Characterisation of the possible effect on birthweight following frequent prenatal ultrasound

examinations

Sharon Evans, John Newnham*, William MacDonald, Catrina Hall

Foundation for Women ‘s and Infants ’ Health, University Department of Obstetrics and Gynaecology. King Edward Memorial Hospital for Women and Princess Margaret Hospital, Bagot Road, Subiaco.

Western Australia 6(#8. Australia

Received 29 May 1995; revised 14 December 1995; accepted 1X December 1995

Abstract

The objective of this study was to evaluate and character&e by study of newborn biometry a possible effect on birthweight which we observed previously in a randomised controlled tial of multiple prenatal ultrasound examinations. A total of 2743 women with single pregnancies had been allocated at random to either a protocol of ultrasound imaging and continuous wave Doppler studies at 18, 24, 28, 34 and 38 weeks gestation (intensive group), or to a protocol of a single imaging examination at 18 weeks and further imaging scans only as clinically indicated (regular group). When compared with those in the regular group, and adjusted for other confounding variables, normally formed babies of term gestational age in the intensive group tended to be shorter when measured at birth (P = 0.123) and on day 2-3 of age (P = O.Ot%). There were statistically insignificant reductions in the circumferences of the chest, abdomen and mid-arm; and in the skinfold thicknesses of the triceps, parascapular and subscapular regions. Principal component analysis showed a trend for a reduction for the skeletal component (p = 0.085) but not for the soft tissue component (P = 0.332). Comparison of the neonatal biometry in the two groups is not conclusive, but the differential effects on the various growth parameters suggest that if multiple scans do indeed restrict fetal growth, the mechanism is more likely to be an effect on bone growth rather than a reduction in nutrient supply from the placenta.

Keywords: Prenatal ultrasonography, adverse effects; Fetal growth

*Corresponding author. Tel.: + 61 9 3401393; fax: + 61 9 3401319.

037X-37X2/96/$15.00 0 1996 Elsevier Science Ireland Ltd. All rights reserved PII SO378-3782(96)01728-S

204 S. Evans et al. I Early Humun Development 45 (1996) 203-214

1. Introduction

The increasing use of diagnostic ultrasound is making a valuable contribution to the management of human pregnancy, but its expanding role is dependent on our certainty there are no harmful effects. We recently reported the results of a randomised controlled trial which was designed to investigate the hypothesis that offering pregnant women a protocol of frequent ultrasound imaging and Doppler flow velocity waveform examinations would improve the outcome of pregnancy [9,10]. The additional ultrasound tests did not result in improvements in pregnancy outcome but were associated with increased numbers of newborns with birthweights in the lower centile groups. This effect could not be explained by any differences between the two study groups in maternal or paternal size, complications of pregnancy or obstetric intervention. These findings have confirmed the need to investigate the hypothesis that multiple ultrasound scans may restrict fetal growth in a proportion of cases. Ideally, testing of this hypothesis would require a controlled trial with fetal growth as the primary question, Conduct of such a trial would need to gain acceptance from an appropriate ethics committee, together with a population of pregnant women and their clinicians, both of whose approval could not be taken for granted. If the definitive trial is not to be performed, further investigation of the relationships between multiple ultrasound exposure and fetal growth will require thorough description of the effect observed in our controlled trial, appropriate animal studies and exploration of the mechanisms by which insonation could possibly influence growth.

The purpose of this study was to characterise the birthweight effect observed in our previously reported trial in order to assist in the direction of future research. Data collected prospectively during the pregnancies have been used to investigate the possible effect on birthweight of maternal stress and physical activity while controlling for known confounding variables, and detailed measurements of the newborns have been evaluated to describe the patterns by which growth may have been altered.

2. Methods

2.1. Trial design

Details of the design of this trial have been reported previously [9,10]. Briefly, this was a randomised controlled trial designed to investigate the hypothesis that frequent ultrasound imaging and Doppler flow velocity waveform examinations would reduce neonatal morbidity and preterm birth. Pregnant women with single pregnancies were enrolled at 16-20 weeks gestation and then allocated at random to either an intensive group for ultrasound imaging and continuous wave Doppler studies at approximately 18, 24, 28, 34 and 38 weeks gestation, or to a regular group to have a single imaging examination at approximately 18 weeks gestation with further scans only as clinically indicated. This latter group was in essence a control group. The study was approved

S. Evans et al. / Early Human Development 45 (1994) 203-214 20.1

by the Institutional Ethics Committee and written consent was obtained from each woman at the time of enrolment.

