fetal growth z-velocity

8/2/2019 Fetal growth Z-velocity

http://slidepdf.com/reader/full/fetal-growth-z-velocity 1/15

Z-velocity Yields Low False Positive Rates in Screening for

IUGR Independent of Gestational Age

Adrian Mondry, Liu Pengbo, Marie Loh, Max Mongelli

Abstract

Background: Ultrasound scans provide the basis for detection of intrauterine growthrestriction (IUGR) but often fail to distinguish IUGR from small for gestational age

(SGA) fetuses. The present study introduces the concept of z-velocity, calculated aschanges in z-scores over time, as an additional criterion in the diagnosis of IUGR.

Methods: A computer program simulated 50,000 fetal abdominal circumference (FAC)

scans based on published growth formulas. False positive rates (FPR) were calculated todetermine optimal scan time and scan intervals. Using data from 325 pregnancies or an

independent simulation of 32,000 FAC scans where real data did not allow statisticallymeaningful analysis, the two methods were compared using receiver operator

characteristics (ROC).

Results: Longer scan intervals generate lower FPR. A scan interval of three weeks withcut off point dz/ dt < -0.5 generates an optimal FPR of about 2%. ROC showed areas

under the curve > 0.74 over the complete range of scan intervals. The positive predictivevalue of growth arrest as only diagnostic criterion, however, is too low to recommend it

as exclusive or first diagnostic criterion.

Discussion: Z- velocity may provide a viable addendum to 10th percentile as diagnosticcriterion for IUGR by reducing FPR well below the published 16- 49%. It can be used to

decide whether further diagnostic measures such as umbilical chord Doppler are calledfor in fetuses that fall below the 10

thpercentile. The diagnostic gain of combined

diagnostic approaches can be calculated from large databases that include ponderal indexas gold standard.

Introduction

Birth weight lower than normal, commonly understood as below the 10th

percentile, may be associated with increased morbidity and mortality, both in the perinatal period and in

the later course of life1-3

. Monitoring of intrauterine fetal growth by measuringstandardized parameters of ultrasound morphometry allows the consulting physician to

assess the potential risk of perinatal stress and to plan appropriate birth strategies2, andsome causes for inadequate growth may be susceptible to treatment if recognized in



time4.

Unfortunately, two commonly used terms in the process of decision making remaininsufficiently defined and are often used synonymously: “Small for Gestational Age”

(SGA) and “Intrauterine Growth Restriction” (IUGR).

For the sake of clarity, in the following, SGA is defined as any fetus whose development parallels the normal growth curve, but is below the 10

thpercentile, while IUGR is

attributed to any fetus whose growth departs from this parallel, even within the boundaries of the 10th percentile (see figure 1).

Figure 1: Comparison of SGA, IUGR and normal growth fetus by customized

growth chart

The potential for intervention necessitates a correct and early diagnosis of potentially low

birth weight. Technically more demanding procedures such as umbilical Doppler scan

can help differentiate between a constitutionally small fetus and one that is trulycompromised. Convincing parents of the necessity to conduct supplementary diagnosticmeasures, however, may increase their level of anxiety which can have negative

implications on the course of the pregnancy5,6. In view of this, the consultant obstetricianmust carefully compromise between sensitivity and specificity of his initial suspicion.

Given the narrow definition as defined above, the aim of this study is to describe a

mathematical model for the diagnosis of IUGR based on growth velocity calculated fromchanges in z-scores per unit of time, discuss its limitations and translate it into practical

recommendations.

Materials and Methods

Definition of terms and objective of the present study: The present study compares twomethods for the diagnosis of intrauterine growth restriction (IUGR) based on fetal

abdominal circumference (FAC) measurements.

The first method is the classical method. Comparing individual fetus’ FAC scans to“normalized growth charts”, it uses as cutoff value for diagnosis of IUGR the lower 10

th



percentile, i.e. if the FAC value for the fetus falls below the 10th

percentile value for itscorresponding gestational age, it is classified as IUGR. In this study, the 10

thpercentile

cutoff values published by Chitty et al.7 were used.

The second method proposed here uses growth velocity arrest, that is assessment of the

fetus’s growth rate calculated as z- velocity, as the factor for IUGR diagnosis.

The z- velocity, dz/dt, is calculated as follows.

------------ (1)

where Zscore(1) and Zscore(2) are the Z-scores calculated at the time of the first and the

second ultrasound scan respectively, and scaninterval is the scan interval between thefirst and second ultrasound scan in weeks.

