Materials and Methods
We performed a retrospective cohort study of all patients who delivered between January 2008 and January 2013 at the University of Cincinnati Medical Center. The study was approved by local institutional review board. All data were collected from patient charts by trained abstractors from the electronic medical record.
The study was performed prior to the American College of Obstetricians and Gynecologists Task Force on Hypertension in Pregnancy guidelines, and therefore, the diagnosis was based on a combination of elevated blood pressures greater than 140/90 mm Hg on at least 2 measures and proteinuria (0.3 g total urinary protein excreted over a 24 hour period). Factors that also were considered in diagnosing preeclampsia and determining severity as well as candidacy for expectant management were presence of neurological symptoms, epigastric pain, the HELLP syndrome, pulmonary edema, and renal compromise. All patients diagnosed with preeclampsia were considered for inclusion in our study.
We excluded multiple pregnancies, given the strong effect this may have on latency. We also excluded anomalous fetuses and stillbirths. We considered only those with a diagnosis prior to 34 weeks, the point in which expectant management is more aggressively pursued. Finally, we included only patients who had a management plan that explicitly stated expectant management was going to be attempted, excluding those who had a plan for delivery upon admission or after completion of the steroid window.
Two study groups were determined based on the presence or absence of fetal growth restriction at the time of diagnosis. Determination of the sonographic estimation of the fetal weight (EFW) was performed by registered sonographers, and the Hadlock 84 formula was used. Patients were considered to have FGR if the estimated fetal weight was less than the 10th percentile.
In addition, patients were included in the FGR study group if they had an isolated abdominal circumference less than the fifth percentile and also had abnormal umbilical artery (UA) Doppler findings, defined as a pulsatility index greater than the 95th percentile or absent/reversed end diastolic flow. This categorization was chosen to best distinguish those fetuses with poor growth secondary to placental insufficiency from constitutionally small fetuses. Doppler of the UA was performed on any patient with an EFW less than the 10th percentile or an abdominal circumference less than the fifth percentile but was not otherwise performed, according to institutional protocol.
Patients admitted with suspected preeclampsia were admitted and treated with magnesium for 24 hours while undergoing evaluation. Patients were given betamethasone to induce fetal lung maturity if they were less than 34 weeks, and rescue steroids were given if they were greater than 2 weeks from an initial course of steroids and under 32 weeks. The majority of patients remained inpatient until delivery once a diagnosis of preeclampsia was made. The decision to proceed forth with delivery was made for either worsening maternal or fetal status at the discretion of the managing obstetrical team, or women were delivered when they reached 37 weeks’ gestation.
Our primary outcome was interval to delivery between diagnosis of preeclampsia and delivery (days), measured as a continuous variable. Delivery timing is dictated by departmental protocol but is ultimately at the discretion of the managing physician. Common indications for delivery included attainment of 37 weeks, inability to control blood pressures, persistent neurological symptoms, development of the HELLP syndrome, worsening fetal status, onset of labor, or rupture of membranes.
Secondary outcomes included development of severe morbidities of preeclampsia including the development of severe hypertension, the need for intravenous antihypertensive medication, and the development of a composite of severe morbidities of preeclampsia that included neurological symptoms, HELLP syndrome, renal insufficiency, pulmonary edema, or eclampsia. In addition, we analyzed the indication for delivery, rate of attempted induction, and rate of cesarean delivery when FGR was present vs absent. Neonatal outcomes evaluated included the rate of neonatal intensive care unit (NICU) admission and neonatal length of stay (days).
A multivariate logistic regression was performed to estimate the crude odds ratio and the adjusted ORs with 95% confidence intervals after inclusion of pertinent covariates assessing the following categorical outcomes: interval to delivery of less than 1 week, NICU admission, prolonged neonatal hospitalization over 4 days, and neonatal mortality. A stepwise backward regression analysis was used to evaluate the impact of multiple potential confounding variables. After backward elimination maternal race, tobacco use, chronic hypertension, gestational age at delivery, severe preeclampsia, and fetal sex were included in the final regression model.
Study data were collected and managed using Redcap electronic data capture tools posted at the University of Cincinnati. Categorical variables were compared using a χ 2 analysis or a Fisher exact test when any cell in a 2 × 2 table contained fewer than 10 subjects. Continuous variables were compared using a Student t test when normally distributed, a Wilcoxon rank sum when nonnormally distributed. Data normally distributed are presented as mean ± SD, nonnormally distributed data are presented at median (25th and 75th percentile). A two-sided value of P < .05 or 95% confidence interval not inclusive of the null value 1.0 was considered statistically significant. All data analyses were performed using NCSS 8 statistical software (release 8; NCSS LLC, Kaysville, UT).
Results
Of 851 patients with preeclampsia, 199 met inclusion criteria and were included in the analyses. The most common reasons for exclusion were a diagnosis beyond 34 weeks (n = 554), multiple pregnancy (n = 2), major fetal anomaly (n = 8), stillbirth (n = 6), and plan for immediate delivery or delivery once the patient received betamethasone to induce fetal lung maturity (n = 82). Of the 199 patients included, 60 of the eligible patients (31%) had FGR and 139 (69%) had an ultrasound demonstrating a normal growth profile ( Figure 1 ).
