Association of Prenatal Exposure to Antiseizure Medications With Creative and Executive Function at Age 4.5 Years

Abstract

Background and Objectives

Neurodevelopmental effects of fetal antiseizure medication (ASM) exposure on creativity and executive functions are poorly understood. We previously found fetal valproate exposure to adversely affect measures of creativity and executive functions. In this study, we examine fetal exposure of newer ASMs on these functions in children of women with epilepsy (WWE) compared with children of healthy women (HW).

Methods

The Maternal Outcomes and Neurodevelopmental Effects of Antiepileptic Drugs study is a multicenter NIH-funded prospective observational cohort study of WWE and HW enrolled in pregnancy and their offsprings. This report examines blindly assessed creativity and executive functions in 4.5-year-old children of WWE vs HW. In addition, exposure-dependent ASM effects during the third trimester were examined in children of WWE, using a ratio of maximum observed ASM concentrations and ratio of defined daily dose (ratio DDD). For polytherapy, ratios were summed across ASMs. Linear regression models adjusted for multiple potential confounding factors were conducted for all analyses. The primary outcome for 4.5-year-old children was the Torrance Test of Creative Thinking—Figural Creativity Index. Secondary outcomes included the Global Executive Composite Score from the Behavior Rating Inventory of Executive Function—Preschool Version and subscales and other indexes of both measures.

Results

The primary analysis included 251 children of WWE and 73 of HW. No differences in creativity or executive function were found between children of WWE vs HW. No ASM exposure-dependent effects were found for the creativity measures, but exposure-dependent effects for executive function were present for ratio ASM concentration and ratio DDD.

Discussion

Our findings at 4.5 years show no differences in creative thinking between children of WWE vs HW (−3.2 [−9.0 to 2.7], p = 0.286) or associations with fetal exposure to ASMs (−2.6 [−11.0 to 5.7], p = 0.530). Secondary analyses revealed fetal exposure-dependent effects for executive function in children of WWE (7.0 [2.9–11.2], p = 0.001), which are most marked for levetiracetam (12.9 [4.2–21.6], p = 0.004). Our findings suggest that even for relatively safe ASMs, dosing needs to be adjusted to concentrations that prevent seizures, but balance risks to the fetus that high concentrations may pose.

Trial Registration Information

The study is registered at ClinicalTrials.gov as NCT01730170.

Introduction

Creativity is conceptualized as the ability to produce original and worthwhile content¹ and is a major force for positive innovations in our society. The neural mechanisms of creativity are uncertain, but likely involve dynamic interactions of large-scale brain systems. Although creativity is related to IQ, executive functions, and other cognitive functions, it is an independent construct.^2-4

Fetal exposure to antiseizure medications (ASMs) can produce neurodevelopmental effects and congenital malformations.⁵ IQ and specific cognitive abilities can be adversely affected by fetal exposure to some ASMs. Few studies have examined the effects of fetal ASM exposure on creative and executive abilities. In our prior study, the Neurodevelopmental Effects of Antiepileptic Drugs investigation examining the neurodevelopmental effects of fetal ASM exposure to monotherapies of carbamazepine, lamotrigine, phenytoin, and valproate, we found that valproate impaired creative fluency and originality at 4.5 years old as measured by the Torrance Thinking Creatively in Action and Movement (TCAM),^6,7 which can be predictive of personal and public achievement.⁸ In this new cohort of the Maternal Outcomes and Neurodevelopmental Effects of Antiepileptic Drugs (MONEAD) study, ASM prescribing patterns have changed,^9,10 and we examined the effects of fetal ASM exposure on creativity functions in 4.5-year-old children of mothers with and without epilepsy, as well as fetal ASM exposure effects in children of women with epilepsy. We also investigated fetal ASM exposure effects on executive functions at 4.5 years old and the relationship of creativity functions to executive functions and IQ.

Methods

Study Design

The MONEAD study is an ongoing multicenter NIH-funded prospective observational parallel-group cohort study of pregnant women with epilepsy (WWE) and pregnant healthy women (HW) without epilepsy and their children and a cohort of non-pregnant women with epilepsy. Enrollment occurred from December 2012 through January 2016. Inclusion criteria were female patients aged 14–45 years. Exclusion criteria included IQ <70,¹¹ drug/alcohol abuse within the prior year, history of psychogenic non-epileptic spells, progressive cerebral disease, other major medical illness, planned surgical intervention for epilepsy, or switching ASMs during pregnancy before enrollment (WWE). Additional exclusion criteria for pregnant women were >20 weeks gestational age (WGA), exposure to other known teratogens, detection of fetal congenital malformations before enrollment, or known genetic disorder. Twenty epilepsy centers across the United States with a specialty focus on treatment of WWE were enrollment sites.

Standard Protocol Approvals, Registrations, and Patient Consents

Signed informed consent was obtained from all adult participants before participation. This study was approved by the individual site institutional review boards and is registered on ClinicalTrials.gov as NCT01730170. The central site institutional review board is Stanford University.

Outcome Measures

The primary outcome for 4.5-year-old children was the Torrance⁷ Test of Creative Thinking—Figural (TTCT-F) Creativity Index by a blinded assessor. Secondary outcomes included the TTCT-F subscales and the Global Executive Composite Score, Inhibitory Self Control Index, Flexibility Index, and Emergent Metacognition Index from the Behavior Rating Inventory of Executive Function—Preschool Version (BRIEF-P),¹² completed by the child's mother/father at age 4.5 years. Both evaluations were scored by blinded assessors to the child study group. Note that higher scores are better for TTCT-F and worse for BRIEF-P.