2.2. Data collection

Data on standard clinical outcomes were collected from the hospital records by study midwives. These outcomes included measurements of birthweight and crown- heel length made immediately after birth by hospital midwives. On the second or third day of newborn life, a specially trained research midwife examined the infant and recorded the crown-heel length to the nearest 0.1 cm using a Harpenden neonatometer (Holtain Ltd., Crymych, Dyfed, UK); and head, chest and mid-arm circumferences were measured to the nearest 0.1 cm with a paper tape measure using standard techniques [3,7]. The chest circumference was measured at the level of the nipple, and abdominal circumference immediately above the umbilicus, with both measurements made at the end of expiration. Skinfold thicknesses from the left triceps and subscapular regions were measured using a Holtain calliper with the infant lying prone; the calliper was left in place until the needle gauge had stopped moving. In addition, a parascapular skinfold was measured adjacent to the medial border of the scapula at the level of the scapular spine. More than 95% of the measurements were made by a single observer, the remainder by a consultant paediatrician.

Inter- and intraobserver coefficients of variation were estimated when the study was one-third completed on two separate groups of 16 term infants, measured 6 h apart. For body length the interobserver variation was 0.8% and the intraobserver variation was 0.9%. The variations for head circumference were 0.4% and 0.20/c, respectively, chest circumference 1.3% and 1 .O%, abdominal circumference 3.2% and 1.3%, triceps skinfold 11.1% and 5.1%, parascapular skinfold 10.78% and 4.0%, subscapular skinfold 8.4% and 4.2% and mid-arm circumference 4.3% and 1.8%. All data were computerised using the Statistical Analysis System (SAS): Gary, North Carolina.

2.3. Statistical methods and analyses

Univariate analyses on numeric variables were calculated by Student’s t-test or Wilcoxon Rank Sum Test as appropriate. Group testing of categorical data was by Fisher’s Exact Test or Mantel-Haenszel Chi Square Test for linear association of ordinal variables. Birthweight percentiles were calculated using locally derived birthweight charts which account for maternal height, parity and fetal sex [2]. The birthweight ratio was calculated as the infant’s birthweight divided by the median birthweight for the sex and gestational age from these standards [S].

Socio-economic index was calculated at 16-20 weeks gestation by maternal and paternal levels of education, employment, family income and marital status. Physical activity during the first 3 months of pregnancy was assessed at the time of enralment by the method of Taylor et al. [ 14) and expressed as kilocalories expended per day. A further questionnaire was administered at 34 weeks gestation which included measurement of the stress of recent life events [lS].

206 S. Evans et al. I Early Human Development 45 (1996) 203-214

Neonatal biometry in the two groups was compared in analyses of covariance using a sample which included only term pregnancies (37-42 completed weeks) and with the exclusion of major congenital abnormalities including cardiac anomalies (n = 6), Down syndrome (4), chromosomal disorders (3), major limb defects (2), exomphalos (l), diaphragmatic hernia (1) and arthrogryposis multiplex congenita (1). The covariates chosen were maternal height, pre-pregnancy weight and age; the rate of maternal weight gain between 18 and 34 weeks gestation; so&o-economic index; maternal parity, physical activity, smoking practice, diabetes and history of previous fetal growth restriction; infant sex, and gestational age at birth expressed as a quadratic function to account for the growth profile within this gestational age range. The dependent variables were the weight and length at birth, and biometry measured on the second or third day of life as described previously.

Principal components analysis [6] was employed. This is a method of dimension reduction in multivariate analysis which aims to transform a group of observed variables which are highly correlated into a new reduced set of uncorrelated variables which simplify subsequent analyses. If the original set of variables is highly correlated, then they are effectively ‘saying the same thing’ about the data, and so may be replaced with one single principal component of the variance.

