A nationwide analysis of maternal morbidity and acute postpartum readmissions in women with epilepsy

Abstract

Objective:

To compare maternal delivery hospitalization characteristics and postpartum outcomes in women with epilepsy (WWE) versus women without common neurological comorbidities.

Methods:

We performed a retrospective cohort analysis of index characterizations and short-term postpartum rehospitalizations after viable delivery within the 2015–2017 National Readmissions Database using International Classification of Diseases, Tenth Revision codes. Wald chi-squared testing compared baseline demographic, hospital and clinical characteristics and postpartum complications between WWE and controls. Multivariable logistic regression models examined odds of non-elective readmissions within 30 and 90 days for WWE compared to controls (alpha = 0.05).

Results:

A total of 38,518 WWE and 8,136,335 controls had a qualifying index admission for delivery. Baseline differences were most pronounced in Medicare/Medicaid insurance (WWE: 58.2%, controls: 43%, p < 0.0001), alcohol/substance abuse (WWE: 8.3%, controls: 2.5%, p < 0.0001), psychotic disorders (WWE: 1.2%, controls 0.1%, p < 0.0001), and mood disorder (WWE:15.5%, controls: 3.7%, p < 0.0001). At the time of delivery, WWE were more likely to have edema, proteinuria, and hypertensive disorders (WWE: 19%, controls: 12.9%, p < 0.0001); a history of recurrent pregnancy loss (WWE: 1%, controls: 0.4%, p < 0.0001); preterm labor (WWE: 7.3%, controls: 4.8%, p < 0.0001), or presence of any Center for Disease Control severe maternal morbidity indicator (WWE: 3.2%, controls: 0.6%, p < 0.0001; AOR 5.16, 95% CI 4.70–5.67, p < 0.0001). A higher proportion of WWE were readmitted within 30 days (WWE: 2.4%, controls: 1.1%) and 90 days (WWE: 3.7%, controls: 1.6%). After adjusting for covariates, the odds of postpartum non-elective readmissions within 30 days (AOR 1.86, 95% CI 1.66–2.08, p-value < 0.0001) and 90 days (AOR 2.04, 95% CI 1.83–2.28, p-value < 0.0001) were higher in WWE versus controls.

Interpretation:

WWE experienced critical obstetric complications and higher risk of severe maternal morbidity indicators at the time of delivery. Although relatively low, non-elective short-term readmissions after delivery were higher in WWE than women without epilepsy or other common neurological comorbidities. Further research is needed to address multidisciplinary care inconsistencies, improve maternal outcomes, and provide evidence-based guidelines.

1. Introduction:

Studies have demonstrated that WWE are at significant risk for adverse obstetric outcomes during pregnancy and delivery, such as preeclampsia, preterm labor, inviable deliveries, fetal hypoxia at birth, and maternal death.[1–5] Downstream effects include increased healthcare utilization and prolonged length of stay.[1,4] However, several potentially important ensuing outcomes remain understudied and underreported in literature. Accurate measurements to gauge maternal morbidity are challenging due to insufficient monitoring and missed documentation. Literature on postpartum outcomes in WWE remains especially sparse. Unplanned maternal readmissions after delivery are costly, disruptive to maternal-newborn care, and potentially preventable. Knowing whether national rates of postpartum readmissions are significantly greater in WWE is critical to support further development and refine recommendations for pregnancy care, management, and maternal-fetal heath research in WWE.[1,6] Furthermore, understanding factors contributory to these readmissions could allow identification of the subpopulation at greatest risk to be readmitted and create appropriate interventions.

To address this gap in data and inform maternal-fetal outcomes and healthcare services research, we produced national estimates of patient-hospital characteristics and short-term non-elective readmissions after a viable delivery in WWE versus women without common neurological diseases utilizing the National Readmissions Databases (NRD) to provide benchmark readmissions data. We hypothesized that WWE would have higher short-term readmission rates compared to controls and distinct differences in demographic, hospital, pregnancy-related and medical characteristics between the cohorts would be identified.

2. Materials and Methods:

Standard protocol approvals, registrations, and patient consents:

The Institutional Review Board at the University of Pennsylvania reviewed the study protocol, and a waiver of patient consent was granted due to the deidentified nature of the data. Our study design and reporting are also in accordance with the data use agreement between the authors and the Healthcare Cost and Utilization Project.