Primary Objectives

The primary objective of the study was (1) to compare 4.5-year-old TTCT-F Creativity Index between children of WWE and children of HW and (2) to assess the association between 4.5-year-old TTCT-F Creativity Index and in utero third-trimester ASM blood concentration for children of WWE on ASM. To standardize ASM blood concentrations across ASMs, the ratio of the ASM concentration was calculated for each ASM by dividing the mother's ASM concentration by the upper limit of the suggested therapeutic ranges, mostly taken from published ranges,¹³ or clinical laboratory references used by Stanford University clinic, as was done in our age 3 neurocognitive analyses.¹⁴ The sum of the ratio ASM concentrations across all ASMs was used for mothers on polytherapy. The maximum observed ratio ASM concentration during the third trimester (including delivery) was used for analysis.

Risk Factors

Potential risk factors affecting 4.5-year-old TTCT-F Creativity Index scores assessed in secondary analyses included mother’s breastfeeding status, periconceptual folate use and folate dose (none, >0–0.4, >0.4–1.0, >1.0–4.0, >4.0 mg), average maternal anxiety (Beck Anxiety Inventory),¹⁵ depression (Beck Depression Inventory-2),¹⁶ and perceived stress (Perceived Stress Scale-14)¹⁷ during pregnancy and/or after birth through the child's 4.5-year-old visit, and average maternal sleep quality (Pittsburgh Sleep Quality Index)¹⁸ during pregnancy and/or postpartum. Additional potential risk factors for children of WWE on ASM included maximum third-trimester ASM dose, ASM group (monotherapy vs polytherapy), ASM category (lamotrigine monotherapy, levetiracetam monotherapy, other monotherapy, lamotrigine + levetiracetam polytherapy, or other polytherapy), and specific ASM. ASM dose was standardized across ASMs by dividing the prescribed daily dose for a specific ASM by its World Health Organization–defined daily dose value¹⁹ or derived from package inserts to calculate the mother's ratio of defined daily dose (ratio DDD). The sum of the ratio DDDs across all ASMs was used for mothers on polytherapy, and the maximum ratio DDD in the third trimester was used for analysis.

Statistical Analysis

Analysis Populations

Primary analysis 1 was restricted to the subset of children with non-missing TTCT-F Creativity Index scores, and primary analysis 2 was further restricted to children of WWE on ASM during the third trimester with available blood concentrations.

Primary Analyses

Unadjusted and adjusted linear regression models were used to compare TTCT-F Creativity Index scores between children of WWE and children of HW for primary analysis 1 and to assess the association between TTCT-F Creativity Index scores and mothers' maximum observed third-trimester ratio ASM concentration for primary analysis 2. Mother's IQ¹¹ was included in both adjusted models a priori. Additional covariates were chosen using separate stepwise selection algorithms for each primary analysis. Akaike information criterion was used to compare models. The significance level for variable inclusion was p = 0.10 and that for staying in the model was p = 0.15. Potential maternal covariates consisted of age, WGA, employment status, education level, and household income at enrollment, planned and/or welcomed pregnancy, lifetime history of mood and/or anxiety disorders, any major depressive episode, smoking, alcohol use, and illicit substance use during pregnancy, and the risk factors described above. Additional maternal covariates for primary analysis 2 included epilepsy type (focal, generalized, unclassified/mixed), >5 convulsive seizures during pregnancy, number of convulsions during pregnancy, ASM group, and ASM category. Potential child-related covariates were WGA at birth, birth weight, sex, race, ethnicity, small for gestational age, and major congenital malformations. Variables with >5% missingness were excluded from consideration. Assumptions for the linear regression models were assessed graphically. Significance was considered at the p < 0.05 level (see eMethods for additional details of statistical analyses).

Secondary Analyses

Secondary analyses for risk factors and additional outcomes used similar linear regression models as the primary analyses and incorporated the same covariates in the adjusted models. The BRIEF-P Global Executive Composite Score was used as the primary measure of executive function, and the other BRIEF-P Indexes were subsequently analyzed. Interaction models were used to assess whether the association between the TTCT-F or BRIEF-P outcomes and ASM concentration or ASM dose in the third trimester differed by ASM group or ASM category. Correlations between the 4.5-year-old TTCT-F creativity outcomes, the BRIEF-P Global Executive Composite Score, and the 3-year-old Differential Abilities Scale II (DAS-II) General Conceptual Ability Standard Score were measured using Pearson correlation coefficient. Age 3 years was used because DAS-II was not given at age 4.5 years.

Exploratory Analyses

An exploratory analysis was run to assess whether the coronavirus disease 2019 (COVID-19) pandemic affected the TTCT-F Creativity Index. Scores before and after the start of the COVID-19 shutdown, defined as April 1, 2020, were compared for all children and separately in children of WWE on ASM in the third trimester.

Sensitivity Analyses

Sensitivity analyses were run to compare demographic variables, risk factors, covariates, and outcomes between children included in the primary analyses and those who were excluded. The impact of twins on the primary analyses was assessed by rerunning the primary analyses excluding 1 twin, excluding both twins, and using generalized estimating equations to account for the correlation between twin pairs. Primary analysis 1 was rerun removing children of WWE whose mothers were not on ASM during pregnancy. As a post hoc sensitivity analysis, the primary adjusted analyses were rerun including age at TTCT-F assessment as an additional covariate because of our sample consisting of mostly children younger than 5 years and the TTCT-F normalized scores being defined for 5-year-old children.

Data Availability

All data included in these analyses will be shared as deidentified data by request from any qualified investigator.

