Early modification to study protocol

Modifications to original study protocol were made soon after commencement with respect to inclusion criteria, methods o f recmitment and neonatal anthropometry. These are discussed in this section.

5.3.1.1 Definition o f SGA

An SGA infant was initially defined as one whose birthweight was < 3*^^ centile according to the Child Growth Foundation (CGF) algorithm. This criterion was originally proposed to limit recmitment to the mainly growth restricted infants and to minimise recmitment o f small but appropriately grown infants. However, it was recognised that with this definition we would not recmit sufficient full-term infants to achieve the necessary power o f study required during the study period. Definition of an SGA infant was therefore amended to include those infants whose birthweight was < 10^^ centile according to the CGF program.

It was acknowledged that more infants who were small but appropriately grown might be recmited by this amendment. However, the use o f the GROW program, which calculates an individualised birthweight ratio (Wilcox et al. 1993), may assist in identifying growth restricted infants. Thus we proposed, once data collection was completed, to conduct final analyses for all SGA infants as a group, and also separately for those under the 3*^** birthweight centiles to examine the extent to which potential misclassification of lUGR might influence study findings.

5.3.1.2 Gestational age

Recmitment o f all infants was originally limited to infants who were ^ 37 weeks gestation. This might also limit recmitment o f SGA infants, as infants in whom

growth restriction is suspected are often delivered electively before 37 weeks gestation. In order to ensure that such infants, who were otherwise eligible, were not excluded, gestational age for eligible infants was extended to > 35 weeks gestation.

5.3.1.3 Antenatal steroids

Similarly, infants whose mothers received antenatal steroids were excluded in the original protocol. However, if fetal growth restriction is identified during the second trimester and growth faltering continues to be o f concern to both obstetrician and paediatrician, antenatal steroids are often given to these mothers before a planned early delivery (often before 37 weeks gestation). While the effect o f antenatal steroids on fetal lung development is unclear, it was decided that such infants would be recruited if they were > 3 5 weeks gestation. However, pending the final number of such infants recruited to the study, final analyses would be performed for this group separately or, if the number was low, after their exclusion from the final analysis.

5.3.1.4 Recruitment

An audit o f the recruitment process (Section 3.1.5) following the first six months showed no difference in the success rate between face to face and postal recruitment (Appendix J). In view o f the time consuming nature o f trying to contact mothers on the wards, all subsequent recruitment was initiated by postal contact.

Accrual o f eligible subjects was limited due to the large number o f women fi*om ethnic minority groups delivering at Homerton Hospital. To ensure adequate numbers, recruitment was extended to infants delivered at University College Hospital, London. Ethical approval for this was granted and recruitment at the second site commenced September 1998.

5.3.1.5 Prenatal recruitment

Prenatal diagnosis of suspected fetal growth restriction may be made following confirmation o f reduced growth velocity by ultrasound scan. Hence, SGA infants

who were diagnosed antenatally were originally proposed as the 5^^ group o f infants in the study population. However, despite various attempts to recruit these infants over the first 12 months of the study period, referrals for this group o f infants were not forthcoming from the Fetal Medicine Unit. In view o f this, it was decided to omit this group from the study protocol.

5,3,L6 Anthropometric measurement

Accurate birth length measurement is essential for the calculation o f pondéral indices [birthweight (g) / length^ (cm) x 100] and for further classification o f SGA infants, as symmetric or asymmetric in terms o f body proportions. However, length is a difficult measurement in routine practice, and problems include equipment, training and availability o f personnel, as two people are needed to perform this measurement (Section 3.1.6). Our research team offered training on the use o f a stadiometer, for the accurate measurement of infant length to ward staff. In order to validate the accuracy o f infant body length measurement following the change in practice, two researchers performed length measurements on at least 10 infants prior to their discharge home in random order on the postnatal wards at three to six monthly intervals. The measurements obtained by the researchers were then compared with those obtained by staff on the Delivery Suite. Regrettably, the validation o f length measurements showed that ward staff were inconsistent with the method used (Appendix K), as mothers reported that tape measures were frequently used instead o f the stadiometer. Hence the decision was taken to accept that birth length, while still recorded, was unlikely to be as accurate as measurements taken in our laboratory due to unreliable technique.

With the extended recruitment to University College Hospital, there were slight policy differences between the hospitals with respect to neonatal anthropometric measurements. At University College Hospital, birth length was measured using a neonatal Rolameter (CMS). Length measurements using the stadiometer were found to be in close agreement with measurements made using the Rolameter. Chest circumference and mid-arm circumference were not measured at University College Hospital. Thus, although birth anthropometry was not routinely available for all the

infants studied, anthropometric measurements were performed on all infants at time of test by the researchers.