Research www.AJOG .

org

OBSTETRICS
Incidence of preeclampsia: risk factors and outcomes
associated with early- versus late-onset disease
Sarka Lisonkova, MD, PhD; K. S. Joseph, MD, PhD

OBJECTIVE: The population-based incidence of early-onset (<34 unmarried status, and male fetus. African-American race, chronic
weeks) and late-onset preeclampsia (
34 weeks) has not been hypertension, and congenital anomalies were more strongly associ-
adequately studied. We examined the gestational ageespecific inci- ated with early-onset preeclampsia, whereas younger maternal age,
dence of preeclampsia onset and identified the associated risk factors nulliparity, and diabetes mellitus were more strongly associated with
and birth outcomes. late-onset disease. Early- but not late-onset preeclampsia conferred a
high risk of fetal death (AOR, 5.8; 95% confidence interval [CI],
STUDY DESIGN: All singleton deliveries in Washington State, 2003-
4.0e8.3 vs AOR, 1.3; 95% CI, 0.8e2.0, respectively). The AOR for
2008 (n ¼ 456,668), were included, and preeclampsia onset was
perinatal death/severe neonatal morbidity was 16.4 (95% CI,
determined from hospital records linked to birth certificates. Cox and
14.5e18.6) in early-onset and 2.0 (95% CI, 1.8e2.3) in late-onset
logistic regression models were used to obtain adjusted hazard ratios
preeclampsia.
and odds ratios (AORs) for risk factors and birth outcomes,
respectively. CONCLUSION: Early- and late-onset preeclampsia shares some
etiological features, differ with regard to several risk factors, and
RESULTS: The overall preeclampsia rate was 3.1% and the incidence
lead to different outcomes. The 2 preeclampsia types should be
increased sharply with gestation; early- and late-onset preeclampsia
treated as distinct entities from an etiological and prognostic
rates were 0.38% and 2.72%, respectively. Among women with early-
standpoint.
onset preeclampsia, 12% delivered at a gestation of 34 weeks or
longer. Risk/protective factors common to both diseases included Key words: birth outcomes, early onset, eclampsia, fetuses at risk,
older maternal age, Hispanic and Native-American race, smoking, late onset, preeclampsia

Cite this article as: Lisonkova S, Joseph KS. Incidence of preeclampsia: risk factors and outcomes associated with early- versus late-onset disease. Am J Obstet Gynecol
2013;209:xx-xx.

P reeclampsia is characterized by
elevated blood pressure and pro-
teinuria or involvement of other organs
failure.4-10 Infants of mothers with pre-
eclampsia are at approximately 2-fold
higher risk of neonatal death11 and at
cerebral palsy, and neonatal mortality
compared with infants born very pre-
term for other reasons.2,15-17
in an exaggerated systemic inflama- increased risk of neonatal morbidity Preeclampsia is a heterogeneous dis-
tion.1-3 In industrialized countries, pre- including low Apgar scores, seizures, order with 2 distinct subtypes that have
eclampsia complicates approximately neonatal encephalopathy, and neonatal been described based on the timing of
3-5% of pregnancies and represents one intensive care admission.11-14 However, disease onset: early-onset preeclampsia
of the most common causes of maternal some previous studies have shown that occurring before or at 33 weeks’ gesta-
mortality and severe maternal morbidity infants born at very preterm gestation tion and late-onset preeclampsia that
including eclampsia, placental abrup- because of preeclampsia have a reduced occurs at 34 weeks’ gestation or later.18-20
tion, pulmonary edema, and acute renal risk of retinopathy of prematurity, Early-onset disease, in particular, confers
a high risk of life-threatening maternal
complications and fetal demise, and
From the Department of Obstetrics and Gynecology (both authors) and the School of Population and early delivery is the only treatment.20-22
Public Health (Dr Joseph), University of British Columbia, and the Children’s and Women’s Hospital
and Health Centre of British Columbia (both authors), Vancouver, BC, Canada.
Although previous publications22-25
have described early-onset preeclamp-
Received May 1, 2013; revised July 23, 2013; accepted Aug. 16, 2013.
sia and associated neonatal outcomes,
S.L. is supported by a grant on severe maternal morbidity (MAH-114445) from the Canadian
Institutes of Health Research. K.S.J. is supported by a chair in maternal, fetal, and infant health
most reports have been based on small,
services research from the Canadian Institutes of Health Research (grant APR-126338). hospital-based studies and clinical
The authors report no conflict of interest. trials focusing on obstetric manage-
Presented at the 26th annual meeting of the Society for Pediatric and Perinatal Epidemiologic
ment.22,23,25 The gestational age-specific
Research, Boston, MA, June 17-18, 2013. incidence of preeclampsia, based on the
Reprints not available from the authors. onset of symptoms and not gestational
0002-9378/$36.00  ª 2013 Mosby, Inc. All rights reserved.  http://dx.doi.org/10.1016/j.ajog.2013.08.019
age at delivery, has not been documented
to date at the population level. We

