Epigraph Vol. 25 Issue 2, Spring 2023

Postpartum morbidity and mortality in women with epilepsy: A 28-year study

Reported by Bruna Nucera | Produced by Nancy Volkers

Listen below or download the episode.

Find Sharp Waves episodes on SpotifyApple PodcastsGoogle Podcasts, or Amazon Music.

Dr. Bruna Nucera interviewed Dr. Andrea Cheng-Hakimian about a retrospective US study including 28 years of data from Washington State. The study found increased risks of preeclampsia, gestational diabetes, ICU admission during hospitalization for delivery, and postpartum hospitalization, as well as an increased risk of mortality, compared with pregnant women who did not have epilepsy.


 Sharp Waves episodes are meant for informational purposes only, and not as clinical or medical advice.

Podcast Transcript

Dr. Bruna Nucera: Thank you for accepting our invitation for this podcast interview.

Dr. Andrea Cheng-Hakimian: Thank you. It’s my pleasure and my honor and I’m happy to be talking with you today.

Dr. Nucera: Thank you. As you know we’re talking about a recent publication in Epilepsy & Behavior. The title is “Morbidity and rehospitalization postpartum among women with epilepsy and their infants: A population-based study.” Please introduce yourself.

Headshot of Andrea Cheng-Hakimian
Andrea Cheng-Hakimian (USA)

Dr. Cheng-Hakimian: So my name is Andrea Cheng-Hakimian. I’m a clinical associate professor of neurology at the University of Washington in Seattle, Washington, United States. I am particularly interested within the field of epilepsy in women’s health in epilepsy, and I came upon that interest because when I finished my fellowship, I was one of a handful of female epileptologists in the United States. And people who wanted a female practitioner to take care of them during their pregnancy were limited in who they could contact. And from the other end, undergoing my own pregnancy at the end of my own training, I felt a special interest in all things related to that. 

Currently, I have of course my clinic in general epilepsy unit, but also I have a clinic in the department of OB/GYN where I see patients who are pregnant who also have epilepsy, and this is the clinic that I particularly enjoy and a population that I love working with because I feel like everything that I do, I’m rewarded twice over – once for the mom and once for the baby.

Dr. Nucera: Thank you. Could you provide some background regarding this topic and talk about why this work was done?

Dr. Cheng-Hakimian: Previous studies in the literature have identified a whole range of adverse outcomes in these women, such as eclampsia, premature rupture of the membranes, preterm labor, placental abruption, C-section, et cetera, et cetera. These studies sometimes conflict and are not consistent. There’s a possibility that the findings have shifted over time as AED prescription patterns have shifted. So we wanted to form a study that would further answer these controversies but also I think one of the strengths of our study is that we are a population-based study, so we could have nearly 100% capture of these patients and give a more global answer to some of these issues.

Dr. Nucera: Could you please describe the study methods? How were the data extracted and evaluated?

Dr. Cheng-Hakimian: Of course, I’d be glad to. Our study is a retrospective and population-based cohort study. In the United States, since the late 1980s, all birth records and fetal death records have been linked to hospital discharge records. And we identified all delivery records in Washington state from 1987 to 2014 and linked them to hospital discharge records. 

From that data, we pulled information on all women with epilepsy, with ICD-9 codes consistent with epilepsy, and got their hospital discharge records. We identified a control population of randomly selected women with no epilepsy, matched by year of delivery, in a 10:1 ratio, so there’s 10 control women for every 1 study woman. We looked at hospitalization, rehospitalization records for the first 2 years after delivery among both groups, and we also identified maternal and infant death records in both groups.

In the end we had a little under 3,000 women with epilepsy—2,714 women with epilepsy—and 27,062 women without epilepsy in the control group. For the analysis, we looked at maternal outcomes, identified through hospital discharge and birth records. These are all outcomes that have unique value and are of previous concern of women and their caretakers, such as pre-eclampsia, gestational diabetes, placental abruption, preterm labor, labor induction, malpresentation, C section, post-partum hemorrhage, ICU admission, length of hospitalization, and death. We also looked at epilepsy type; we had ICD-9 codes focusing on focal vs generalized, tractable vs intractable, and convulsive vs non-convulsive epilepsy.

The infant outcomes include major congenital malformation, fetal distress, meconium aspiration, low birthweight, preterm delivery, small for gestational age, 5-minute APGAR less than 7, requiring assisted ventilation, breastfeeding rate and NICU admission. We also looked at rehospitalization and deaths, and time to rehospitalization. 

We then identified the relative risk ratios between these two groups for each adverse outcome using relative risk ratios with a 95% confidence interval with Poisson regression analysis. Adjustments include the year of delivery, but we also adjusted for maternal age and parity. We looked at the use of tobacco, race, and socioeconomic status markers, such as education and insurance type, but those factors did not alter results, so we did not include them in the analysis. 

