Интеллектуальные расстройства и расстройства в развитии диагностируются в Калифорнии и неблагоприятно сказываются на материнском здоровье.

INTRODUCTION

Intellectual and developmental disability (IDD) is an umbrella term covering conditions defined by atypical intellectual growth trajectories present from birth or early childhood. IDD includes but is not limited to conditions such as autism spectrum disorder (ASD), cerebral palsy (CP), chromosomal differences such as Trisomy 21, and intellectual disability (ID). Although recent estimates from a Medicaid birth cohort estimated the birth rate for women with IDD from 2011 to 2022 to be 9–10 births per 1000 [ 1 ], maternal health outcomes among pregnant people with IDD remain poorly documented [ 2 ].

In the small literature to date examining IDD and maternal health, concerning findings have emerged. Population‐based studies drawing on healthcare claims from Ontario [ 3 ], Wisconsin Medicaid [ 4 ], Washington state [ 5 ], and Sweden [ 6 ] have each found elevated risk of severe maternal morbidity (SMM), preeclampsia, and cesarean delivery among women with IDD. These adverse maternal outcomes, including SMM, cesarean delivery, preeclampsia, and hospital readmission, can have significant short‐ and long‐term health consequences for both mother and child [ 7 , 8 ]. Most of this evidence, however, analyzed IDD as a single category or focused more narrowly on ID, including a study by our group [ 9 ]. Few studies have examined individual subtypes. Subtype‐specific analyses are important because IDD conditions differ substantially in the underlying physiology, comorbidity burden, and healthcare access patterns, all of which contribute differently to pregnancy risk [ 10 ]. Swedish birth registry data suggest women with ASD face elevated risk of preeclampsia and cesarean delivery [ 11 ]. US Healthcare Cost and Utilization Project (HCUP) data point to increased cesarean delivery and preeclampsia risks among women with Trisomy 21 [ 12 ]. A single study has examined pregnancy outcomes among women with CP [ 13 ]. The evidence base for subtypes remains fragmented, and no study to date has examined all major IDD subtypes within a single analytic framework.

Another methodologic consideration of IDD is the distinction between confounders and mediators, as adjustment for the latter can attenuate effect estimates and obscure intervention targets [ 14 ]. Clarifying which prenatal clinical factors, such as chronic hypertension or gestational diabetes, mediate these associations may help identify intervention targets during pregnancy. We have previously examined causal mediation analyses of maternal ID and adverse maternal outcomes [ 9 ]. The present analyses extend that work by including all IDD subtypes, using an additional year of data, and employing an analytic framework to allow for more complex confounding structures in mediation analyses. This present study had two objectives. First, we aimed to estimate the association between maternal IDD, including specific IDD subtypes, and SMM, preeclampsia, cesarean delivery, and hospital admission within one year of delivery. Second, we aimed to assess the extent to which select prenatal clinical characteristics mediate the relationship between IDD and adverse maternal outcomes.

MATERIALS AND METHODS

Source of the sample

The present study used data from the Study of Outcomes in Mothers and Infants (SOMI), a population‐based cohort and near complete collection of all deliveries ≥22 weeks gestation in the state of California between 2007 and 2021, with maternal postpartum data through 2022. SOMI contains vital statistics data probabilistically linked onto maternal and infant inpatient hospital and emergency department discharge summaries from the California Department of Healthcare Access and Information (HCAI), and has been described previously [ 15 ]. In the present analysis, we included singleton deliveries with both live born infants and fetal deaths.

Exposure, outcomes, and covariates

Maternal intellectual and developmental disabilities were defined and grouped via International Classification of Diseases (ICD9/10) codes defined previously [ 3 ]. To disaggregate IDD into subtypes based on etiology, we used the following categorization: ASD, CP, ID, chromosomal differences, and a group for other IDDs. The chromosomal differences group included Trisomy 21, autosomal deletion syndromes, partial trisomy, duplications with other complex rearrangements, and other conditions due to chromosomal anomalies. To reduce the chance of misclassification from the child's ICD codes appearing in the maternal record, we removed any ICD codes incompatible with reproduction or life (e.g., Trisomy 18). Frequencies of specific genetic conditions in this group are available in Supplement S1 . Other IDDs included but were not limited to fetal alcohol syndrome (FAS), tuberous sclerosis, and congenital malformation syndromes. IDD was analyzed as both an umbrella term and as disaggregated groups by the above categorizations. IDD subtypes were not mutually exclusive. Mothers were included in the exposed group of all analyses for which they had a diagnosis code. All diagnostic codes used to identify maternal IDD are listed in Table S1 .