2.1. Study design:

We performed a retrospective cohort study of short-term, postpartum, non-elective rehospitalizations in the 2015–2017 National Readmissions Database (NRD) limited to ICD-10 criteria.

2.2. Database:

As part of the Healthcare Cost and Utilization Project (HCUP), the National Readmissions Database is a population-wide administrative claims database that provides information on inpatient readmissions from community and academic hospitals, containing data pertaining to 36 million weighted discharges. Sample weighting provided by the HCUP-NRD allows for national discharges estimates and accounts for approximately 60 percent of total U.S. hospitalizations, regardless of payer type.[7] Patients admitted to these qualifying hospitals may be followed longitudinally within respective calendar years but not across state boundaries, due to the state-level acquisition of data.

Diagnoses were captured using the International Classification of Diseases, Tenth Revision, Clinical Modification/Procedure Coding System to evaluate the most recently available data.[8] All diagnoses codes were captured using previously validated ICD-9-CM codes or Clinical Classification Software codes (CCS) when available and translated to equivalent ICD-10 coding as necessary using Healthcare Cost and Utilization Project (HCUP) publicly available software tools.[9] The diagnoses codes used in this study are listed in supplementary table 1.

2.3. Study Population

Our study cohort included all women ages 12–55 discharged after an index admission for a viable delivery. WWE (G40.XX, G41.XX)[10–12] were compared to women without common neurological comorbidities of childbearing age [to include epilepsy, migraine (G43.XX), pre-existing stroke (Z86.73, I69.0X- I69.3X, I69.8X, I69.9X), myasthenia gravis (G70.0X), and multiple sclerosis (G35.XX)[12]], herein referred to as controls. These neurological comorbidities were specifically excluded in our control cohort due to similarly observed obstetric complications and/or therapeutic medications that could act as effect modifiers or mediate outcomes.[13–17]

Patients with encounters for the following were excluded: 1) Missing baseline variables (e.g., undocumented age, primary payer, median zip code income, etc.) and 2) Index hospitalizations which resulted in non-viable deliveries[18] which resulted in respective cohorts eligible for readmission. Patients were further excluded if 3) Index hospitalizations ended in maternal mortality 4) Index hospitalizations occurred after December and October of each calendar year for 30-day and 90-day readmissions, respectively, due to insufficient follow up time and 5) Individuals were not a resident of the state in which they were treated. Details of the study populations are illustrated in Figure 1.

Figure 1:

Flow chart of study population

2.4. Outcomes:

Our primary outcome were the rates of any non-elective readmissions within 30 days and 90 days after discharge from index hospitalization for delivery, identified by the primary or secondary diagnosis. The 30-day time point was selected as a commonly used as a metric in readmission quality measures. We chose to examine up to 90 days to encompass the postpartum period after delivery. Only an initial readmission was considered for patients with multiple readmissions within this time frame. Secondary outcomes of interest included: 1) demographic, hospital, clinical characteristics, and obstetric complications in WWE compared to controls at index delivery hospitalization and 2) readmission indications.

2.5. Statistical Analysis

We performed descriptive analyses on baseline patient and hospital characteristics for each cohort at the time of index delivery, 30-day, and 90-day readmissions. Wald chi-squared test was used to compare baseline demographic, hospital, clinical characteristics and obstetric complications between cohorts. Readmission rates for any non-elective readmission within 30 days and 90 days of index delivery hospitalization discharge were calculated.

Logistic regression was used to calculate adjusted odds ratio and 95% CI for all binary outcomes. Multivariable logistic regression models, adjusted for all baseline patient characteristics (listed in table 1 and supplementary table 2) and hospital variables (listed in table 2), examined the odds of 30-day and 90-day non-elective readmissions between WWE and controls. All hypothesis testing was two-sided and statistically significant if p < 0.05, and all analyses used survey weights to generate nationally weighted results as well as appropriate estimates of variance. Statistical analysis was performed using the SAS system software version 9.4 (SAS Institute Inc, Cary, NC).

