Published Ahead of Print on August 10, 2018 as 10.1212/WNL.

0000000000006173
ARTICLE

Unintended pregnancy, prenatal care, newborn

outcomes, and breastfeeding in women with
epilepsy
Emily L. Johnson, MD, Anne E. Burke, MD, MPH, Anqi Wang, MHS, and Page B. Pennell, MD Correspondence

®
Dr. Johnson
Neurology 2018;00:1-9. doi:10.1212/WNL.0000000000006173 ejohns92@jhmi.edu

Abstract
Objective
To compare the proportions of unintended pregnancies, prenatal vitamin or folic acid (PNVF)
use, adequate prenatal care visits, and breastfeeding among women with epilepsy (WWE) to
women without epilepsy (WWoE).

Methods
The Pregnancy Risk Assessment Monitoring System (PRAMS) is an annual survey of randomly
sampled postpartum women administered by the Centers for Disease Control and Prevention.
We used PRAMS data from 13 states from 2009 to 2014 to compare the primary outcomes in
WWE and WWoE, as well as our secondary outcomes of contraception practices, newborn
outcomes, and time to recognition of pregnancy. We adjusted for maternal age, race, ethnicity,
and socioeconomic status (SES), and we calculated odds ratios for these outcomes using
logistic regression.

Results
This analysis included 73,619 women, of whom 541 (0.7%) reported epilepsy, representing
3,442,128 WWoE and 26,635 WWE through weighted sampling. In WWE, 55% of pregnancies
were unintended compared to 48% in WWoE. After adjustment for covariates, epilepsy was not
associated with unintended pregnancy or with inadequate prenatal care. WWE were less likely
to report breastfeeding but more likely to report daily PNVF use. Newborns of WWE had
higher rates of prematurity.

Conclusions
Although planning for pregnancy is of utmost importance for WWE, more than half the
pregnancies in WWE were unintended. Maternal age and SES differences likely contribute to
the higher rates in WWE compared to WWoE. The proportion of women reporting breast-
feeding is lower in WWE despite studies indicating the safety of breastfeeding in WWE.

From the Department of Neurology (E.L.J.) and Department of Gynecology and Obstetrics (A.E.B.), Johns Hopkins School of Medicine, Baltimore; Johns Hopkins Bloomberg School of
Public Health (A.E.B., A.W.), Baltimore, MD; and Department of Neurology (P.B.P.), Brigham and Women’s Hospital, Boston, MA.

Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.

Copyright © 2018 American Academy of Neurology 1

Copyright ª 2018 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
 Glossary
AED = antiepileptic drug; CDC = Centers for Disease Control and Prevention; CI = confidence interval; IUD = intrauterine
device; MCM = major congenital malformation; NEAD = Neurodevelopmental Effects of Antiepileptic Drugs; NES =
nonepileptic seizures; NICU = neonatal intensive care unit; OR = odds ratio; PNVF = prenatal vitamin or folic acid; PRAMS =
Pregnancy Risk Assessment Monitoring System; SES = socioeconomic status; SGA = small for gestational age; WIC = Women,
Infants, and Children; WWE = women with epilepsy; WWoE = women without epilepsy.

Children born to women with epilepsy (WWE) on anti- Methods

epileptic drugs (AEDs) are at increased risk for adverse
outcomes, including major congenital malformations The Centers for Disease Control and Prevention (CDC)
(MCMs),1,2 neurodevelopmental delays,3 and low birth conducts the standardized Pregnancy Risk Assessment
weights.4 AED use and dose before conception influence Monitoring System (PRAMS), an ongoing survey-based
these risks; some AEDs such as valproic acid have sub- surveillance program of women during the postpartum period
stantially higher risks of these outcomes (with a dose- (sampled from birth certificate data). PRAMS surveys are
dependent effect), while others have much lower risks of currently administered in 47 states in English and Spanish,
adverse outcomes.1 Thus, pregnancy planning is of utmost representing ≈83% of all live births in the United States.12
importance for WWE to ensure AED optimization before Each state samples between 1,300 and 3,400 women annually,
pregnancy. However, because some AEDs interact with depending on population. PRAMS uses stratified, weighted
hormonal contraceptives, WWE may be at increased risk population sampling to ensure that adequate data are available
for unintended pregnancies.5 Although breastfeeding is for some high-risk populations (e.g., neonates with low birth
generally thought to be safe for WWE,6 some women are weight, mothers with advanced maternal age). The number of
still actively discouraged from breastfeeding while taking targeted respondents in each category (e.g., low-birth-weight
AEDs.6 Prenatal vitamin or folic acid (PNVF) use before infants born to mothers in a rural area) is determined from the
conception and adequate prenatal care are also vital number of births in each category, proportion of all births that
in WWE. each category represents, and predicted response rates, ad-
justed by the finite population correction when necessary.
One study of 1,144 WWE who self-enrolled in a birth Each participant has a weight assigned by the CDC according
control registry and completed a retrospective survey to the representative population-based geographic and de-
found that 65% of reported pregnancies in WWE were mographic sampling.
unintended.7 In this group, younger women, racial minor-
ities, and Hispanic women had higher proportions of un- Data are collected via mailed survey ≈3 months postpartum.
intended pregnancies. In a survey of 148 reproductive-age Multiple contacts (mailing and telephone) are attempted to
WWE cared for in a tertiary medical center, 50.3% of the maximize participation; participation rates are >65% in most
181 reported pregnancies in 77 women had been states. The PRAMS survey consists of a set of core questions
unplanned.8 administered by every state and optional standardized questions
(available at cdc.gov/prams/questionnaire.htm). From 2009 to
However, no comparisons have been made of unintended 2014 (the most recent PRAMS data available at time of data
pregnancies in a population including both WWE and analysis), 13 centers asked mothers whether they carried a di-
women without epilepsy (WWoE), which limits the con- agnosis of epilepsy or seizures in the 3 months before their
clusions that can be drawn about the effects of epilepsy per pregnancy.
se vs the effects of maternal age, race, and socioeconomic
status (SES). Much of the knowledge about WWE and We conducted a retrospective analysis using deidentified
pregnancy outcomes has been obtained through careful data from the 12 states and 1 city that collected self-
population-based studies, particularly from European reported epilepsy status through PRAMS (Delaware,
countries.9–11 To the best of our knowledge, no US Hawaii, Florida, Maryland, Michigan, Minnesota, Missouri,
population–based studies have examined breastfeeding, New York City, Utah, Washington, Wisconsin, West Vir-
pregnancy intention, or PNVF use. Therefore, to de- ginia, and Wyoming).
termine whether the prevalence of unintended pregnancy
and other outcomes is different between WWE and WWoE, Our primary outcomes were intendedness of pregnancy, daily
we compared the proportions of unintended pregnancies, preconception use of PNVF, adequate prenatal care (assessed
PNVF use, prenatal care, and breastfeeding in WWE and with the Kotelchuck13 index, with inadequate care defined as
WWoE, adjusting for age, race, ethnicity, and SES, in <50% of recommended visits, intermediate care as 50%–79%
a rigorously sampled population. Our secondary outcomes of recommended visits, adequate care as 80%–110% of rec-
were contraceptive practices, time to recognition of preg- ommended visits, and “adequate-plus” care as >110% of
nancy, and neonatal outcomes. recommended visits), and any breastfeeding. Outcomes were

