Birth weight, prematurity and accuracy of gestational age

GYNECOLOGY &OBSTETRICS International

Journal

of Gynecology

Birth weight, prematurity

& Obstetrics

aDepamnent of Obstetrics Research and Monitoring

Received

30 April

251-256

and accuracy of gestational age

M. Mongelli”*, bPerinatal

56 (1997)

J. Gardosib

and Gynoecology Prince of Wales Hospital, Shatin, Hong Kong Unit, Department of Obstelrics and Gynaecology, Queens’ Medical Nottingham, UK

1996; revised

2 December

1996; accepted

20 December

Centre,

1996

Abstract Objective: To investigate commonly used birth weight categories in relation to gestational age assignment by menstrual (LMP) or ultrasound dates. Methods: A total of 34249 cases were retrieved from the East Midlands Obstetric Database. Of these, 2281 (6.7%) delivered preterm (< 37 weeks) by ultrasound dating. The percentage of preterm and post-term cases was calculated for birth weight categories from < 1500 g to > 4499 g at 500 g intervals. The incidence of preterm delivery was estimated for birth weights from < 1500 g to 3400 g. Results: For the ‘low birth weight’ cut-off of 2500 g, only 64% were actually preterm as assessed by ultrasound, as opposed to 59% as judged by menstrual dates. Nearly 95% of infants weighing over 3000 g are full-term. For birth weights under 2900 g, there is a trend for LMP-derived gestational age to underestimate preterm delivery, by up to 35.5%. By ultrasound dating, 90% of infants under 1800 g are preterm, rising to 98.2% for those under 1500 g. Conclusions: As nearly 40% of ‘low birth weight’ infants are born at term, the old classification of ‘low birth weight’ should be replaced by gestational age-specific percentile categories. Menstrual dates systematically underestimate the prevalence of preterm delivery. Previous estimates of preterm delivery rates based on LMP data have under-stated their true incidence. 0 1997 International Federation of Gynecology and Obstetrics Keywords:

Preterm

delivery; Post-maturity;

Birth weight

1. Introduction The strong link between low birth weight and perinatal mortality formed the basis for classifying underweight infants into categories such as

* Corresponding OO20-7292/97/$17.00 PII SOO20-7292(96)

author.

Fax:

+ 852 26360008.

Q 1997 International 02835-X

Federation

of Gynecology

‘low birth weight’ (LBW, < 2500 g> and ‘very low birth weight’ (VLBW, < 1500 g). These were intended to identify those infants most likely to need special care in the neonatal period [l]. The reason birth weight was selected in preference to gestational age is probably because the latter relied on menstrual dates, which were often missing or unreliable. and Obstetrics

252

M. Mongelli,

J. Gardosi

/International

Journal

As a result, vital statistics such as survival figures in neonatal intensive care units continue to be quoted in relation to birth weight [2], even though gestational age has been shown to be a better predictor of survival [3]. Gestational age estimated from menstrual dates can be very inaccurate, due to the great variability in the timing of ovulation in relation to the menstrual cycle [4]. Furthermore, in up to 20% of cases the LMP is not known. For those who book late, their menstrual dates cannot be confirmed by ultrasound. We have found that in relation to ultrasound-estimated gestational age, the 95% confidence interval of gestational age derived from menstrual data is -27 to +9 days, which indicates a marked tendency to overestimate the length of pregnancy by LMP [5]. In contrast, the accuracy of gestational age estimated from the biparietal diameter in the mid-trimester results in a 95% confidence interval of t- 6.3 days [6], which we were able to confirm [7]. The greater accuracy of ultrasound is now well established, and the inclusion of menstrual dates such as the lo-day rule confers no clear advantage [8]. In this study we present the significance of commonly used birth weight categories in terms of gestational age assessed by both ‘certain’ menstrual dates and ultrasound examination. This information should be useful in the interpretation of a large body of historical data that has been expressed in terms of birth weight. 2. Subjects and methods