The z-score gives the number of standard deviations that a measurement is from themean. In this case, the measurement used is the fetal abdominal circumference (FAC).The formula for calculating Z-score for a given fetus i of gestational age j, denoted by

Zscorei,j, is:

j

j ji

ji FAC SD

FAC Mean FAC Observed Zscore

_

_ _ ,

,

−

= ------------ (2)

where Observed_FAC i,j is the actual observed measured FAC value for fetus i of

gestational age j while Mean_FAC and SD_FAC are the mean and standard deviation of simulated “observed” FAC measurements of gestational age j respectively.

Theoretically, a normally growing fetus will have dz/dt values close to zero. IUGR can be

diagnosed when dz/dt <0, as this is indicative of a lack of growth of the fetus. Here, the performance of the method is analyzed for various cutoff points between –1 and 0.

The objective of the study is to answer the question whether the use of z- velocity instead

or in addition to the classical 10th percentile method will improve the ultrasound baseddiagnosis of IUGR.

Patients: Ultrasound scans from 325 British pregnancies identified as low risk for IUGR

were used for assessment of the reliability of growth velocity arrest as diagnosticcriterion for IUGR. After informed consent was obtained, each of the candidates

underwent observations for different periods of time ranging from 14 to 105 days. Theultrasound measurements for the fetal abdominal circumference were taken every 1~6

weeks from 26 weeks of gestation onwards. All pregnancies led to singleton births. Table1 and 2 summarized clinical characteristics of the maternal and infant data respectively.

The patient cohort has been characterized more in detail previously8.



Mean: 26.74Age (year)

SD: 4.81

Mean: 163.59Height (cm)

SD: 6.31

Mean: 66.48Weight (kg)

SD: 11.7European: 93.8%

Indo-Pakistani: 4.3%

Afro-Caribbean: 1.2%Ethnic Groups

Others: 0.6%

Mean: 0.73Parity

SD: 0.91

Table 1: Maternal clinical characteristics

Mean: 3406.09Birth Weight (g)SD: 552.55

Mean: 277.56Delivery time (day)

SD: 12.31

Table 2: Newborn clinical characteristics

Simulations: Two independent simulations of fetal abdominal circumference werecomputer generated as described below. The first simulation of 50,000 cases was used as

a reference population, while the second simulation of 32,500 cases was used as samplesto assess the efficiency of growth arrest calculated by z- score for the diagnosis of IUGR.

To obtain the mean and standard deviation of simulated “observed” FAC measurements

of gestational age j, a simulation consisting of 50,000 cases was performed. Using a published growth formula

9for British women and a fixed coefficient of variation of 5%

for ultrasound error, computer software is used to generate a normal growth chart for FAC according to gestational age, then an error value is introduced to simulate “observed

values”.

A java program was written for the simulation, which runs on a Pentium IV PC. Arandom number generator was employed to produce a range of random numbers between

0 and 1. Normally distributed z values were then derived from these random numbers

using a rational approximation for the standard distribution of Odeh and Evans 10. Notethat these are not the final z values used for the calculation of dz/dt. These derived zvalues are then used as the seed to generate the two ultrasound scan simulations.

The computer first generates the “true value” of FAC at the initial scan according to

Chitty’s formula7. The “observed” FAC value is then calculated based on the true value

plus an error term, which is randomly allocated to the known ultrasound error for this



variable. For the simulation runs, a coefficient of variation for ultrasound error 5% wasentered. For simplicity, a constant coefficient is chosen for all gestational ages.

For comparison of the two diagnostic methods, Receiver Operating Characteristics

(ROC) were calculated using data from a British population. The samples were 325

British pregnancies identified as low risk population for IUGR.With respect to sample size in ROC analysis, it has been suggested that meaningfulqualitative conclusions can be drawn from ROC experiments performed with at least a

total of about 100 observations11. Hence, the given data from real pregnancies allowed to perform ROC for the following combinations of gestational age and scan interval (Table

3) only.

Gestational Age at First Scan (weeks) Scan Interval (weeks)

26-28 6

329-31

63

432-34

5

35-37 3

Table 3: Combinations of Gestational Age and

Scan Interval for which ROC was performed

In order to overcome the numerical limitations of the data from real pregnancies, anindependent population of 32,500 cases was simulated to provide sufficient sample size

for statistical analysis.

Wherever mathematically meaningful, results from the analysis of data from real pregnancies was used, or analysis of simulated cases was compared to real data.All statistical analysis was done using SPSS 11.5 for Windows (SPSS Inc., Chicago, Il,

USA).