2 Neurology | Volume , Number  | Month 0, 2018 Neurology.org/N

Copyright ª 2018 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
taken from responses to the PRAMS questionnaire; Results
respondents indicated whether pregnancy was intended,
whether and how often they took a PNVF in the month This study included 73,619 women, of whom 548 (0.7%)
before pregnancy, at what gestational age they had their first reported epilepsy (WWE). As a result of the proportional
prenatal visit, how many prenatal visits they had, and whether stratified, weighted sampling described above, these sampled
they breastfed at any time postpartum. Secondary outcomes women represent a population of 3,442,128 WWoE and
were approximate gestational age at which pregnancy was 26,635 WWE (0.7% with epilepsy).
recognized, neonatal outcomes (neonatal intensive care unit
[NICU] admission, small for gestational age [SGA; weight WWE were younger than WWoE (p = 0.001), had lower
<10th percentile for age], and premature birth [<37 weeks]), household incomes (p < 0.001), and were more likely to be on
birth control methods used at the time of conception, and WIC or Medicaid (p < 0.001; table 1). WWE had a younger
birth control methods used after delivery. mean age than WWoE for pregnancies reported as intended
and as unintended (p = 0.001). There was no difference in
The covariates in our analysis were age, self-reported race race between WWE and WWoE.
(white, black, Asian, Native American, other), ethnicity
(Hispanic or not Hispanic), and SES factors: yearly house- Pregnancy intention
hold income, use of Medicaid, and use of Women, Infants, and In WWE, 54.6% of pregnancies were unintended, which was
Children (WIC) during pregnancy. higher than the 47.6% of pregnancies that were unintended
in WWoE. WWE had an unadjusted OR of 1.40 for un-
We excluded women missing data on history of epilepsy or intended pregnancy compared to WWoE (95% CI
covariates or with age <12 or >50 years from primary analysis. 1.18–1.66, p < 0.001). Unintended pregnancies were more
We conducted sensitivity analyses comparing the results of common in women with younger ages (p < 0.001), with
analyses including women of all age values and including pregnancies most likely to be unintended in the <20-year-old
those missing data on history of epilepsy covariates (race, age group and most likely to be intended in the 30- to 34-
ethnicity, income) with dummy categorical variables used for year-old group. Unintended pregnancies were more com-
the missing variables. mon in women with lower income; pregnancies were most
likely to be intended in the group with annual household
We used Stata 15.0 (StataCorp, College Station, TX) to ac- income >$50,000 (p < 0.001). Use of Medicaid or WIC was
count for the survey sampling methodology applied by also associated with unintended pregnancy (p < 0.001).
PRAMS with multiple strata and sample weights. We first
compared participant characteristics of WWE and WWoE
with univariate analysis (using χ 2 or t tests as appropriate) and Table 1 Characteristics of WWE and WWoE
univariate logistic regression for unadjusted odds ratios WWoE
(ORs). For categorical independent variables such as race, the WWE (n = 548), (n = 73,071), p
category with the largest number of respondents was selected mean or % mean or % Value

as the reference group. We then used multivariable logistic Age, y 27.1 28.6 0.001
regression to adjust for the confounders of age, race, and SES
Race, % 0.827
and to estimate the ORs and 95% confidence intervals (CIs)
for the outcomes of interest in WWE compared to WWoE. White 70.3 67.6
For comparisons of continuous outcomes, we used a t test. A Black 15.9 15.9
value of p < 0.05 was considered significant.
Asian 6.7 8.3