And we conducted analyses separately for maternal and infant outcomes for each epilepsy type. And for birth year categorization. So for birth year we put that into three separate bins, the first bin is from 1987 to 1996, when the US prescribing patterns were predominantly for phenytoin, carbamazepine and valproic acid. From 1997 to 2006, which was more of a transition period, and also when the second-generation anti-epileptic drugs (AEDs) start to be prescribed. And then the third bin is 2007 to 2014, when prescription patterns in the US show predominantly lamotrigine and levetiracetam.

Dr. Nucera: What are the main findings of the study?

Dr. Cheng-Hakimian: The reassuring thing that I love to see was that most women with epilepsy, in the order of high 90%s, for most of the outcomes, did not have any serious adverse outcomes. And their infants did very well long term, with no increased risk of mortality. But there definitely were some adverse outcomes. If I may I’d like to talk about it in two lumps, and one lump was for me what were the expected and not terribly surprising adverse outcomes, and then I’d like to talk about those and then I’d like to talk about what surprised me and what immediately changed my practice.

So first to just quickly talk about the expected adverse outcomes. We found consistent with prior studies that infants of women with epilepsy had modestly increased risk, and by modestly, I mean the relative risk ratios were between 1 and 2, for longer hospitalization, increased major malformations, low birthweight, small for gestational age, requirement for assisted ventilation, and NICU admission, and were less likely to be breastfed. The moms had increased risk of pre-eclampsia and eclampsia, gestational diabetes, preterm rupture of membranes, preterm labor, labor induction, C-section, ICU admission, longer hospitalizations, greater rehospitalizations, and mortality risk, which I’ll talk about again.

There’s generally greater risk for adverse outcomes for convulsive vs non-convulsive seizures, and intractable vs tractable seizures. Unfortunately, our numbers are too small to draw a strong statistical conclusion for these categorizations—for example, 90% of our women had tractable epilepsy and 80% were not classified as focal vs generalized. So we cannot draw a strong conclusion on that analysis.

But there was also a temporal trend that was not significant toward decreasing risk of major malformations and gestational diabetes over time, that I think would be consistent with what we know about prescription patterns. There was also increased risk of preterm labor, low birthweight, and preterm delivery over that time, increasing, so that would be something we could look at in more detail in another study.

So I’m going to transition now if I may to the findings that surprised me and changed my clinical practice. Far and away the number-one thing that surprised me was that women with epilepsy have a 7 times higher risk of mortality around their pregnancy and delivery, up to 2 years post-partum. Seven times. Let me give you some specific numbers. In our group of 2,708 women with epilepsy we had 6 deaths. In our control group of 27,054 women we had 8 deaths. So this was very much a statistically significant relative risk of 7.11, with a 95% confidence interval, the lower end of which is 2.47.

Dr. Nucera: How does this elevated mortality rate compare with findings from other studies? We spoke with a clinician and researcher in the field who was not involved in this publication.

Headshot of Jakob Christensen
Jakob Christensen (Denmark)

Dr. Jakob Christensen: My name is Jakob Christensen, I’m a consultant at Aarhus University in Denmark. I’m also a clinical professor in epilepsy, and a clinical specialist in clinical pharmacology.

The finding of the relatively high mortality in in women with epilepsy is not a new finding. It’s actually similar in size to what has been described in the US population and also in the UK population, and also in a smaller study that we did in Denmark. There seems to be an increased mortality during pregnancy and during delivery and now this finding extends 2 years after birth. Usually you would say maternal mortality, that would be during pregnancy and the first 42 days after delivery, so that’s the [World Health Organization] definition of maternal mortality. 

We studied this in the past, in a smaller Danish study and women who are in the fertile age, so not necessarily those who give birth but those who are in the same age category, they actually have a much higher mortality compared to the background population, around 16 times increased risk. So actually, the women that become pregnant have a lower relative mortality than those in the general population of persons with epilepsy at that age, if that makes sense.

So it shows that women with epilepsy who become pregnant are healthier than those who do not become pregnant. When you think about it, it’s not surprising that those who become pregnant or decide to have a child are healthier than those who do not, but women in this age group with epilepsy have a relatively high mortality.

Dr. Cheng-Hakimian: So what could be causing this? I have two hypotheses. The one that jumps into all of our minds right now is SUDEP (sudden unexpected death in epilepsy). In the US, the latest SUDEP rates are cited as 1.16 per 1,000 person-years. Using that statistic we’d expect 3 deaths in our study among women with epilepsy. So at a minimum, SUDEP could possibly explain a giant portion of the deaths reported in our study.