Results

Of 345 children born to WWE and 106 children born to HW, 251 and 73 were included in primary analysis 1, respectively. Primary analysis 2 consisted of 228 children of WWE whose mothers had third-trimester ASM blood concentration data. The most prevalent third-trimester ASM regimen for WWE in primary analysis 1 was monotherapy (72.5%), followed by polytherapy (21.9%), while mothers not on ASM during pregnancy were only a minority (5.2%). One mother was not on ASM during the third trimester because of delivery in the second trimester. Lamotrigine and levetiracetam were the most common monotherapies (44.5% and 37.4% of all monotherapies, respectively), and lamotrigine + levetiracetam was the most common polytherapy (41.8% of all polytherapies) (see eTable 1 for the full list of individual ASMs). More information on individual ASM doses and blood concentrations are available in our prior publication.¹⁴ A flowchart of enrollment and exclusions is provided (eFigure 1). Demographics, baseline characteristics, and risk factors for children of WWE vs HW from primary analysis 1 are summarized in Table 1 and for children of WWE from primary analysis 2 in eTables 2 and 3.

Table 1.

Demographic, Baseline Characteristics, and Risk Factors for Children of WWE vs HW in Primary Analysis 1 (N = 324)

Categorical variables	Children of WWE (N = 251), n (%)	Children of HW (N = 73), n (%)	p Valuea
Child's sex: male	112 (44.6)	38 (52.1)	0.262
Child's race			0.011
White	204 (81.3)	49 (67.1)
Black or African American	12 (4.8)	10 (13.7)
Other/unknown	35 (13.9)	14 (19.2)
Child's ethnicity: Hispanic or Latino	52 (20.7)	17 (23.3)	0.637
Small for gestational age	16 (6.5)	5 (6.8)	1.000
Missing	6	0
Major congenital malformations	10 (4.0)	1 (1.4)	0.467
Any breastfeeding	194 (77.6)	65 (89.0)	0.031
Missing	1	0
Periconceptional folate use	223 (88.8)	48 (65.8)	<0.001
Periconceptional folate dose			<0.001
0 mg	28 (11.3)	25 (34.7)
>0–0.4 mg	20 (8.1)	13 (18.1)
>0.4–1 mg	46 (18.5)	27 (37.5)
>1–4 mg	138 (55.6)	7 (9.7)
>4 mg	16 (6.5)	0 (0.0)
Missing	3	1
Pregnancy planned/welcomed			0.300
Planned	181 (72.1)	46 (63.0)
Unplanned/welcome	63 (25.1)	25 (34.2)
Unplanned/unwelcome	7 (2.8)	2 (2.7)
Major depressive episode during pregnancyb	11 (4.4)	2 (2.7)	0.740
Mother's education			0.245
College degree (advanced)	69 (27.5)	26 (35.6)
College degree (not advanced)	112 (44.6)	25 (34.2)
No college degree	70 (27.9)	22 (30.1)
Mother's employment			0.767
Employed full-time	143 (57.0)	40 (54.8)
Employed part-time	36 (14.3)	13 (17.8)
Unemployedc	72 (28.7)	20 (27.4)
Household income			0.950
Subject desires not to answer	10 (4.0)	4 (5.5)
$24,999 or less	45 (17.9)	14 (19.2)
$25,000–49,999	23 (9.2)	7 (9.6)
$50,000–74,999	37 (14.7)	8 (11.0)
$75,000–99,999	46 (18.3)	12 (16.4)
$100,000 and up	90 (35.9)	28 (38.4)
Smokingd	13 (5.2)	5 (6.8)	0.567
Alcohol used	58 (23.1)	25 (34.2)	0.055
Illicit substance used	7 (2.8)	4 (5.5)	0.276
Lifetime history of mood disordere	42 (16.9)	9 (13.0)	0.436
Missing	3	4
Lifetime history of anxiety disordere	41 (16.5)	11 (15.9)	0.907
Missing	3	4

Continuous variables	N	Mean (95% CI)	Median (IQR)	Min–max	N	Mean (95% CI)	Median (IQR)	Min–max	p Valuef
WGA at birth	251	38.5 (38.3–38.8)	39 (38–40)	24–42	73	38.5 (38.0–38.9)	39 (38–40)	26–41	0.831
WGA at enrollment	251	13.2 (12.6–13.8)	13 (10–17)	4–29	73	15.3 (14.4–16.3)	16 (13–19)	5–23	<0.001
Birth weight (kg)	246	3.24 (3.16–3.31)	3.3 (2.9–3.6)	0.6–4.9	73	3.28 (3.13–3.42)	3.3 (3.0–3.5)	1.1–5.0	0.606
Mother's age at enrollment	251	31.1 (30.5–31.7)	31 (28–34)	18–46	73	29.8 (28.6–31.0)	31 (27–33)	15–38	0.056
Mother's IQ	251	98.5 (96.9–100.0)	100 (90–107)	62–123	73	104.6 (101.6–107.7)	106 (96–114)	72–129	<0.001
Father's IQg	178	103.9 (102.0–105.9)	105 (95–113)	58–139	45	103.2 (99.2–107.2)	101 (95–114)	75–123	0.753
Maternal relative's IQg	102	100.9 (97.7–104.1)	102 (93–113)	51–133	25	102.2 (98.0–106.5)	103 (97–109)	74–123	0.693
BDI-IIh
Enrollment: 4.5-y-old visit	251	6.00 (5.35–6.66)	4.9 (2.2–7.6)	0.2–34.0	73	5.16 (4.21–6.12)	4.7 (2.3–6.3)	0.0–26.1	0.449
Pregnancy	251	6.95 (6.22–7.67)	5.3 (3.3–8.7)	0.0–38.7	73	5.97 (5.07–6.87)	5.0 (3.0–8.3)	0.0–17.7	0.419
Post-birth	251	5.63 (4.92–6.34)	4.3 (1.7–7.5)	0.0–32.9	73	4.75 (3.66–5.85)	3.9 (1.6–6.0)	0.0–30.6	0.415
BAIh
Enrollment: 4.5-y-old visit	251	4.93 (4.32–5.53)	3.3 (1.9–6.6)	0.0–34.2	73	3.79 (2.76–4.82)	3.0 (1.3–4.6)	0.0–26.8	0.028
Pregnancy	251	6.29 (5.56–7.02)	4.7 (2.3–8.0)	0.0–38.3	73	5.40 (4.25–6.55)	4.0 (2.3–7.7)	0.0–29.3	0.309
Post-birth	251	4.38 (3.74–5.02)	2.6 (1.2–5.5)	0.0–32.4	73	3.08 (2.00–4.15)	1.9 (0.6–3.7)	0.0–25.7	0.006
PSS-14h
Enrollment: 4.5-y-old visit	251	18.05 (17.24–18.86)	18.0 (13.0–22.3)	5.0–39.2	73	17.35 (16.20–18.51)	17.4 (12.9–20.6)	8.1–28.7	0.494
Pregnancy	251	17.98 (17.07–18.90)	17.7 (12.0–23.0)	2.7–40.0	73	16.65 (15.28–18.03)	16.3 (12.7–21.0)	3.3–28.7	0.272
Post-birth	251	18.10 (17.25–18.95)	18.1 (13.2–22.6)	2.1–40.3	73	17.64 (16.38–18.89)	16.9 (12.8–21.6)	8.9–33.0	0.639
PSQIg,i
Pregnancy + postpartum	200	5.66 (5.31–6.02)	5.6 (3.7–7.0)	0.1–13.0	56	5.47 (4.82–6.11)	5.0 (3.9–6.7)	1.2–12.3	0.605
Pregnancy	229	5.75 (5.37–6.12)	5.3 (3.6–7.6)	0.1–14.0	65	4.87 (4.23–5.52)	4.3 (3.4–5.8)	1.0–11.8	0.029
Postpartum	202	5.60 (5.22–5.98)	5.1 (3.6–7.0)	0.0–13.6	56	5.89 (5.15–6.62)	6.1 (3.7–7.7)	0.8–12.5	0.484

