Impact of social, health, and disability-related factors on pregnancy outcomes in women with intellectual and developmental disabilities: A population-based latent class analysis

Abstract

Background:

Studies have shown women with intellectual and developmental disabilities (IDD) have elevated risks of perinatal complications, but few studies have examined how social, health, and disability-related factors affect these risks.

Objectives:

To identify and describe subgroups of pregnant women with IDD according to social, health, and disability-related factors and examine the risks of perinatal complications in these subgroups compared to women without IDD.

Methods:

We performed a population-based cohort study in Ontario, Canada, of women with (n=1,922) and without (n=1,126,854) IDD, with a singleton birth in 2003-2018. We used latent class analysis (LCA) to identify subgroups of women according to social (e.g., age), health (e.g., chronic medical conditions), and disability-related (e.g., IDD type) characteristics. Modified Poisson regression was then used to compare the risks of hypertensive disorders of pregnancy, caesarean delivery, and preterm birth across identified subgroups to women without IDD.

Results:

The LCA identified 4 classes of women with IDD: (1) young women who were mostly healthy and had little primary care before pregnancy (n=253); (2) older women who were mostly healthy (n=795); (3) young to mid-aged women who had significant comorbidities (n=181); and (4) young women, many of whom were autistic, who had some medical comorbidities and significant psychiatric comorbidities (n=693). Class 3 consistently had the greatest risks of perinatal complications, across all IDD groups, compared to women without IDD.

Conclusions:

These findings underscore the importance of multidisciplinary care approaches tailored to the needs of at-risk women with IDD, in the preconception and perinatal periods.

INTRODUCTION

Historically, women with intellectual and developmental disabilities (IDD) experienced many barriers to pregnancy, including institutionalization and involuntary sterilization.¹ The establishment of community-based living² and international affirmation of the reproductive rights of people with disabilities³ have resulted in increasing opportunities for childbearing for women with IDD. About 1.3% of reproductive-aged women in North America have IDD,⁴ and pregnancy rates in young women with IDD are comparable to those without IDD.⁵ As such, an understanding of their pregnancy outcomes is critical for informing accessible pregnancy care.

Women with IDD experience a range of social and health disparities, including high rates of poverty, chronic physical and mental health conditions, and systemic barriers accessing health care.^6,7 These social and health disparities are well-known risk factors for maternal and neonatal complications.^8,9 Meta-analyses have shown higher risks of perinatal complications in women with IDD compared to those without IDD, with complications most frequently examined being hypertensive disorders of pregnancy, caesarean delivery, and preterm birth.^10,11 Yet, few studies have examined the collective impact of social and health disparities, or of disability-related factors, on perinatal outcomes in women with IDD. This is a critical omission: information on heterogeneity in perinatal outcomes according to social, health, and disability-related factors would be useful for developing data-driven guidance on who needs enhanced perinatal supports and what factors such supports might target. Instead, studies tend to examine social and health disparities as outcomes among women with IDD⁷ or perinatal outcomes among women with IDD controlling for social and health disparities.^10,11 Only two studies have investigated heterogeneity in perinatal outcomes in women with IDD according to other factors, finding that risks of preterm birth in women with IDD were greatest among those with mental illness¹² and Australian Aboriginal status.¹³ Further research is needed to understand how a broader range of social, health, and disability-related factors act together to influence outcomes.

The perinatal health framework for women with disabilities proposed by Mitra et al.¹⁴ informs consideration of social, health, and disability-related factors in the perinatal outcomes of women with IDD. This framework integrates multiple determinants of pregnancy-related health, and stresses the importance of interactions across disability, comorbidities, and the larger social context in producing perinatal outcomes. This framework aligns with a life-course perspective on perinatal health,¹⁵ and with social determinants of health frameworks.¹⁶ The perinatal health framework has never been operationalized in population-based data as a way to identify subgroups of women with IDD who may benefit most from perinatal supports. Using this framework, we (1) identified and described clinically relevant subgroups of pregnant women with IDD according to social, health, and disability-related characteristics and (2) examined the risks of hypertensive disorders of pregnancy, caesarean delivery, and preterm birth across these subgroups, compared to women without IDD.

