This population-based retrospective birth cohort study used mother-child pairs of singleton births at Kaiser Permanente Southern California (KPSC) hospitals between January 1, 2001, and December 31, 2014. Of 414,463 mother-child pairs, 49,875 were excluded for lack of Kaiser Permanente membership by the age of 1 year or death of the child by the age of 1 year. The final cohort for this study was 364,588 mother-child pairs. Children were followed through KPSC EMR from birth until age 5 years. The KPSC health care system includes a diverse population of 4.5 million members throughout Southern California, and member demographic data reflect that of local census tracts [25]. Maternal social demographic and pregnancy health data, as well as the child’s ASD diagnosis, were extracted from KPSC’s integrated EMR.

The outcome of this study was whether a child had an ASD diagnosis before or at the age of 5 years and the age of the initial diagnosis in KPSC EMR. We chose the follow-up of children up to the age of 5 years because the majority of ASD cases were diagnosed by the age of 5 years, and prenatal exposure is likely to manifest its adverse effect on early developmental disorders. The diagnosis of ASD is recorded in EMR using the International Classification of Diseases (ICD) codes developed by the World Health Organization. KPSC transitioned from ICD-9 codes to ICD-10 on October 1, 2015. In EMR records dated before October 1, 2015, ASD diagnoses were identified by ICD-9 codes 299.0, 299.1, 299.8, and 299.9. After that date, ASD diagnosis was identified by ICD-10 codes F84.0, F84.3, F84.5, F84.8, and F84.9. The diagnostic codes included autistic disorders, Asperger syndrome, and pervasive developmental disorder not otherwise specified but excluded Rett syndrome or childhood disintegrative disorder. ASD diagnosis was determined if codes were present in the EMR at 2 or more separate health visits. This method was previously validated by an expert chart review with a positive predictive value of 88% [8,26-28].

Maternal preeclampsia during pregnancy was identified by ICD-9 codes 642.50-642.54, 642.60-642.64, and 642.40-642.44 and included eclampsia and hemolysis, elevated liver enzymes, low platelet count (HELLP) syndrome, a serious form of preeclampsia characterized by a low platelet count and elevated liver enzymes [29]. Preeclampsia is diagnosed when blood pressure measurements are >140 systolic or 90 diastolic and when a urine test reports >0.3 g of protein in 24 hours [24,30]. Severe cases involve blood pressure >160 systolic or 110 diastolic and 1 or more of the following: persistent headache, vision changes, thrombocytopenia (platelet count <100,000/mcL), impaired liver function, progressive renal insufficiency (serum creatinine >1.1 mg/dL or doubling of serum creatinine not explained by other known renal disease), and pulmonary edema [24,30]. The date of first occurrence (ie, onset defined for this study) of preeclampsia during pregnancy was extracted, and the corresponding gestational week of preeclampsia onset was calculated by subtracting the date of last menstrual period (LMP) from the date of first occurrence recorded in the EMR. Preeclampsia onset was categorized into the following three groups: (1) <34 weeks (diagnosis between 24 weeks 0 days and 33 weeks 6 days); (2) 34-37 weeks (34 weeks 0 days to 36 weeks 6 days); and (3) ≥37 weeks (≥37 weeks 0 days). This categorization reflected known gestational ages for fetal viability at birth and clinical management guidance for preeclampsia. The gestation of 24 weeks has been reported to be the lower boundary for infant survival [31,32], and expectant management guidelines aim to maintain pregnancies until 34 or 37 weeks, depending on the timing of preeclampsia onset and the severity of symptoms [30]. Expectant management duration was calculated as the number of days between the date of preeclampsia onset and the date of delivery.

Covariates selected to adjust for potential confounding were maternal age, self-reported race, ethnicity, and education; prepregnancy obesity, diabetes, and smoking during pregnancy; history of comorbidity (≥1 diagnosis of heart, lung, kidney, liver disease, or cancer); offspring sex; and census tract–level household income at child’s first birthday. These covariates were potential risk factors associated with child’s ASD risk, as shown in previous studies [21,24]. The birth year was also included as a covariate to account for trends of increasing ASD prevalence over the study period [8]. Maternal obesity was defined as a prepregnancy BMI ≥30 kg/m². Maternal prepregnancy BMI was calculated using maternal height and weight recorded in EMR from the date closest to LMP, with a window of 6 months before and 3 months after LMP [33]. Diabetes during pregnancy included preexisting type 1 or type 2 diabetes and gestational diabetes mellitus diagnosed before 26 weeks, as these were previously associated with ASD risk in this study sample [26,27].