MONTH 2013 American Journal of Obstetrics & Gynecology 1.e1

Research Obstetrics www.AJOG.org

therefore carried out a population-based database) and maternal hospitalization the 10th percentile of the sex- and gesta-
study to describe the gestational age- data (CHARS database) were excluded tional ageespecific birthweight reference
specific incidence of preeclampsia onset (5.7%, n ¼ 27,443). for the United States,26 whereas large-for-
among women with singleton pregnan- Maternal characteristics and clinical gestational age infants were those weigh-
cies and to examine risk factors and birth risk factors examined for potential as- ing over the 90th percentile. We used the
outcomes associated with early-onset sociation with preeclampsia included clinical estimate of gestation provided in
and late-onset disease. maternal age (younger than 20 and the data source because this is more ac-
35 years old or older vs 20-34 years); curate than gestational age estimated by
M ATERIALS AND M ETHODS parity (number of previous live births, the last menstrual period.27,28
We included all singleton deliveries in none vs 1 or more); marital status (single/ Gestational ageespecific rates of pre-
Washington State during the period from widowed/separated vs married/common eclampsia were calculated using ongoing
2003 to 2008, utilizing information from law); education (less than high school pregnancies as the denominator. c2 tests
2 large population databases: (1) the vs high school education or greater); were used to assess the differences be-
Comprehensive Discharge Abstract race (non-Hispanic white vs Hispanic, tween rates of early-onset and late-onset
Database (CHARS) which included all African-American, Native-American, and preeclampsia across maternal and clin-
hospitalizations in Washington State, and other); smoking during pregnancy (yes/ ical characteristics. The Cox regression
(2) the Birth Events Record Database no); infertility treatment (yes/no); dia- model, with preeclampsia onset as the
(BERD), which included birth records of betes mellitus (yes/no); chronic hyper- outcome and gestational age as the time
all live born infants and fetal deaths. tension prior to pregnancy (yes/no); axis, was used to estimate adjusted hazard
Women with a diagnosis of preeclampsia infant’s sex (male/female); and congenital ratios (AHRs) and 95% confidence in-
or eclampsia (henceforth referred to as anomalies (yes/no). tervals (CIs). This enabled us to create the
preeclampsia), including preeclampsia Fetal death was defined as in utero or appropriate risk sets, with censoring of
superimposed on chronic hypertension intrapartum death of a fetus delivered at subjects who developed preeclampsia or
were identified from the CHARS data- 20 weeks’ gestation or later, neonatal who delivered at any particular gestation.
base (International Classification of Dis- death was defined as a death of an infant When the proportional hazards as-
eases, ninth revision [ICD-9] diagnostic within 28 days after birth, and perinatal sumption was not satisfied, we examined
codes 642.4, 642.5, 642.6, and 642.7). death included fetal or neonatal death. the interaction term between the risk
Hospitalization records with a diag- Using birth hospitalization data for in- factor and gestational age at diagnosis
nosis of preeclampsia were linked to birth fants (obtained from the linked infants’ categorized as less than 34 weeks and 34
records to obtain information about birth and hospitalization records), the weeks or longer and obtained AHRs for
gestational age at delivery, maternal following adverse birth outcomes were both early-onset and late-onset pre-
characteristics, clinical risk factors, and identified based on ICD-9 codes: bron- eclampsia separately. The Wald statistic
birth outcomes. The number of weeks chopulmonary dysplasia (BPD; code was used to assess statistical significance
between the hospitalization when the 770.7), intraventricular hemorrhage of the interaction terms. Shoenfeld re-
preeclampsia diagnosis was made and (IVH) grade III and IV (codes 772.13 siduals were used to evaluate the pro-
birth hospitalization was calculated based and 772.14), periventricular leukomala- portional hazards assumption of the
on the CHARS and BERD record linkage. cia (PVL; code 779.7), retinopathy of final model.29
Preeclampsia occurring at less than prematurity (ROP; code 362.2), necro- Birth outcomes including fetal death,
34 weeks of gestation was identified as tizing enterocolitis (NEC; code 777.5), perinatal death, and severe neonatal
early-onset disease, whereas preeclamp- and neonatal sepsis (code 771.81). Other morbidity were analyzed using the
sia that occurred at 34 weeks or later was neonatal outcomes were identified from fetuses-at-risk approach. Under this
labeled late-onset disease, irrespective of birth records, namely, neonatal seizures, formulation, all fetuses at a specific
the gestational week at delivery. Infant Apgar score at 5 minutes, and neonatal gestation were considered at risk for
birth records were also linked to CHARS intensive care unit (NICU) admission. adverse outcomes at that gestation.
(infant) hospitalization records to iden- Severe neonatal morbidity included Thus, for example, all fetuses at 28 weeks
tify cases of severe neonatal morbidity any of the following: a 5-minute Apgar with early-onset preeclampsia were
(see the following text). score of 3 or less, neonatal seizures, BPD, considered to be at risk of live birth,
There were 484,111 women who were IVH grade III or IV, PVL, ROP, NEC, and stillbirth, neonatal death, or severe
residents of Washington State and who neonatal sepsis. The composite out- neonatal morbidity at 28 weeks, irre-
delivered a singleton stillbirth or live come of neonatal mortality/morbidity spective of whether they actually deliv-
birth in a Washington State hospital be- included both neonatal death and severe ered at 28 weeks or at a subsequent
tween 2003 and 2008. Women with a neonatal morbidity, whereas perinatal gestational week.30,31 Two fetuses-at-
missing estimate of gestation or gesta- mortality/morbidity included perinatal riskebased logistic regression models
tional age at delivery less than 20 weeks death and severe neonatal morbidity. were used to estimate causal associations
and women without a linkage between Small-for-gestational-age (SGA) infants between early-onset and late-onset pre-
the birth/fetal death certificate (BERD were defined as those weighing less than eclampsia and birth outcomes.

1.e2 American Journal of Obstetrics & Gynecology MONTH 2013

www.AJOG.org Obstetrics Research

TABLE 1
Maternal characteristics and clinical factors associated with early- and late-onset preeclampsia, singleton
deliveries, Washington State, 2003-2008
Ongoing Ongoing
Maternal pregnancies pregnancies
characteristics/clinical at 20 weeks Early-onset preeclampsia at 34 weeksa Late-onset preeclampsia
factors (n [ 456,668) (n [ 1752) Rate per 1000 (n [ 447,822) (n [ 12,449) Rate per 1000
Age, y
<20 39,584 160 4.0 38,561 1627 42.2
20-34 347,105 1256 3.6 340,850 9001 26.4

35 69,979 336 4.8 68,411 1821 26.6
Race
Non-Hispanic white 323,552 1154 3.6 317,775 8914 28.1
African-American 20,045 158 7.9 19,356 690 35.6
Hispanic 56,615 210 3.7 55,533 1555 28.0
Native-American 10,625 47 4.4 10,313 372 36.1
Other 44,032 164 3.7 43,192 851 19.7
Education
Less than high school 89,238 324 3.6 87,255 2465 28.3
High school or more 359,160 1357 3.8 352,795 9752 27.6
Smoking during pregnancy
Yes 46,936 162 3.5 45,737 1131 24.7
No 403,471 1527 3.8 396,224 11,185 28.2
Marital status
Unmarried 149,369 657 4.4 145,677 4740 32.5
Married 305,666 1076 3.5 300,669 7654 25.5
Number of prior live births
0 186,980 964 5.2 182,906 8131 44.5

1 258,135 673 2.6 254,003 4043 15.9
Diabetes mellitus
Yes 25,815 207 8.0 25,110 1353 53.9
No 430,853 1545 3.6 422,712 11,096 26.2
Chronic hypertension
Yes 5560 237 42.6 5163 699 135.4
No 451,108 1515 3.4 442,659 11,750 26.5
Infertility treatment
Yes 3455 26 7.5 3339 133 39.8
No 453,213 1726 3.8 444,483 12,316 27.7
Infant sex
Male 234,224 914 3.9 229,314 6647 29.0
Female 222,441 838 3.8 218,508 5802 26.6
Lisonkova. Early- vs late-onset preeclampsia. Am J Obstet Gynecol 2013. (continued)