3. Results

Details of the mean birthweight and cumulative birthweight centiles are shown in Table 1. The mean birthweight of infants alive at birth was 25 g less in the intensive group than in the regular group but this difference was not statistically significant by univariate methods (P = 0.45). Examination of the centile groups indicates a significantly greater number of birthweights below the 10th and 3rd centiles without differences below the 50th and 90th centiles. The effect of gestational age on the proportion in each group with birthweights less than 10th centile are shown in Fig. 1. Increased numbers of growth restricted babies in the intensive group were observed in 12 of the 17 gestational age groups with equal numbers in two age groups and fewer in only three age groups. This effect of increased proportions of growth restricted

Table 1 Mean birthweight and birthweight centiles of infants alive at birth in the two groups

Intensive Regular P-value Relative Risk (95% CI)

n 1375 Mean birthweight (g) 3320 (582) Birthweight centiles

<9oth 1239 (90.1) <5Oth 674 (49.0) <lOth 179 (13.0) < 3rd 58 (4.2)

Birthweight ratio 0.99 (0.13)

Data are n (standard deviation).

1368 3345 (573) 0.45

1215 (88.8) 0.29 1.01 (0.99-1.04) 643 (47.0) 0.30 1.04 (0.97-1.13) 132 (9.6) 0.006 1.35 (1.09-1.69) 35 (2.6) 0.020 1.65 (1.09-2.49)

1.00 (0.13) 0.013

S. Evans et al. I Early Human Development 4.5 (1996) 203-213 207

22 24 26 28 30 3i? 34 3tt 40 42

Fig. 1. The number of infants with birthweight < 10th centile in the intensive (black) and regular I grey) groups plotted against gestational age in completed weeks.

newborns across the whole gestational age spectrum was not accompanied by a difference between the two groups in age at birth. Fig. 2 displays the number of births at each gestational age and confirms that in this trial the protocol of frequent ultrasound scans did not alter the age at birth (Z’ = 0.694).

In order to display the differences in birthweights between the two groups, and accounting for the effects of gestational age, maternal height, parity and infant gender, each birth was expressed as the birthweight ratio estimated from locaily derived charts. Fig. 3 displays the number of live born infants plotted against the birthweight ratio. The differences between the two groups can be seen to occur at the left end of the curve, with similar distributions in the higher fractions (Kolmogorov Smimov P = 0.085). The number of birthweights in the intensive group relative to the regular group remained in excess up to the 0.9 cumulative fraction.

In order to investigate the possible mechanism by which frequent ultrasound scans could influence growth, the neonatal biometry was compared on a sample consisting of all live born infants of term gestational age (37-42 completed weeks inclusive) and without major congenital abnormalities. This left 1219 in the Intensive group and 1196 in the Regular group. Table 2 displays characteristics of these women and their histories which may have influenced fetal growth and which were included in the mathematical modelling. There were similar proportions between the two groups in the number of women with a previous history of having delivered an infant with

208 S. Evans et al. I Early Human Development 45 (1996) 203-214

50

0 -I-

22 24 28 28 30 32 34 36 38 40 42

Fig. 2. The number of infants plotted against gestational age at birth in the intensive (black) and regular (grey) groups. Age is presented as completed weeks.

birthweight less than the 10th centile, their socio-economic assessment, their physical activity level as recorded by questionnaire at 18 weeks gestation, and the stress of recent life events reported at 34 weeks gestation.

A total of 336 (13.9%) neonates were measured immediately after birth but were not measured again on day 2 or 3 of life because their mothers had been discharged home early (intensive 97, regular 11 l), had delivered elsewhere (intensive 46, regular 60) or had withdrawn from the study (intensive 13, regular 9). Analysis of the data collected immediately after birth has revealed no differences in birthweight between these infants and those which did have subsequent measurements on day 2 or 3 of life (intensive P = 0.501, regular P = 0.978). Details of infants included in the study at birth and on the second or third day of newborn life are shown in Fig. 4 and Fig. 5. The P-values quoted are those calculated for univariable comparisons (t-test) and for multivariate analysis of variance controlling for the confounding variables as outlined in Section 2.3.