Data availability Other 5.3 6.7

Limited PRAMS data are publically available at cdc.gov/
Hispanic ethnicity, % 8.0 12.5 0.056
prams/index.htm. Full PRAMS data are available to researchers
who submit a project application, abstract, and data-sharing Annual household 0.001
income, %
agreement to the CDC.
≤$15,000 44.3 26.4

Standard protocol approvals, registration, and $15,001–$25,000 13.8 15.3

patient consents
$25,001–$50,000 16.9 22.4
The PRAMS program is approved by the Institutional Re-
view boards of each participating state and of the CDC. This >$50,000 25.2 35.9
study was approved by the central PRAMS proposals com- WIC or Medicaid use 65.9 52.7 <0.001
mittee and separately by the Florida Department of Health
Institutional Review Board and was reviewed and de- Abbreviations: WIC = Women, Infants, and Children supplemental nutrition
program; WWE = women with epilepsy; WWoE = women without epilepsy.
termined to be exempt by the Johns Hopkins Institutional Values are means or percentages of respondents.
Review Board.

Neurology.org/N Neurology | Volume , Number  | Month 0, 2018 3

Copyright ª 2018 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
 Pregnancies were most likely to be intended in Asian and interaction between state and epilepsy (i.e., results of as-
white participants and least likely to be intended in Native sociation of epilepsy with pregnancy intention did not
American, black, or other race participants. Pregnancies were differ by state).
less likely to be intended in Hispanic than in non-Hispanic
participants (p < 0.001). Breastfeeding
We found that 69.1% of WWE reported breastfeeding at any
After adjustment for age, race, ethnicity, and SES (income time compared to 84.6% of WWoE (OR 0.40, 95% CI
and WIC or Medicaid use), epilepsy was not associated 0.30–0.55, p < 0.001). After adjustment for covariates, WWE
with unintended pregnancy (OR 1.00, 95% CI 0.76–1.32, p remained less likely to report any breastfeeding than WWoE
= 0.989; table 2). Higher income, age of 30 to 34 years, and (OR 0.42, 95% CI 0.26–0.68, p < 0.001; table 3). Higher
no use of Medicaid or WIC were all associated with income, Asian and other race, Hispanic ethnicity, and no use
pregnancy intention in multivariable analysis (table 2). of Medicaid or WIC were associated with breastfeeding.
Although the proportion of intended pregnancies varied Among women who reported breastfeeding, there was no
by state (with intended pregnancies least common in difference in the length of time that infants were breastfed: the
Hawaii after adjustment for covariates), there was no mean was 6.6 weeks in WWoE compared to 6.4 weeks in
WWE (p = 0.607). There was no interaction between state
and epilepsy.

Table 2 Adjusted ORs for unintended pregnancy

compared to intended pregnancy
Table 3 Adjusted ORs for reporting breastfeeding
Adjusted OR for unintended p compared to not reporting breastfeeding
Characteristic pregnancy (95% CI) Value
Adjusted OR for
WWE (compared to 1.00 (0.76–1.32) 0.989 Characteristic Breastfeeding (95% CI) p Value
WWoE)
WWE (compared to WWoE) 0.42 (0.26–0.68) <0.001
Age, y
Age, y
<20 3.01 (2.62–3.47) <0.001
<20 0.71 (0.57–0.89) 0.003
20–24 1.50 (1.39–1.61) <0.001
20–24 0.87 (0.75–1.00) 0.053
25–29 1.0 (Reference group)
25–29 1.0 (Reference group)
30–34 0.92 (0.86–0.99) 0.023
30–34 1.08 (0.92–1.24) 0.327
35–39 1.09 (1.00–1.18) 0.049
35–39 0.98 (0.82–1.17) 0.843
≥40 1.26 (1.12–1.42) <0.001
≥40 0.78 (0.59–1.03) 0.081
Race
Race
White 1.0 (Reference group)
White 1.0 (Reference group)
Black 2.05 (1.92–2.19) <0.001
Black 0.81 (0.72–0.92) 0.001
Asian 0.95 (0.86–1.04) 0.268
Asian 1.33 (1.09–1.63) 0.006
Native American 0.75 (0.65–0.87) <0.001
Native American 1.33 (0.93–1.92) 0.121
Other 1.11 (0.99–1.25) 0.080
Other 1.38 (1.04–1.83) 0.025
Hispanic ethnicity 0.98 (0.89–1.07) 0.651
Hispanic ethnicity 2.02 (1.66–2.46) <0.001
Annual household
income, $ Annual household income, $

≤15,000 0.58 (0.54–0.63) <0.001 ≤15,000 0.59 (0.51–0.69) <0.001

15,001–25,000 0.73 (0.68–0.79) <0.001 15,001–25,000 0.89 (0.75–1.06) 0.187

25,001–50,000 1.0 (Reference group) 25,001–50,000 1.0 (Reference group)

>50,000 1.66 (1.53–1.79) <0.001 >50,000 1.14 (0.95–1.37) 0.164

WIC or Medicaid use 0.60 (0.56–0.64) <0.001 WIC or Medicaid use 0.51 (0.43–0.61) <0.001

Abbreviations: CI = confidence interval; OR = odds ratio; WIC = women, Abbreviations: CI = confidence interval; OR = odds ratio; WIC = Women,
infants, and children supplemental nutrition program; WWE = women with Infants, and Children supplemental nutrition program; WWE = women with
epilepsy; WWoE = women without epilepsy. epilepsy; WWoE = women without epilepsy.
OR for WWE compares unintended pregnancy in WWE to WWoE. ORs are OR for WWE compares breastfeeding in WWE to WWoE. ORs are adjusted for
adjusted for all covariates in table. all covariates in table.