MONTH 2013 American Journal of Obstetrics & Gynecology 1.e3

Research Obstetrics www.AJOG.org

TABLE 1
Maternal characteristics and clinical factors associated with early- and late-onset preeclampsia, singleton
deliveries, Washington State, 2003-2008 (continued)
Ongoing Ongoing
Maternal pregnancies pregnancies
characteristics/clinical at 20 weeks Early-onset preeclampsia at 34 weeksa Late-onset preeclampsia
factors (n [ 456,668) (n [ 1752) Rate per 1000 (n [ 447,822) (n [ 12,449) Rate per 1000
Congenital anomalies
Yes 2249 23 10.2 1949 66 33.9
No 454,419 1729 3.8 445,873 12,383 27.8
The number of pregnancies does not add up to the total in some categories because of missing values (missing values exceeding 3% were 4.1% for education, 3.6% for smoking, and 6.6% for
number of prior live births in the early-onset preeclampsia group).
a
Ongoing pregnancies without early-onset preeclampsia.
Lisonkova. Early- vs late-onset preeclampsia. Am J Obstet Gynecol 2013.

These models were constructed using denominator and provided a predictive and 13.9% in the early-onset, late-onset,
ongoing pregnancies (ie, fetuses at risk) (noncausal) model comparing the odds and no preeclampsia groups, respec-
as the denominator32,33; all ongoing of adverse neonatal outcomes among tively) were imputed using multiple
pregnancies at 20 weeks’ gestation were mothers with and without preeclampsia, imputation procedures (proc MI, SAS
included in models examining birth conditional on delivery of a live-born software, version 9.2; SAS Institute Inc,
outcomes following early-onset pre- infant at a specific gestational age.32,33 Cary, NC). In addition, we adjusted for
eclampsia, whereas all ongoing preg- We further compared birth outcomes time period (year of delivery) to address
nancies at 34þ0 weeks’ gestation (among between mothers with early-onset pre- the potential effects of changes in ob-
women without early-onset preeclamp- eclampsia who delivered at 34 weeks or stetric and neonatal practices.
sia) were included in the denominator longer, and mothers with late-onset All analyses were performed on pub-
for birth outcomes following late-onset preeclampsia (who, by definition, all licly accessible de-identified data. An
preeclampsia. delivered at
34 weeks). exemption from ethics approval was
In addition, we compared neonatal We performed sensitivity analyses granted by the Department of Social and
outcomes between infants born to examining the effect of obesity on the risk Health Services, State of Washington.
mothers with and without early-onset or of early-onset and late-onset pre- Analyses were carried out using SAS
late-onset preeclampsia, adjusting for eclampsia and its association with birth software, version 9.2 (SAS Institute Inc.,
gestational age at delivery (traditional outcomes. Obesity was defined as a body Cary, NC). A 2-tailed P < .05 was
analysis). This analysis used live births at mass index (BMI) greater than 30 kg/m2. considered significant.
a particular gestational age as the Missing values for BMI (27.7%, 14.6%,
R ESULTS
The study included 456,668 women who
FIGURE delivered a singleton live birth or still-
Gestational ageespecific incidence of preeclampsia, singleton birth between 2003 and 2008. The rate of
deliveries, Washington State, 2003-2008 preeclampsia was 3.11 per 100 singleton
deliveries (14,201 of 456,668), and the
rate of eclampsia was 4.12 per 10,000
singleton deliveries (188 of 456,668).
The frequency of early-onset pre-
eclampsia was 0.38 per 100 deliveries,
and the frequency of late-onset pre-
eclampsia was 2.72 per 100 deliveries
(Table 1). The gestational ageespecific
incidence of preeclampsia increased with
pregnancy duration, from 0.01 per 1000
ongoing pregnancies at 20 weeks’ gesta-
tion to 9.62 per 1000 ongoing pregnan-
cies at 40 weeks’ gestation (Figure).
Lisonkova. Early- vs late-onset preeclampsia. Am J Obstet Gynecol 2013. Women who were at the extremes of
maternal age (younger than 20 or 35 years

1.e4 American Journal of Obstetrics & Gynecology MONTH 2013

www.AJOG.org Obstetrics Research

TABLE 2
Crude and adjusted hazard ratios for early- and late-onset preeclampsia, singleton deliveries, Washington State,
2003-2008
Preeclampsia, unadjusted analysis Preeclampsia, adjusted analysis
Early onset Late onset Early onset Late onset
Demographic/clinical factors HR 95% CI HR 95% CI AHR 95% CI AHR 95% CI
Age, y
<20a 1.12 0.95e1.32 1.56 1.48e1.65 0.84 0.70e1.00 1.07 1.01e1.14
20-34 Ref Ref Ref Ref

35 a
1.33 1.18e1.50 1.04 0.99e1.10 1.15 1.10e1.21 1.15 1.10e1.21
Race
Non-Hispanic white Ref Ref Ref Ref
a
African-American 2.21 1.88e2.61 1.26 1.16e1.36 1.75 1.45e2.12 1.20 1.11e1.31
Hispanic 1.01 0.96e1.06 1.01 0.96e1.06 1.07 1.01e1.13 1.07 1.01e1.13
Native-American 1.27 1.54e1.59 1.27 1.54e1.59 1.36 1.22e1.51 1.36 1.22e1.51
a
Other 1.03 0.88e1.22 0.73 0.68e0.79 0.98 0.82e1.16 0.68 0.63e0.73
Maternal education less 0.98 0.94e1.03 0.98 0.94e1.03 0.98 0.93e1.03 0.98 0.93e1.03
than high school
Smoking during pregnancy 0.91 0.86e0.96 0.91 0.86e0.96 0.87 0.82e0.93 0.87 0.82e0.93
Unmarried 1.27 1.23e1.32 1.27 1.23e1.32 1.14 1.10e1.19 1.14 1.10e1.19
a
No prior live births 1.98 1.80e2.19 2.59 2.49e2.69 2.13 1.92e2.37 2.67 2.57e2.78
a
Diabetes mellitus 2.45 2.32e2.58 2.45 2.32e2.58 1.87 1.60e2.18 2.46 2.32e2.61
a
Chronic hypertension 13.06 11.39e14.97 6.72 6.22e7.25 11.72 10.11e13.59 5.83 5.39e6.32
Infertility treatment 1.60 1.37e1.87 1.60 1.37e1.87 1.17 0.99e1.37 1.17 0.99e1.37
Infant sex (male) 1.10 1.07e1.14 1.10 1.07e1.14 1.10 1.06e1.14 1.10 1.06e1.14
a
Congenital anomalies 2.91 1.93e4.39 1.50 1.18e1.92 2.59 1.66e4.02 1.49 1.16e1.91
AHR, adjusted hazard ratio; CI, confidence interval; HR, hazard ratio; Ref, referent.
a
The hazard ratios differed significantly between early- and late-onset preeclampsia. The Wald statistic was used to assess statistical significance of the interaction term between the risk factor and
the gestational age at preeclampsia onset.
Lisonkova. Early- vs late-onset preeclampsia. Am J Obstet Gynecol 2013.