The computerised obstetric records of the East Midlands Obstetric Database [9] were derived from three major maternity units in the East Midlands (Nottingham City and University Hospitals, and Derby City Hospital). Excluded from analysis were stillbirths, multiple pregnancies, pregnancies with congenital abnormality, and late booking pregnancies (over 24 weeks). This left a total of 34249 cases dated by both menstrual and ultrasound criteria for analysis, of whom 2281 (6.7%) delivered preterm (< 37 weeks) by ultrasound dating. The menstrual histories were taken by midwives, and were entered only if the patient was

of Gynecology

& Obstem’cs 56 (1997) 251-256

certain of her dates, the menstrual cycle was regular and oral contraceptives had not been used in the preceding 3 months. Ultrasound examinations at the three maternity units were carried out with Aloka 650, ATL Ultramark 9 and Toshiba Tosbee, 250, 410, 450, and 565 machines. The vast majority of cases were examined trans-abdominally with curvilinear transducers. Gestational age at booking was estimated in 95.5% of cases from the BPD using Campbell’s dating standard [lo], the remainder being dated by either the crown-rump length (for gestational age less than 12 weeks) or the femur length (for cases where the BPD could not be visualized optimally). The mean gestational age at the time of ultrasound examination was 15.6 weeks, with 3.4% of cases over 20 weeks at booking. 2.1. Statistical methods The percentage of cases whose gestational age at birth was less than 259 days, less than 224 days and over 294 days (post-term) by both menstrual and ultrasound dating was calculated for birth weight categories at 500 g increments. A more detailed analysis of the rates of preterm deliveries (< 259 days) by both dating methods was carried out at 100 g birth weight increments from < 1500 gto 3400 g. We examined the relationship between error of menstrual dates (as assessed by ultrasound) in relation to the apparent length of pregnancy as estimated by menstrual dates. For each case, the difference between LMP-estimated and ultrasound-estimated gestational age was calculated in days and converted into exact weeks. A positive value indicates overestimation of gestational age by the LMP. The Spearman rank correlation coefficient was calculated between these variables. Statistical analyses were performed with SPSS for Windows (version 6, SPSS Inc., Chicago). 3. Results

The characteristics of the population are described in Table 1. This is a predominantly European population (90%) with reproductive histories fairly typical of Western societies.

M. Mongelli, Table 1 Population shown with

J. Gardosi

characteristics (N = 34 249). standard deviations in brackets

Maternal age (years) Weight at booking (kg) Height (cm) Parity Primiparae (%) Ethnic groups (%) White Afro-Caribbean Indo-Pakistani Other Missing Gestational age by ultrasound Gestational age by LMP

/International

Mean

Journal

values

are

26.9 (5.1) 64.5 (12.0) 162.4 (6.4) 43.0 89.7 2.2 4.7 1.1 2.2 276.7 (13.1) 279.5 (14.8)

Table 2 Birth weight distribution in relation to gestational age group and method of gestational age estimation. The groups include ‘very preterm’ with gestational age < 32 weeks, preterm with gestational age of 32-36 weeks and post-term, with gestational age > 42 weeks Group

Birth

weight

fg)

N

Median

Mean

S.D.

Ultrasound dating Very preterm Preterm Post-term

276 1983 1238

1300 2550 3740

1314 2547 3762

406 502 472

LMP dating Very preterm Preterm Post-term

280 1844 3822

1330 2600 3560

1506 2609 3575

718 581 475

Table 2 shows the birth weight distribution in relationship to gestational age categories. Mean birth weights are lower when ultrasound is used to define the gestational age categories. Table 3 shows the relationship between birth weight categories and the percentage of cases classified as either preterm or post-term. Almost 95% of infants weighing over 3000 g are born at term. Of the very large infants ( > 4500 g), 11.4% are post-term by WHO criteria [ll]. Fig. 1 is a graph plotting the percentage of preterm delivery in relation to birth weight categories, using either menstrual dates or ultrasound-estimated gestational age to define prema-

of

Gynecology

& Obstetrics

56 (1997)