4 Neurology | Volume , Number  | Month 0, 2018 Neurology.org/N

Copyright ª 2018 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
Vitamin use and prenatal care 0.202); WWE had more adequate-plus prenatal care (>110%
The majority (52.3%) of women reported no preconception of recommended prenatal visits; adjusted OR 1.50, 95% CI
PNVF use, and only 43.6% of WWE and 31.8% of WWoE 1.00–2.27, p = 0.051; table 5), although this was not a statis-
reported daily PNVF use. After adjustment for covariates, tical difference. Women <20 years of age, those with income
WWE were more likely to report daily PNVF use before <$25,000, women of races other than white or other, and
conception than were WWoE (OR 2.03, 95% CI 1.31–3.13, p women with WIC/Medicaid use were less likely to have at
= 0.002). PNVF use was more common in women with in- least adequate prenatal care.
creasing age, race other than black or Asian, and higher
incomes and without Medicaid or WIC use (table 4). There WWE reported being sure that they were pregnant at a mean
was an interaction between state and epilepsy: PNVF was of 6.5 weeks’ gestational age (median 5 weeks, interquartile
reported less among WWE in Delaware, Florida, Maryland, range 4–8 weeks, range 1–24 weeks). After adjustment for
Utah, Wisconsin, and West Virginia than in other states. covariates, there was no difference at time of pregnancy rec-
ognition between WWE and WWoE, who recognized they
WWE and WWoE were equally likely to have at least adequate were pregnant at a mean of 5.9 weeks.
prenatal care (adjusted OR 0.74, 95% CI 0.47–1.17, p =
Contraception before pregnancy
and postpartum
Table 4 Adjusted ORs for reported daily use of PNVF in A large proportion (44.5%) of the 34,220 women who
the month before pregnancy compared to responded to queries on contraception reported using some
reporting less than daily use of PNVF
method of birth control when they became pregnant. Detailed
Adjusted OR for daily information on the type of birth control used before con-
PNVF use before
Characteristic conception (95% CI) p Value
ception was available in 5.6% (4,135) of WWoE and 7.2%
(39) of WWE (table 5). The most commonly cited methods
WWE (compared to WWoE) 2.03 (1.31–3.13) 0.002 of birth control used at the time of conception were condoms
Age, y (70.2% of WWE, 49.4% of WWoE) and withdrawal (38.5% of
WWE, 42.6% of WWoE). No WWE reported using in-
<20 0.56 (0.43–0.71) <0.001
trauterine devices (IUDs) or contraceptive implants at the
20–24 0.78 (0.69–0.88) <0.001 time of conception. There was no difference in the use of
25–29 1.0 (Reference group) hormonal birth control at the time of conception (pill, ring,
injectable, patch, implant) by WWE (31.6%) and WWoE
30–34 1.24 (1.13–1.37) <0.001
(30.7%) (table 6).
35–39 1.27 (1.13–1.43) <0.001

≥40 1.46 (1.21–1.77) <0.001

After delivery, the majority of women (82.6%) reported using
birth control (table 6). Condoms were the most common
Race form of postpartum birth control reported by both WWE and
White 1.0 (Reference group) WWoE. The proportion of WWE reporting a long-acting or
permanent birth control method after delivery was 35.8%,
Black 0.70 (0.59–0.74) <0.001
including 17.6% reporting an IUD. Among WWoE, 28%
Asian 0.75 (0.66–0.86) <0.001 reported a long-acting or permanent birth control after de-
Native American 0.87 (0.66–1.15) 0.325
livery, including 16.8% reporting an IUD.

Other 0.90 (0.74–1.09) 0.289

Neonatal Outcomes
Hispanic ethnicity 0.92 (0.80–1.06) 0.264 In univariate analysis, epilepsy was associated with higher
rates of prematurity (p < 0.001), NICU stay (p < 0.001), and
Annual household income, $
SGA (p < 0.001). In multivariable analysis, epilepsy was as-
≤15,000 0.74 (0.65–0.84) <0.001 sociated with a higher risk of prematurity (adjusted OR 1.63,
15,001–25,000 0.82 (0.72–0.94) 0.004 95% CI 1.23–2.16), although not with SGA or with NICU
stay after adjustment for prematurity. In univariate analysis,
25,001–50,000 1.0 (Reference group)
unintended pregnancies were more likely than intended
>50,000 1.58 (1.43–1.76) <0.001 pregnancies to result in a neonate that was SGA (p < 0.001) or
WIC or Medicaid use 0.62 (0.56–0.69) <0.001
had NICU stay (p = 0.023); there was no difference in pre-
maturity (p = 0.249). After adjustment for epilepsy and de-
Abbreviations: : CI = confidence interval; PNVF = prenatal vitamins, vitamins, mographics, there were no interactions between epilepsy and
or folic acid; OR = odds ratio; WIC = Women, Infants, and Children supple-
mental nutrition program; WWE = women with epilepsy; WWoE = women pregnancy intention for neonatal outcomes.
without epilepsy.
OR for WWE compares daily PNVF use in WWE to WWoE. ORs are adjusted
for all covariates in the table. In sensitivity analyses, results including women with missing
epilepsy history (7,482 of 73,619) and of all reported ages (n

Neurology.org/N Neurology | Volume , Number  | Month 0, 2018 5

Copyright ª 2018 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
 Table 5 Adjusted ORs for at least adequate prenatal care (compared to less than adequate prenatal care) and for
adequate-plus prenatal care (compared to less than adequate and adequate prenatal care)
Adjusted OR for at least adequate Adjusted OR for adequate-plus
Characteristic prenatal care (95% CI) p Value prenatal care (95% CI) p Value