Maternal adverse outcomes were captured via a combination of vital statistics data and ICD9/10 codes. SMM was defined using the US Centers for Disease Control and Prevention (CDC) 21 indicator ICD9/10 coded definitions for SMM [ 16 ]. As a sensitivity analysis, we analyzed SMM excluding blood transfusions as blood transfusions generally make up the majority of SMM cases and are not always indicative of severe outcomes [ 17 ]. Preeclampsia and cesarean delivery were identified using ICD codes from maternal and infant diagnostic codes in discharge data and relevant birth certificate indicators. Maternal hospital admission was identified from inpatient hospital discharge summaries in the 365 days after delivery. All diagnostic codes used to identify adverse maternal outcomes are listed in Table S1 .

Mediators included in causal mediation analyses were selected based on feasibility as a prepregnancy intervention target among pregnant women and clinical expertise of the study team. The exception was preeclampsia, which was added post hoc based on findings as an outcome and its potential to increase the risk of other select outcomes (e.g., cesarean delivery). Mediators were all perinatal characteristics and included tobacco use, pre‐existing hypertension, pre‐existing diabetes mellitus, preeclampsia, prepregnancy body mass index (BMI), epilepsy, depression/anxiety, bipolar disorder or schizophrenia, and adequate prenatal care. Adequate prenatal care was operationalized using the Kotelchuck Index and dichotomizing the index as inadequate or intermediate care versus adequate and adequate plus care [ 18 ]. All other mediators were defined using a combination of diagnostic codes (Table S1 ) or vital statistics data. Mediators were included in each model only when cell sizes of the joint distribution of exposure‐mediator‐outcome were greater than 5.

Covariates collected to describe the analytic cohorts and adjust for mediator‐outcome confounding in the mediation analyses included maternal age at delivery (continuous), year of delivery, maternal education (< 12 years, ≥12 years), county‐level rurality, maternal race/ethnicity, prepregnancy BMI (≥25 kg/m 2 and < 25 kg/m 2 ) and maternal supplemental nutrition program (Women, Infants, and Children [WIC]) enrollment as a proxy for socioeconomic status. Rurality was defined at the county level using the Federal Information Processing Standards (FIPS) urban/rural county code continuum where 1 is the most urban and 6 is the most rural. Race was self‐reported and defined into five categories: Asian, Hispanic, non‐Hispanic Black, non‐Hispanic White, and an ‘Other’ category. The ‘Other’ category was composed of Hawaiian or Pacific Islander, Native American or Alaska Natives, and self‐reported ‘other’ race. All covariates were collected from the birth certificate.

Statistical analysis

To reduce computational burden in the mediation analyses, six 1:30 (IDD:no IDD) exposure‐matched datasets were constructed, one for each type of IDD and one for overall IDD. Maternal age at delivery (< 35 and ≥35) and year of delivery were used as matching factors. Although not traditional confounders (they did not cause IDD), maternal age at delivery and year of delivery may be associated with diagnostic trends and were therefore included as matching factors. The matched datasets were used for all analyses including the outcome models, causal mediation models, and quantitative bias analysis. We estimated adjusted risk ratios (aRR), matched only on maternal age and year of delivery, in each matched dataset to estimate the association between maternal IDD and its subtypes and adverse maternal outcomes. Adjusted risk ratios and accompanying 95% confidence intervals (CIs) were estimated using modified Poisson regression [ 19 ].

We conducted both single and multiple causal mediation analyses. Causal mediation analysis allows us to deconstruct the excess risk of adverse maternal outcomes relative to each hypothesized mediator [ 20 , 21 ]. Additional information and assumptions underlying causal mediation analyses are detailed in Supplement S2 . We used marginal structural models to account for exposure‐induced confounding of the mediator‐outcome relationships and a potential outcomes framework to estimate effects and bootstrapping to estimate 95% CIs [ 22 ]. Joint mediation models included all mediators that met cell size inclusion for the single mediation models. We did not include tests of interaction between mediators in the joint mediation analyses due to sample size limitations. In all mediation models, maternal age (continuous) and year of delivery were included as exposure‐outcome confounders, while covariates included as confounders of the mediator outcome were maternal education, county‐level rurality, maternal race, year of delivery, maternal age, preeclampsia (when preeclampsia was not the outcome), and WIC enrollment. WIC enrollment and preeclampsia were included as exposure‐induced confounders. For single mediation models where preeclampsia was the mediator, the following variables were included as exposure‐induced confounders: maternal BMI, pre‐existing diabetes, pre‐existing hypertension, depression or anxiety diagnoses, maternal epilepsy, bipolar disorder or schizophrenia diagnoses, and maternal tobacco use during pregnancy. A complete case analysis was used in all regression models.