Results

The simulated true FAC measurements and observed FAC measurements have a verysimilar mean value and standard deviation (table 4.) compared with the published data for

British women12

.

'Observed' fetal

AC values

'True' fetal

AC values

Reference values

[A]

Reference values

[B]GA

(wk)Mean SD Mean SD Mean SD Mean SD

23 178.66 11.03 178.71 6.47 190 10.2 181.17 11.09

24 189.40 11.63 189.34 6.81 206 12.7 191.51 11.58

25 199.77 12.28 199.82 7.14 214 11.8 201.84 12.08



26 210.17 12.90 210.12 7.50 228 9.9 212.17 12.57

27 220.50 13.55 220.50 7.87 238 10.1 222.50 13.07

28 230.76 14.18 230.65 8.27 247 11.7 232.83 13.56

29 240.54 14.76 240.62 8.64 255 12.2 243.16 14.05

30 250.36 15.48 250.43 9.04 269 13.4 253.49 14.54

31 259.95 16.06 260.09 9.35 278 12.6 263.82 15.0432 269.66 16.57 269.68 9.70 291 13.5 274.16 15.53

33 279.17 17.17 279.12 10.02 300 12.1 284.49 16.03

34 288.39 17.83 288.35 10.46 314 13.9 294.82 16.52

35 297.55 18.42 297.50 10.79 322 12.5 305.15 17.02

36 306.60 18.97 306.49 11.13 329 15.5 315.48 17.51

37 315.05 19.43 315.18 11.53 338 14.8 325.82 18.00

38 323.87 20.02 323.82 11.90 343 16.8 336.15 18.50

39 332.06 20.67 332.17 12.33 349 20.6 346.48 18.99

40 340.55 21.09 340.40 12.55 369 14.5 356.81 19.49

Table 4.: Simulated “Observed” and “True” FAC value compared to publishedmeasurements by [A] Larsen

12and [B] Chitty

7. GA: gestational age. Wk: week. AC:

abdominal circumference. SD: standard deviation

Receiver operator characteristics were calculated comparing the growth velocity arrestmethod against the classical method. Due to restricted sample size of the pregnancies

described in8, only the combinations shown in figure 2 could be calculated.

1 - Specificity

1.0.9.8.7.6.5.4.3.2.10.0

S e n s

i t i v i t y

1.0

.9

.8

.7

.6

.5

.4

.3

.2

.1

0.0

1 - Specificity

1.0.9.8.7.6.5.4.3.2.10.0

S e n s

i t i v i t y

1.0

.9

.8

.7

.6

.5

.4

.3

.2

.1

0.0

26-28wks; 6wks 29-31wks; 3wks 29-31wks; 6wks 32-34wks; 3wks

32-34wks; 4wks 32-34wks; 5wks 35-37wks; 3wks



1 - Specificity

1.0.9.8.7.6.5.4.3.2.10.0

S e n

s i t i v i t y

1.0

.9

.8

.7

.6

.5

.4

.3

.2

.1

0.0

1 - Specificity

1.0.9.8.7.6.5.4.3.2.10.0

S e n

s i t i v i t y

1.0

.9

.8

.7

.6

.5

.4

.3

.2

.1

0.0

1 - Specificity

1.0.9.8.7.6.5.4.3.2.10.0

S e n

s i t i v i t y

1.0

.9

.8

.7

.6

.5

.4

.3

.2

.1

0.0

Figure 2: ROC curves using data from 325 British pregnancies. The individual

indices tell at what gestational age the first scan was performed, and the scan

interval.

The overall test performance of each measurement was assessed by examining the area

under the ROC curve (AUC, table 5), which is considered the best discriminator of diagnostic performance11. In the present study, the area under the curve represents the

probability that the growth velocity (dz/dt) value for a randomly chosen positive IUGR

case will be less than the result for a randomly chosen negative IUGR case.

Most of the AUC values were above 0.7 and had reasonably small standard errors. All the

ROC yielded significances of less than 0.05 (except for gestational age 26-28 weeks andscan interval 6 weeks), which indicates a rejection of the null hypothesis of true AUC =

0.5 at 5% significance level.