Table 1:

Baseline demographics at index delivery hospitalization

	Epilepsy	Control	p-value1
Variable	N (%)	N (%)
Age, years (Median, IQR)	27.28, 23.10–31.70	28.16, 23.82–32.35	<0.0001
Age group
12–18	1167 (3)	261338 (3.2)	<0.0001
19–24	10442 (27.1)	1845128 (22.7)
25–39	11676 (30.3)	2382344 (29.3)
30–34	9484 (24.6)	2279328 (28)
35–39	4779 (12.4)	1123029 (13.8)
40–55	970 (2.5)	245168 (3)
Primary payer
Medicare	1761 (4.6)	61804 (0.8)	<0.0001
Medicaid	20647 (53.6)	3436281 (42.2)
Private insurance	14664 (38.1)	4286157 (52.7)
Self-Pay/no-charge/other	1446 (3.8)	352092 (4.3)
Median household income by ZIP code			<0.0001
$1 – $37,999	13620 (35.4)	2291910 (28.2)
$38,000 – $47,999	10358 (26.9)	2105710 (25.9)
$48,000 – $63,999	8769 (22.8)	2042420 (25.1)
$64,000+	5773(15)	1696294 (20.8)

¹Wald chi-squared testing.

Table 2:

Hospital characteristics at index delivery hospitalization

	Epilepsy	Control	p-value1
Variable	N (%)	N (%)
Hospital ownership
Government, nonfederal	4880 (12.7)	888561 (10.9)	<0.0001
Private, not-for-profit	29770 (77.3)	6288260 (77.3)
Private, for-profit	3868 (10)	959513 (11.8)
Hospital bed size
Small	5676 (14.7)	1309465 (16.1)	<0.0001
Medium	10262 (26.6)	2368153 (29.1)
Large	22581 (58.6)	4458715 (54.8)
Teaching status
Metropolitan non-teaching	7726 (20.1)	1913283 (23.5)	<0.0001
Metropolitan teaching	27289 (70.8)	5449629 (67)
Non-metropolitan	3504 (9.1)	773422 (9.5)
Length of stay (Median, IQR)
	2.10, 1.45–2.89	1.85, 1.31–2.62	<0.0001

¹Wald chi-squared testing.

2.6. Data Availability:

The full dataset is available through HCUP.

3. Results:

3.1. Baseline demographic and hospital characteristics

A total of 38,518 WWE and 8,136,335 controls met a qualifying index admission for delivery and were eligible for non-elective readmission. Figure 1 details the study cohort. Baseline demographic and hospital characteristics at the time of index admission are summarized in Tables 1 and 2.

3.2. Clinical characteristics

Specific medical comorbidities were assessed due to their increased prevalence with epilepsy and/or increased likelihood of maternal morbidity and mortality at index delivery hospitalization.[1,19] WWE more frequently had a diagnosis of pre-existing diabetes mellitus (WWE: 2.1%, controls: 1%, p < 0.0001), kidney disease (WWE: 0.3%, controls: 0.1%, p < 0.0001) and hypertension (WWE: 6%, controls: 3.1%, p < 0.0001). The most marked differences between WWE and controls related to psychiatric comorbidities and substance abuse. Differences included alcohol/substance abuse (WWE: 8.3%, controls: 2.5%, p < 0.0001), psychotic disorders (WWE: 1.2%, controls 0.1%, p < 0.0001), mood disorders (WWE:15.5%, controls: 3.7%, p < 0.0001), and anxiety and other nonpsychotic mental disorders (WWE:13%, controls: 3.7%, p < 0.0001). Full results are summarized in Table 3, and adjusted odds ratios are illustrated in Figure 2.

Table 3:

Clinical characteristics at index delivery hospitalization

	Epilepsy	Control	p-value1
Clinical Characteristics	N (%)	N (%)
Multiple Gestations	700 (1.8)	143321 (1.8)	0.6141
Previous Cesarean	4170 (10.8)	806102 (9.9)	0.0004
Pre-existing Diabetes Mellitus	802 (2.1)	80184 (1)	<0.0001
Pregnancy-induced Diabetes	2553 (6.6)	591080 (7.3)	0.0011
Kidney Disease	126 (0.3)	7187 (0.1)	<0.0001
Hypertension	2319 (6)	252965 (3.1)	<0.0001
Alcohol/Substance abuse	3195 (8.3)	203130 (2.5)	<0.0001
History of recurrent pregnancy loss	302 (0.8)	31833 (0.4)	<0.0001
Schizophrenia, schizotypal, delusional, and other non-mood psychotic disorders	460 (1.2)	9199 (0.1)	<0.0001
Mood disorders	5970 (15.5)	298627 (3.7)	<0.0001
Anxiety, dissociative, stress-related, somatoform and other nonpsychotic mental disorders	5004(13)	301797 (3.7)	<0.0001
Specific personality disorders	210 (0.5)	4100 (0.1)	<0.0001
Eating disorders	29 (0.1)	2283 (0)	0.0557

¹Wald chi-squared testing.