We conducted deterministic quantitative bias analyses to estimate the level of misclassification of IDD and its subtypes that would be needed to attenuate the risk ratios we found to the null (RR = 1.0). Specifically, we assumed reduced maternal IDD sensitivity among deliveries without an adverse maternal outcome (i.e., differential misclassification of the exposure), and estimated the sensitivity needed to explain the observed associations. To examine robustness to the possibility of the child's ICD code being included on the maternal record for cases of chromosomal differences, we performed a non‐differential misclassification analysis, setting sensitivity to 1.0 and reducing specificity to test findings.

All analyses other than the mediation analyses were completed using StataNow/SE 19.5 [ 23 ]. Causal mediation analysis was performed using the R user‐created program, “CMAverse” [ 24 ]. The study was approved by the Committee for the Protection of Human Subjects within the Health and Human Services Agency of the State of California, and by the Institutional Review Board at the University of California, San Diego. A waiver of consent was granted for secondary use of data not collected for research. This study follows Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [ 25 ] and the Guideline for Reporting Mediation Analyses of Randomized Trials and Observational Studies (AGReMA) recommendations [ 26 ]. Throughout this paper, we use “maternal,” “mother,” and “women” to align with epidemiologic terminology while recognizing that pregnancy is not limited to cisgender women.

RESULTS

Cohort characteristics

The full cohort included 6,435,742 singleton births, of which 4492 (0.07%) were mothers with IDD (Figure  1 ). Among women with IDD, 47 pregnancies (1%) resulted in a fetal death. Among women with IDD, the most common subtype was chromosomal differences ( n  = 1350), followed by ID ( n  = 1187), CP ( n  = 1019), other IDD ( n  = 600), and ASD ( n  = 458). Co‐occurring IDD diagnoses were present in a subset: 49 individuals had both ID and CP, 33 had both ID and ASD, 20 had both ID and chromosomal differences, and 18 had other IDD combinations. Maternal IDD prevalence rose across the study period, from 5 per 10,000 live births in 2007 to 9 per 10,000 in 2021. In the matched cohort, compared to women without IDD, those with IDD were more likely to be non‐Hispanic White or non‐Hispanic Black, publicly insured, WIC‐enrolled, use tobacco during pregnancy, have a BMI ≥25 kg/m 2 , have chronic/pre‐existing hypertension, and diabetes (Table  1 ). Psychiatric and neurological comorbidities, including anxiety, depression, bipolar disorder, schizophrenia, and epilepsy, were also more prevalent among women with IDD (Table  1 ).

STROBE Diagram. ASD, autism spectrum disorder; CP, cerebral palsy; ID, intellectual disability; IDD, intellectual and developmental disability; Other IDD, fetal alcohol syndrome, tuberous sclerosis, and various congenital malformations noted in Table S1 . Counts include fetal deaths and IDD subtypes are not mutually exclusive.

TABLE 1 Demographic and clinical characteristics among people with intellectual and developmental disability and those without during pregnancy. No IDD
a

IDD
N  = 134,760
N  = 4492 Maternal demographics Maternal age at birth, mean (SD) 30 (6) 29 (7) Race/ethnicity, n (%) Asian 19,436 (14%) 259 (6%) Hispanic 65,537 (49%) 1859 (41%) Non‐Hispanic Black 6174 (5%) 427 (10%) Non‐Hispanic White 36,192 (27%) 1584 (35%) Other
b

7421 (6%) 363 (8%) Year of delivery, mean (SD) 2014 (4) 2014 (4) Maternal education at birth, n (%) <12 years 22,440 (17%) 845 (19%) ≥12 years 106,195 (79%) 3338 (74%) Missing 6125 (5%) 309 (7%) WIC status, n (%) No WIC 70,847 (53%) 1840 (41%) WIC 62,838 (47%) 2585 (58%) Missing 1075 (1%) 67 (1%) Payer, n (%) Private 73,508 (55%) 1833 (41%) Public 61,236 (45%) 2630 (59%) Other 16 (0%) 29 (1%) Rurality (see notes), median (IQR) 1 (1‐2) 1 (1‐3) Maternal IDD subtypes
c