95% CIGestational Age at

First Scan (weeks)

Scan Interval

(weeks)

AUC

(significance)

Standard

Error Lower Upper

26-28 6 0.675 (0.048) 0.083 0.512 0.839

3 0.730 (0.093) 0.099 0.536 0.92529-31

6 0.715 (0.028) 0.088 0.543 0.887

3 0.705 (0.045) 0.082 0.545 0.865

4 0.755 (0.006) 0.074 0.610 0.899

32-34

5 0.876 (0.000) 0.050 0.778 0.974

35-37 3 0.801 (0.001) 0.060 0.684 0.919

Table 5: Area under the ROC curve (AUC) values for various gestational ages and

scan intervals

The corresponding false positive rates (FPR), that is the percentage of samples

incorrectly identified as IUGR as described previously8,13, are summarized in table 6.Across all gestational age and scan intervals, as the dz/dt cutoff decreases from 0 to -1,

FPR decreases from above 28% to around 1%. However, at the same time, true positiverates (TPR) also decrease from above 60% to below 4%. For further analysis, the median

cutoff value of dz/dt = -0.5 was chosen as it maintains clinically acceptable FPR values of less than 2.2% across all gestational ages and scan intervals tested.

Gestational Age at Scan Interval dz/dt cutoff



First Scan (weeks) (weeks) 0 -0.25 -0.5 -0.75 -1

26-28 6 28.3 8.7 2.0 1.7 0.5

29-31 3 37.7 10.8 2.2 1.8 1.3

6 55.8 8.8 1.2 0.8 0.5

32-34 3 50.0 13.6 2.1 1.5 1.0

4 37.7 7.3 0.0 0.0 0.05 44.6 5.1 0.0 0.0 0.0

35-37 3 50.5 18.9 1.7 1.5 1.2

Table 6: False Positive Rates (FPR) values for identification of IUGR cases using z-

velocity for diagnosis

When the ROC computation is repeated using the much larger dataset of 32,000

simulated cases, the analysis can be carried out over the full range of gestational age andscan intervals as all cells are sufficient in number (table 7). Comparing growth velocity

arrest against the classical method, all AUC values were at least 0.74 and with

corresponding p-values of < 0.001. Standard error was also low at 0.003 across allcombinations. AUC increases with larger scan interval and is generally independent of the gestational age.

Scan Interval

Gestational Age (wk) 1 2 3 4 5 623 0.752 0.767 0.767 0.777 0.792 0.799

24 0.750 0.759 0.768 0.775 0.787 0.789

25 0.749 0.756 0.764 0.773 0.787 0.785

26 0.742 0.752 0.763 0.770 0.783 0.788

27 0.746 0.756 0.760 0.767 0.776 0.786

28 0.743 0.756 0.761 0.771 0.774 0.781

29 0.750 0.755 0.759 0.768 0.778 0.783



30 0.743 0.754 0.752 0.761 0.773 0.780

31 0.747 0.754 0.759 0.767 0.771 0.776

32 0.742 0.751 0.757 0.764 0.770 0.772

33 0.748 0.750 0.758 0.760 0.767 0.772

34 0.747 0.752 0.759 0.760 0.766 0.770

35 0.741 0.758 0.757 0.762 0.768 0.76836 0.747 0.747 0.757 0.759 0.762 0.770

37 0.743 0.751 0.759 0.764 0.766 NA

38 0.748 0.754 0.752 0.761 NA NA

39 0.748 0.750 0.754 NA NA NA

40 0.748 0.753 NA NA NA NA

Table 7: Area under ROC curve (AUC) values using simulated data across

gestational ages (23-40 wks) and scan intervals (1-6 wks)

(NA: Time at 2nd

scan is more than 42 weeks of gestation)

Discussion

A fetus is commonly considered as “small for gestational age” (SGA) if ultrasoundmeasurements show an abdominal circumference and estimated birthweight below the

10th

percentile14,15

. Within the heterogenous group of SGA fetuses, 50- 70% areconstitutionally small16. The lower the percentile cutoff point is set, however, the more

likely a SGA fetus is not constitutionally small, but suffers from fetal growth restriction(IUGR), and the high incidence of truly IUGR fetuses is thought to explain the poor

perinatal outcome of several studies examining SGA births17,18. Numerous studies (summary in

15) show that lower than average birth weight may be

associated with increased morbidity and mortality in later life. While it seems likely thatgenetic and environmental factors play a pivotal role

19,20, estimates of lower birth weight

during pregnancy may allow intervention on various levels. Interventions may addressexternal social factors such as smoking, alcohol consumption and use of illicit drugs that

contribute to pregnancy outcome21

, or medical problems such as obstructed umbilicalchord, thrombophilia, insufficient placental perfusion, and placenta previa.