Women with epilepsy (compared to WWoE) 0.74 (0.47–1.17) 0.202 1.50 (1.00–2.27) 0.051

Age, y

<20 0.68 (0.56–0.83) <0.001 1.00 (0.82–1.22) 0.982

20–24 0.90 (0.80–1.01) 0.078 0.91 (0.82–1.02) 0.117

25–29 1.0 (Reference group) 1.0 (Reference group)

30–34 0.98 (0.87–1.10) 0.682 1.00 (0.91–1.10) 0.977

35–39 0.95 (0.83–1.09) 0.485 1.13 (1.00–1.28) 0.045

≥40 0.98 (0.78–1.24) 0.893 1.17 (0.96–1.42) 0.119

Race

White 1.0 (Reference group) 1.0 (Reference group)

Black 0.57 (0.51–0.64) <0.001 0.87 (0.78–0.97) 0.011

Asian 0.68 (0.59–0.77) <0.001 0.75 (0.65–0.86) <0.001

Native American 0.71 (0.54–0.94) 0.016 1.00 (0.77–1.32) 0.961

Other 0.85 (0.70–1.03) 0.088 0.93 (0.77–0.98) 0.419

Hispanic ethnicity 0.87 (0.76–1.01) 0.069 0.85 (0.74–0.98) 0.024

Annual household income

≤15,000 0.66 (0.58–0.75) <0.001 0.96 (0.85–1.09) 0.537

15,001–25,000 0.84 (0.73–0.97) 0.016 1.06 (0.93–1.20) 0.382

25,001–50,000 1.0 (Reference group) 1.0 (Reference group)

>50,000 0.95 (0.83–1.09) 0.461 1.02 (0.91–1.14) 0.725

WIC or Medicaid use 0.86 (0.75–0.98) 0.020 1.09 (0.98–1.22) 0.110

Abbreviations: CI = confidence interval; OR = odds ratio; WIC = Women, Infants, and Children supplemental nutrition program; WWE = women with epilepsy;
WWoE = women without epilepsy.
OR for WWE compares adequate or adequate-plus prenatal care in WWE to WWoE. ORs are adjusted for all covariates in table.

= 5 with age <12 years, n = 749 with age >50 years) and with prematurity, SGA, or NICU stay after adjusting for de-
including women with missing covariates (race, n = 580; mographics and epilepsy. Ideally, unintended pregnancies
ethnicity, n = 16,090; income, n = 16,548) were similar. would be lower in WWE than in WWoE, given the impor-
tance of pregnancy planning and optimization of AED med-
ication regimen and dose before pregnancy because of the
Discussion risks for MCM, low birth weight, and cognitive deficits as-
The proportion of WWE in our study population is 0.7%, which sociated with some AEDs.
falls within the accepted estimates of the prevalence of epilepsy
as 4 to 10 per 1,000 adults.14 We found that 55% of pregnancies Our findings are complementary to those from the self-
in WWE were unintended but that epilepsy was not a risk factor selecting Epilepsy Birth Control Registry that 65% of preg-
for unintended pregnancy after adjusting for age, race, ethnicity, nancies in WWE were unintended and that 78% of WWE
and SES. After adjusting for covariates, we found that WWE treated at a tertiary epilepsy center had ever had an un-
were less likely to breastfeed than were WWoE, but they had intended pregnancy.7 Because our study reports intention
higher proportions of daily preconception PNVF use. regarding only the woman’s most recent pregnancy, some
WWE may have had unintended pregnancies in the past.
Unplanned pregnancies have been associated with higher Similar to our study, the Epilepsy Birth Control Registry study
health care costs and a risk of adverse infant and maternal also found that unintended pregnancies were more common
outcomes,15 although in this study we found no association in younger WWE and racial minorities.7 Our results show that

6 Neurology | Volume , Number  | Month 0, 2018 Neurology.org/N

Copyright ª 2018 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
Table 6 Birth control methods reported at the time of conception and after delivery
Before conception After delivery

Method WWE %WWE WWoE %WWoE WWE %WWE WWoE %WWoE

Any 133/322 41.3 15,354/33,597 45.7 362/460 78.7a 47,021/56,886 82.7a

Condoms 26/37 70.3a 2,024/4,100 49.4a 56/187 30.0 7,744/22,610 34.2

Withdrawal 15/39 38.5 1,763/4,135 42.6 20/187 10.7 3,257/22,600 14.4

a
Abstinence 5/22 22.7 129/2,375 5.4* 21/187 11.2 2,185/22,602 9.7

Birth control pill 7/38 18.4 985/4,045 24.3 37/187 19.8 5,519/22,623 24.4

a
Injection 4/37 10.8 178/4,033 4.4 35/187 18.7 2,198/22,620 9.72a

Ring or patch 1/22 4.5 71/2,377 3.0 2/157 1.6 272/14,669 1.9

Rhythm/natural family planning 1/37 2.7 552/4,054 13.6 2/187 1.1a 980/22,598 4.3a

IUD 0/37 0 89/4,027 2.2 33/187 17.6 3,810/22,615 16.8

a
Implant 0/37 0 20/2025 0.5 11/187 5.9 692/22,609 3.1a

Tubal ligation 0/22 0 8/2,376 0.3 23/187 12.3a 1,833/22,610 8.1a

Abbreviations: IUD = intrauterine device; WWE = women with epilepsy; WWoE = women without epilepsy.
The number of women answering each question varied because some respondents skipped some questions.
a
p < 0.05 on χ2 comparison.