Outcome variables were child’s ASD diagnosis by the age of 5 years and the age of ASD diagnosis. Exposure variables of interest were the gestational age of preeclampsia onset and days from preeclampsia onset to delivery. The gestational age of preeclampsia onset was analyzed as a continuous variable as well as a categorical variable categorized as <34 weeks, 34-37 weeks, and ≥37 weeks. Duration from preeclampsia onset to delivery was analyzed both as a continuous variable and a categorical variable, using the median number of days between onset and delivery for each onset group as the cutoff. Maternal and child characteristics by preeclampsia exposure status were reported as median and IQR for continuous variables and total number (n) and proportion (%) for categorical variables. Wilcoxon rank-sum tests and chi-square tests were used to assess differences in maternal and child characteristics between preeclampsia exposures.

Associations between preeclampsia exposure and child’s ASD risk were assessed using Cox regression models. Robust standard errors were used to correct for potential correlation between siblings born to the same mothers. Associations were quantified as hazard ratios (HRs) with 95% CI. We first assessed the risk of ASD associated with preeclampsia exposure (yes vs no), followed by assessing ASD risk associated with gestational age of onset within the preeclampsia group and comparing the risk of ASD in the <34 weeks, 34-37 weeks, and ≥37 weeks onset groups relative to the unexposed. We then examined ASD risk associated with duration from onset to delivery among those with preeclampsia exposure, adjusting for gestational weeks of the first occurrence of preeclampsia. All models adjusted for birth year, maternal age, self-reported race and ethnicity, educational qualifications, prepregnancy obesity, diabetes, smoking during pregnancy, census tract–level household income at child’s first birthday, and child’s sex. Birth year was modeled as a penalized spline to account for the nonlinear relationship between birth year and outcomes. We also assessed the roles of gestational age at delivery and birth weight as pathways to risk associated with early onset of preeclampsia and duration from onset to delivery by further adjusting these 2 variables.

Statistical significance was set at P<.05. All statistical analyses were performed in R software (version 3.6; R Foundation for Statistical Computing).

This study was approved by KPSC Institutional Review Boards (review #12075), with individual participant consent waived. All data analyzed were deidentified. No compensation was offered to individual participants. There was no community involvement in the study.

In this large sample and multiethnic clinical cohort study, children exposed to preeclampsia in utero were at higher risk of ASD than children who were not exposed, with a greater risk for children exposed to preeclampsia earlier in pregnancy. However, among children exposed to preeclampsia, there were no significant associations of risk of ASD with the number of days from first onset to delivery. This was analyzed both as a continuous variable adjusting for gestational age of preeclampsia first onset and as a categorical variable cut at the median number of days and stratified by gestational age of preeclampsia first occurrence. These results suggest that a child’s risk of ASD associated with preeclampsia is not increased by expectant management of preeclampsia in standard clinical practice.

To our knowledge, this study is the first to assess whether the reported ASD risk associated with maternal preeclampsia was due to preeclampsia onset or clinical management of preeclampsia during pregnancy. Current clinical management of preeclampsia advises maintenance of pregnancies until 34 or 37 weeks gestation, depending on the week of diagnosis and preeclampsia severity [24]. Expectant management of preeclampsia requires balancing maternal medical needs with intrauterine developmental requirements and reduction of infant comorbidities associated with preterm birth [34,35]. Planned preterm delivery of pregnancies complicated by preeclampsia improves maternal outcomes but increases the risk of neonatal admissions for prematurity [36]. Expectant management of moderate preeclampsia before 37 weeks gestation has been reported to extend pregnancies without increasing neonatal comorbidities; however, studies have reported the risk of stillbirth [37] and neonatal mortality [38] associated with expectant management for severe preeclampsia diagnosed before 34 weeks gestation. The results presented in our study demonstrate that early onset of preeclampsia is associated with a greater risk of ASD in offspring; however, clinical management for preeclampsia does not add additional risk, potentially eliminating one comorbidity of concern associated with preeclampsia exposure and expectant management.