Figure 2:

Forest plot of adjusted odds ratios of obstetric and clinical characteristics in WWE versus controls

3.3. Pregnancy characteristics and obstetric complications at index admission

Pregnancy characteristics and obstetric complications were selected based on literature demonstrating an increased association with epilepsy and/or peripartum morbidity and mortality, which included the validated and widely used Center for Disease Control (CDC) severe maternal morbidity indicators (SMM).[1,4,14,19–22]. Frequencies of pregnancy characteristics and obstetric complications are summarized in Table 4 (comprehensive listing of assessed variables is included in supplementary table 2). At the time of delivery, WWE were more likely to have edema, proteinuria, and hypertensive disorders (WWE: 19%, controls: 12.9%, p < 0.0001); a history of recurrent pregnancy loss (WWE: 1%, controls: 0.4%, p < 0.0001); and experience preterm labor (WWE: 7.3%, controls: 4.8%, p < 0.0001). WWE were more commonly supervised as a high-risk pregnancy for any reason (WWE: 13.3%; controls: 11.7%, p < 0.0001). WWE were also more frequently to receive maternal care for known or suspected poor fetal growth (WWE: 5.8%, controls: 3.4%, p < 0.0001). WWE more commonly had the presence of any Centers for Disease Control (CDC) severe maternal morbidity indicator (SMM)[20,23,24] (WWE: 3.2%, controls: 0.6%, p < 0.0001; AOR 5.16, 95% CI 4.70–5.67, p < 0.0001).

Table 4:

Obstetric outcomes at index delivery hospitalization^*

	Epilepsy	Control	p-value1
Obstetrical Complication	N (%)	N (%)
Edema, proteinuria and hypertensive disorders	7324 (19)	1052977 (12.9)	<0.0001
High risk pregnancy, any reason	5115 (13.3)	953744 (11.7)	<0.0001
High risk pregnancy due to history of preterm labor	365 (0.9)	36768 (0.5)	<0.0001
High risk pregnancy due to insufficient antenatal care or social problems	1238 (3.2)	143432 (1.8)	<0.0001
Maternal care for other known or suspected poor fetal growth	2242 (5.8)	273721 (3.4)	<0.0001
Preterm labor	2793 (7.3)	389754 (4.8)	<0.0001
CDC Severe Maternal Morbidity Indicators
Acute renal failure	123 (0.3)	8049 (0.1)	<0.0001
Adult respiratory distress syndrome	257 (0.7)	5707 (0.1)	<0.0001
Eclampsia	642 (1.7)	6830 (0.1)	<0.0001
Presence of any maternal morbidity indicator	1232 (3.2)	45606 (0.6)	<0.0001

^*See supplementary table 2 for comprehensive list of all obstetric variables assessed.

¹Wald chi-squared testing.

3.4. Readmission Rates

As illustrated in Table 5, higher maternal age of 35 years or older was associated with readmission for WWE within 30 days, and maternal age of 40 years or older for readmission within 90 days. For WWE, primary payer government funded insurance was associated with increased odds of readmission within 30 days (Medicare: AOR 2.18, 95% CI 1.97–2.42, p < 0.0001; Medicaid: AOR 1.36, 95% CI 1.32–1.39, p < 0.0001) and 90 days (Medicare: AOR 2.69, 95% CI 2.42–2.43, p < 0.0001; Medicaid: AOR 1.48, 95% CI 1.45–1.52, p < 0.0001). WWE who resided in the zip income quartiles below $64,000 were more likely to be readmitted within 30 and 90 days, most frequently in the lowest quartile ($1 – $37,999, 30 days: AOR 1.37, 95% CI 1.31–1.43, p < 0.0001; 90 days: AOR 1.36, 95% CI 1.30–1.41, p < 0.0001). Patients treated at non-metropolitan hospitals had lower odds to be readmitted when compared to metropolitan teaching hospitals (30 days: AOR: 0.67, 95% CI 0.63–0.71, p < 0.0001; 90 days: AOR: 0.60, 95% CI 0.65–0.73, p < 0.0001). Given psychiatric comorbidities were markedly different between WWE and controls, we ran a multivariable logistic analysis to assess for interaction between psychiatric comorbidities and 30-day readmissions (utilizing CCS codes listed in supplementary table 3). We found no significant difference between adjusted odds ratio for 30-day readmission between the two groups (p = 0.12).