although WWE were more likely to have an unintended WWoE (84.6%). Moreover, this disparity persisted after ad-
pregnancy than were WWoE in the overall study population, justment for covariates. This may indicate that WWE are less
this association did not persist after controlling for age, race, likely to breastfeed their children because of concerns for the
and SES factors. These findings indicate that the demographic effects of AEDs and that there is a need for continued edu-
characteristics of WWE help explain the excess proportion of cation of health care providers and patients on breastfeeding
unintended pregnancies among WWE. safety in WWE. In the widely used resource Medications &
Mothers’ Milk, none of the 5 most commonly prescribed
WWE had a younger mean age than WWoE; this finding was AEDs in pregnancy17 receive the highest rating of compati-
true for both intended and unintended pregnancies. Further bility with breastfeeding, despite the studies listed above.18 In
study of factors driving the childbearing choices and the birth the NEAD study, 42.9% of WWE taking AEDs reported
control choices of WWE is needed to investigate the reasons breastfeeding at 3 months postpartum.6 The higher pro-
for this finding. Future studies should also include clinical portion of WWE reporting any breastfeeding in the current
details to allow stratification by epilepsy types and severity study is likely due to women who breastfed <3 months.
and types of AEDs used, which are likely factors in both
intended and unintended pregnancies. Folic acid is important for all women of childbearing potential,
who are advised to take at least 400 μg folic acid daily be-
Breastfeeding has numerous documented health benefits to ginning before pregnancy to reduce the risk of neural tube
both the infant and mother.16 The Norwegian Mother and defects and other adverse neonatal outcomes.19,20 This may
Child Cohort Study, a prospective study that included 223 be especially important in WWE; in the NEAD study, peri-
mothers taking AEDs (including polytherapies), found im- conceptional folic acid use in WWE was associated with
proved infant weight at 6 weeks in breastfed infants and higher child IQ at 6 years of age.3 Recent prospective data
suggested a tendency toward improved motor and social skills from Norway showed a lower risk of autistic traits in children
at 6 months in children who were breastfed for at least 6 of WWE taking AEDs if the mother took a periconceptional
months.10 The Neurodevelopmental Effects of Antiepileptic folic acid supplement,9 although definitive evidence on
Drugs (NEAD) study, a prospective multicenter observa- whether folic acid reduces MCMs in WWE is lacking. En-
tional study of women taking carbamazepine, lamotrigine, couragingly, WWE were more likely to report daily PNVF use
phenytoin, and valproate, found higher IQ and cognitive before pregnancy than were WWoE after adjustment for
abilities in breastfed children of women taking these AEDs at confounders. However, the majority (52.3%) of women
up to 6 years of age and found no adverse effects, supporting reported no PNVF use, and only 43.6% of WWE and 31.8% of
the recommendation of breastfeeding for WWE who choose WWoE reported daily PNVF use. Continued public health
to do so.6 In the current study, the proportion of 69.1% education and educational efforts by health care providers on
reporting breastfeeding among WWE was lower than that in the importance of prenatal vitamins and folic acid are needed,