Abbreviations: BAI = Beck Anxiety Inventory; BDI-II = Beck Depression Inventory II; HW = healthy women; IQR = interquartile range; PSQI = Pittsburgh Sleep Quality Index; PSS-14 = Perceived Stress Scale 14; TTCT-F = Torrance Test of Creative Thinking—Figural; WGA = weeks gestational age; WWE = women with epilepsy.

Missing not used in calculation of percentages.

^aFor categorical variables: p values from the χ² test except for small for gestational age, major congenital malformations, major depressive episode during pregnancy, smoking, and illicit substance, which used the Fisher exact test.

^bDepression based on a SCID mood module at any depression assessment during pregnancy.

^cUnemployed includes stay-at-home parents without outside job, unemployed because of disability, full-time student, and looking for work.

^dSelf-reported smoking, alcohol use, or illicit substance use (including marijuana) at any time during pregnancy.

^eDiagnosis from SCID.

^fp Values from t test (WGA at birth, WGA at enrollment, birth weight, age, IQ scores, and PSQI scores) and Wilcoxon rank-sum test (BDI, BAI, and PSS scores).

^gFather's IQ, maternal relative's IQ, and PSQI scores not assessed for inclusion in the regression model because of extensive missing data.

^hAverage of all assessments completed during the time period. Enrollment: 4.5-year-old visit: enrollment through the 4.5-year-old visit; pregnancy: enrollment through the day of delivery; post-birth: after delivery through the 4.5-year-old visit.

ⁱWeighted monthly average of PSQI assessments completed during the time period. At most 1 PSQI score per day was used in the calculation. Pregnancy: enrollment through the day before delivery; postpartum: 6 weeks through 9 months after delivery; pregnancy + postpartum: assessments completed during either period (only women with scores in both periods included).

Primary Analysis 1

There were no statistically significant differences found between children of WWE vs HW in the mean 4.5-year-old TTCT-F Creativity Index standard score (unadjusted mean [95% CI] 90.2 [87.4–93.0] vs 94.9 [89.8–100.1]; adjusted least squares mean [95% CI] 90.5 [87.8–93.2] vs 93.7 [88.6–98.8]) (Figure 1). Higher maternal IQ was significantly associated with higher 4.5-year-old TTCT-F Creativity Index in the adjusted model (Table 2). Education level and illicit substance use during pregnancy were also selected as additional covariates for the adjusted model, but were not significant at the 0.05 level in the adjusted model.

Figure 1. Age 4.5 Years TTCT-F Creativity Index by Mother's Study Group—Children of WWE vs HW (N = 324).

Adjusted least squares mean scores (95% CI) 90.5 (87.8–93.2) for children of WWE vs 93.7 (88.6–98.8) for children of HW (nonsignificant). HW = healthy women; TTCT-F = Torrance Test of Creative Thinking—Figural; WWE = women with epilepsy.

Table 2.

Age 4.5 Years TTCT-F Creativity Index Scores: Full Model Summaries for Primary Analyses

Model parameter	Parameter estimate (95% CI)	p Value
Children of WWE vs HW (N = 324)a
Mother's study group: WWE vs HW	−3.2 (−9.0 to 2.7)	0.286
Mother's IQ	0.3 (0.0 to 0.5)	0.024
Education level		0.062
College degree (advanced)	Ref	—
College degree (not advanced)	−5.2 (−11.1 to 0.8)	0.087
No college degree	−8.9 (−16.4 to −1.4)	0.020
Mother's illicit substance use during pregnancy: any vs noneb	−12.3 (−25.6 to 1.0)	0.070
Children of WWE with third-trimester ASM blood concentrations (N = 228)c
Mother's third trimester max observed ratio ASM concentrationd	−2.6 (−11.0 to 5.7)	0.530
Mother's IQ	0.3 (0.1 to 0.6)	0.007

Abbreviations: ASM = antiseizure medication; HW = healthy women; TTCT-F = Torrance Test of Creative Thinking—Figural; WWE = women with epilepsy.