Autism spectrum, n (%) NA 458 (10%) Intellectual disability, n (%) NA 1187 (26%) Cerebral palsy, n (%) NA 1019 (23%) Chromosomal differences, n (%) NA 1350 (30%) Other IDD
d
, n (%) NA 600 (13%) Maternal clinical characteristics Adequate Prenatal Care (Dichotomized Kotelchuck Index), n (%) Inadequate or intermediate 33,235 (25%) 1241 (28%) Adequate or Adequate+ 97,298 (72%) 3002 (67%) Missing 4227 (3%) 249 (6%) Body mass index (dichotomized) <25 65,723 (49%) 1905 (42%) ≥25 63,602 (47%) 2326 (52%) Missing 5435 (4%) 261 (6%) Pre‐existing hypertension, n (%) 3004 (2%) 281 (6%) Pre‐existing diabetes, n (%) 1688 (1%) 156 (3%) Tobacco use during pregnancy, n (%) 560 (< 1%) 113 (3%) Anxiety or depression diagnosis, n (%) 5154 (4%) 709 (16%) Bipolar disorder, n (%) 530 (< 1%) 303 (7%) Schizophrenia, n (%) 90 (< 1%) 181 (4%) Bipolar disorder or schizophrenia, n (%) 578 (< 1%) 382 (9%) Epilepsy, n (%) 464 (< 1%) 316 (7%) Fetal death
e

0 (0%) 47 (1%) Abbreviations: BMI, body mass index; IDD, intellectual and developmental disabilities; NICU, neonatal intensive care unit; Rurality, Federal Information Processing Standards (FIPS) urban/rural county code continuum, 1 (most urban) to 6 (most rural), measured at the county level; WIC, USDA's Special Supplemental Nutrition Program for Women, Infants, and Children.
a
Matched 30:1 on maternal age and year of delivery.
b
Other race category included American Indian/Alaska Native, Hawaiian/Pacific Islander, self‐reported “other” race group, two or more races, not stated/unknown.
c
IDD subtypes are not mutually exclusive.
d
Other IDD is composed of fetal alcohol syndrome, tuberous sclerosis, and various congenital malformations noted in Table S1 .
e
Fetal death (autism: n  < 11; ID: n  = 12; CP: n  < 11; chromosomal differences: n  = 21; other IDD: n  < 11).

Demographic and clinical characteristics among people with intellectual and developmental disability and those without during pregnancy.

Abbreviations: BMI, body mass index; IDD, intellectual and developmental disabilities; NICU, neonatal intensive care unit; Rurality, Federal Information Processing Standards (FIPS) urban/rural county code continuum, 1 (most urban) to 6 (most rural), measured at the county level; WIC, USDA's Special Supplemental Nutrition Program for Women, Infants, and Children.

Matched 30:1 on maternal age and year of delivery.

Other race category included American Indian/Alaska Native, Hawaiian/Pacific Islander, self‐reported “other” race group, two or more races, not stated/unknown.

IDD subtypes are not mutually exclusive.

Other IDD is composed of fetal alcohol syndrome, tuberous sclerosis, and various congenital malformations noted in Table S1 .

Fetal death (autism: n  < 11; ID: n  = 12; CP: n  < 11; chromosomal differences: n  = 21; other IDD: n  < 11).

Characteristics differed across IDD subtypes (Table S2 ). Women with chromosomal differences were oldest at delivery (median: 32 years), while those with ASD were youngest (median: 26 years). Educational attainment of at least 12 years was highest among women with CP (82%) and lowest among those with ID (62%). Adequate prenatal care was most common among women with chromosomal differences or other IDD (75%) and least common among those with ID (52%). Anxiety and/or depression were most prevalent among women with ASD (34%) and ID (23%), and least prevalent among those with chromosomal differences (9%). A similar pattern was observed for bipolar disorder and/or schizophrenia, with the highest rates among women with ASD and ID (both 21%) and the lowest among those with chromosomal differences (1%). Epilepsy was most frequent among women with CP and ID (13% and 12%, respectively), with a notably elevated rate also among those with ASD (9%).