Neurology.org/N Neurology | Volume , Number  | Month 0, 2018 7

Copyright ª 2018 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
 as well as improved access for women who cannot afford them, study is a survey-based, cross-sectional study, and limitations
to increase the proportion of all women meeting this guideline. of the PRAMS methodology include recall and response bias
The increased daily use of PNVF in WWE may reflect the fact (although multiple attempts are made for each selected re-
that WWE generally see a physician at least yearly for their spondent), incomplete data available for each woman (e.g.,
epilepsy and may receive extra education on the importance of prescribed medications), limited generalizability because not
folic acid use given increased teratogenic (structural and neu- all states are represented, and lack of inclusion of women who
rodevelopmental) risks in women on AEDs. have experienced a spontaneous or therapeutic pregnancy
termination or stillbirth (because only live births are in-
Prior studies have shown increased prematurity, neonatal cluded). Our specific study has limitations in that we rely on
complications, and SGA in WWE4,11,21; in particular, the use women’s self-report of epilepsy, which may inappropriately
of topiramate has been linked to SGA infants.4,22 Similarly, categorize WWE; however, the bias due to misclassification is
this study found a higher proportion of SGA, NICU stays, and expected to be toward the null.28,29 Women of childbearing
prematurity in neonates born to WWE; the association with age make up a large percentage of patients with nonepileptic
prematurity persisted after adjustment for covariates. These seizures (NES), and some may be included in WWE in this
outcomes were not the primary focus of our study, given the study. However, on the basis of the estimated prevalence of 2
lack of information on medications, number of seizures during to 33 per 100,000 of patients with NES,30 of which up to 71%
pregnancy, and other factors specific to epilepsy that may are women of childbearing age,31 and on the estimated
influence neonatal outcomes. number of WWE in the United States of 1.3 million, only 5.7%
of the women included in our WWE group are expected to
After confounders were controlled for, WWE were equally have NES; thus, our findings likely accurately represent WWE
likely to report at least adequate prenatal care, suggesting that in the included states. Another limitation is the lack of in-
epilepsy alone does not meaningfully limit women’s access to formation about whether WWE had neurology or sub-
prenatal care. A higher, but not statistically different, proportion specialty care in addition to their obstetric and primary care
of WWE reported “supra-adequate” prenatal care compared to and on specific AEDs used. In addition, the subanalysis of
WWoE. While standard “adequate” prenatal care is defined for contraceptive use was limited by small cell size.
all pregnant women, WWE may require more visits because of
a concern for intrauterine growth restriction, MCMs, AED The majority of pregnancies in WWE are unintended. However,
blood level monitoring, and seizure control. WWE often see after adjusting for age, race, ethnicity, and SES, we found no
a high-risk obstetrician and/or maternal–fetal medicine sub- increased proportion of unintended pregnancy among WWE
specialist because the risks of morbidity and mortality of compared to WWoE. The proportion of women reporting
pregnancy and delivery are elevated in WWE compared to breastfeeding is lower in WWE, while daily preconception
WWoE.23,24 The number of visits observed in this study may PNVF use is higher in WWE; these findings persisted after
also reflect a concern for conditions leading to the increased adjustment for covariates. Prospective studies of WWE are
premature births, SGA, and NICU stays in WWE we observed. needed to determine more accurate rates of unintended preg-
nancies, PNVF use, prenatal care, neonatal outcomes, and
In our study, pregnancy recognition occurred at a mean of 6.5 breastfeeding. Effective education on the importance of preg-
weeks’ gestational age in WWE, emphasizing the need for folate nancy planning and folic acid use and the risk of unintentional
supplementation and careful consideration of AED regimen, pregnancy is particularly needed in groups at higher risk. In
e.g., avoiding valproic acid when alternatives exist, in all WWE addition, continued education of WWE and health care pro-
of childbearing potential (because pregnancy may not be rec- viders who care for WWE (including neurologists, primary care
ognized before some vital embryonic development). doctors, and obstetricians) is needed regarding the safety of
breastfeeding, which has been established in long-term follow-
A higher proportion of WWE (35.8%) reported using a long- up studies of cognition in children of WWE.6
acting or permanent birth control method after delivery than
did WWoE (28.0%). The use of IUDs was similar in WWE Author contributions
and WWoE, while the use of tubal ligation and implant was Emily Johnson designed the study, obtained the data, con-
higher in WWE. IUDs are often a preferred option for WWE ducted preliminary analyses, and drafted the manuscript.
who desire a reversible form of birth control because of the Anne Burke provided critical feedback on the study design
high effectiveness, reversibility, longevity, and absence of and substantially edited the manuscript for content. Anqi
hormonal interactions with AEDs.25–27 Whether the differ- Wang performed data analyses and substantially edited the
ences in the use of long-acting birth control are due to patient manuscript for content. Page B. Pennell provided critical
choice, affordability, or counseling regarding long-acting feedback on the study design and substantially edited the
forms of birth control is unknown. manuscript for content.

The strengths of this study include the large sample size, Acknowledgment
rigorous sampling methodology, and use of a control group of The authors acknowledge the CDC and PRAMS working
WWoE. This study also has several limitations. The PRAMS group (CDC PRAMS Team, Applied Sciences Branch,