old or older), African-American, un- than non-Hispanic white, Hispanic, with early- vs late-onset preeclampsia.
married, and nulliparous had higher rates African-American and Native-American; African-American race, chronic hyper-
of early-onset preeclampsia (Table 1). Table 2) had lower rates of late-onset tension, and congenital anomalies were
Similarly, women who had diabetes disease. more strongly associated with early-onset
mellitus or chronic hypertension, used Several risk factors were associated with disease, whereas young maternal age
infertility treatment to conceive, and had preeclampsia, without a significant dif- (younger than 20 vs 20-35 years), other
an infant with a congenital anomaly also ference in adjusted hazard ratios for early- race (not including African-American,
had higher rates of early-onset pre- and late-onset disease (Table 2). These Hispanic or Native-American vs non-
eclampsia. Women who were very young included smoking during pregnancy Hispanic white), nulliparity, and dia-
(younger than 20 years of age), unmar- (AHR, 0.87; 95% CI, 0.82e0.93 for both betes mellitus were more strongly
ried, nulliparous, had diabetes mellitus early- and late-onset disease), unmarried associated with late-onset disease. Other
or chronic hypertension, used infertility status (AHR, 1.14; 95% CI, 1.10e1.19), race had a protective effect on late-onset
treatment to conceive, and were pregnant older maternal age (AHR, 1.15; 95% CI, disease compared with non-Hispanic
with a male fetus also had higher rates of 1.10e1.21), and infant’s sex (AHR for white race (AHR, 0.68; 95% CI,
late-onset preeclampsia. On the other male sex, 1.10; 95% CI, 1.06e1.14). 0.63e0.73).
hand women who smoked or belonged In contrast, several risk factors dif- Women with chronic hypertension
to the “other” race category (ie, other fered significantly in their association had the highest risk for preeclampsia,

MONTH 2013 American Journal of Obstetrics & Gynecology 1.e5

Research Obstetrics www.AJOG.org

TABLE 3
Birth outcomes associated with early- and late-onset preeclampsia, singleton deliveries, Washington State,
2003-2008
Early-onset No early-onset Late-onset No late-onset
preeclampsia preeclampsia preeclampsia preeclampsia
Rate per Rate per Rate per Rate per
Birth outcomes n 1000 FAR n 1000 FAR n 1000 FAR n 1000 FAR
Ongoing pregnancies 1752 454,916 12,449 435,373
Gestational age at delivery, wks
20-33 1539 878.4 7094 15.6 n/a n/a
34-36 128 73.1 26,062 57.3 2911 233.8 23,151 53.2
37-43 85 48.5 421,760 927.1 9538 766.2 412,222 946.8
Birthweight, g
<1500 867 494.9 3208 7.1 42 3.4 149 0.3
1500-2499 641 365.9 18,020 39.6 2020 162.3 12,972 29.8
2500-4499 183 104.5 424,395 932.9 10,162 816.3 413,747 950.3

4500 0 0.0 7733 17.0 166 13.3 7567 17.4
SGA (<10th percentile) 563 321.3 29,439 64.7 2007 161.2 26,800 61.6
LGA (>90th percentile) 23 13.1 52,702 115.8 1171 94.1 51,372 118.0
Apgar score at 5 min 3 78 44.5 2064 4.5 99 8.0 1328 3.1
Neonatal seizures 5 2.9 186 0.4 10 0.8 159 0.4
Neonatal sepsis 200 114.2 3468 7.6 170 13.7 2409 5.5
NICU admission 1202 686.1 22,434 49.3 1658 133.2 16,843 38.7
Fetal death 58 33.1 1648 3.6 28 2.2 631 1.4
Neonatal death 50 28.5 1071 2.4 15 1.2 402 0.9
a
Severe neonatal morbidity 367 209.5 5748 12.6 274 22.0 3758 8.6
Neonatal death/severe morbidity 381 217.5 6198 13.6 283 22.7 4020 9.2
Perinatal death 108 61.6 2719 6.0 43 3.5 1033 2.4
Perinatal death/severe morbidity 439 250.6 7846 17.2 311 25.0 4651 10.7
All differences were statistically significant (P < .001). For early-onset preeclampsia comparisons, all ongoing pregnancies at 20 weeks of gestation were included in the denominator; for late-onset
preeclampsia comparisons, all ongoing pregnancies at 34 weeks of gestation were included in the denominator.
FAR, fetuses at risk; NICU, neonatal intensive care unit.
a
Includes any of the following: a 5 minute Apgar score of 3 or less, neonatal seizures, neonatal sepsis, bronchopulmonary dysplasia, necrotizing enterocolitis, intraventricular hemorrhage grades 3
and 4, periventricular leukomalacia, and retinopathy of prematurity.
Lisonkova. Early- vs late-onset preeclampsia. Am J Obstet Gynecol 2013.

with more than a 10-fold higher rate without early-onset disease (Table 3). The rate of adverse birth outcomes
of early-onset disease (AHR, 11.7; 95% Among women with early-onset pre- remained severalfold higher among
CI, 10.1e13.6) and an approximately eclampsia, approximately 12%, deliv- mothers with early-onset preeclampsia
5-fold higher rate of late-onset disease ered at 34 weeks’ gestation or later, and as compared with mothers without early-
(AHR, 5.8; 95% CI, 5.4e6.3) as almost one half of births (49.5%) were onset disease after adjustment for risk
compared with women without chronic very low birthweight (<1500 g). With factors (Table 4). For example, the rate of
hypertension. the exception of neonatal death rates, the fetal death was approximately 6 times
The rates of all adverse birth rates of all adverse birth outcomes were higher (AOR, 5.8; 95% CI, 4.0e8.3), and
outcomes, except for large for gesta- significantly higher among mothers the rate of perinatal death or serious
tional age (LGA), were significantly with late-onset preeclampsia compar- neonatal morbidity was 16 times higher
higher among women with early-onset ed with those without preeclampsia (AOR, 16.4; 95% CI, 14.5e18.6) among
preeclampsia compared with women (Table 3). women with early-onset disease.