The incidence of all adverse maternal outcomes was elevated among women with IDD and its subtypes versus women without (Tables  2 and  3 ). This included SMM (5% vs. 2%, aRR: 2.77, 95% CI: 2.40, 3.19), cesarean delivery (44% vs. 30%, aRR: 1.44, 95% CI: 1.39, 1.50), preeclampsia (10% vs. 4%, aRR: 2.52, 95% CI: 2.30, 2.77), admission to the hospital within 1 year of delivery (13% vs. 3%, aRR: 4.64, 95% CI: 4.26, 5.05). Among IDD subtypes, several notable patterns emerged. Women in the Other IDD group had the highest incidence of SMM (7%), which remained the case when transfusions were excluded (4%). Women with ID or Other IDD shared the highest risk of preeclampsia (12% each). Cesarean delivery was most common among women with cerebral palsy (49%), while women with ID had the highest rate of hospital admission within one year of delivery (25%). The highest RRs were observed for maternal hospital admission within one year among both women with ID (aRR 8.55, 95% CI: 7.59, 9.63) and women with ASD (RR 8.04, 95% CI: 6.50, 9.93). All other estimates for each subgroup and outcome were elevated above the null.

TABLE 2 Incidence of maternal adverse pregnancy outcomes (California, 2007–2021), by maternal IDD type. No IDD Any IDD Autism spectrum Cerebral palsy Chromosomal differences Intellectual disability Other IDD
N  = 134,760
N  = 4492
N  = 458
N  = 1019
N  = 1350
N  = 1187
N  = 600 SMM 2210 (2%) 204 (5%) 19 (4%) 37 (4%) 49 (4%) 62 (5%) 43 (7%) SMM (no transfusion) 957 (1%) 111 (2%) 11 (2%) <11 (< 1%) 31 (2%) 39 (3%) 23 (4%) Cesarean delivery 40,907 (30%) 1969 (44%) 168 (37%) 501 (49%) 560 (41%) 531 (45%) 266 (44%) Preeclampsia 5517 (4%) 464 (10%) 46 (10%) 109 (11%) 107 (8%) 141 (12%) 71 (12%) Maternal admission
a

3679 (3%) 569 (13%) 90 (20%) 96 (9%) 62 (5%) 292 (25%) 70 (12%) Abbreviations: IDD, Intellectual and developmental disability; SMM, severe maternal morbidity.
a
Maternal admission within 1 year of delivery.

Incidence of maternal adverse pregnancy outcomes (California, 2007–2021), by maternal IDD type.

Abbreviations: IDD, Intellectual and developmental disability; SMM, severe maternal morbidity.

Maternal admission within 1 year of delivery.

TABLE 3 Risk ratios for maternal outcomes among pregnant people with and without IDD, by IDD type (California, 2007–2021). Matched risk ratio [95% CI] SMM SMM (no transfusions) Cesarean delivery Preeclampsia Maternal admission
a

Any IDD 2.77 3.48 1.44 2.52 4.64 [2.40, 3.19] [2.86, 4.23] [1.39, 1.50] [2.30, 2.77] [4.26, 5.05] Autism spectrum 2.51 3.37 1.33 2.32 8.04 [1.56, 4.05] [1.82, 6.23] [1.17, 1.51] [1.74, 3.08] [6.50, 9.93] Intellectual disability 3.36 5.06 1.51 3.00 8.55 [2.60, 4.34] [3.63, 7.07] [1.41, 1.62] [2.54, 3.54] [7.59, 9.63] Cerebral palsy 2.33 1.53 1.61 2.77 3.37 [1.67, 3.27] [0.77, 3.05] [1.50, 1.74] [2.27, 3.37] [2.75, 4.12] Chromosomal differences 2.11 2.85 1.29 1.74 1.65 [1.59, 2.80] [1.98, 4.10] [1.21, 1.38] [1.43, 2.10] [1.29, 2.13] Other IDD 4.02 4.54 1.50 2.62 4.14 [2.94, 5.49] [2.95, 6.99] [1.36, 1.66] [2.08, 3.32] [3.28, 5.23]
Note : Samples were matched 30:1 (no IDD:IDD) on maternal age (≥35, <35) and year of delivery. Abbreviations: CI, confidence interval; IDD, intellectual and developmental disability; SMM, severe maternal morbidity.
a
Maternal admission within one year of delivery.

Risk ratios for maternal outcomes among pregnant people with and without IDD, by IDD type (California, 2007–2021).

Note : Samples were matched 30:1 (no IDD:IDD) on maternal age (≥35, <35) and year of delivery.