Variables selected into adjusted models using a stepwise selection algorithm. Mother's IQ and study group or third-trimester ratio ASM concentration included in the model a priori. Akaike information criterion was used to compare models. Significance level for covariate entry was set to p = 0.01, and significance level to remain in the model was set to p = 0.15.

^aN = children of WWE and HW with age 4.5 years TTCT-F Creativity Index. Full model R²: 0.091.

^bSelf-reported illicit substance use (including marijuana) at any time during pregnancy.

^cN = children of WWE with age 4.5 years TTCT-F Creativity Index and mothers with third-trimester ASM concentrations. Full model R²: 0.036.

^dRatio ASM concentration calculated as the of ratio the upper limit for therapeutic range. For mothers on polytherapy, ratio ASM concentration calculated by summing the ratio ASM concentration for each ASM. Maximum observed value during the third trimester, including day of delivery.

Primary Analysis 2

No significant relationship was found between 4.5-year-old TTCT-F Creativity Index and mother's third-trimester maximum observed ratio ASM concentration (unadjusted parameter estimate [95% CI] −4.1 [−12.5 to 4.3]; adjusted parameter estimate [95% CI] −2.6 [−11.0 to 5.7]) (Table 2 and Figure 2). Higher maternal IQ was significantly related to greater creativity in the adjusted model. No additional covariates were selected into the adjusted analyses.

Figure 2. Age 4.5 Years TTCT-F Creativity Index Score vs Third-Trimester Maximum Observed Ratio ASM Concentration in Children of WWE With Third-Trimester Blood Concentrations (N = 228).

Adjusted parameter estimate (95% CI) −2.6 (−11.0 to 5.7) (nonsignificant). ASM = antiseizure medication; TTCT-F = Torrance Test of Creative Thinking—Figural.

Secondary Analyses

Children of WWE and HW did not differ on any subscales of the 4.5-year-old TTCT-F Creativity Index (eTable 4). The associations between third-trimester ASM concentration with 4.5-year-old TTCT-F Creativity Index subscales were not significant (Table 3) by either ASM group or ASM category (eTables 5–8). The results with ASM dose were similar (eTables 9–14).

Table 3.

Association Between Age 4.5 Years TTCT-F Subscales and BRIEF-P Scores and Maximum Observed Ratio ASM Concentration^a in the Third Trimester for Children of WWE With Third-Trimester Blood Concentrations

	N	Unadjusted analysisb		Adjusted analysisc
		Parameter estimate (95% CI)	p Value	Parameter estimate (95% CI)	p Value
TTCT-F Fluency Standard Score	228	−1.6 (−9.1 to 6.0)	0.684	−1.0 (−8.6 to 6.6)	0.799
TTCT-F Originality Standard Score	228	−4.8 (−14.3 to 4.6)	0.315	−3.4 (−12.8 to 6.0)	0.483
TTCT-F Elaboration Standard Score	228	−2.4 (−7.6 to 2.8)	0.356	−1.1 (−6.2 to 4.0)	0.668
TTCT-F Abstractness of Titles Standard Score	228	−0.0 (−11.8 to 11.7)	0.993	1.8 (−10.0 to 13.5)	0.766
TTCT-F Resist to Closure Standard Score	228	−8.2 (−20.1 to 3.7)	0.176	−7.6 (−19.6 to 4.3)	0.211
BRIEF-P Global Executive Composite Score T-score	230	7.6 (3.5 to 11.8)	<0.001	7.0 (2.9 to 11.2)	0.001
BRIEF-P Inhibitory Self-Control Index T-score	230	7.4 (3.3 to 11.5)	<0.001	7.1 (3.0 to 11.2)	<0.001
BRIEF-P Flexibility Index T-score	230	7.0 (3.0 to 11.1)	<0.001	6.6 (2.5 to 10.6)	0.002
BRIEF-P Emergent Metacognition Index T-score	230	7.1 (3.4 to 10.9)	<0.001	6.7 (2.9 to 10.4)	<0.001

Abbreviations: ASM = antiseizure medication; BRIEF-P = Behavior Rating Inventory of Executive Function—Preschool Version; HW = healthy women; TTCT-F = Torrance Test of Creative Thinking—Figural; WWE = women with epilepsy.

N = children of WWE with non-missing outcome score and mothers with max observed ratio ASM concentration in the third trimester (includes day of delivery).

Note positive parameter estimates indicated association of higher ratio ASM concentrations, which are better for TTCT-F scores but worse for BRIEF-P scores.

^aRatio ASM concentration calculated as the ratio of the upper limit for the therapeutic range. For mothers on polytherapy, ratio ASM concentration calculated by summing the ratio ASM concentration for each ASM. Maximum observed value recorded during the third trimester, including the day of delivery used for analysis.

^bLinear regression model.

^cLinear regression model adjusted for mother's IQ.

Executive Function

The mean T-scores of children of WWE and HW did not differ on any BRIEF-P indexes (eTable 4). The BRIEF-P Global Executive Composite Score at age 4.5 years was associated with third-trimester ASM concentration in adjusted analyses, as were all the other indexes (Table 3). Worse outcomes of executive functions (i.e., higher scores on the BRIEF-P) were observed with increasing third-trimester ASM exposure. Analyses of BRIEF-P scores stratified by ASM category and group were conducted (Table 4 and eTable 15). Higher blood concentrations of levetiracetam were associated with poorer performance on the BRIEF-P Global Executive Composite Score, Inhibitory Self Control Index, and Emergent Metacognition Index, and higher concentrations of lamotrigine + levetiracetam were associated with poorer performance on the Flexibility Index and Emergent Metacognition Index (Table 4 and eFigure 2). The results with ASM dose were similar (eTable 12–14). Lamotrigine had no exposure-dependent effects, and samples sizes for other ASMs were inadequate to assess individually.