METHODS

Study Design and Setting

We performed a population-based cohort study using health administrative data in Ontario, Canada, according to REporting of studies Conducted using Observational Routinely-collected health Data guidelines.¹⁷ Ontario is Canada’s most populous province, with a universal health insurance plan that covers all medically necessary health care, including perinatal care, for all 14.7 million residents.¹⁸

Data Sources

Data were accessed and analyzed at ICES, a not-for-profit organization that holds health administrative data for the province of Ontario.¹⁹ Primary hospital diagnoses, physician billing claims, and sociodemographic data recorded in ICES datasets are complete and accurate.²⁰ We used the MOMBABY dataset to identify obstetrical deliveries to women with and without IDD. MOMBABY captures maternal-newborn records for hospital births—i.e., 98% of births.¹⁸ This information was linked to hospitalizations, emergency department visits, physician visits, and Census data using a unique encoded identifier (Table S1). The data were authorized for research use under section 45 of Ontario’s Personal Health Information Protection Act, which does not require research ethics board approval.

Study Population

The study cohort was comprised of women with and without IDD aged 15 to 49 years with a singleton livebirth with a conception date between April 1, 2003 and March 31, 2018, and who delivered in hospital after 20 weeks gestational age. One birth per woman was randomly selected from this cohort for inclusion to avoid issues of clustering in the analyses.²¹

Variables

Maternal IDD status was identified using an algorithm developed in previous research,²² and applied in other pregnancy studies,^23,24 where IDD was considered present if a relevant diagnosis (i.e., autism spectrum disorder, chromosomal anomalies causing intellectual disability, fetal alcohol spectrum disorder, other intellectual disabilities) was recorded in ≥ 2 physician visits, ≥ 1 emergency department visits, or ≥ 1 hospitalizations, from database inception to conception.²⁵ The comparator group was women without IDD.

Outcomes were indicators of maternal, delivery, and neonatal complications that are elevated in women with IDD^10,11 and that have important implications for long-term maternal and child health:^8,9 (1) hypertensive disorders of pregnancy (i.e., gestational hypertension, preeclampsia, or eclampsia), (2) caesarean delivery, and (3) preterm birth < 37 weeks gestation.

Social, health, and disability-related characteristics were identified to address Mitra et al.’s perinatal health framework.¹⁴ Social characteristics were: age, parity, neighbourhood income quintile (derived by linking residential postal codes to dissemination area-level income), and rural residence (based on the Rurality Index of Ontario).²⁶ Health characteristics were chronic medical conditions (a composite of type 1 or type 2 diabetes,²⁷ chronic hypertension,²⁸ cardiovascular disease,^29-31 or asthma,³² ascertained using validated algorithms < 2 years before conception); pre-pregnancy mental illness (a composite of psychotic, mood/anxiety, substance use, and other mental disorders, or self-harm in the 2 years before pregnancy); and continuity of primary care (i.e., the proportion of primary care visits made to the regular primary health care provider < 2 years before conception³³). Disability-related characteristics were comorbid physical disabilities (a composite of congenital anomalies, musculoskeletal disorders, neurologic disorders, or permanent injuries) and sensory disabilities (a composite of hearing or vision impairments) ascertained using algorithms applied from database inception to conception,^23,24 as well as IDD type (i.e., autism spectrum disorder only or intellectual disability).

Statistical Analyses

Women with and without IDD were described using frequencies and percentages and compared using standardized differences,³⁴ which are appropriate for large cohorts since they are not influenced by sample size as p-values are; values > 0.10 indicate imbalance.

To address the first objective, we performed a latent class analysis (LCA) to identify clinically relevant subgroups of women with IDD according to social (age, parity, neighbourhood income quintile, rural residence), health (chronic medical conditions, mental illness, continuity of primary care), and disability-related characteristics (comorbid physical and sensory disabilities, IDD type). LCA identifies subgroups within a population, given a set of variables.³⁵ It assumes the existence of an underlying latent categorical variable, which accounts for the division of the population into homogenous subgroups. Since no previous studies have used LCA to identify subgroups of pregnant women with IDD, we used LCA as an exploratory technique (i.e., unconstrained LCA), making no assumptions about the number of classes. We estimated a sequence of LCA models, increasing the number of classes using a stepwise approach. The best model fit was determined using the Akaike information criterion (AIC), Bayesian information criterion (BIC), g-squared test statistic (G²), and Chi-squared goodness of fit (X²),³⁷ and the final model was chosen after also considering the clinical relevance of the identified classes. After identifying the best model, each class was described and characterized by an “IDD profile”. The LCA was performed using the poLCA package in R.³⁵

To address the second objective, modified Poisson regression was used to estimate relative risks (RR) for the outcomes according to each IDD profile compared to women without IDD.³⁷ In sensitivity analyses, women with physical or sensory disabilities were removed from the referent group. Regression analyses were performed using SAS Enterprise Guide, version 7.1.