1.e6 American Journal of Obstetrics & Gynecology MONTH 2013

www.AJOG.org Obstetrics Research

TABLE 4
Crude and AORs for birth outcomes following early- and late-onset preeclampsia, singleton deliveries,
Washington State, 2003-2008
Early-onset preeclampsia Late-onset preeclampsia
Birth outcomes OR 95% CI AOR 95% CI OR 95% CI AOR 95% CI
SGA (<10th percentile) 7.19 6.49e7.96 6.08 5.43e6.80 2.94 2.80e3.09 2.68 2.54e2.82
LGA (>90th percentile) 0.11 0.07e0.16 0.10 0.07e0.16 0.78 0.73e0.83 0.81 0.76e0.86
Fetal death 9.42 7.22e12.3 5.79 4.03e8.33 1.55 1.06e2.27 1.26 0.81e1.96
Neonatal death 12.84 9.63e17.1 11.44 8.07e16.4 1.31 0.78e2.19 1.09 0.61e1.96
Perinatal death 10.93 8.97e13.3 8.38 6.48e10.8 1.46 1.07e1.98 1.19 0.83e1.69
a
Perinatal death/morbidity 19.07 17.08e21.29 16.41 14.48e18.60 2.37 2.11e2.67 2.02 1.78e2.28
For early-onset preeclampsia comparisons, all ongoing pregnancies at 20 weeks of gestation were included in the denominator; for late-onset preeclampsia comparisons, all ongoing pregnancies at
34 weeks of gestation were included in the denominator.
AOR, adjusted odds ratio (adjusted for race, parity, maternal age, maternal education, infant’s sex, marital status, infertility treatment, chronic hypertension, diabetes, and congenital anomalies); CI,
confidence interval; LGA, large for gestational age; OR, odds ratio; SGA, small for gestational age.
a
Perinatal death/morbidity includes any of the following: perinatal death, 5 minute Apgar of 3 or less, neonatal seizures, neonatal sepsis, bronchopulmonary dysplasia, necrotizing enterocolitis,
intraventricular hemorrhage grades 3 and 4, periventricular leukomalacia, and retinopathy of prematurity.
Lisonkova. Early- vs late-onset preeclampsia. Am J Obstet Gynecol 2013.

The adjusted rates of adverse birth covariates increased the relative odds Additional adjustment for time period
outcomes were also higher among (AOR, 1.09; 95% CI, 1.02e1.16). (year of birth) did not change the results.
mothers with late-onset disease as Birth outcomes among women with Women excluded from the study were
compared with mothers without pre- early-onset preeclampsia who delivered different from the study population
eclampsia, although the differences in at a gestation of 34 weeks or longer and with regard to several risk factors for
rates were substantially less and some women with late-onset preeclampsia preeclampsia. Some risk factors (such
were not statistically significant (Table 4). were similar in terms of fetal and as African-American race, no prior
The rates of fetal, neonatal, and perinatal neonatal death (Table 5). However, the live births, chronic hypertension, and
death, for example, were not signifi- rates of the most common outcomes, congenital anomalies) were overrep-
cantly higher among women with late- such as NICU admission and SGA, were resented among the excluded women,
onset preeclampsia (AOR, 1.29; 95% significantly elevated among the early- whereas other risk factors (such as older
CI, 0.81e1.96; AOR, 1.09; 95% CI, onset group (AOR, 2.22; 95% CI, maternal age, Hispanic and Native-
0.61e1.96, and AOR, 1.19; 95% CI, 1.60e3.07; and AOR, 1.66; 95% CI, American race, single parent status, and
0.83e1.69, respectively). Rates of SGA, 1.24e2.23, respectively). diabetes mellitus) were less frequent than
in contrast, were significantly elevated Sensitivity analyses showed that high expected (Appendix; Supplementary
among mothers with late-onset disease BMI was a stronger risk factor for early- Table 2).
(AOR, 2.68; 95% CI, 2.54e2.82). onset than for late-onset disease (AHR,
From the prognostic perspective, live- 2.10; 95% CI, 1.91e2.32; and AHR, 1.71; C OMMENT
born infants of mothers with early-onset 95% CI, 1.65e1.77, respectively), similar Our population-based study showed
preeclampsia were less likely to die in the to the association between chronic hy- that the gestational ageespecific inci-
neonatal period (AOR, 0.51; 95% CI, pertension and the preeclampsia sub- dence of preeclampsia among women
0.36e0.74) compared with those born at types. The AHR for chronic hypertension with singleton pregnancies increased
the same gestation to mothers without decreased after additional adjustment for sharply with gestational age. The rate of
preeclampsia (Appendix; Supplementary obesity (AHR, 9.4; 95% CI, 8.2e10.9; and early-onset disease (<34 weeks’ gesta-
Table 1). However, these infants had AHR, 2.24; 95% CI, 2.11e2.38, for early- tion) was substantially lower than the
higher odds of severe neonatal morbidity and late-onset preeclampsia, respec- rate of late-onset disease (gestation of
(AOR, 1.35; 95% CI, 1.16e1.57), NICU tively). The associations between early-
34 weeks): 0.38 vs 2.72 per 100 de-
admission (AOR, 2.44; 95% CI, and late-onset preeclampsia and birth liveries, respectively. Several factors,
2.13e2.80), and SGA (AOR, 2.78; 95% outcomes were not appreciably affected including chronic hypertension, African-
CI, 2.46e3.13). The unadjusted odds of by additional adjustment for BMI except American race, and congenital anomalies
LGA were lower among infants of for LGA, which was no longer signifi- conferred a relatively higher risk for
mothers with late-onset disease (odds cantly elevated among live-born infants early-onset (as opposed to late-onset)
ratio, 0.78; 95% CI, 0.73e0.83), although of mothers with late-onset preeclampsia disease, whereas other factors such as
adjustment for gestational age and other (AOR, 1.03; 95% CI, 0.97e1.10). diabetes mellitus, nulliparity, and young