Table 4.

Association Between Age 4.5 Years BRIEF-P Scores and Maximum Observed Ratio ASM Concentration^a in the Third Trimester by ASM Category, Children of WWE With Third-Trimester Blood Concentrations

Third-trimester ASM categoryd	N	Unadjusted analysisb		Adjusted analysisb,c
		Parameter estimate (95% CI)	p Value	Parameter estimate (95% CI)	p Value
BRIEF-P Global Executive Composite Score T-score
Lamotrigine	79	5.2 (−7.9 to 18.4)	0.432	5.4 (−7.6 to 18.4)	0.415
Levetiracetam	70	12.5 (3.7 to 21.3)	0.005	12.9 (4.2 to 21.6)	0.004
Other monotherapy	33	−4.7 (−22.1 to 12.7)	0.598	−5.1 (−22.3 to 12.1)	0.562
Lamotrigine + levetiracetam	24	12.0 (0.3 to 23.8)	0.045	11.6 (−0.1 to 23.2)	0.052
Other polytherapy	24	5.6 (−7.9 to 19.1)	0.416	5.4 (−8.0 to 18.7)	0.429
p Value for interaction terme	—	—	0.435	—	0.395
BRIEF-P Inhibitory Self-Control Index T-score
Lamotrigine	79	5.3 (−7.6 to 18.3)	0.421	5.4 (−7.6 to 18.3)	0.415
Levetiracetam	70	9.9 (1.2 to 18.5)	0.026	10.0 (1.4 to 18.7)	0.024
Other monotherapy	33	−2.9 (−20.1 to 14.2)	0.738	−3.1 (−20.3 to 14.0)	0.720
Lamotrigine + levetiracetam	24	10.6 (−1.0 to 22.2)	0.073	10.4 (−1.2 to 22.0)	0.079
Other polytherapy	24	6.1 (−7.2 to 19.4)	0.367	6.0 (−7.3 to 19.3)	0.375
p Value for interaction terme	—	—	0.708	—	0.697
BRIEF-P Flexibility Index T-score
Lamotrigine	79	8.3 (−4.5 to 21.1)	0.201	8.4 (−4.3 to 21.1)	0.193
Levetiracetam	70	7.0 (−1.5 to 15.6)	0.108	7.3 (−1.2 to 15.8)	0.093
Other monotherapy	33	1.8 (−15.1 to 18.7)	0.832	1.5 (−15.3 to 18.3)	0.862
Lamotrigine + levetiracetam	24	12.5 (1.1 to 24.0)	0.032	12.1 (0.7 to 23.5)	0.037
Other polytherapy	24	5.8 (−7.3 to 18.9)	0.381	5.7 (−7.4 to 18.7)	0.393
p Value for interaction terme	—	—	0.867	—	0.872
BRIEF-P Emergent Metacognition Index T-score
Lamotrigine	79	4.5 (−7.2 to 16.2)	0.449	4.6 (−7.0 to 16.2)	0.435
Levetiracetam	70	14.2 (6.3 to 22.0)	<0.001	14.5 (6.7 to 22.2)	<0.001
Other monotherapy	33	−9.8 (−25.3 to 5.6)	0.211	−10.2 (−25.5 to 5.2)	0.192
Lamotrigine + levetiracetam	24	11.0 (0.5 to 21.4)	0.040	10.6 (0.2 to 21.0)	0.046
Other polytherapy	24	4.2 (−7.7 to 16.2)	0.486	4.1 (−7.8 to 15.9)	0.501
p Value for interaction terme	—	—	0.075	—	0.061

Abbreviations: ASM = antiseizure medication; BRIEF-P = Behavior Rating Inventory of Executive Function—Preschool Version; WWE = women with epilepsy.

N = children of WWE with non-missing outcome score and mothers with maximum observed ratio ASM concentration in the third trimester (includes the day of delivery).

^aRatio ASM concentration calculated as the ratio of the upper limit for the therapeutic range. For mothers on polytherapy, ratio ASM concentration calculated by summing the ratio ASM concentration for each ASM. Maximum observed value recorded during the third trimester, including the day of delivery.

^bParameter estimate and p value from a linear regression model including mother's max observed ratio ASM concentration in the third trimester, ASM category, and their interaction. Note that higher BRIEF scores are worse, so a positive parameter estimate indicates worse score with higher ASM concentration.

^cAdjusted for mother's IQ.

^dBased on ASM regimen taken at the time of blood draw for maximum observed ratio ASM concentration during the third trimester (includes the day of delivery).

^ep Value for the interaction between max observed ratio ASM concentration in the third trimester and ASM category. If significant, the association between max observed ratio ASM concentration in the third trimester and age 4.5-year-old score differs by mother's ASM category.

Risk Factors

No risk factors were found to be significantly associated with the 4.5-year-old TTCT-F Creativity Index in adjusted models in either analysis population (eTables 16–21). Mother's ASM group, ASM category, and specific ASM were also not related to creativity in the primary analysis 2 population (eTable 22).

Exploratory Analyses

There were no differences in the TTCT-F Creativity Index in children evaluated before vs after the COVID-19 shutdown (eTables 23 and 24).

Sensitivity Analyses

Comparisons of children included vs excluded in the primary analyses are presented in eTables 25–31. Conclusions from primary analysis 1 remained the same after removing children of WWE whose mothers were not on ASM (eTables 32 and 33). The presence of twins did not affect the results of either primary analysis (eTables 34 and 35). Age was significantly associated with creativity, but adjusting for age in the primary analysis models did not affect the main results of the analysis (eTables 36 and 37).