The potential for intervention obliges the consultant obstetrician to secure the diagnosis,and the particularly vulnerable psychological environment of pregnancy calls for an

optimal balance of sensitivity and specificity in the decision- making process.Fetal abdominal circumference (FAC) and estimated fetal weight (EFW) are the most

accurate diagnostic measurements to predict SGA. In high-risk women, FAC at less thanthe tenth percentile has sensitivities of 72.9–94.5% and specificities of 50.6–83.8% in the

prediction of fetuses with birthweight at less than the tenth percentile. This allows to

calculate false positive rates (FPR) of 16.2- 49.4%. The respective figures for EFW aresensitivities of 33.3–89.2% and specificities of 53.7–90.9%14, with FPR 9.1- 46.3%. Arecent Cochrane collaboration review22 that examined routine late pregnancy (after week

24) ultrasound examinations with regards to altered perinatal outcome, however, foundno conclusive benefit in maternal populations of unspecified or low risk. In low risk

populations, it may be inferred that the FPR is relatively higher, as it has been postulatedthat in high risk populations, the high number of



In view of these findings, optimization of diagnostic procedures to increase specificityand sensitivity with the aim to reduce false positive diagnosis of IUGR as a trigger of

anxiety seems a worthwhile endeavor.Monitoring of fetal growth velocity has been suggested as an alternative diagnostic

criterion to the classical 10th

percentile cutoff point before23

, and is carried out routinely

in late pregnancies with unclear gestational age

24

, but the efficiency of this approach isquestionable25

.This is to a large extent due to a high rate of false positive diagnoses. Using a

mathematical model13 that assumed a coefficient of variation of 5% for error inultrasonographic measurement (variability within and between observers) and a

definition of IUGR as no apparent growth in abdominal circumference between twoconsecutive scans, false positive rates were shown to be high (12- 31% for scan intervals

of one or two weeks started between week 28 and 38) and, as expected, they increasedwith shorter time intervals between scans and increasing gestational age. Accordingly, a

cutoff point of 2 standard deviations was recommended25

for decision making whether additional diagnostic procedures such as umbilical Doppler scan should be initiated.

This suggestion has met with critique

26

as being too strict, thus reducing the sensitivity of the procedure, and the traditionally used 10th

percentile cut off was recommended. As this

is the generally accepted diagnostic criterion, the present study uses it for calculation of false positive rates as previously reported

8,13.

For clinical purposes, it would be ideal to assess the gain in diagnostic performance if z-velocity is used in addition to the classical 10

thpercentile method. Such a combination

would necessarily result in a trade- off in sensitivity and specificity of the combined testrelative to the individual component tests. This trade- off can be used as for deciding

whether the combined test is advantageous over the individual tests27

on the basis of their likelihood ratios if sensitivity and specificity can be calculated for the individual tests by

comparing them against a “true- true” gold standard.In the setting of the current study, comparison of diagnosis to ponderal index as absolute

true/ gold standard was not possible: the data from real pregnancies8 did not include birthlength, which could be used to calculate the ponderal index and thus distinguish better

between SGA and IUGR pregnancies.The growth formula used for simulation did not account for birth length either, so that no

birth lengths could be simulated that corresponded to the simulated FAC values. In fact,an adequate simulation seems not viable as the times of peak length velocity and peak

weight velocity differ in utero, and therefore, it is generally assumed that alterations inthe patterns of growth at different stages in gestation will lead to different anthropometric

phenotypes at birth28

. Even more disadvantageous for meaningful simulation of estimated birth length are recent findings in a small cohort of 44 Belgian women that length growth

occurs in a gender- specific, pulsatile way in healthy fetus29.The use of simulation, however, is justified in order to allow for statistical meaningful

numbers. The simulation is acceptably close to real pregnancies (table 4 andreferences

7,12). While the analysis of z- velocity is novel to the present study, the

simulation is based on a well established model13

that has been used for the assessment of growth velocity (but not z- velocity) in the diagnosis of IUGR