The decrease of 25 g in birthweight is not significant by univariate methods but with 33% of the variance in the model explained by the covariates, the group effect showed a trend towards significance (P = 0.080). When compared with those in the regular group, infants in the intensive group were not significantly shorter when measured immediately after birth (P = 0.123) and showed a trend on day 2-3 of age (P = 0.068). Mean head circumference in the intensive group was significantly less

E 1 100-I

Evans et al. I Early Human Development 45 (1996) 203-214 3oY

Oh 4

0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5

Birth weight ratSo Fig. 3. The number of infants plotted against the birthweight ratio for the regular group (grey) superimposed over the intensive group (black).

when measured immediately after birth (P = 0.032) but this measurement was not included in further statistical analyses because of the confounding effects of moulding. There was no significant difference between the two groups in head circumference when measured again 2-3 days later (P = 0.197). Measurements of the circumferences of the chest, abdomen and mid-arm, and skinfold measurements from the triceps, parascapular and subscapular regions, were not statistically different. There was no significant difference between mean ponderal index at birth in the intensive group (2.84, S.D. 0.289) and in the regular group (2.82, S.D. 0.282) (P =T 0.145) nor when measured at 2 days of age (intensive 2.74, SD. 0.240; regular 2.74, S.D. 0.254; P = 0.795). The difference in ponderal index at the two times of measurement may have been due to greater precision of measurement of length by the neonatometer at 2 days of age when compared with the regular tape measure used on the day of birth.

To investigate further the relative reductions in bone and soft tissue growth. principal components analyses were performed on the measurements of skeletal size (length, head circumference and chest circumference) and on the measurements of soft tissue size (abdominal circumference, and skinfold thicknesses from the triceps, parascapular and subscapular regions). These analyses allow an assessment of the combinatory effect of several highly correlated variables. The first principal com- ponent for the skeletal measurements explained 73% of the total variance with

210

Table 2

5’. Evans et al. I Early Human Development 45 (1996) 203-214

Characteristics of the women and their lifestyles of all pregnancies ending between 37 and 42 completed weeks of gestation and without major congenital abnormality

Intensive Regular P-value

N Maternal height (cm) Median pre-pregnancy weight (kg) Maternal age Parity

0 1 >l

Smoking practice 0 l-10 > 10

Diabetes History of previous birthweight < 10th centile

0 1 >l

Mean socioeconomic index Mean physical activity at 18 weeks gestation (kCal/day) Stress at 34 weeks gestation

0 l-2 3-4

>4

Male fetus Birthweight Birthweight ratio

1219 1196 163.6 (6.63) 163.7 (6.50)

57 (52-64) 57 (52-65)

27.5 (5.9) 27.5 (6.0)

569 517

366 354

284 265

906 864

191 185

122 147

41 39

1151 1136 51 49 11 11

1.88 (0.83) 1.87 (0.83) 2554 (267) 2559 (262)

473 452

524 498

129 116

11 12

632 592

3404 (466) 3430 (459)

0.99 (0.124) 1.00 (0.120)

0.740

0.103

0.914

0.406

0.202

0.910

0.629

0.904

0.682

0.370

0.255

0.170

0.103

Data are n, mean (standard deviation) or median (interquartile range) as appropriate

approximately equal weighting of the three variables. Linear modelling of the first principal component, with the independent variables of gestational age, gestational age squared, fetal gender, maternal height, pre-pregnancy weight, parity and history of previous growth restriction, although not significant, showed a strong trend for the intensive group relative to the regular group to have smaller measures of size (P = 0.085). Similar analysis of the soft tissue measurements also revealed the first principal component to explain 74% of the variance with approximately equal weighting of the four measures. Linear modelling showed this first principal component was dependent on only gestational age, pre-pregnancy maternal weight, parity and past history of growth restriction, and an insignificant difference between the intensive and regular groups (P = 0.332).