MONTH 2013 American Journal of Obstetrics & Gynecology 1.e7

Research Obstetrics www.AJOG.org

Although the causes of preeclampsia

TABLE 5 are not known, placental dysfunction has
Crude and adjusted odd ratios for birth outcomes contrasting been implicated in its origins,18,20,34 and
early-onset vs late-onset preeclampsia, singleton deliveries placental morphology studies suggest
at gestation of ‡34 weeks, Washington State, 2003-2008 that preeclampsia is a heterogeneous
Early-onset preeclampsia Late-onset entity.34-36 Our results show that effect of
with delivery at gestation preeclampsia risk factors such as race/ethnicity, nulli-
Birth outcomes ‡34 weeks (n [ 213) (n [ 12,449) P valuea
parity, chronic hypertension, and dia-
Gestational age at delivery, wks betes vary according to the subtype of
34-36 128 (60.1) 2911 (23.4) < .01 preeclampsia. For example, congenital
anomalies were more strongly associated
37-43 85 (39.9) 9538 (76.6)
with early-onset disease, suggesting the
Birthweight, g presence of associated placental abnor-
<1500 2 (0.95) 42 (0.34) < .01 malities that affect perfusion and result
1500-2499 70 (33.3) 2020 (16.2) in early-onset disease. In contrast, a
stronger positive association between
2500-4499 138 (65.7) 10,162 (81.6) diabetes mellitus and late-onset pre-

4500 0 (0.00) 166 (1.33) eclampsia suggests that relative placental
SGA (<10th percentile) 44 (21.0) 2007 (16.1) < .01 insufficiency is more likely to occur in
diabetic pregnancies with a larger fetus.
LGA (<90th percentile) 11 (5.24) 1171 (9.41) < .01
We also observed a stronger association
Apgar score at 5 min 3 b
2 (0.94) 99 (0.80) between early-onset disease and SGA (as
NICU admissionb 55 (25.8) 1658 (13.4) < .01 compared with late-onset disease and
Fetal death 0 (0.00) 28 (0.22) .49
SGA), likely because of the profound
b
effects of poor placental perfusion early
Neonatal death 0 (0.00) 15 (0.12) .61 in gestation and differences in disease
b,c
Severe neonatal morbidity 7 (3.29) 274 (2.21) .41 severity.20,37
Neonatal death/severe 7 (3.29) 283 (2.28) .46 Our study has a few limitations.
morbidityb Gestational age at the onset of pre-
Perinatal death 0 (0.00) 43 (0.35) .79 eclampsia was estimated based on the
time between hospital admission for
Perinatal death/severe 7 (3.29) 311 (2.50) < .047
morbidity
preeclampsia and hospital admission for
delivery. We were not able to capture
LGA, large for gestational age; NICU, neonatal intensive care unit; SGA, small for gestational age.
women with preeclampsia who were not
a
P value based on c2 or Fisher exact test; b Excludes fetal deaths; c Includes any of the following: a 5 minute Apgar score of 3
or less, neonatal seizures, neonatal sepsis, bronchopulmonary dysplasia, necrotizing enterocolitis, intraventricular hemor- hospitalized and those who did not deliver
rhage grades 3 and 4, periventricular leukomalacia, and retinopathy of prematurity. in the hospital. However, such missed
Lisonkova. Early- vs late-onset preeclampsia. Am J Obstet Gynecol 2013. cases of preeclampsia were likely milder
cases that did not result in serious com-
plications requiring hospitalization. In
maternal age were associated with a (Table 4). Furthermore, the rates of some cases, the onset of preeclampsia may
higher risk of late-onset preeclampsia. NICU admission and SGA were signifi- have occurred a few days before the
Early-onset preeclampsia conferred a cantly higher among infants born to admission to the hospital. The delay be-
substantially higher risk for adverse birth mothers with early-onset preeclampsia. tween the onset of preeclampsia and
outcomes than late-onset preeclampsia. Approximately 12% of the women with admission to the hospital may have
In contrast, the prognosis for neonatal early-onset preeclampsia delivered at a resulted in some misclassification of early-
death among infants born to women gestation of 34 weeks or longer. onset disease as late-onset preeclampsia.
with early-onset preeclampsia was better Our population-based study exam- As with any administrative database,
than the prognosis for infants born at the ined gestational ageespecific rates of the accuracy of diagnoses was contingent
same gestation because of other causes. preeclampsia in a large cohort of sin- on documentation and abstraction from
The lower risk of neonatal death among gletons. Overall rates of preeclampsia medical records. However, it has been
infants born to mothers with pre- were consistent with previous estimates shown that linkage between hospitali-
eclampsia at preterm gestation has been from industrialized countries, which zation data and birth/infant death cer-
reported previously.2,16,17 This differ- have reported preeclampsia rates be- tificates increases data accuracy and that
ence in prognosis, however, was condi- tween 3 and 5 per 100 deliveries2,3 and the accuracy of major obstetric diagnoses
tional on a live birth and not evident eclampsia rates between 2.7 and 8.2 per and procedures is relatively high in
in the causal fetuses-at-risk analysis 10,000 deliveries.2,7 Washington State’s linked data file.38

