Study population

Data were obtained from the Netherlands Perinatal Registry (PRN registry) between January 2000 and December 2006. Since 2000 all gestation/delivery records (National Delivery Record [LVR]), both home deliveries (LVR-1 registry) and hospital deliveries (LVR-2 registry), are combined with neonatal admission records (National Neonatal Register [LNR]) into a national perinatal register (PRN). Méray et al and Tromp et al have extensively described the technical approach and subsequent validation of the probabilistic linkage of these 3 anonymous population-based registries of the midwives, obstetricians, and neonatalogists.

The LVR-1 and LVR-2 register have a 96% national coverage on all births (approximately 180,000 deliveries per year at more than 16 completed weeks of gestation in The Netherlands). The LNR registry has a 68% coverage of all hospitals in The Netherlands of whom the 10 perinatal centers have a coverage of 100% and the other hospitals have 58%.

For the present study, we included all women who delivered between 34 ⁺⁰ and 36 ⁺⁶ weeks of gestation. Exclusion criteria were chronic hypertension, multiple pregnancies, noncephalic presentation, congenital malformations, mothers diagnosed with AIDS, diabetes, or drug use (drugs, not cannabis), and more than 24 hours of rupture of membranes (in The Netherlands an expectant monitoring management is often followed for this latter group in the late preterm period).

These exclusion criteria were selected because these specific conditions themselves are related to neonatal morbidities. If neonatal follow-up was not available, data were excluded. From the available data, subjects whom the mothers were diagnosed with gestational hypertension (GH group) or preeclampsia (PE group) were selected in. The other patients were the normotensive control group (preterm delivery without a reason).

GH was defined as de novo hypertension, occurring after 20 weeks’ gestational age. Hypertension is defined as a diastolic blood pressure of 90 mm Hg or greater or a systolic blood pressure of 140 mm Hg or greater. PE was defined as de novo hypertension after 20 gestational weeks and proteinuria (≥300 mg/day or a spot urine protein/creatinine ratio of ≥30 mg/mmol). Patients could be included based only on their diastolic blood pressure and on the amount of proteinuria. The systolic blood pressure is not an item in this national database. This means that we might have missed a small group of patients that did not reach a diastolic blood pressure of 90 mmHg or greater during their entire pregnancy. According to the national guidelines, calculation of gestational age was based on the first day of the last menstrual period and verified by a routinely performed first-trimester ultrasound in all patients.

Maternal baseline characteristics recorded were maternal age, ethnicity, and parity. Variables recorded were gestational age at delivery, birthweight, small for gestational age (SGA; defined as birthweight below the 10th percentile adjusted for gestational age based on a local reference population), and a 5 minute Apgar score less than 7.0.

Poor neonatal outcome included admission to the neonatal intensive care unit (NICU), metabolic and gastrointestinal morbidity subdivided in hypoglycemia (defined as glucose serum or plasma level <2.5 mmol/L), hyperbilirubinemia (indirect hyperbilirubinemia defined as above the phototherapy level, direct hyperbilirubinemia defined as >10% of the total serum bilirubine), or any stage of necrotizing enterocolitis (NEC).

Respiratory problems were subdivided in the need for oxygen therapy more than 24 hours, any grade of infant respiratory distress syndrome (RDS; based on radiographic thorax findings according to Giedeon grade I-IV), bronchopulmonary disease (BPD; defined as more than 28 days of oxygen therapy or oxygen therapy after 36 postmenstrual weeks), transient tachypnoe of the newborn (TTN; based on the typical clinical picture [respiratory support that is rapidly weaned within the first 24 hours of life]), and chest radiograph (high lung volume).

Neurologic morbidity was subdivided into intracranial hemorrhage including intraventricular hemorrhage (defined according to Papile [grade I: hemorrhage subependymal–germinal matrix; grade II: intraventricular hemorrhage with normal ventricle size; grade III: intraventricular hemorrhage with ventricular dilation; and grade IV: parenchymal hemorrhage]); cerebral ischemia subdivided in any grade of periventricular leucomalacia (PVL; based on ultrasound images or magnetic resonance imaging) or ischemia other than PVL any stage of hypoxic ischemic encephalopathy (HIE); or convulsions. For this study only the first admission data after birth were used.

Analysis

The study population was categorized in 3 groups: GH, PE, and the normotensive control group. In later analysis, the 3 study groups were further subdivided by 3 groups of gestational age (GA) per week: delivered at 34 ⁺⁰ to 34 ⁺⁶ GA; 35 ⁺⁰ to 35 ⁺⁶ and 36 ⁺⁰ to 36 ⁺⁶ GA.

Data are presented as n (percentage) and in mean (±SD), as appropriate. We used χ ² to test for categoric data and Student t test or 1-way analysis of variance (in case of more than 2 variables) for continuous variables. Two-sided probability of less than .05 was considered statistically significant. Data are expressed in odds ratio (ORs) and 95% confidence intervals (CIs).

A multivariable logistic regression analysis was performed to examine respiratory morbidity outcomes across maternal subgroups (GH, PE, and control) controlling for gestational age in 3 weeks groups (34, 35, and 36 weeks) and for onset or mode of delivery (in 7 groups). We checked for 2 possible interaction effects; first, we tested for interaction between the study group (GH, PE, and control) and gestational age in weeks and second, between the study group and mode of delivery.

Reference groups were the groups with the lowest risk. In addition, we adjusted in the multivariable analysis for parity, maternal ethnicity, and SGA. Data of the multivariable analysis are expressed in ORs with 95% CIs. All analyses were performed using SAS software (version 9.2; SAS Institute, Cary, NC).