4. Discussion

The purpose of this study was to investigate further the effects on birthweight observed in our randomised controlled trial of frequent prenatal ultrasound examina-

S. Evans et al. / Early Human Development 4.5 (1996) X3-214 211

9 350 4

i cm cm c

50- i 35.- 33- 1 T

49- 34 -I 321

I I I /

Birthweight -----Lwth Head am chest arc Day 0 Day 2-3 Day 2-3 Dsy 2-3

n Int 1219 1210 1071 1143 1142 R@GJ 1196 1178 1019 1076 1076

p Uni 0.380 0.007 0.083 0.269 0.216 Multi 0.080 0.123 0.068 0.197 0.177

Fig. 4. Mean and standard errors for infants in the intensive (W) and regular (0) groups showing birthweight and length immediately after birth, and length, head circumference and chest circumftrence on day 2 or 3. Shown in tabular form are the number of infants measured in the intensive and regular _mups and the P-values from univariate and multivariate analyses.

tions [9,10]. As reported previously, significantly more infants were of birthweight less than the 10th and 3rd centiles in the group which had been allocated to receive frequent scans, when compared with those allocated to receive a single ultrasound examination at 18 weeks gestation with further scans only if clinically in&a&d. Mean birthweight in the intensive group was 25 g less, but this difference was not statistically significant. The increased number of babies in the lower birthweight centiles in the inknsive group was not associated with a shift in gestational age at birth and the effect appeared to occur throughout the range of gestational ages. In statistical terms, the differences between the two groups was significant when tested either by comparison of the number of birthweights in the lower centile groups, or by testing the residual differences between the lower ends of the two birthweight ratio curves.

As discussed elsewhere [9,10], it remains possible that the effect observed on the lower birthweight centiles was due to chance. The trial was not designed to study the bioeffects of ultrasound and a reduction in birthweight was not the hypothesis u&r

investigation. Moreover, we cannot exclude the possibility that a component of the intervention other than ultrasound exposure itself may have produced the effect. Some of the other*‘&ssible factors which could have influenced fetal growth in this trial would include increased stress in the women by more constant awareness of the fetus. or, conversely, a reduction in stress from the frequent reassurance offered by

212 S. Evans et al. / Early Human Development 45 (1996) 203-214

cm Cll

31.7- 10.6,

1

30.71 10.2.

- Abdom Mldarm

n Int 1142 989 Reg 1075 923

p Uni 0.360 0.483 Multi 0.419 0.168'

cr

4.6

4.2

___ - Trtwps Paftmc Subscap

1135 1141 1141 1072 1074 1075

0.253 0.288 0.540 0.227 0.256 0.539

Fig. 5. Mean and standard errors for infants from the intensive (m) and regular (0) groups showing measurements done 2 or 3 days after birth of the circumferences of the abdomen and mid-arm, and skinfold thickness from the triceps, parascapular and subscapular regions. Shown in tabular form are the number of infants measured in the intensive and regular groups and the P-values from univariate and multivariate analyses.

the scans resulting in increased physical activity and less rest. There were no differences between the groups in the stress of recent life events reported by the women at 34 weeks gestation, nor in their physical activity level when assessed by questionnaire at 18 weeks gestation. We did not collect data specifically on the response of the women to their ultrasound examinations, nor their physical activity during the third trimester.

Since the hypothesis that multiple ultrasound scans may influence birthweight is supported by animal studies [ 1 l-131, we have thought it important to characterise the birthweight reduction in order to obtain clues of a possible mechanism and thus assist in the direction of future research. Accordingly, we have analysed a sample of the population restricted to normally formed infants delivered at term gestation, so as to remove the confounding effects of those pregnancy complications which result in preterm birth and the possible alterations in growth which occur in fetuses with major anomalies. In the monkey model, Tarantal and colleagues [12,13] observed significant reductions after multiple scans in birthweight and crown-rump length, with similar measurements to control newborns in biparietal diameter, occipitofrontal diameter, head circumference, hand and foot lengths, humerus and femur lengths, arm circumference, chest circumference, tail length and skinfold thicknesses. The results of the present study have some similarities with significant effects on neonatal length,