25and in highlighting the

importance of customized growth charts for assessing normal growth8,26.Calculation of z- scores depends on a sizeable and customized reference database of



meaningful composition. Here, reference database and sample simulation were based onthe same growth formula, so that a correct customization can be assumed. In practical

application, the need to customize reference databases cannot be stressed enough8,26. Assuch, the use of simulation lies in highlighting probable results from real data and

pointing out what information is needed not only to verify the prediction, but more

importantly so answer the clinical want for more secure diagnosis of IUGR.In view of the findings presented here, the following diagnostic proceedings seemrecommendable: In fetus smaller than 10

thpercentile found accidentally in a low risk

pregnancy, a second scan after three weeks with calculation of z- velocity should becarried out to reduce the potential for false positive diagnosis of IUGR unless other

clinical parameters indicate the need for more rapid diagnosis.The data from 325 real pregnancies can in this context be regarded to have the value of

extended case descriptions. As such, however, it may have a value quite different from, but equal to that of a randomized, controlled study. Aronson recently discussed the value

of medical anecdotes30

; of the eight reasons listed there, the present study meets four: itgenerates a hypothesis (i.e., use of z- velocity reduces the rate of false positive diagnosis

of IUGR and therefore rate of high- level interventions), it suggests a method of management (i.e., stepwise approach with regular scan and use of z- velocity in low risk

cases where the fetal growth lies below the 10th percentile), it reminds and educates (i.e.,of the benefit of early ultrasound, the importance of customized growth charts, and the

lack of evidence for a benefit of ultrasound screens in low risk populations), and it hopesto stimulate a systematic review (i.e., the authors hope that the results will entice large

scale data collectors such as national/ international obstetrics associations to design andcarry out a prospective, randomized study).

More often than not, observational studies give similar results to controlled randomizedtrials31,32, and critical appraisal of this notion leads to the conclusion that, while good

controlled randomized trials do provide the highest level of evidence, a flexible approachmay be taken "in which randomised controlled trials and observational studies have

complementary roles. High quality observational studies may extend evidence over awider population and are likely to be dominant in the identification of harms and when

randomised controlled trials would be unethical or impractical"33

.In view of this, only analysis of a very large database of real pregnancies that holds the

required information (FAC, birth weight, birth length for this specific purpose) wouldallow to first calculate the ponderal index as a gold standard against which the individual

and combined tests sensitivity and specificity could then be calculated and thusevaluated. If such data is available for retrospective analysis, the diagnostic benefit of

sequential ultrasound with calculation of z- velocity compared to other diagnostictechniques can be calculated

27, and definite practical recommendations can be made.

Reference

1. Kady MS, Gardosi J. Perinatal mortality and fetal growth restriction. Best PractRes Clin Obstet Gynaecol 2004;18(3):397.

2. Thornton JG, Hornbuckle J, Vail A, et al. Infant wellbeing at 2 years of age in theGrowth Restriction Intervention Trial (GRIT): multicentred randomised

controlled trial. Lancet 2004;364(9433):513.



3. A randomised trial of timed delivery for the compromised preterm fetus: shortterm outcomes and Bayesian interpretation. Bjog 2003;110(1):27.

4. Nicolaides KH, Campbell S, Bradley RJ, et al. Maternal oxygen therapy for intrauterine growth retardation. Lancet 1987;1(8539):942.

5. Andersson L, Sundstrom-Poromaa I, Wulff M, et al. Implications of antenatal

depression and anxiety for obstetric outcome. Obstet Gynecol 2004;104(3):467.6. Green JM, Baston HA. Feeling in control during labor: concepts, correlates, andconsequences. Birth 2003;30(4):235.

7. Chitty LS, Altman DG, Henderson A, et al. Charts of fetal size: 3. Abdominalmeasurements. Br J Obstet Gynaecol 1994;101(2):125.

8. Mongelli M, Gardosi J. Reduction of false-positive diagnosis of fetal growthrestriction by application of customized fetal growth standards. Obstet Gynecol

1996;88(5):844.9. Manning FA. Intrauterine growth retardation. In: Romero R, editor. Sonography

in obstetrics and gynecology. 5 ed. Englewood Cliffs: Prentice-Hall InternationalInc; 1996. p 518.

10. Odeh RE, Evans JO. Algorithm AS 70: Percentage points of the normaldistribution. Applied Statistics 1974;23:96.

11. Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978;8(4):283.12. Larsen T, Petersen S, Greisen G, et al. Normal fetal growth evaluated by

longitudinal ultrasound examinations. Early Hum Dev 1990;24(1):37.13. Mongelli M, Ek S, Tambyrajia R. Screening for fetal growth restriction: a

mathematical model of the effect of time interval and ultrasound error. ObstetGynecol 1998;92(6):908.

14. Chang TC, Robson SC, Boys RJ, et al. Prediction of the small for gestational ageinfant: which ultrasonic measurement is best? Obstet Gynecol 1992;80(6):1030.