1.e8 American Journal of Obstetrics & Gynecology MONTH 2013

www.AJOG.org Obstetrics Research
Other potential weaknesses of the preeclampsia had far greater adverse ef- eclampsia and febrile convulsions. Arch Dis
study include limited information about fects on the fetus and infant compared Child 2003;88:726-7.
15. McElrath TF, Allred EN, Boggess KA, et al.
risk factors, such as BMI. Sensitivity with late-onset disease. Our study thus Maternal antenatal complications and the risk of
analyses showed that high BMI had a confirms the heterogeneity of pre- neonatal cerebral white matter damage and later
stronger association with early- vs late- eclampsia and shows that the timing of cerebral palsy in children born at an extremely
onset preeclampsia (similar to chronic disease onset is one important indicator low gestational age. Am J Epidemiol 2009;170:
hypertension), suggesting that metabolic of disease severity and possibly of disease 819-28.
16. Chen XK, Wen SW, Smith G, Yang Q,
syndrome (associated with high BMI) etiology. Research studies should treat
Walker M. Pregnancy-induced hypertension is
may play a stronger role in early-onset vs the 2 preeclampsia subtypes as distinct associated with lower infant mortality in preterm
late-onset disease. entities from an etiological and prog- singletons. BJOG 2006;113:544-51.
We did not have detailed information nostic standpoint. 17. Yu XD, Branch DW, Karumanchi A, Zhang J.
about antenatal screening, and obstetric Preeclampsia and retinopathy of prematurity in
preterm births. Pediatrics 2012;130:e101-7.
and neonatal care practices in the REFERENCES 18. Roberts JM, Hubel CA. The two stage
different hospitals in Washington state 1. Steegers EAP, Von Dadelszen P, Duvekot JJ, model of pre-eclampsia: variations on the
including use of aspirin among women Pijnenborg R. Pre-eclampsia. Lancet 2010;376: theme. Placenta 2009;30(Suppl A):S32-7.
at high risk for preeclampsia. Potential 631-44. 19. Ness RB, Roberts JM. Heterogeneous
inaccuracies in diagnosis and differences 2. Hutcheon JA, Lisonkova S, Joseph KS. The causes constituting the single syndrome of pre-
epidemiology of preeclampsia and the hyper- eclampsia: a hypothesis and its implications. Am
in intervention between physicians and
tensive disorders of pregnancy. Best Pract Res J Obstet Gynecol 1996;175:1365-70.
hospitals may have resulted in some Clin Obstet Gynaecol 2011;25:391-403. 20. Von Dadelszen P, Magee LA, Roberts JM.
nondifferential misclassification of risk 3. Sibai B, Dekker G, Kupferminc M. Pre- Subclassification of pre-eclampsia. Hypertens
factors and outcomes and led to some eclampsia. Lancet 2005;365:785-99. Pregnancy 2003;22:143-8.
dilution in observed associations. 4. Khan KS, Wojdyla D, Say L, Gulmezoglu AM, 21. Paruk F, Moodley J. Maternal and neonatal
Van Look PF. WHO analysis of causes of outcome in early- and late-onset pre-eclampsia.
Finally, subjects excluded from the
maternal death: a systematic review. Lancet Semin Neonatol 2000;5:197-207.
study because of missing information or 2006;367:1066-74. 22. Bombrys AE, Barton JR, Nowacki EA, et al.
unlinked records were significantly 5. Centre for Maternal and Child Enquiries Expectant management of severe preeclampsia
different from those included in the (CMACE). Saving Mothers’ Lives: reviewing at less than 27 weeks’ gestation: maternal and
study, with respect to some maternal maternal deaths to make motherhood safer: perinatal outcomes according to gestational age
2006-08. The Eighth Report on Confidential by weeks at onset of expectant management.
characteristics, although the fraction
Enquiries into Maternal Deaths in the United Am J Obstet Gynecol 2008;199:247.e1-6.
excluded was small (5.7%). Kingdom. BJOG 2011;118(Suppl1):1-203. 23. Hall DR, Grove D, Carstens E. Early pre-
The strengths of our study include a 6. Kuklina EV, Ayala C, Callaghan WM. Hyper- eclampsia: what proportion of women qualify
large study population and gestational tensive disorders and severe obstetric morbidity for expectant management and if not, why not?
ageespecific incidence rates defined us- in the United States. Obstet Gynecol 2009;113: Eur J Obstet Gynecol 2006;128:169-74.
1299-306. 24. Jenkins SM, Head BB, Hauth JC. Severe
ing the onset of disease. The population
7. Knight M. Eclampsia in the United Kingdom, preeclampsia at <25 weeks of gestation:
perspective obtained by using statewide 2005. BJOG 2007;114:1072-8. maternal and neonatal outcomes. Am J Obstet
information minimized potential selec- 8. Liu S, Joseph KS, Bartholomew S, et al. Gynecol 2002;286:790-5.
tion bias that may be present in hospital- Temporal trends and regional variations in severe 25. Sibai BM, Mercer BM, Schiff E,
based studies, especially those that maternal morbidity in Canada, 2003 to 2007. Friedman SA. Aggressive versus expectant
J Obstet Gynecol Canada 2010;32:847-55. management of severe preeclampsia at 28 to 32
include selected hospital patients.30-33
9. Ananth CV, Savitz DA, Williams MA. Placental weeks’ gestation: a randomized controlled trial.
The large study size offered statistical abruption and its association with hypertension Am J Obstet Gynecol 1994;171:818-22.
power for analysis of rare adverse birth and prolonged rupture of membranes: a meth- 26. Alexander G, Himes J, Kaufman R, et al.
outcomes. Finally, the use of the fetuses- odologic review and meta-analysis. Obstet A United States reference for fetal growth.
at-risk approach in addition to tradi- Gynecol 1996;88:309-18. Obstet Gynecol 1996;87:163-8.
10. MacKay AP, Berg CJ, Atrash HK. Preg- 27. Mustafa G, David RJ. Comparative accu-
tional perinatal modeling provides the
nancy-related mortality from pre-eclampsia and racy of clinical estimate versus menstrual
causal and prognostic perspectives on eclampsia. Obstet Gynecol 2001;97:533-8. gestational age in computerized birth certifi-
the relation between preeclampsia and 11. Basso O, Rasmussen S, Weinberg CR, cates. Public Health Rep 2001;116:15-21.
birth outcomes. Wilcox AJ, Irgens LM, Skjaerven R. Trends in 28. Ananth CV. Menstrual versus clinical esti-
In summary, population-based data fetal and infant survival following pre-eclampsia. mate of gestational age dating in the United
JAMA 2006;296:1357-62. States: temporal trends and variability in indices
on singleton deliveries in Washington
12. Badawi N, Kurinczuk JJ, Keogh JM, et al. of perinatal outcomes. Paediatr Perinat Epi-
State between 2003 and 2008 showed Antepartum risk factors for newborn encepha- demiol 2007;21(Suppl 2):22-30.
that the incidence of preeclampsia lopathy: the Western Australian case-control 29. Alison PD. Survival analysis using SAS: a
increased sharply with gestational age. study. BMJ 1998;317:1549-53. practical guide, 2nd ed. Cary, NC: SAS Institute
Even though some risk factors were 13. Friedman SA, Schiff E, Kao L, Sibai BM. Inc; 2010.
Neonatal outcome after preterm delivery for pre- 30. Yudkin PL, Wood L, Redman CW. Risk of
common to both early- and late-onset
eclampsia. Am J Obstet Gynecol 1995;172: unexplained stillbirth at different gestational
disease, several risk factors had quanti- 1785-8; discussion 8-92. ages. Lancet 1987;1:1192-4.
tatively different associations with the 2 14. Vestergaard M, Basso O, Henriksen TB, 31. Joseph KS. Incidence-based measures of
subtypes of preeclampsia. Early-onset Østergaard J, Secher NJ, Olsen J. Pre- birth, growth restriction and death can free