Abbreviations: CI, confidence interval; IDD, intellectual and developmental disability; SMM, severe maternal morbidity.

Maternal admission within one year of delivery.

In the QBA, estimates were robust to low sensitivity among women without adverse outcomes (sensitivity range: 0.11–0.67; Table S3 ). Misclassification of specificity would only tolerate a reduction to 96%–98% (Table S4 ) due to small cell sizes becoming quickly reduced below zero. Notably, however, risk ratios were increasing, not decreasing, with reduced specificity.

The results of the causal mediation analyses are presented in Figure  2 and Tables S5–S9 . The strongest single mediators between IDD and its subtypes with adverse outcomes were pre‐existing/chronic hypertension and mental health conditions, particularly for preeclampsia and SMM. For SMM, depression/anxiety mediated 35% of the effect of ASD and 31% of the effect of ID. For preeclampsia, pre‐existing hypertension mediated 35% of the effect of ID, 40% of other IDDs, and 45% of the effect of CP. Preeclampsia, when considered as a mediator in the SMM model, was also a particularly strong mediator, explaining 23%–36% of the observed excess risk across the subgroups. The proportion mediated did not meaningfully change when blood transfusions were excluded from the SMM outcome (Table S9 and Figure S1 ).

Proportion of total effect of maternal intellectual and developmental disabilities (IDDs) mediated by maternal perinatal characteristics. Mediators are included when the sample is stratified by the exposure, outcome, and mediator, and all cells are ≥5; cells in gray indicate insufficient sample; * =  p  < 0.05; BMI, body mass index, and dichotomized as underweight or normal BMI versus overweight or obese BMI.

Results for multiple mediation analyses (joint estimates, Figure  2 ) varied greatly by IDD/IDD‐subtype and outcome. For women with autism, perinatal characteristics explained 65% of the excess risk of cesarean delivery, 54% of the excess risk of preeclampsia, and 42% of the excess risk of hospital admission. For women with ID, 39% of the excess risk of cesarean delivery, 52% of the excess risk of hospital admission, and 71% of the excess risk of SMM were explained by selected perinatal characteristics.

DISCUSSION

This study presents the largest analysis to date of maternal IDD and adverse maternal pregnancy outcomes. Using longitudinal data from California between 2007 and 2021, we found that women with IDD documented in birth records, compared with those who do not have documented IDD, experienced higher risks of all adverse maternal outcomes examined. Elevated risk persisted across all IDD subtypes, and risk ratios were particularly high among women with ASD and women with ID when maternal hospital admission within 1 year was the outcome. Hypertensive disorders and depression/anxiety were important mediators and represent potential targets for intervention.

Prior studies document a consistent pattern of elevated risk across multiple outcomes and geographic populations, with evidence drawn from both IDD broadly and specific IDD subtypes [ 3 , 4 , 5 , 6 , 11 , 12 , 13 ]. Existing studies consistently find elevated risks across conditions, including ID [ 5 , 6 ], CP [ 13 ], and autism [ 11 ]. The one exception is Trisomy 21 [ 12 ], where a small prior study found no elevated preeclampsia risk. Additional discussion of previous studies is provided in Supplement S3 .

Causal mediation analysis has not, to our knowledge, been previously applied to IDD and adverse maternal outcomes, apart from our own prior work on ID using a different analytic framework [ 9 ]. In the present study, we found that hypertensive disorders and depression/anxiety were important mediators of the relationship between maternal IDD and adverse pregnancy outcomes, suggesting that treatment of these conditions may represent meaningful intervention targets in this population. The American College of Obstetricians and Gynecologists (ACOG) recommends regular monitoring of women with pre‐existing hypertension, particularly in the second half of pregnancy, with assessment for end‐organ damage, avoidance of excessive sodium and caffeine intake, and smoking cessation [ 27 ]. Oral antihypertensives may be indicated depending on the degree of blood pressure elevation, and treatment before or early in pregnancy optimizes outcomes [ 28 ]. For depression and anxiety, ACOG recommends psychotherapy as first‐line treatment and pharmacotherapy where indicated, and advises ongoing evaluation of treatment response throughout pregnancy [ 29 ]. Untreated or undertreated depression and anxiety in pregnancy are associated with negative outcomes for the mother and fetus [ 29 ]. Our study was unable to evaluate whether women with IDD are more likely to enter pregnancy with untreated mental health or hypertension.