Correlative Analyses

Correlations between various measures of the 4.5-year-old TTCT-F with the 4.5-year-old BRIEF-P Global Executive Composite Score and the 3-year-old DAS-II General Conceptual Ability Standard Score were significant but low, ranging from −0.16 through −0.02 and 0.21 through 0.38, respectively (Table 5). The correlation between the BRIEF-P and DAS-II measures was −0.28.

Table 5.

Correlation Between Age 4.5 TTCT-F Subscales, Age 4.5 Years BRIEF-P Global Executive Composite Score T-Score, and Age 3 Years DAS-II General Conceptual Ability Standard Score, Children of WWE and HW (N = 449)

Outcome	Summary statistics			Pearson correlation
				4.5-y-old BRIEF-P Global Executive Composite score T-score			3-y-old DAS-II General Conceptual Ability Standard Score
	N	Mean (95% CI)	Median (min–max)	n	ρ	p Value	n	ρ	p Value
4.5-y-old TTCT-F Creativity Index	324	91.2 (88.8–93.7)	94.0 (46–137)	314	−0.13	0.027	303	0.38	<0.001
4.5-y-old TTCT-F Fluency Standard Score	324	84.6 (82.3–86.9)	85.0 (40–147)	314	−0.10	0.066	303	0.32	<0.001
4.5-y-old TTCT-F Originality Standard Score	324	85.8 (83.0–88.6)	89.0 (40–141)	314	−0.08	0.139	303	0.32	<0.001
4.5-y-old TTCT-F elaboration standard score	324	83.7 (82.2–85.3)	84.0 (53–138)	314	−0.16	0.006	303	0.37	<0.001
4.5-y-old TTCT-F Abstractness of Titles Standard Score	324	88.3 (84.9–91.7)	91.0 (40–149)	314	−0.12	0.030	303	0.30	<0.001
4.5-y-old TTCT-F Resist to Closure Standard Score	324	90.7 (87.3–94.2)	96.0 (40–147)	314	−0.02	0.785	303	0.21	<0.001
4.5-y-old BRIEF-P Global Executive Composite Score T-score	327	47.2 (45.8–48.5)	44.0 (33–95)	—	—	—	301	−0.28	<0.001
3-y-old DAS-II General Conceptual Ability Standard Score	361	105.4 (103.8–107.0)	106.0 (42–155)	—	—	—	—	—	—

Abbreviations: ASM = antiseizure medication; BRIEF-P = Behavior Rating Inventory of Executive Function—Preschool Version; DAS-II = Differential Abilities Scale II; HW = healthy women; TTCT-F = Torrance Test of Creative Thinking—Figural; WWE = women with epilepsy.

N = children of WWE and HW with non-missing outcome measure; n = children of WWE and HW with non-missing measures for both outcomes in correlation.

Discussion

Our study demonstrates that there are no significant differences in measures of creativity or parent ratings of executive functions at age 4.5 years between children of WWE vs HW. Furthermore, no ASM exposure-dependent effects were found for the creativity measure, although exposure-dependent effects for executive function were present for ASM concentrations and ASM doses. Analyses of risk factors and sensitivity analyses support these findings.

The absence of ASM effects may be because of the young age of the children as the TTCT-F is only standardized down to age 5 years. However, we previously demonstrated adverse effects of valproate vs carbamazepine and lamotrigine on creativity measures in children with a mean age of 4.2 years.⁶ Furthermore, the specific tasks from the TTCT-F differed from the creativity measure used in our prior study, which used the Torrance TCAM. The TCAM is a tactile and kinesthetic measure that may be more sensitive for capturing creativity in young children compared with the TTCT-F used in this study because young children more easily express their creativity through this modality.⁷ However, fetal exposure to the currently used ASMs may not affect creativity as valproate does.

Although children of WWE did not differ from children of HW on executive functions, there was an exposure-dependent effect on executive functions, such that increasing third-trimester ASM concentrations were associated with poorer performance on parent ratings of executive functions. This effect seemed to be primarily due to levetiracetam. Lamotrigine had no exposure-dependent effects, and sample sizes for other ASMs were inadequate to assess individually. We observed similar findings for levetiracetam in our prior analyses of cognition at age 3 years and of adaptive and behavioral functioning at 4.5 years.^14,20 In addition, a recent retrospective population-based study found that fetal levetiracetam exposure is associated with increased risk of anxiety and attention-deficit/hyperactivity disorder, although ASM concentration and dose effects were not assessed.²¹ Nevertheless, levetiracetam has low risks of major congenital malformations and adverse neurodevelopmental outcomes, so it is considered one of the safest ASMs during pregnancy.²² All ASMs are potential teratogens, and teratogens act in an exposure-dependent manner. Even lamotrigine has shown a small but significant dose-dependent risk of increased malformations.²³ Thus, management of WWE during pregnancy requires ASM dosing to be sufficiently high enough to control seizures to protect the mother and the fetus, but at the lowest effective dose to minimize risks to the fetus.⁵ Furthermore, dosing needs to take into account clearance changes for many ASMs during pregnancy.²⁴

Our creativity measure (TTCT-F at age 4.5 years) was correlated with IQ (DAS-II at age 3 years) and to a lesser degree with executive function (BRIEF-P at age 4.5 years). These correlations are expected based on prior studies.^2-4

Strengths of this study include the large prospective design with detailed data for WWE and HW and their children, allowing control for multiple factors. Our study also assessed ASM blood concentrations to control for clearance changes during pregnancy. Despite being one of largest prospective studies of WWE in pregnancy with detailed information on potential confounders, our sample size is limited. Patients were recruited from tertiary centers, so there is a concern for generalizability. Assessments for a portion of our children aged 4.5 years occurred during the COVID-19 epidemic, which could have affected results; however, no differences in TTCT-F Creativity Index scores were found between children evaluated before vs after the COVID-19 shutdown. Finally, because this is an observational study, like all studies of pregnancy outcomes in WWE, there may be unmeasured confounding because of the observational nature of the study design. While adjusted models were run, few potentially confounding variables were selected into these models because of nonsignificant associations with the primary outcome measure, which may have resulted in biased results because of potential unmeasured confounding. Selection bias may also be present because of the lack of randomization of ASM regimens for practical and ethical reasons. Thus, these findings require replication in separate cohorts.