MONTH 2013 American Journal of Obstetrics & Gynecology 1.e9

Research Obstetrics www.AJOG.org

perinatal epidemiology from erroneous con- and vascular abnormalities in early- and underperfusion in preeclampsia. J Perinat Med
cepts of risk. J Clin Epidemiol 2004;57:889-97. late-onset pre-eclampsia with and without 2011;39:641-52.
32. Auger N, Delezire P, Harper S, Platt RW. fetal growth restriction. BJOG 2006;113: 37. Ødegård RA, Vatten LJ, Nilsen ST,
Maternal education and stillbirth: estimating 580-9. Salvesen KÅ, Austgulen R. Preeclampsia
gestational-age-specific and cause-specific as- 35. Valensise H, Vasapollo B, Gagliardi G, and fetal growth. Obstet Gynecol 2000;96:
sociations. Epidemiology 2012;23:247-54. Novelli GP. Early and late pre-eclampsia: two 950-5.
33. Cole SR, Platt RW, Schisterman EF, et al. different maternal hemodynamic states in the 38. Lydon-Rochelle MT, Holt VL, Nelson JC,
Illustrating bias due to conditioning on a collider. latent phase of the disease. Hypertension et al. Accuracy of reporting maternal in-hospital
Int J Epidemiol 2010;39:417-20. 2008;52:873-80. diagnoses and intrapartum procedures in
34. Egbor M, Ansari T, Morris N, Green CJ, 36. Ogge G, Chaiworapongsa T, Romero R, Washington state linked records. Paediatr Peri-
Sibbons PD. Morphometric placental villous et al. Placental lesions consistent with maternal nat Epidemiol 2005;19:460-71.

1.e10 American Journal of Obstetrics & Gynecology MONTH 2013

www.AJOG.org Obstetrics Research

A PPENDIX

SUPPLEMENTARY TABLE 1
Crude and AORs for neonatal outcomes among live newborns, following early- and late-onset preeclampsia,
singleton deliveries, Washington State, 2003-2008a
Early-onset preeclampsia Late-onset preeclampsia
Neonatal outcomes OR 95% CI AORa 95% CI OR 95% CI AORa 95% CI
SGA (<10th percentile) 6.51 5.62e7.54 2.78 2.46e3.13 2.94 2.80e3.09 2.13 2.02e2.24
LGA (>90th percentile) 0.27 0.14e0.52 0.46 0.29e0.74 0.78 0.73e0.83 1.09 1.02e1.16
Apgar score at 5 min 3 0.46 0.36e0.59 0.84 0.64e1.10 2.62 2.14e3.22 1.98 1.59e2.46
NICU admission 1.90 1.66e2.17 2.44 2.13e2.80 3.82 3.62e4.03 1.65 1.55e1.76
Severe neonatal morbidityb 0.82 0.72e0.94 1.35 1.16e1.57 2.59 2.29e2.93 1.57 1.37e1.79
Neonatal death 0.29 0.22e0.39 0.51 0.36e0.74 1.31 0.78e2.19 0.71 0.39e1.28
Neonatal death/severe morbidity 0.75 0.66e0.85 1.18 1.01e1.37 2.50 2.21e2.82 1.51 1.32e1.72
Regression models adjusted for gestational age, race, parity, maternal age, maternal education, infant’s sex, marital status, infertility treatment, chronic hypertension, diabetes, and congenital
anomalies.
AOR, adjusted odds ratio; CI, confidence interval; LGA, large for gestational age; NICU, neonatal intensive care unit; OR, odds ratio; SGA, small for gestational age.
a
Prognostic model comparing neonatal outcomes conditional on live birth, adjusted for gestational age at delivery. Fetal death was not included in the prognostic model because such deaths occur
before birth; b Includes any of the following: a 5 minute Apgar score of 3 or less, bronchopulmonary dysplasia, necrotizing enterocolitis, neonatal seizures, neonatal sepsis, intraventricular
hemorrhage grades 3 and 4, periventricular leukomalacia, and retinopathy of prematurity.
Lisonkova. Early- vs late-onset preeclampsia. Am J Obstet Gynecol 2013.

MONTH 2013 American Journal of Obstetrics & Gynecology 1.e11

Research Obstetrics www.AJOG.org

SUPPLEMENTARY TABLE 2
Demographic and clinical characteristics, singleton deliveries with and
without inclusion criteria, Washington State, 2003-2008a
Included Excluded
Demographic/clinical risk factors n [ 456,668 (%) n [27,443 (%)a P valueb
Age, y
<20 39,584 (8.67) 2285 (8.33) < .01
20-34 347,105 (76.01) 22,217 (80.96)

35 69,979 (15.32) 2687 (9.79)
Race
Non-Hispanic white 323,552 (70.85) 19,056 (69.44) < .01
African-American 20,045 (4.39) 2455 (8.95)
Hispanic 56,615 (12.4) 2735 (9.97)
Native-American 10,625 (2.33) 488 (1.78)
Other 44,032 (9.64) 2480 (9.04)
Maternal education less 89,238 (19.54) 3683 (13.42) < .01
than high school
Smoking during pregnancy 46,936 (10.28) 2766 (10.08) < .01
Single parent 149,369 (32.71) 5213 (19.00) < .01
No prior live births 186,961 (40.94) 12801 (46.65) < .01
Diabetes mellitus 25,815 (5.65) 1279 (4.66) < .01
Chronic hypertension 5560 (1.22) 499 (1.82) < .01
Infertility treatment 3455 (0.76) 145 (0.53) < .01
Infant sex (male) 234224 (51.29) 14,063 (51.24) .98
Congenital anomalies 2249 (0.49) 353 (1.29) < .01
Some numbers do not add because of missing values; all missing values were less than 3%.
a
Excluded were women who delivered outside the hospital, those without a linkage between birth/fetal death certificate and
hospitalization file, and those without a missing estimate of gestation at delivery; b P values are based on c2 test.
Lisonkova. Early- vs late-onset preeclampsia. Am J Obstet Gynecol 2013.

1.e12 American Journal of Obstetrics & Gynecology MONTH 2013