Table 5:

Adjusted odds ratio of readmission in WWE versus controls within 30 and 90 days

Characteristic	30-day AOR (95% CI)	p-value	90-day AOR (95% CI)	p-value
WWE	1.86 (1.66–2.08)	<0.0001	2.04 (1.83–2.28)	<0.0001
Age
12–18	Ref		Ref
19–24	0.96 (0.9–1.03)	0.2088	0.95 (0.89–1.01)	0.0764
25–39	0.95 (0.89–1.01)	0.1212	0.87 (0.82–0.92)	<0.0001
30–34	1.05 (0.98–1.13)	0.1668	0.88 (0.83–0.94)	0.0001
35–39	1.33 (1.24–1.43)	<0.0001	1.06 (1–1.14)	0.0633
40–55	1.66 (1.53–1.81)	<0.0001	1.31 (1.21–1.41)	<0.0001
Primary Payer
Medicare	2.18 (1.97–2.42)	<0.0001	2.69 (2.42–3)	<0.0001
Medicaid	1.36 (1.32–1.39)	<0.0001	1.48 (1.45–1.52)	<0.0001
Private insurance	Ref		Ref
Self-Pay/no-charge/other	1.03 (0.96–1.1)	0.3885	1.08 (1.01–1.14)	0.0222
ZIP Income Quartile
$1 – $37,999	1.37 (1.31–1.43)	<0.0001	1.36 (1.3–1.41)	<0.0001
$38,000 – $47,999	1.21 (1.16–1.26)	<0.0001	1.25 (1.2–1.3)	<0.0001
$48,000 – $63,999	1.12 (1.07–1.16)	<0.0001	1.13 (1.09–1.18)	<0.0001
$64000+	Ref		Ref
LOS	1.05 (1.05–1.06)	<0.0001	1.05 (1.05–1.06)	<0.0001
Hospital Teaching Status
metropolitan non-teaching	0.86 (0.82–0.89)	<0.0001	0.89 (0.86–0.93)	<0.0001
metropolitan teaching	Ref		Ref
non-metropolitan	0.67 (0.63–0.71)	<0.0001	0.69 (0.65–0.73)	<0.0001
Year	1.04 (1.01–1.07)	0.0147	1.04 (1–1.08)	0.0592

The 30-day non-elective readmission rate was 2.4% in WWE versus 1.1% in controls. Controlling for baseline demographic, hospital and clinical differences between cohorts, the odds of post-partum non-elective readmission were higher in WWE (AOR 1.86, 95% CI 1.66–2.08, p < 0.0001). The 90-day non-elective readmission rate was 3.7% in WWE versus 1.6% in controls. After covariate adjustments, the odds of post-partum non-elective readmission were higher in WWE (AOR 2.04, 95% CI 1.83–2.28, p < 0.0001).

Most women requiring non-elective readmission within 30 and 90 days of discharge were readmitted for puerperium affecting management (WWE: 54.1%; controls: 61% and WWE: 36.1%; controls: 48% within 30 and 90 days, respectively). Hypertension complicating pregnancy, childbirth and the puerperium was the second most common indication for 30-day readmission in both cohorts (WWE: 17.5%; controls: 22.7%), and 90-day readmission for controls (16%). Notably, 3.5% of WWE were readmitted for epilepsy/convulsions within 30 days; however, this frequency increased to 10.6% within 90 days. Results are summarized in Tables 6 and 7.

