Umbilical cord blood samples were collected in a prospective study of pregnancy outcome that took place from January 1993 to December 1995 at Rogaland Central Hospital in Stavanger, Norway. The maternity clinic at this hospital serves a region of approximately 239
000 inhabitants. Deliveries (12
804) took place during the study period. The Norwegian Medical Birth Registry records information on all deliveries that take place in the country (Lie et al, 1998
), and we used this information to identify cases of pre-eclampsia and to select appropriate controls, as previously described (Ødegård et al, 2000
From the Medical Birth Registry, we initially identified approximately 1300 cases with clinical information indicating possible pre-eclampsia or eclampsia. After verifying and supplementing this information with details from the hospital records, we identified 307 singleton pregnant women with definite pre-eclampsia. We used a previously described definition of pre-eclampsia in this study (CLASP Collaborative Group, 1994
). Briefly, for pre-eclampsia to be diagnosed, persistent diastolic blood pressure of at least 90
mmHg had to develop after 20 weeks of gestation, and diastolic blood pressure had to increase by at least 25
mmHg. In addition, proteinuria had to be present, and cut-off was defined as 0.3
(semiquantitative dipstick 1+) in at least one urine sample after 20 weeks of gestation without simultaneous urinary infection.
Pre-eclampsia was classified as severe (n
=66) if diastolic blood pressure increased to at least 110 mmHg, along with proteinuria 3+ on dipstick, or at least 500
mg per 24
h. Cases with eclampsia and suspected HELLP (haemolysis elevated liver enzymes, low platelets) syndrome were regarded interchangeable with severe pre-eclampsia. Cord plasma was analysed for all 66 cases of severe pre-eclampsia.
For comparison, women without pre-eclampsia were selected from the cohort of women who gave birth at Rogaland Central Hospital, as previously described (Ødegård et al, 2000
). Among 619 control women, cord blood was available for analysis from 609. Information on baseline data was obtained at the first maternal visit at around 12 weeks of pregnancy. All infant data were compiled from hospital records.
Blood samples were collected after delivery from the placental side of the umbilical cord in syringes containing heparin, and chilled to 4°C up to 60
h before being centrifuged at 3000
r.p.m. for 15
min. Plasma was stored at −80°C until analysed.
Birth weight was standardised as the ratio between the observed and expected birth weight, the latter being adjusted for sex and gestational age at birth. We used standards of expected birth weights derived from the results of weight curves based on ultrasonographic measurements in a large Scandinavian population (Marsal et al, 1996
). Gestational age at birth was calculated from routine ultrasonographic measurements at 18 weeks of gestation. Small-for-Gestational-Age (SGA) was defined as an observed birth weight two standard deviations or more below the expected, which corresponds to a ratio lower than 0.76, or to a birth weight reduction of approximately 840 g for a term infant.
We measured oestradiol and AFP using commercially available fluoroimmunoassays (Wallac Oy, Turku, Finland). Single samples were analysed, and procedures suggested by the manufacturer were followed. Oestradiol and AFP were detected in all plasma samples. In both assays, the intra-assay coefficients of variation at high and low levels were always less than 9%.
Oestradiol and AFP had positively skewed distributions, and we used Mann–Whitney U-test for comparisons between the groups. Differences between proportions were assessed by chi-square tests. The standardised birth weight was divided into four categories: <0.76 corresponds to a strict definition of small for gestational age (SGA), and 0.76–0.89 is a broad category of relatively small infants. The category 0.90 to 1.09 includes infants with appropriate weight for their gestation, and the category >1.09 includes large babies. For each level of birth weight, we estimated values of oestradiol and AFP in the pre-eclampsia group and in controls. In multiple linear regression analyses, we assessed whether the contribution to AFP or oestradiol differed between cases of severe pre-eclampsia and controls, after adjustment for length of gestation, birth weight and offspring sex. All statistical analyses were calculated using the Statistical Package for the Social Sciences (SPSS), version 10.05 (SPSS, Inc., Chicago, IL, USA).