251-256

253

turity. For birth weights under 2900 g, there is a consistent trend for LMP-derived gestational age to underestimate the prevalence of preterm delivery, by up to 35.5% in the range of 1800 g to 1900 !3 By ultrasound dating, 95% of infants whose birth weight is less than 1800 g are preterm; this percentage rises to 98.2% for those under 1500 g. For the commonly used cut-off of 2500 g (‘low birth weight’), only 64% are actually preterm as assessed by ultrasound. There was a highly significant positive correlation between gestational age error of menstrual dates (as assessed by ultrasound) and the apparent duration of pregnancy as assessed by menstrual dates CR = 0.48, P < 0.0001). 4. Discussion

A vast amount of perinatal and birth weight data has been collected from populations in whom the date of the last menstrual period is the only means to estimate gestational age. For women in many parts of the world the basis of gestational age assignment remains the date of the last menstrual period and clinical examination, and often these are not obtainable. Given the wide error of menstrual dates, such perinatal data is difficult to interpret. In addition to improving the accuracy of gestational age assignment, ultrasound dating has improved our understanding of fetal growth considerably. For instance, in contrast to earlier beliefs, it has shown that significant fetal growth continues in the post-term period [12]. Our findings suggest that 36% of LBW infants are born at term. As 2472 g corresponds to the 10th percentile at 37 weeks in our population [9], many of these infants could be growth retarded or small for gestational age. Hence this definition is of limited use, since it cannot distinguish prematurity from fetal growth retardation. If birth weight is to be used as an epidemiologic index of preterm delivery, a cut-off of 1800 g could be more meaningful, since about 95% of infants weighing below this value were preterm by ultrasound dates. Another useful value would be the

254

M. Mongelli,

J. Gardosi

/International

Journal

of Gynecology

& Obstetrics

56 (1997) 251-256

Table 3 Birth weight categories and the percentage of cases classified aseither preterm or post-term according to dating method (ultrasound vs. LMP)

Gestational age (weeks)

Birth weight categories
2500-2999 N = 5674

3000-3499 N= 12840

3500-3999 N= 10114

4000-4499 N = 3208

> 4500 N=559

< 32 ultrasound < 32 LMP < 37 ultrasound < 37 LMP > 42 ultrasound > 42 LMP

83.3 77.9 98.2 96.8 0 0

0.1 0.2 13.7 11.3 0.7 5.5

0 0.1 2.2 2.6 1.5 8.6

0 0.1 0.5 1.1 3.5 11.4

0 0 0.2 0.7 6.5 14.9

0 0 0.5 0 11.4 19.0

22.2 17.4 88.9 80.6 0.3 0.3

0.9 1.5 50.4 42.5 0 2.7

3000 g threshold: 95% of these infants over this birth weight are full term. There is a significant risk of prematurity being

missed if menstrual dates are employed. By ultrasound dating, 64% of infants were classified as preterm, as opposed to 59% as judged by LMP.

Percent 100 90

60

Fig. 1. The percentage of infants deemed preterm (y-axis) by eitherultrasound line) is shown for each 100 g birth weight category (x-axis).