There was a paper coming from the UK in 2014 that reviewed 30 years’ worth of maternal death records in the UK, and that paper identified SUDEP as the biggest cause of maternal mortality among women with epilepsy, followed by aspiration of gastric contents during a seizure and drowning during bathing. And to any epileptologist we look at that and think, “Oh, these are all consequences of inadequate seizure control.” 

For the support of that concern, in the UK study, most women with epilepsy that had anti-epileptic drug levels checked at the time of death had levels that were subtherapeutic or even undetectable.

I think it’s reasonable to hypothesize that a good percentage of the deaths come from uncontrolled seizures, and that we as practitioners need to be aware of this high risk of mortality, not only during pregnancy but during the first year or two after delivery. At least here in the United States, there is almost a belief that epilepsy during pregnancy is not too bad, it’s relatively benign compared with other comorbidities. And I think that attitude needs to be more informed as to the severity of epilepsy during pregnancy. 

Another question that I think is, do moms with epilepsy need closer follow up post-partum to maintain that tight seizure control? Routinely, women with epilepsy post-partum will have a six-week follow-up and then revert to yearly follow-ups as dictated by seizure control. But with this information, I have changed my clinical practice so that I contact the mother at 2 weeks, 4 weeks, 3 months, 6 months, and a year. And I coordinate with the OB/GYN, who typically sees the mom at 3 weeks and 3 months. So that there is extensive medical contact during this time, and any issues can be addressed before they become serious problems. Would this sort of approach reduce mortality and serous morbidity in the population? I think that’s something that should be studied.

Another thought I had, or we had, I can’t take credit for this myself, as far as why there is such an elevated maternal mortality, is the question of post-partum depression. I think all of us in our practice have seen that there is a high risk of depression among people with epilepsy in general, and when you compound that with the risk of post-partum depression, does that cause higher rates of post-partum depression or more severe post-partum depression? Is there an increased risk of suicide attempts? These have not been looked at and should be investigated further.

 Another finding that correlates with this pattern is that there is five times the risk of maternal ICU admission during delivery for women with epilepsy. It’s possible this reflects increased risk of status epilepticus of course, but I also wonder if there are effects of AEDs on cardiovascular health or on mental health that are causing these ICU admissions, or is it a reflection of possibly lower health status or socioeconomic determinants of health in this population? 

These concerns also play into the last finding I’ll talk about, which is that there’s an increased risk of rehospitalization. The relative risk of this was 2.34 for 2 or more rehospitalizations, with the highest relative risk of rehospitalization in the first year post-partum. Does this reflect for instance the reverting pharmacokinetics of AEDs post-partum? As a concrete example, someone who’s taking a higher dose of lamotrigine during pregnancy may not know to decrease their doses post-partum, become toxic, have some ataxia, fall, injure themselves, end up in the hospital. Could more information about peri-partum management decrease hospitalizations?

As a general summary, what I learned in my clinical practice is that women with epilepsy need just as much care in the year after delivery as they do during pregnancy. Patients need to be aware of that increased risk and take their own health care seriously. I think any mom or any dad who’s been through the neonatal period will know that often adult personal care falls by the wayside to take care of the baby, and perhaps we need to emphasize to these moms that you have to take time out to participate in your health care, get your blood draws, see your doctor, prioritize your own sleep habits and your own diet and nutrition so that you can remain seizure free as a first priority. 

I think there’s also further work to be done around breastfeeding. It seems that women with epilepsy are choosing not to breastfeed at higher rates, likely due to concern for AEDs, adverse effects on outcomes in their babies. So we probably should be doing more counseling and more reassurance in that area.

Dr. Nucera: Thank you so much. Could you briefly discuss the study’s strengths and limitations?

Dr. Cheng-Hakimian: Of course. I think our biggest strength is our large sample size and we have population-based data with nearly 100% capture. We also have a relatively long follow-up of 2 years, that’s fairly rare for these types of studies in this area. And we performed a sub-study where we reviewed the charts of all women with epilepsy who delivered at a certain hospital within the state of Washington, and that suggested we did indeed capture all women with epilepsy with deliveries in our state. 

Our study does have limitations however, and one of them is that we depended on ICD-9 coding. We’re depending on coding accuracy and at the same time we have no indicators of severity from the ICD-9 code. We don’t have indicators of disease duration, drug compliance or drug dose. And we don’t have any information on which AEDs these women were using. We also don’t have any outpatient data, and if there were any women who moved out of state before the delivered, or any homebirths that did not end up in this hospital, these women were not captured. 

I’m also questioning whether or not our statistical control for socioeconomic status is adequate; in my clinical practice I see every day the adverse influence of lower socioeconomic status, and I believe that is a big factor in the poorer outcomes, but I think that a bigger study would need to be performed before we could draw conclusions on that.