8 Neurology | Volume , Number  | Month 0, 2018 Neurology.org/N

Copyright ª 2018 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
Division of Reproductive Health) for providing access to the 12. Centers for Disease Control and Prevention. Pregnancy Risk Assessment Monitoring
System (PRAMS) [online]. 2017. Available at: cdc.gov/prams/index.htm. Accessed
PRAMS dataset used in this analysis. June 19, 2017.
13. Kotelchuck M. An evaluation of the Kessner Adequacy of Prenatal Care Index and
a proposed Adequacy of Prenatal Care Utilization Index. Am J Public Health 1994;84:
Study funding 1414–1420.
This study was supported with institutional funds (Johns 14. Duncan JS, Sander JW, Sisodiya SM, Walker MC. Adult epilepsy. Lancet 2006;367:
1087–1100.
Hopkins Department of Neurology). 15. Curtis KM, Tepper NK, Jatlaoui TC, et al. U.S. medical eligibility criteria for con-
traceptive use, 2016. MMWR Recomm Rep 2016;65:1–103.
16. Ip S, Chung M, Raman G, Trikalinos TA, Lau J. A summary of the Agency for
Disclosure Healthcare Research and Quality’s evidence report on breastfeeding in developed
The authors report no disclosures relevant to the manuscript. countries. Breastfeed Med 2009;4(suppl 1):S17–S30.
17. Wen X, Meador KJ, Hartzema A. Antiepileptic drug use by pregnant women enrolled
Go to Neurology.org/N for full disclosures. in Florida Medicaid. Neurology 2015;84:944–950.
18. Hale T. Medications & Mothers’ Milk [online]. 2018. Available at: medsmilk.com/.
Received January 24, 2018. Accepted in final form June 15, 2018. Accessed April 6, 2018.
19. Harden CL, Pennell PB, Koppel BS, et al. Practice parameter update: management
issues for women with epilepsy: focus on pregnancy (an evidence-based review):
References vitamin K, folic acid, blood levels, and breastfeeding. Neurology 2009;73:142–149.
1. Hernandez-Diaz S, Smith CR, Shen A, et al. Comparative safety of antiepileptic drugs 20. Centers for Disease Control and Prevention. Folic acid recommendations [online].
during pregnancy. Neurology 2012;78:1692–1699. Natl Cent Birth Defects Dev Disabil 2017. Available at: cdc.gov/ncbddd/folicacid/
2. Tomson T, Battino D, Bonizzoni E, et al. Dose-dependent risk of malformations with recommendations.html. Accessed September 22, 2017.
antiepileptic drugs: an analysis of data from the EURAP epilepsy and pregnancy 21. Pennell PB, Meador KJ, May R, et al. Obstetric and Neonatal Outcomes in the
registry. Lancet Neurol 2011;10:609–617. Monead Study. Washington, DC: American Epilepsy Society; 2017:1.217.
3. Meador KJ, Baker GA, Browning N, et al. Fetal antiepileptic drug exposure and 22. Hernández-Dı́az S, McElrath TF, Pennell PB, Hauser WA, Yerby M, Holmes LB. Fetal
cognitive outcomes at age 6 years (NEAD study): a prospective observational study. growth and premature delivery in pregnant women on antiepileptic drugs. Ann
Lancet Neurol 2013;12:244–252. Neurol 2017;82:457–465.
4. Veiby G, Daltveit AK, Engelsen BA, et al. Fetal growth restriction and birth defects with 23. MacDonald SC, Bateman BT, McElrath TF, et al. Mortality and morbidity during
newer and older antiepileptic drugs during pregnancy. J Neurol 2014;261:579–588. delivery hospitalization among pregnant women with epilepsy in the United States.
5. Reimers A, Brodtkorb E, Sabers A. Interactions between hormonal contraception and JAMA Neurol 2015;72:981–988.
antiepileptic drugs: clinical and mechanistic considerations. Seizure 2015;28:66–70. 24. Kapoor D, Wallace S. Trends in maternal deaths from epilepsy in the United King-
6. Meador KJ, Baker G, Browning N, et al. Breastfeeding in children of women taking dom: a 30-year retrospective review. Obstet Med 2014;7:160–164.
antiepileptic drugs: cognitive outcomes at age 6 years. JAMA Pediatr 2014;168: 25. Davis AR, Saadatmand HJ, Pack A. Women with epilepsy initiating a progestin
729–736. IUD: a prospective pilot study of safety and acceptability. Epilepsia 2016;57:
7. Herzog AG, Mandle HB, Cahill KE, Fowler KM, Hauser WA. Predictors of un- 1843–1848.
intended pregnancy in women with epilepsy. Neurology 2017;88:728–733. 26. Espinera AR, Gavvala J, Bellinski I, et al. Counseling by epileptologists affects con-
8. Davis AR, Pack AM, Kritzer J, Yoon A, Camus A. Reproductive history, sexual behavior traceptive choices of women with epilepsy. Epilepsy Behav 2016;65:1–6.
and use of contraception in women with epilepsy. Contraception 2008;77:405–409. 27. Pennell PB. Pregnancy, epilepsy, and women’s issues. Continuum 2013;19:697–714.
9. Bjørk M, Riedel B, Spigset O, et al. Association of folic acid supplementation during 28. Wacholder S, Hartge P, Lubin JH, Dosemeci M. Non-differential misclassification and
pregnancy with the risk of autistic traits in children exposed to antiepileptic drugs in. bias towards the null: a clarification. Occup Environ Med 1995;52:557–558.
utero. JAMA Neurol 2018;75:160–168. 29. Höfler M. The effect of misclassification on the estimation of association: a review. Int
10. Veiby G, Engelsen BA, Gilhus NE. Early child development and exposure to anti- J Methods Psychiatr Res 2005;14:92–101.
epileptic drugs prenatally and through breastfeeding: a prospective cohort study on 30. Benbadis SR, Allen Hauser W. An estimate of the prevalence of psychogenic non-
children of women with epilepsy. JAMA Neurol 2013;70:1367–1374. epileptic seizures. Seizure 2000;9:280–281.
11. Razaz N, Tomson T, Wikstrom AK, Cnattingius S. Association between pregnancy and 31. Dworetzky BA, Baslet GC. Psychogenic nonepileptic seizures. In: Toward the In-
perinatal outcomes among women with epilepsy. JAMA Neurol 2017;74:983–991. tegration of Care. 1st ed. Oxford: Oxford University Press; 2017.

Neurology.org/N Neurology | Volume , Number  | Month 0, 2018 9

Copyright ª 2018 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
Unintended pregnancy, prenatal care, newborn outcomes, and breastfeeding in women
with epilepsy
Emily L. Johnson, Anne E. Burke, Anqi Wang, et al.
Neurology published online August 10, 2018
DOI 10.1212/WNL.0000000000006173

This information is current as of August 10, 2018

Updated Information & including high resolution figures, can be found at:
Services http://n.neurology.org/content/early/2018/08/10/WNL.0000000000006
173.full
Subspecialty Collections This article, along with others on similar topics, appears in the
following collection(s):
All Clinical Neurology
http://n.neurology.org/cgi/collection/all_clinical_neurology
All epidemiology
http://n.neurology.org/cgi/collection/all_epidemiology
All Epilepsy/Seizures
http://n.neurology.org/cgi/collection/all_epilepsy_seizures
Risk factors in epidemiology
http://n.neurology.org/cgi/collection/risk_factors_in_epidemiology
Permissions & Licensing Information about reproducing this article in parts (figures,tables) or in
its entirety can be found online at:
http://www.neurology.org/about/about_the_journal#permissions
Reprints Information about ordering reprints can be found online:
http://n.neurology.org/subscribers/advertise

Neurology ® is the official journal of the American Academy of Neurology. Published continuously since
1951, it is now a weekly with 48 issues per year. Copyright © 2018 American Academy of Neurology. All
rights reserved. Print ISSN: 0028-3878. Online ISSN: 1526-632X.