dates (top thick line) or menstrual dates (lower thin

M. Mongelli,

J. Gardosi

/International

Journal

This implies that without ultrasound scanning some babies may be incorrectly diagnosed as growth retarded. This is clearly illustrated in the figure; for most birth weight categories, a higher percentage of infants are recognized as pre-term by ultrasound dates than menstrual dates. It is of interest here that the ultrasound-dating curve is much smoother that the LMP dating curve, and this is probably because of the greater error associated with LMP dates combined with the smaller numbers in the low birth weight categories. Our results will be affected by the confounding effect of induction of labor. In our hospitals the overall rate is about 20%, the majority being carried out for post-maturity. Although there is no single induction policy for this condition, most consultants will induce at 10 days beyond the EDC. The effect would be an artificial reduction in the observed mean length of pregnancy, and fewer infants being born post-term than would be the case without medical intervention. We decided to include these pregnancies, since to do otherwise would lead to a degree of population selection which would alter the birth weight distribution. This issue should not materially affect our conclusions regarding the preterm period. With ultrasound dating, the incidence of postmaturity in large newborns (> 4500 g) is significantly lower than if menstrual dates are used (11.4% vs. 19.0%). The significant trend between error in menstrual dates and the length of pregnancy suggests that the longer the apparent duration of pregnancy by LMP, the more likely it is that gestational age has been overestimated by the latter. Post-maturity should nevertheless be considered in the management of macrosomia in pregnancies in which gestational age is not known accurately. Birth weight is significantly affected by variables such as ethnic group, maternal size, parity and fetal gender [131. The interpretation of birth weight data needs to take into account population characteristics. For example, the percentage of preterm infants among babies born below 2500 g is likely to be significantly smaller in ethnic groups with lower mean term birth weights. Since perinatal mortality attributable to prematurity differs from that due to growth retardation, the LBW

of Gynecology

& Obstetrics

56 (1997) 251-256

255

category should not be used in making conclusions regarding ethnic differences in preterm mortality rates. Our findings suggest that published figures on preterm delivery rates based on LMP-dated pregnancies may have underestimated the true extent of this problem. In the analysis of trends in preterm delivery rates, the method of gestational age assignment can be a significant confounding variable. Acknowledgments

We would like to thank Dr. Mark Wilcox for providing access to the East Midlands Obstetric Database. References Ounsted M, Ounsted C. On fetal growth rate. Clin Dev Med 1973; 46:40. affecting labour. In: Dl Beazley JM. Special circumstances Whitfield CR, editor, Dewhurst’s Textbook of Obstetrics and Gynaecology for Postgraduates. 1995: 317. SP, Venvey RA, Brand R, Graven[31 Verloove-Vanhorick hors JB, Keirse MJNC, Ruys JH. Neonatal mortality risk in relation to gestational age and birthweight. Lancet 1986; I (8472): 55-57. RT. Ultrasound instead of last menstrual pe141 Geirsson riod as the basis of gestational age assignment. Ultrasound Obstet Gynecol 1991; 1: 212-219. Gardosi J, Mongelli M. Risk assessment adjusted for El gestational age in maternal serum screening for Down’s syndrome. Br Med J 1993; 306: 1509. PH, Weldner BM. Reliability of ultrasound 161 Persson fetometty in estimating gestational age in the second trimester. Acta Obstet Gynecol Stand 1986; 65: 481-483. J, Mu1 T, Mongelli M, Wilcox M. Ultrasound [71 Gardosi dating and birth weight at term. Br Med J 1994; 308: 1635. M, Wilcox M, Gardosi J. Estimating the date 181 Mongelli of confinement: ultrasound biometry versus certain menstrual dates. Am J Obstet Gynecol 1996; 174: 278-281. J, Mongelli M, Ray C, Johnson I. [91 Wilcox M, Gardosi Birth weight from pregnancies dated by ultrasonography in a multicultural British population. Br Med J 1993; 307: 588-591. Campbell S, Newman GB. Growth of the fetal biparietal [lOI diameter during normal pregnancy. J Obstet Gynaecol Br Cwlth 1971; 78: 513-519. HI

256

[ll]

M. Mongelli,

J. Gardosi

/International

Journal

Manual of the International Statistical Classification of Diseases, Injuries and Causes of Death. Based on the Recommendations of the Ninth Revision Conference, 1975, and adopted by the 29th World Health Assembly. Geneva: World Health Organisation. 1977; 1: 773. [12] McLean FH, Boyd ME, Usher RH, Kramer MS. Post-

of Gynecology

& Obstetrics

56 (1997)

251-256

term infants: too big or too small? Am J Obstet Gynecol 1991; 164: 619-624. [13] Gardosi J, Mongelli M, Wilcox M, Chang A. Screening and assessment of fetal growth. In: Van Geijn HP, Copray FJA, editors, A Critical Appraisal of Fetal Surveillance. Amsterdam: Elsevier. 1994, Chapter 11.