RESULTS

A total of n=1,128,776 singleton livebirths were included in the cohort, of which 0.2% were delivered to women with IDD. Among women with IDD, 20.5% were autistic with no intellectual disability. Women with IDD were more likely than those without IDD to be young and primiparous, to have a low neighbourhood income quintile, and to have a medical comorbidity, psychiatric comorbidity, or comorbid physical or sensory disability (Table 1).

Table 1. Baseline characteristics of women with and without IDD with a singleton livebirth, in Ontario, Canada, 2003-2018.

Data presented as n (%).

Characteristics	IDD (n=1,922)	No IDD (n=1,126,854)	Standardized difference
Age group, years
15-24	863 (44.9)	188,031 (16.7)	0.64
25-34	808 (42.0)	708,371 (62.9)	0.43
35-49	251 (13.1)	230,452 (20.5)	0.20
Primiparous	1,041 (54.2)	537,183 (47.7)	0.13
Low neighbourhood income (Q1-Q2)	1,131 (59.3)	477,593 (42.5)	0.34
Rural residence	212 (11.1)	107,380 (9.5)	0.05
Medical comorbidity *	649 (33.4)	200,353 (17.8)	0.36
Type 1 or type 2 diabetes	78 (4.1)	19,122 (1.7)	0.14
Chronic hypertension	51 (2.7)	28,264 (2.5)	0.01
Cardiovascular disease	< 6 †	817 (0.1)	---
Asthma	580 (30.2)	163,066 (14.5)	0.38
Psychiatric comorbidity ‡	791 (41.2)	157,560 (14.0)	0.64
Psychotic disorders	91 (4.7)	2043 (0.18)	0.30
Mood or anxiety disorders	715 (37.2)	148,862 (13.2)	0.58
Substance use disorders	155 (8.0)	10,944 (1.0)	0.34
Other mental disorders	108 (5.6)	3137 (0.28)	0.32
Self-harm	791 (41.1)	157,560 (14.0)	0.63
Continuity of primary care
<3 primary care visits in 2 years	338 (17.6)	200,520 (17.8)	0.01
Low continuity (≤ 50%)	535 (27.8)	268,512 (23.8)	0.09
Moderate or high continuity (≥ 51%)	1,049 (54.6)	657,822 (58.4)	0.08
Comorbid disability
Physical disability	453 (23.6)	93,932 (8.3)	0.43
Sensory disability	200 (10.4)	32,631 (2.9)	0.30
Hearing	164 (8.5)	23,220 (2.1)	0.29
Vision	44 (2.3)	9763 (0.9)	0.11
IDD type
Autism spectrum disorder only	395 (20.5)	----	----
Intellectual disability	1,527 (79.5)	----	----

^*Medical comorbidity included type 1 or type 2 diabetes, chronic hypertension, cardiovascular disease, and asthma.

^†Values < 6 are suppressed to protect patient identities.

^‡Psychiatric comorbidity included a psychotic, mood/anxiety, substance use, or other mental disorder (e.g., personality disorder), or self-harm.

Table S2 provides fit statistics for LCA models with up to 10 classes. The fit statistics improved with an increasing number of classes up to 5 classes and did not substantially improve thereafter. Comparing the 3, 4, and 5-class solutions, the 4-class solution was deemed to be the most clinically meaningful. This solution divided women with IDD into: (1) young women who were mostly healthy and had little primary care contact before pregnancy (Class 1; n=253); (2) older women who were mostly healthy (Class 2; n=795); (3) young to mid-aged women who had significant medical, psychiatric, and physical disability comorbidities (Class 3; n=181); and (4) young women, many of whom were autistic, and who had some medical comorbidities and significant psychiatric comorbidities (Class 4; n=693) (Table 2; Table S3). Given the underlying characteristics of women with IDD as a whole; all classes were mostly primiparous and poor. These factors, along with rural residence and sensory disability contributed less to the division of classes than age; medical, psychiatric, and physical disability comorbidities; and IDD type.