Prenatal preeclampsia may affect child neurodevelopment by altering the maternal environment during fetal maturation. Preeclampsia has been observed to influence maternal immune activation [39], increasing circulation of maternal proinflammatory cytokines, and contributing to higher levels of oxidative stress, which have been associated with divergent neurodevelopment in children [40]. Studies have reported differences in the etiology of early and late-onset preeclampsia [41,42], as well as varying risks of adverse outcomes associated with the timing of exposure [43,44]. Early-onset preeclampsia may be driven by placental dysfunction [45], which occurs when a placental abnormality restricts blood flow, potentially due to suppression of estrogen-related receptor-gamma leading to vascular abnormalities [46]. Late-onset preeclampsia could be primarily a maternal hypertensive condition [45]. Placental inflammation and vascular dysfunction have been linked to an increased risk of ASD in children [47,48]. Therefore, differences in ASD risk by gestational week of onset may be explained by the influence of preeclampsia on specific stages of fetal neurodevelopment [19].

In our study, earlier onset of preeclampsia had larger HRs than onset in later pregnancy, which is consistent with results from a study of ASD risk and preeclampsia, which reported an increased risk of ASD in children exposed to preeclampsia earlier in pregnancy [19]. Adjustment for gestational age at delivery and birth weight attenuated the risk of ASD after early preeclampsia in our study, suggesting that there may be different mechanisms underlying the risk of ASD by the timing of preeclampsia exposure. Exposure to preeclampsia has been reported to be associated with an increased risk of ASD in offspring [49]. Preeclampsia is also associated with preterm birth, another risk factor for ASD [20]. Our results are consistent with previous results that prematurity and low birth weight are risk factors for ASD [6,20] and extend them to show that early preeclampsia is one of the root causes for prematurity, low birth weight, and associated ASD risk.

Our results concerning the increased risk of ASD associated with maternal preeclampsia during pregnancy are consistent with previous findings. However, the novel finding in our study is that the number of days between preeclampsia’s first occurrence and delivery was not associated with ASD risk in offspring after taking the gestational age of preeclampsia onset into account. Thus, the results of our study suggest that clinical expectant management of preeclampsia after diagnosis does not increase the risk of ASD in offspring. As this is the first study to report these results, future research is needed to determine whether this finding is consistent in other populations and to assess whether these associations vary by type of preeclampsia treatment or severity of illness. Our results provide evidence demonstrating the need to identify effective approaches to prevent the onset of preeclampsia, especially during early pregnancy, to mitigate risk not only to mothers but also to offspring.

A strength of this study is its large, clinical, and longitudinal birth cohort with characteristics reflecting census tract–level social and demographic information of Southern California. Comprehensive EMR data with detailed pregnancy history and dates allowed us to assess preeclampsia by gestational week of diagnosis and duration from diagnosis to delivery, and to adjust for relevant covariates. The continuity of care at KPSC minimizes the risk of ascertainment bias in exposures and outcomes. We think the assessment of expectant management of preeclampsia on ASD risk is novel.

This observational study has some limitations. The results presented here do not establish a causal link between preeclampsia exposure and ASD risk. The severity of preeclampsia was not explicitly considered due to the lack of a clear definition of severity in EMR. However, diagnosis to delivery periods allowed some inference about severity, as severe cases were likely to be delivered more quickly than moderate cases [24]. Genetic information was unavailable; therefore, we were unable to control for genetic contributions to ASD risk. There may be other perinatal risk factors or postnatal environmental exposures not adjusted for in these analyses.

In this population-based retrospective clinical birth cohort study, exposure to preeclampsia in utero was associated with an increased risk of ASD in offspring, with a greater risk for children of mothers with preeclampsia occurring earlier during pregnancy. However, among children of mothers with preeclampsia, the number of days between preeclampsia diagnosis and delivery was not associated with increased ASD risk. Our study suggests that clinical management of pregnancies with preeclampsia does not increase the risk of ASD in offspring. Future research into the prevention of preeclampsia is still needed.