15. Coomarasamy A, Fisk NM, Gee H, et al. Royal College of Obstetricians andGynaecologists guideline No. 31:The investigation and management of the small

for gestational age fetus. London: Royal College of Obstetricians andGynaecologists; 2002 November 2002. 1 p.

16. Ott WJ. The diagnosis of altered fetal growth. Obstet Gynecol Clin North Am1988;15(2):237.

17. Gardosi J, Mul T, Mongelli M, et al. Analysis of birthweight and gestational agein antepartum stillbirths. Br J Obstet Gynaecol 1998;105(5):524.

18. Beattie RB, Johnson P. Practical assessment of neonatal nutrition status beyond birthweight: an imperative for the 1990s. Br J Obstet Gynaecol 1994;101(10):842.

19. Hubinette A, Cnattingius S, Ekbom A, et al. Birthweight, early environment, andgenetics: a study of twins discordant for acute myocardial infarction. Lancet

2001;357(9273):1997.20. Milani S, Bossi A, Bertino E, et al. Differences in Size at Birth Are Determined

by Differences in Growth Velocity during Early Prenatal Life. Pediatr Res 2004.21. Tambyrajia RL, Mongelli M. Sociobiological variables and pregnancy outcome.

Int J Gynaecol Obstet 2000;70(1):105.22. Bricker L, Neilson JP. Routine ultrasound in late pregnancy (after 24 weeks

gestation). The Cochrane Database of Systematic Reviews 2000(1).23. Gardosi J. Customized growth curves. Clin Obstet Gynecol 1997;40(4):715.



24. Jahn A, Razum O, Berle P. Routine screening for intrauterine growth retardationin Germany: low sensitivity and questionable benefit for diagnosed cases. Acta

Obstet Gynecol Scand 1998;77(6):643.25. Bobrow CS, Soothill PW. Fetal growth velocity: a cautionary tale. Lancet

1999;353(9163):1460.

26. Mongelli M, Gardosi J. Fetal growth velocity. Lancet 1999;353(9170):2156.27. Macaskill P, Walter SD, Irwig L, et al. Assessing the gain in diagnostic performance when combining two diagnostic tests. Stat Med 2002;21(17):2527.

28. Hindmarsh PC, Geary MP, Rodeck CH, et al. Intrauterine growth and itsrelationship to size and shape at birth. Pediatr Res 2002;52(2):263.

29. Lampl M, Jeanty P. Timing is everything: a reconsideration of fetal growthvelocity patterns identifies the importance of individual and sex differences. Am J

Hum Biol 2003;15(5):667.30. Aronson JK. Anecdotes as evidence. Bmj 2003;326(7403):1346.

31. Benson K, Hartz AJ. A comparison of observational studies and randomized,controlled trials. N Engl J Med 2000;342(25):1878.

32. Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observationalstudies, and the hierarchy of research designs. N Engl J Med 2000;342(25):1887.

33. Barton S. Which clinical studies provide the best evidence? The best RCT stilltrumps the best observational study. Bmj 2000;321(7256):255.

34. Bernstein PS, Divon MY. Etiologies of fetal growth restriction. Clin ObstetGynecol 1997;40(4):723.

35. Lin CC, Su SJ, River LP. Comparison of associated high-risk factors and perinataloutcome between symmetric and asymmetric fetal intrauterine growth retardation.

Am J Obstet Gynecol 1991;164(6 Pt 1):1535.36. Resnik R. Intrauterine growth restriction. Obstet Gynecol 2002;99(3):490.

37. Stahl K, Hundley V. Risk and risk assessment in pregnancy - do we scare becausewe care? Midwifery 2003;19(4):298.

38. Dole N, Savitz DA, Hertz-Picciotto I, et al. Maternal stress and preterm birth. AmJ Epidemiol 2003;157(1):14.

39. Andersson L, Sundstrom-Poromaa I, Wulff M, et al. Neonatal outcome followingmaternal antenatal depression and anxiety: a population-based study. Am J

Epidemiol 2004;159(9):872.40. Misri S, Oberlander TF, Fairbrother N, et al. Relation between prenatal maternal

mood and anxiety and neonatal health. Can J Psychiatry 2004;49(10):684.41. O'Connor TG, Heron J, Glover V. Antenatal anxiety predicts child

behavioral/emotional problems independently of postnatal depression. J Am AcadChild Adolesc Psychiatry 2002;41(12):1470.