No differences in creativity or executive functions at age 4.5 years were found for children of WWE vs HW. No ASM exposure-dependent effects were found for creativity, but secondary analyses revealed exposure-dependent effects for executive function.

Glossary

ASM

antiseizure medication

BRIEF-P

Behavior Rating Inventory of Executive Function—Preschool Version

COVID-19

coronavirus disease 2019

DAS-II

Differential Abilities Scale II

HW

healthy women

MONEAD

Maternal Outcomes and Neurodevelopmental Effects of Antiepileptic Drugs

ratio DDD

ratio of defined daily dose

TCAM

Thinking Creatively in Action and Movement

TTCT-F

Torrance Test of Creative Thinking—Figural

WWE

women with epilepsy

WGA

weeks gestational age

Appendix 1. Authors

Name	Location	Contribution
Kimford J. Meador, MD	Stanford University, Palo Alto, CA	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data
Morris J. Cohen, EdD	Pediatric Neuropsychology International, Augusta, GA	Drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data
David W. Loring, PhD	Emory University School of Medicine, Atlanta, GA	Drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data
Abigail G. Matthews, PhD	The Emmes Company, Rockville, MD	Drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data
Carrie A. Brown, PhD	The Emmes Company, Rockville, MD	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Chelsea Robalino, MS	The Emmes Company, Rockville, MD	Drafting/revision of the manuscript for content, including medical writing for content; study concept or design
Andrea Carmack, MB	The Emmes Company, Rockville, MD	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Sarah Sumners, PhD	Piedmont University, Athens, GA	Drafting/revision of the manuscript for content, including medical writing for content
Angela K. Birnbaum, PhD	University of Minnesota, Minneapolis	Drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data
Laura A. Kalayjian, MD	University of Southern California, Los Angeles	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Evan Gedzelman, MD	Emory University School of Medicine, Atlanta, GA	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Paula E. Voinescu, MD	Brigham and Women's Hospital, Harvard Medical School, Boston, MA	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Elizabeth E. Gerard, MD	Northwestern University, Chicago, IL	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Julie Hanna, MD	Minnesota Epilepsy Group, Roseville	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Jennifer Cavitt, MD	University of Cincinnati, OH	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Maria Sam, MD	Wake Forest University, Winston-Salem, NC	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Sean T. Hwang, MD	Northwell Health, Great Neck, NY	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Alison M. Pack, MD	Columbia University, New York, NY	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Jeffrey J. Tsai, MD	University of Washington, Seattle	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Page B. Pennell, MD	University of Pittsburgh, PA	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data

Appendix 2. Coinvestigators

Coinvestigators are listed at Neurology.org.

Study Funding

Funded by NIH, National Institute of Neurological Disorders and Stroke/NICHD U01-NS038455; ClinicalTrials.gov number NCT01730170.

Disclosure

K.J. Meador has received research support from the NIH, Eisai, and Medtronic Inc. The Epilepsy Study Consortium pays his university for his research consultant time related to Eisai, GW Pharmaceuticals, NeuroPace, Novartis, Supernus, Upsher-Smith Laboratories, UCB Pharma, and Vivus Pharmaceuticals. In addition, K.J. Meador is Co-I and Director of Cognitive Core of the Human Epilepsy Project for the Epilepsy Study Consortium, and is on the editorial boards for Neurology, Cognitive & Behavioral Neurology, Epilepsy & Behavior, and Epilepsy & Behavior Case Reports. M.J. Cohen receives royalties from Multi Health Systems, Inc. as author of the Children's Memory Scale. D.W. Loring receives research support from the NIH, is a consultant for Medtronic, serves on the editorial boards for Neuropsychology Review and Epilepsia for which he receives editorial stipends, and is on the editorial board for Archives of Clinical Neuropsychology. A.K. Birnbaum reports receiving grant support, paid to her institution, from Supernus Pharmaceuticals and Veloxis Pharmaceuticals, holding patent US9770407B2 on parenteral carbamazepine formulation, licensed to Lundbeck, and patent EP12150783A on novel parenteral carbamazepine formulations, licensed to Lundbeck. P.E. Voinescu received speaking honoraria from Physicians' Education Resource and from Philippines League Against Epilepsy. E.E. Gerard has served as site-PI for clinical trials sponsored by Xenon and Sunovion pharmaceuticals as well as a trial sponsored by Eisai and Stanford University, and has been reimbursed for lectures given to GW Pharmaceuticals Staff and Neurology Week. J. Cavitt received research support from National Institute of Neurological Disorders and Stroke (MONEAD) and from GW Pharmaceuticals, and advisory board fees from Jazz Pharmaceuticals. M. Sam has received advisory board consulting fees for Aquestive. S. Hwang reports no disclosures. A.M. Pack reports funding from NIH, royalties from Up to Date, and travel reimbursement for AAN and ABPN activities. J.J. Tsai reports grant from Institute of Translational Health Sciences (funded by the National Center for Advancing Translational Sciences of the NIH), and has served as site-PI for clinical trials sponsored by Xenon. P.B. Pennell reports grants from NIH, personal fees from NIH for Grant reviews, personal fees from AES and the AAN as speaking honoraria, personal fees from Medical Schools for speaking honoraria and travel, and personal fees from UpToDate, Inc. for royalties outside the submitted work. The other authors report no relevant disclosures. Go to Neurology.org/N for full disclosures.