In the FGR study group, we included 16 patients with an abdominal circumference less than the third percentile and 2 with an abdominal circumference at the fourth percentile, all of which had an abnormal UA Doppler. Demographics, preeclampsia characteristics, delivery outcomes, and neonatal outcomes for eligible patients are described in Table 1 . Maternal demographics in regard to age, race, tobacco use, and history of chronic hypertension were similar between the 2 groups but differed in parity and body mass index (BMI). The group of patients without FGR on average had a higher BMI than those with FGR, with an average BMI of 33.1% and 30.1%, respectively ( P = .04).
| Variables | No FGR (n = 139) | FGR (n = 60) | P value |
|---|---|---|---|
| Demographic data | |||
| Maternal age, mean (SD) | 28.1 ± 6.5 | 27.0 ± 7.7 | .34 |
| Parity, median (IQR) | 1 (0–3) | (0–0) | < .001 |
| Race, n, % | .53 | ||
| White | 54 (64.3) | 30 (35.7) | |
| Black | 75 (64.3) | 27 (26.5) | |
| Hispanic | 4 (66.7) | 2 (33.3) | |
| Other | 3 (75.0) | 1 (25.0) | |
| Tobacco use, n, % | 27 (19.9) | 16 (26.7) | .29 |
| Body mass index, kg/m 2 , mean (SD) | 33.1 ± 10.0 | 30.1 ± 7.9 | .04 |
| Chronic hypertension, n, % | 40 (29.4) | 11 (18.3) | .10 |
| Preeclampsia characteristics | |||
| Gestational age at diagnosis, wks, median (IQR) | 32 (29–33) | 28 (25–31) | < .001 |
| Severe preeclampsia, n, % | 116 (85.9) | 55 (91.7) | .26 |
| Composite of major complication, n, % | 67 (51.1) | 24 (42.9) | .30 |
| Need for intravenous antihypertensive medication, n, % | 64 (47.1) | 28 (46.7) | .96 |
| Delivery outcomes | |||
| Gestational age at delivery, wks, median (IQR) | 33 (30–34) | 29 (26–32) | < .001 |
| Interval to delivery, d, median (IQR) | 5 (2–12) | 3 (1–6) | < .001 |
| Indication for delivery, n, % | |||
| Reached scheduled delivery date | 21 (15.7) | 3 (5.0) | .03 |
| Development of severe preeclampsia | 80 (58.8) | 29 (48.3) | .17 |
| Neurological symptoms | 51 (37.5) | 10 (16.7) | .003 |
| HELLP syndrome | 21 (15.4) | 8 (13.3) | .70 |
| Nonreassuring fetal status | 23 (16.9) | 35 (58.3) | < .001 |
| Induction, n, % | 69 (50.7) | 19 (31.7) | .01 |
| Cesarean delivery, n, % | 86 (63.2) | 48 (80.0) | .02 |
| Neonatal outcomes | |||
| Birth weight, g, mean (SD) | 1810 ± 736 | 992 ± 437 | < .001 |
| Small for gestational age, n, % | 26 (19.1) | 37 (61.7) | < .001 |
| Neonatal mortality, n, % | 6 (4.4) | 8 (13.3) | .02 |
| Neonatal intensive care unit admission, n, % | 106 (78.5) | 58 (96.7) | .001 |
| Neonatal length of stay, d, median (IQR) | 14 (10–18) | 44 (27–64) | < .001 |
Additionally, women with fetal growth restriction were diagnosed with preeclampsia at earlier gestational ages, median (interquartile range [IQR]), 28 (25–31) weeks vs those without fetal growth restriction 32 (29–33) weeks ( P < .001). No differences in presence of severe disease, major complications, or need for intravenous antihypertensive medications were seen between the 2 groups.
Our primary outcome, latency until delivery, was significantly shorter in women with FGR, median (IQR) of 3 (1–6) days vs those with normal EFW, 5 (2–12) days ( P < .001; Figure 2 ). Fewer women with FGR were managed until their scheduled date of delivery, 3 (5.0%) vs those without FGR, 21 (15.7%) ( P = .03). In the presence of FGR, delivery was more likely for nonreassuring fetal status, 35 (58.3%) vs 23 (16.9%). There were no differences in deliveries secondary to development of severe preeclampsia or the HELLP syndrome.
When the decision for delivery was made, patients in the FGR group were less likely to be induced (31.7% vs 50.7%, P = .01) and more likely to have a cesarean delivery (80.0% vs 63.2%, P = .02). After delivery, growth-restricted fetuses had significantly lower birthweight (992 g ± 437 vs 1810 g ± 736, P < .001), a higher percentage of small-for-gestational-age neonates (61.7% vs 19.1%, P < .001), higher neonatal mortality (13.3% vs 4.4%, P < .02), higher rates of NICU admissions (96.7% vs 78.5%, P < .001), and longer median length of stay in the NICU, median (IQR) of 44 (27–64) days vs 14 (10–18) days ( P < .001).
A post hoc logistic regression analysis controlling for confounding variables including race, tobacco use, chronic hypertension, gestational age at delivery, severe preeclampsia, and fetal sex was performed. The association between fetal growth restriction and latency less than 7 days remained significant, adjusted OR 1.66 (95% CI 1.12–2.47). However, the associations with adverse neonatal outcomes were no longer significant ( Table 2 ).