Table 6:

Most common indications for non-elective readmission within 30 days

Indications for WWE	N (%)	Indications for controls	N (%)
Other complications of birth; puerperium affecting management of mother	431 (54.1)	Other complications of birth; puerperium affecting management of mother	50827 (60.9)
Hypertension complicating pregnancy, childbirth and the puerperium	140 (17.5)	Hypertension complicating pregnancy, childbirth and the puerperium	18918 (22.7)
Other complications of pregnancy	49 (6.2)	Other complications of pregnancy	2288 (2.7)
Epilepsy; convulsions	28 (3.5)	Biliary tract disease	1760 (2.1)
Mood disorders	24 (3.0)	Septicemia (except in labor)	763 (0.9)
Miscellaneous mental health disorders	12 (1.5)	Complications of surgical procedures or medical care	656 (0.8)

Table 7:

Most common indications for non-elective readmission within 90 days

Indications for WWE	N (%)	Indications for controls	N (%)
Other complications of birth; puerperium affecting management of mother	338 (36.4)	Other complications of birth; puerperium affecting management of mother	42596(48)
Epilepsy; convulsions	98 (10.6)	Hypertension complicating pregnancy, childbirth and the puerperium	14092 (16)
Hypertension complicating pregnancy, childbirth and the puerperium	87 (9.4)	Biliary tract disease	6581 (7.5)
Mood disorders	46 (5.0)	Septicemia (except in labor)	2271 (2.6)
Other complications of pregnancy	46 (5.0)	Mood disorders	2220 (2.5)
Biliary tract disease	35 (3.8)	Other complications of pregnancy	2105 (2.5)

4. Discussion:

Our representative nationwide analysis found that WWE have slightly higher non-elective short-term readmission rates after a viable delivery as compared to women without common neurological comorbidities for all causes of postpartum readmissions. This benchmark data highlights important associations with readmission that should be noted in WWE to reduce maternal morbidity. Although maternal mortality was not explicitly measured in our study, prior literature has demonstrated that WWE are ten times more likely to suffer maternal death than those without epilepsy.[4,5] Our results also highlight the discrepancy of significantly higher rates of CDC SMM indicators at the time of index delivery despite relatively lower short-term readmission rates within 30 and 90 days.

Attention to multidisciplinary management along the pregnancy continuum and transitional postpartum care are crucial for maternal-fetal outcomes and public health, especially for WWE. The American College of Obstetricians and Gynecologists’ expert committee opinion to optimize postpartum care as a continuous process rather than individual health encounters further highlight that care should be customized to an individual’s needs. This expert opinion fortified the concept of postpartum care as a “fourth trimester” and stressed the importance of coordinating multidisciplinary efforts in women with chronic conditions, such as epilepsy.[25] Our study suggests WWE are at greater risk of adverse outcomes at the time of delivery, consistent with prior data,[1] and several of these risk factors are importantly associated with readmission.

Our analysis identified several distinct demographic, medical, and obstetric differences between WWE and controls. Our finding that WWE were more likely to hold government funded insurance and live in lower household income zip codes is consistent with previous data in WWE at the time of delivery.[1,26] Our study also demonstrated higher psychiatric burden in WWE as compared to controls, which is consistent with prior literature.[27,28]

In our study, the most common indications for 30- and 90-day readmissions for WWE and controls included birth complications/maternal puerperium management and hypertension complicating pregnancy, childbirth and the puerperium. Epilepsy/convulsions was the fourth most common readmission indication within 30 days and second most common readmission indication within 90 days for WWE. Recently published findings using the 2013 NRD assessing 30 days postpartum readmissions demonstrated that eclampsia represented approximately 80% of seizure/epilepsy readmissions, suggesting a high rate of miscoding. Hypertension complicating pregnancy, childbirth and the puerperium as a CCS code includes the diagnosis of eclampsia, suggesting WWE may have experienced higher admissions for epilepsy/convulsions than were captured in this study.[29]

However, WWE more often experienced obstetric complications, such as pre-eclampsia, poor intrauterine growth, CDC SMM indicators and experience longer lengths of stay, in accordance with other population-based data in WWE.[1,30] Furthermore, our study identified the presence of any CDC SMM indicator as five times more likely to be identified in WWE as compared to controls. We found WWE were more likely to have specific CDC SMM indicators of eclampsia, acute respiratory distress syndrome (ARDS), and acute renal failure, whereas the most common CDC SMM indicators in the general obstetric population include blood transfusions, disseminated intravascular coagulation and hysterectomy.[31–34] These distinct SMM indicators in WWE could be in part due to 1) the distinct pathophysiology of epilepsy, 2) antiepileptic drug management related to changing pharmacokinetic/dynamic properties, 3) and/or the need for closer multidisciplinary management.

