and femur measurements–A prospective study. Am J Obstet Gynecol 151:333-337, 1985 Dalsgaard L, Wiberg N, Dragsted N: Quality control of ultrasound weight estimation in a central hospital. Ugeskrift for Laeger 164:2280-2283, 2002 Mongelli M, Gardosi J: Gestation adjusted projection of estimated fetal weight. Acta Obstet Gynecol Scand 75: 28-31, 1996
21. Deter RL, Rossavik IK, Harrist RB, et al: Mathematic modelling of fetal growth: Development of individual growth curve standards. Obstet Gynecol 68:156-161, 1986 22. Shields LE, Huff RW, et al: Fetal growth: A comparison of growth curves with mathematical modeling. J Ultrasound Med 5:271-274, 1993 23. Mongelli M, Gardosi J: Reduction of false-positive diagnosis of fetal growth restriction by application of customized fetal growth standards. Obstet Gynecol 88:844-848, 1996 24. de Jong CLD, Francis A, van Geijn HP, et al: Fetal growth rate and adverse perinatal events. Ultrasound Obstet Gynecol 13:86-89, 1998 25. De Jong CLD, Francis A, van Geijn HP, et al: Customized fetal weight limits for antenatal detection of fetal growth restriction. Ultrasound Obstet Gynecol 15:36-40, 2000 26. Wigglesworth JS: Monitoring perinatal mortality—A pathophysiological approach. Lancet 27:684-687, 1980 27. Maternal and Child Health Consortium. CESDI 8th Annual Report: Confidential Enquiry of Stillbirths and Deaths in Infancy, 2001. www.cemach.org.uk 27a. Gardosi J, Mul T, Mongelli M, et al: Analysis of birthweight and gestational age in antepartum stillbirths. Br J Obstet Gynaecol 105:524-530, 1998. 28. Froen JF, Gardosi J, Thurmann A, et al: Restricted Fetal Growth in Sudden Intrauterine Unexplained Death. Acta Obstet Gynecol Scand (in press) 29. Hepburn M, Rosenberg K: An audit of the detection and management of small-for-gestational age babies. Br J Obstet Gynaecol 93:212-216, 1986 30. Kean LH, Liu DT: Antenatal care as a screening tool for the detection of small for gestational age babies in the low risk population. J Obstet Gynaecol 16:77-82, 1996 31. Backe B, Nakling J: Effectiveness of antenatal care: A population based study. Br J Obstet Gynaecol 100:727732, 1993 32. Sim D, Beattie RB, Dornan JC: Evaluation of biophysical assessment in high risk pregnancy to assess ultrasound parameters suitable for screening in the low risk population. Ultrasound Obstet Gynecol 3:11-17, 1993 33. Jahn A, Razum O, Berle P: Routine screening for intrauterine growth retardation in Germany: Low sensitivity and questionable benefit for diagnosed cases. Acta Obstet Gynecol Scand 77:643-648, 1998 34. Mongelli M, Gardosi J: Symphysis-Fundus height and pregnancy characteristics in ultrasound-dated pregnancies. Obstet Gynecol 94:591-594, 1999 35. Gardosi J, Francis A: Controlled trial of fundal height measurement plotted on customised antenatal growth charts. Br J Obstet Gynaecol 106:309-317, 1999 36. Chang TC, Robson SC, Spencer JA, et al: Prediction of perinatal morbidity at term in small fetuses: Comparison of fetal growth and Doppler ultrasound. Br J Obstet Gynaecol 101:422-427, 1994 37. Alfirevic Z, Nielson JP: Doppler ultrasonography in high risk pregnancies: Systematic review with meta-analysis. Am J Obstet Gynecol 172:1379-1387, 1995 38. Royal College of Obstetricians and Gynaecologists. The investigation and management of the small-for-gestational age fetus. RCOG Green Top Guideline No.31, 2002. www.rcog.org.uk/resources/Public/Small_Gest_ Age_Fetus_No31.pdf