Table 2. Social, health and disability-related characteristics in women with IDD, by latent classes for the best-fit model.

Characteristics	Class 1 (n=253)	Class 2 (n=795)	Class 3 (n=181)	Class 4 (n=693)
Age group, years
15-24	196 (77.5)	8 (1.0)	46 (25.4)	613 (88.5)
25-34	57 (22.5)	545(68.6)	126 (69.6)	80(11.5)
35-49	0	242 (30.4)	9 (5.0)	0
Primiparous	199 (78.7)	250 (31.5)	60 (33.2)	532 (76.8)
Low neighbourhood income (Q1-Q2)	178 (70.9)	350 (44.3)	148 (81.8)	455 (66.2)
Rural residence	74 (29.4)	82 (10.3)	17 (9.4)	39 (5.7)
Medical comorbiditya	73 (28.9)	180 (22.64)	176 (97.2)	220 (31.8)
Psychiatric comorbidityb	52 (20.6)	189 (23.8)	160 (88.4)	390 (56.3)
Continuity of primary care
<3 primary care visits in 2 years	227 (89.7)	107 (13.5)	0-6*	0-6*
Low continuity (≤ 50%)	0	150 (18.9)	70-75*	100-105*
Moderate or high continuity (≥ 51%)	26 (10.3)	538 (67.7)	103 (56.9)	382 (55.1)
Comorbid disability
Physical disability	48 (19.0)	160 (20.1)	109 (60.2)	136 (19.6)
Sensory disability	28 (11.0)	44 (5.5)	41 (22.7)	87 (12.6)
Autism spectrum disorder only	64 (25.3)	105 (13.2)	13 (7.2)	213 (30.7)

^*Values < 6 are suppressed to protect patient identities.

Table 3 displays the RRs for each outcome, by IDD profile, compared to women without IDD. With respect to hypertensive disorders of pregnancy, Class 3 had the greatest risk compared to women without IDD (RR 2.70, 95% CI 1.84-3.95), with the other three classes having elevated, but non-significant risk for the outcome. A similar pattern was observed for caesarean delivery, wherein Class 3 had the greatest risk (RR 1.33, 95% CI 1.10-1.16), with RRs for the other Classes bordering the null value. Finally, all Classes had elevated risk of preterm birth compared to women without IDD, but, again, Class 3 had the greatest risk (RR 2.43, 95% CI 1.75-3.37).

Table 3. Risk of hypertensive disorders of pregnancy, caesarean delivery, and preterm birth in women with disabilities, by IDD class, compared to women without IDD.

Outcomes	N (%) with outcome	RR (95%CI)
Hypertensive disorders of pregnancy
Class 1	16 (6.3)	1.34 (0.84-2.16)
Class 2	45 (5.7)	1.20 (0.90-1.60)
Class 3	23 (12.7)	2.70 (1.84-3.95)
Class 4	37 (5.3)	1.13 (0.83-1.55)
No IDD	53,085 (4.7)	Reference group
Caesarean delivery
Class 1	73 (28.9)	1.02 (0.84-1.24)
Class 2	239(30.1)	1.07 (0.96-1.18)
Class 3	68 (37.6)	1.16 (1.10-1.33)
Class 4	182(26.3)	0.91 (0.82-1.05)
No IDD	317, 921 (28.2)	Reference group
Preterm Birth
Class 1	30 (11.9)	1.74 (1.24-2.43)
Class 2	69 (8.7)	1.27 (1.01-1.59)
Class 3	30 (16.6)	2.43 (1.75-3.37)
Class 4	63 (9.1)	1.33 (1.05-1.69)
No IDD	76,926 (6.8)	Reference group

Note: The IDD classes were: (1) young women who were mostly healthy and had little primary care contact before pregnancy (Class 1; n=253); (2) older women who were mostly healthy (Class 2; n=795); (3) young to mid-aged women who had significant medical, psychiatric, and physical disability comorbidities (Class 3; n=181); and (4) young women, many of whom were autistic, and who had some medical comorbidities and significant psychiatric comorbidities (Class 4; n=693)

Results were unchanged when women with IDD were compared to those without any disability (Table S4).