Despite critical obstetric complications and higher risk of SMM indicators at the time of delivery, inpatient readmission rates in WWE in the postpartum period, albeit slightly higher than controls, were relatively low (2.4% readmitted within 30 days; 3.7% readmitted within 90 days). To provide context with other chronic conditions, a recent multinational cohort study of over 850,000 patients examining short-term readmissions for various chronic diseases found 10.6% were readmitted within 30 days (e.g., 18.4% of COPD patients readmitted within 30 days).[35] Combined with the prior data of increased maternal morbidity for WWE, these unexpected discrepancies could be due to the appropriate management of risk factors, surveillance bias, or other unmeasured influences. Close monitoring of WWE in the peri-postpartum periods with these comorbidities utilizing appropriate multidisciplinary outpatient management could decrease the risk of rehospitalizations and minimize the long term sequalae of morbidities acquired at delivery.

Although administrative data can be advantageous for analyzing large populations and can create generalizability to real world healthcare use, especially of interest in special populations such as pregnancy, limitations exist. Administrative data are not intended for real world clinical interpretation or research purposes and may inherently be less accurate or reliable (i.e., miscoding or underreporting). Although the NRD database provides invaluable insight into readmissions, specific limitations previously enumerated in the Methods sections exist (e.g. reliance on state-level data, discharges limited to each calendar year, and inability to link data to other databases). NRD data acquisition does not allow tracking readmissions between states nor link mother-baby outcomes, which would provide more comprehensive insight into peripartum and postpartum outcomes. Outcomes not related to maternal care and not recorded in the record could have contributed to maternal morbidity at the time of delivery and prompt short term readmissions (i.e., chronic comorbidities). Furthermore, medication information is not available within the NRD, so we were unable to assess how treatment or lack thereof could have contributed to these outcomes. We were also unable to assess if seizure frequency contributed to these outcomes, which could be a significant factor within our population.

Women throughout pregnancy and postpartum often require coordinated multidisciplinary care, and this appears especially crucial for WWE. Identifying the “fourth trimester” as a comprehensive, continuous period of care tailored to the individual may provide appropriate transitional care and coordinated efforts. Furthermore, identifying the subpopulation of WWE at higher risk of short-term readmission may help focus interventions to circumvent costly, burdensome readmissions.

Importantly, maternal mortality rather than morbidity may be a more appropriate marker for inpatient care as the primary outcome of interest for studies of WWE in inpatient care outcome. More research is needed to understand the impact of potentially disabling lifelong morbidities in WWE that may be better captured as outpatient care markers. These SMM indicators require further investigation due to the potentially devastating and preventable impact on women’s health.

These preliminary findings underscore the need for more real-world evidence in postpartum care for WWE to address crucial inconsistencies. Future directions include investigating therapy-specific and disease-specific postpartum outcomes in women with neurologic comorbidities, linking infant-maternal outcomes, and assessing prenatal care patterns using more granular data. Changing the clinical paradigm of neurologic and obstetric care in WWE is pivotal to improving evidence-based guidelines and maternal outcomes.

5. Conclusions:

We found short term non-elective hospital readmissions were relatively higher in WWE than women without epilepsy or other common neurological comorbidities after a viable delivery. This study highlights important discrepancies between WWE versus women without neurological comorbidities. Our findings suggest that WWE have substantial maternal morbidity at the time of delivery which does not necessarily translate into significant readmission rates.

Understanding current postpartum healthcare delivery disparities, in both inpatient and outpatient settings, and subsequent public health impact in WWE is paramount to improving mother-baby outcomes and contributing to crucial knowledge gaps that could provide important multidisciplinary guidelines in care.

Highlights:

• Pregnant women with epilepsy (WWE) have distinct sociodemographic and clinical characteristics at the time of delivery.

• WWE are at higher risk of severe maternal morbidity indicators at the time of delivery than women without common neurological comorbidities.

• WWE have higher short-term postpartum readmission rates.

• More real-world evidence is needed to improve postpartum care for WWE.