This study points to two directions for future research. First, given that one in five women with ASD and one in four with ID were admitted to the hospital within one year of delivery, further studies should identify the timing and causes of admission in this population and approaches for prevention. Hospital admission is disruptive for patients and their families and is designated an important quality of care metric by the Society for Maternal and Fetal Medicine [ 30 ]. Second, because the dataset lacks information on treatment and the timing of comorbid conditions, future studies should examine specific pharmacological and non‐pharmacological treatments, the onset of relevant conditions, and the sequencing of interventions. This is work that would refine treatment targets.

Strengths

This study was population‐based and all deliveries in California between 2007 and 2021 were eligible for inclusion. The population‐based design and inclusion of fetal deaths after 22 weeks reduce the probability of selection bias and may improve generalizability beyond studies relying on Medicaid records alone. The large size of the dataset also enabled disaggregation of IDD into subtypes, providing a more nuanced analysis than has been previously possible. Our data also included information on numerous comorbid conditions and behavioral risk factors such as tobacco use, which enabled the mediation analysis and quantification of potential intervention targets. We also completed a QBA that showed that relatively frequent and differential misclassification would be needed to nullify our findings.

Limitations

These findings should be interpreted in the context of several limitations. We relied on diagnostic coding in maternal hospital discharge records to identify women with IDD and to identify many of our mediators, and the validity of the ICD codes is unknown. This case ascertainment method for IDD may have selectively captured women with more extensive support needs, particularly those relevant to pregnancy and delivery [ 31 , 32 ]. This means our findings may only generalize to populations of women with IDD with the highest support needs, and likely overstate the risk of these outcomes among individuals with less severe IDD. This may also partly explain why our risk ratio estimates are higher than those of previous studies, most of which had access to both outpatient and inpatient diagnostic codes or data from public disability assistance services. Moreover, sociodemographic characteristics, including race/ethnicity, are not causes of IDD and were not included in total effect estimates. However, any sociodemographic biases in the receipt of select diagnoses (e.g., ID or ASD) are subsumed in the total effect estimates and may bias results in either direction. Additionally, for IDD subtypes such as chromosomal differences where fetal diagnosis is possible in utero, some diagnostic codes in the maternal record may reflect the condition of the developing fetus rather than the mother. For example, a Trisomy 21 diagnosis may have been recorded for the fetus, and without prepregnancy diagnostic code data, we were unable to resolve this. However, this potential misclassification should bias our estimates towards the null. If mediators, such as anxiety or depression and tobacco use, are underdiagnosed in hospital discharge records, our estimates of the indirect effects of these mediators and thus the proportion mediated would be underestimated. Underestimation is likely given previous studies showing that inpatient hospital records have high specificity but low sensitivity in classifying some of our mediators, particularly for depression [ 33 ]. Additionally, while we largely prioritized chronic conditions as mediation targets, other pregnancy complications (e.g., gestational hypertension, gestational diabetes) are likely subsumed in the direct effects. Future work to further disaggregate these pathways is warranted. Finally, our study also remains subject to selection bias caused by early fetal loss or reduced fertility among women with IDD [ 1 , 34 ].

CONCLUSION

Women with versus without IDD face substantially elevated risks of adverse maternal outcomes, with the largest disparities observed for hospital admission within 1 year of delivery. Hypertensive disorders and depression/anxiety were the strongest mediators of these disparities, warranting further evaluation as potential intervention targets aimed at reducing morbidity in this population. Taken together, these results highlight the need for targeted perinatal care strategies and support for individuals with IDD.

CONFLICT OF INTEREST STATEMENT

K.P. has research contracts with AbbVie, GSK, Pfizer, and Sanofi, none of which are related to the current study. T.W. serves on the Clinical Scientific Advisory Board for Delfina Care, Inc on work that is not related to the current study. All other authors declare no conflicts of interest.

ETHICS STATEMENT

The study was approved by the Committee for the Protection of Human Subjects within the Health and Human Services Agency of the State of California, and by the Institutional Review Board at the University of California, San Diego. A waiver of informed consent was granted by the institutional review board bodies because the research posed minimal risk to participants, did not adversely affect the rights of participants, and was impractical given the number of individuals.

Supporting information

Supporting Information

ACKNOWLEDGMENTS

This work was funded by an award from the National Institutes of Health (R01HD112525).

REFERENCES

Associated Data

Supplementary Materials

Supporting Information