DISCUSSION

In this population-based study, we found IDD classes defined according to social, health, and disability-related characteristics were largely driven by age, comorbidities, and IDD type. After division of the population of women with IDD into classes based on these characteristics, there was heterogeneity in risk of hypertensive disorders of pregnancy, caesarean delivery, and preterm birth by IDD class, with young women who had significant medical, psychiatric, and physical disability comorbidities having the greatest risk of all outcomes. However, risks of preterm birth were elevated for the other IDD classes as well. These findings have implications for development of tailored supports for women with IDD perinatally.

Our research suggested that age, comorbidities, and IDD type drove the division of IDD classes, with factors such as poverty—which is ubiquitous among women with IDD^6,7—playing a lesser role. There is little research to which to compare these findings. One prior study of 250 people with IDD in the Netherlands used LCA on clinical data to show subgroups of people with IDD were largely driven by age, financial problems, abuse, behaviour problems, and IDD severity.³⁸ Given our focus on pregnant women, and our use of administrative rather than clinical data, it is not surprising that different factors defined the observed classes. Future studies should attempt to replicate our findings using health administrative data from different jurisdictions.

Meta-analyses indicate women with IDD are at elevated risk for perinatal complications compared to those without IDD.^10,11 Only two studies, to our knowledge, have examined heterogeneity in risk according to social, health, and disability-related factors. In a population-based study in Ontario, Brown et al.¹² found comorbid mental illness exacerbated the risk of preterm birth in women with IDD. An Australian study found Indigenous identity exacerbated risks of preterm birth in women with IDD.¹³ These studies focused on one specific health or social characteristic in addition to IDD status; our study adds to the literature by examining a wider range of factors selected based on the perinatal health framework,¹⁴ which provides a comprehensive view of how the contexts of women with IDD may drive perinatal risks.

There are several possible reasons for our findings. Medical and psychiatric illnesses are known risk factors for perinatal complications.^8,9 Prior studies examining the risks of perinatal complications in women with IDD have only been able to hypothesize that observed risks may be due to comorbid conditions.^10,11 Our finding that young women with IDD who had significant medical, psychiatric, and physical disability comorbidities had the greatest risk of the outcomes suggests this may be the case. Nevertheless, all IDD classes showed increased risk of preterm birth compared to women without IDD. This might be explained by the large burden of poverty experienced by all four IDD classes, with poverty being a well-known risk factor for preterm birth.³⁹ It is possible that factors we could not measure, such as prenatal care access, could explain residual risk for this outcome. Prior research has shown women with IDD are more likely to receive prenatal care late, and receive fewer than the recommended number of visits.⁴⁰ Such gaps may result in missed opportunities for strategies to prevent such complications.

Strengths of our study include the use of whole-population data. However, the cohort of women with IDD was relatively small, requiring some characteristics (e.g., specific types of medical, psychiatric, and disability-related comorbidities) to be examined as composites, rather than individually, in the LCA. We also lacked information on characteristics that might have been important in distinguishing IDD classes, such as individual-level socioeconomic data; receipt of disability-related supports; experiences of racism; behavioural factors such as smoking and alcohol consumption; and IDD-related factors such as severity. Future research could benefit from use of survey data, with detailed information on such characteristics, to identify IDD classes and examine heterogeneity in perinatal outcomes.

CONCLUSIONS

Given that this was an exploratory analysis, derivation of IDD classes in other cohorts and examination of associated perinatal outcomes are needed. However, our data have implications for practice. Given medical, psychiatric, and physical-disability comorbidities largely defined the IDD classes and drove the elevated risk of perinatal complications in women with IDD, preconception supports should address their physical and mental health needs. Although reproductive-aged women with IDD are known to experience health disparities, preconception health interventions typically do not consider the unique needs of women with IDD, and reproductive health care for women with IDD focuses on contraception, rather than health promotion and reproductive life planning.⁷ In pregnancy, women with IDD with comorbidities in particular could benefit from multidisciplinary care and increased monitoring in the form of more frequent or longer visits. Such efforts require training for providers to ensure care is delivered in an accessible manner, for example, using plain language resources and in collaboration with social services. Given the burden of poverty across all IDD classes, and the known association between poverty and preterm birth,⁴⁰ perinatal care for all women with IDD requires attention to social disparities, including problems with housing, food security, and transportation. This attention to the broader context of women with IDD before and during pregnancy is consistent with a life course perspective, and may be beneficial for improving their maternal and newborn outcomes.