S. Evans et al. I Early Human Development 45 (1996) 203-214 2 I .3

inconsistent effects on head circumference and less effect on other circumferences and skinfold thicknesses. The principal components analyses have further supported the possibility that there has been a relatively greater reduction in growth of skeletal structures than of soft tissues. Together, the results of these studies are not conclusive and do not present a clear picture, but if anything suggest that if multiple scans do indeed affect fetal growth, the mechanism may be through growth of bones. The biometric findings are not consistent with a symmetrical pattern of growth restriction and the relative lack of differences in abdominal circumference and skinfold thicknesses would appear to preclude the possibility of reduced nutrient supply from placental insufficiency. Ultrasonic energy is selectively absorbed by bone and the temperature rise following insonation is 30 times greater in fetal bone than in soft tissue [5]. There is some evidence that high intensity ultrasound can damage epiphyseal plates and retard bone growth [1,4]. The ultrasound employed in our controlled trial was not of sufficient energy to cause a thermal effect [9:] but the results nevertheless suggest that further research on the effects of ultrasound exposure on fetal bone growth are warranted. Long-term follow-up of this cohort is in progress and the serial measurements are expected to determine the effects, if any. on childhood growth of multiple prenatal ultrasound examinations.

Acknowledgments

Supported by grants from the Raine Research Foundation of The University of Western Australia, National Health and Medical Research Council of Australia and the Research Foundation for Women’s and Infants’ Health at King Edward Memorial Hospital.

References

[ 11 Bender, L.F., Janes, J.M. and Herrick, J.F. (1954): Histological studies following exposure of bone to ultrasound. Arch. Phys. Med. Rehabil., 35, 555-559.

[2] Blair, E. and Stanley, F.J. (1985): Birthweight Charts from a Western Australian Population for Singleton Caucasian Live Births After 21 Weeks Gestation. Australian Government Publishing Service, 1985.

[3] Cameron, N. (1984): The Measurement of Human Growth, pp. 66-91. Croom Helm, London. [4] DeForest, R.E., Herrick, J.F. and Janes, J.M. (1953): Effects of ultrasound on growing bones: an

experimental study. Arch. Phys. Med. Rehabil., 34, 21-31. [S] Drewniak, J.L., Cames, K.I. and Dunn, F. (1989): In vitro ultrasonic heating of fetal bone. J. .4coust.

Sot. Am., 86, 1254-1258. [6] Joliffe, LT. and Morgan, B.J. (1992): Principal component analysis and exploratory factor analysis.

Stat. Methods Med. Res., 1, 69-9.5. [7] Lohman, T.G., Roche, A.F. and Martorell, R. (1988): Anthropometric Standardization Reference

Manual, pp. 5-68. Human Kinetics Books, Champaign, IL. [8] Morley, R., Brooke, O.G., Cole, T.J., Powell, R. and Lucas, A. (1990): Birthweight ratio and outcome

in preterm infants. Arch. Dis. Child., 65, 30-34.

214 S. Evans et al. I Early Human Development 45 (1996) 203-214

[9] Newnham, JR, Evans, S.F., Michael, C.A., Stanley, F.J. and Landau, L.I. (1993): Effect of frequent ultrasound during pregnancy: a randomised controlled trial. Lancet, 342, 887-891.

[lo] Newnham, J.P., Evans, S.F., Michael, C.A., Stanley, F.J. and Landau, L.I. (1993): Effects of frequent ultrasound during pregnancy. Reply to letters to the Editor. Lancet, 342, 1359-1361.

[ 1 I] O’Brien, W.D. Jr. (1983): Dose dependent effect of ultrasound on fetal weight in mice. J. Ultrasound Med., 2, l-8.

[12] Tarantal, A.F. and Hendrickx, A.G. (1989): Evaluation of the bioeffects of prenatal ultrasound exposure in the Cynomolgus Macaque (Macacu fasicularis): I. Neonatal/infant observations. Teratology, 39, 137-147.

[13] Tarantal, A.F., O’Brien, W.D. and Hendrickx, A.G. (1993): Evaluation of the bioeffects of prenatal ultrasound exposure in the Cynomologus Macaque (Macacu fusiculan’s): III. Developmental and hematologic studies. Teratology, 47, 159- 170.

[14] Taylor, H.L., Jacobs, D.R., Schucker, B., Knudsen, J., Leon, A.S. and Debacker, G. (1978): A questionnaire for the assessment of leisure time physical activities. J. Chron. Dis., 31, 741-755.

[15] Tennant, C. and Andrews, G. (1976): A scale to measure the stress of life events. Aust. New Zealand J. Psychiatry, 10, 27-32.