42. O'Connor TG, Heron J, Golding J, et al. Maternal antenatal anxiety and children's behavioural/emotional problems at 4 years. Report from the Avon Longitudinal

Study of Parents and Children. Br J Psychiatry 2002;180:502.

Figures:



Tables:

Text bits:

Over the years three terms associated with pathological intrauterine growth have comeinto common use, but are still not clearly defined in the sense that they are often used

synonymously. These are “Small for Gestational Age” (SGA) and “Intrauterine GrowthRestriction” (IUGR), the latter divided into symmetrical and asymmetrical IUGR.

Looking at the distinctive association of various causes with different patterns of growth,it makes sense to clarify the terminology.

From a physiological point of view, two forms of pathological growth can be

distinguished: symmetrical and asymmetrical growth deficiency34

. Symmetrical IUGR may indicate an early intrinsic insult impairing fetal growth, such as chromosomal

abnormalities and congenital malformations, drugs or other chemical agents, or infection.Growth is symmetrically impaired because the insult happens at a time when fetal growth

occurs primarily by cell division, and is very likely due to an early intrinsic insult, such aschromosomal aberrations and congenital malformations, chemical toxins including drugs,

or infections. These suspicions usually arises from the maternal history and cannot beclarified any further by ultrasound alone but need other diagnostic measures, such as

fetal karyotyping, serum examination for evidence of infection, clinical observation for signs of preecalmpsia, and evaluation of congenital and acquired thrombophilic disorders.

Symmetric growth restriction usually occurs earlier in pregnancy, results more often in preterm labor than asymmetric growth restriction, and affected fetus have a higher

neonatal morbidity rate and attain a lower mean birth weight than those with asymmetric

growth restriction35.

By contrast, asymmetric growth restriction is usually caused by extrinsic factors that

result in inadequate supply for the fetal metabolism, most often due to maternal vascular disease and decreased placentar perfusion. Musculoskeleton and head circumference

develop normally, whereas the abdominal circumference is decreased due to subnormalliver size and lack of subcutaneous fat. Asymmetric growth restriction usually occurs

later in pregnancy, when fetal growth is usually due more to an increase in cell size thanin cell number

36.

The growth pattern of SGA is much alike to the symmetrical type IUGR. “Small for

Gestational Age” should be applied to fetus below the 10th

percentile, which displaynormal anatomy and a normal growth pattern, i.e. parallel the normal growth curve at a

fixed distance. If the fetus is somewhat small, but anatomically normal with anappropriate amniotic fluid volume and growth rate, the outcome will usually be a normal,

constitutionally small neonate. If compared to a local, matched reference population,about 70% of fetus with an estimated birth weight below the 10

thpercentile will be

constitutionally small16

, and thus not be at increased risk.



Given this background, it makes sense to use the term “SGA” for all fetus below the 10th

percentile, unless maternal history arouses suspicion of growth restriction, or any of the

potential causes for symmetric growth restriction have been diagnosed. Under thesecircumstances, “SGA” becomes synonymous with the healthy, but constitutionally small

fetus, while the term “IUGR” is reserved for those fetuses that have reason for or show

signs of a pathological growth pattern. Mathematically, such a pathological growth pattern is best expressed by the z score, which represents the growth velocity. In the present simulation, when dz/dt is less than zero, the growth curve is no more parallel to

the normal growth curve, which leads to IUGR diagnosis.

Nevertheless, this review noted an awareness that repeated ultrasound examinations havea psychological impact on the mothers which at the time was insufficiently examined,

and recommended that further studies focus on this topic.In developed societies with health care systems that provide if not mandate extensive

antenatal monitoring, both parents and doctors often seem to perceive pregnancy as amanageable disease rather than a physiological state. The potential to use additional

diagnostic measures may raise the anxiety level, which in consequence leads to increasedhealth care use during pregnancy. The mothers’ perception of being “at risk” affects their

psychosocial status37. Significant associations were found between depression and/or anxiety and increased nausea and vomiting, prolonged sick leave during pregnancy and

increased number of visits to the obstetrician, specifically, visits related to fear of childbirth and those related to contractions. Planned cesarean delivery and epidural

analgesia during labor were also significantly more common in women with antenataldepression and/or anxiety

5. Maternal anxiety level was found to be associated with

increased risk for preterm birth38

. Whether the neonate is affected by maternal anxiety or mood disorders in general is controversial39,40, but it seems that maternal antenatal

anxiety leads to emotional and behavioral problems in the children at four years of age

41,42.

fetal growth z-velocity

Documents