Maternal Intellectual or Developmental Disability and Newborn Discharge to Protective Services

This was a population-based study in Ontario, Canada, which has 13.6 million residents and 145 000 births annually.²¹ Ontario’s universal health care delivers physician and hospital services to residents at no direct cost to the patient. Child protective services are funded provincially and delivered by 48 local children’s aid societies that operate on the basis of the Child and Family Services Act.²² This study’s cohort comprised newborns of women with and without IDDs who were conceived between April 1, 2002, and March 31, 2012. The study was approved by the Sunnybrook Health Sciences Centre Research Ethics Board (Institute for Clinical Evaluative Sciences [ICES] logged study 2014-0990-5520-000).

Linked databases were accessed and analyzed at the ICES in Toronto. We obtained health data on birth date, postal code, and death date (Registered Persons Database); outpatient physician visits (Ontario Health Insurance Plan database); emergency department visits (National Ambulatory Care Reporting System); hospitalizations (Canadian Institutes of Health Information Discharge Database and Ontario Mental Health Reporting System); and publicly funded drug benefits (Ontario Drug Benefits database). Databases were linked by using a unique encoded identifier. Data are recorded by using the International Statistical Classification of Diseases and Related Health Problems or Diagnostic and Statistical Manual of Mental Disorders for hospital data and physician billing codes for outpatient data and are reliable and valid.²³ We obtained data on Ontario Disability Support Program receipt from the Ministry of Community and Social Services and linked these with health data using name, sex, date of birth, and postal code.²⁴ 

Women with IDDs were identified from a larger population-based cohort comprising all adults with IDDs in Ontario. The entry criterion was an IDD diagnosis (1) recorded in ≥2 outpatient physician visits or ≥1 hospitalizations or emergency department visits, as abstracted from medical records and submitted to physician billing and hospital databases, since database inception or (2) listed in Ministry of Community and Social Services data as the reason for Ontario Disability Support Program receipt.²⁴ Ontario legislation²⁵ and frequently used clinical definitions² were used to conceptualize this definition; diagnoses are listed in Supplemental Table 5.²⁶ For the current study, we identified all 18- to 49-year-old women with IDDs who had a singleton live birth at ≥20 weeks’ gestational age with a conception date of April 1, 2002, to March 31, 2012. These births and their conception dates (date of birth minus gestational age) were identified in MOMBABY, a linked maternal-newborn data set derived from the Canadian Institute for Health Information Discharge Abstract Database that is used to identify 98.0% of Ontario births.²¹ As a referent, we included a 20% random sample of 18- to 49-year-olds without IDDs who had a singleton live birth with a conception date during the study period. This size was chosen to maximize study power while minimizing computing requirements.

The main outcome was newborn discharge to child protective services directly from the birth hospitalization. This information was derived from the discharge disposition variable in the infant’s hospital record, which is used to identify the location to which the patient was discharged. We also had information on maternal and infant postal codes when the infant reached 12 months of age. Sustained separation of mother and infant was defined as newborn discharge to child protective services and discordant maternal and infant postal code 12 months after delivery.

Sociodemographic variables were maternal age, parity, neighborhood income quintile, and region of residence.²⁷ Maternal health variables were a composite of chronic medical conditions that are common in women with IDDs (diabetes, hypertension, thyroid disease, epilepsy),^28,29 psychotic disorder,³⁰ nonpsychotic mental illness,³¹ and substance use disorders. Service variables were social assistance receipt,³² failure to receive an ultrasound by 20 weeks, and receipt of <4 prenatal care visits by 36 weeks. Perinatal variables were a composite of serious pregnancy complications, cesarean delivery, preterm birth, neonatal morbidity, and congenital anomalies (Supplemental Table 6).

To estimate and compare risk of discharge to child protective services in newborns of women with versus without IDDs, we first described their characteristics using frequencies and percentages and compared them using standardized differences (SDs).³³ In preparation for multivariable modeling, covariates were tested for multicollinearity.³⁴ We estimated the relative risk (RR) and 95% confidence interval (CI) of discharge to child protective services comparing newborns of women with versus without IDDs using modified Poisson regression, which allows direct estimation of risk with binary outcomes.³⁵ Because there could be >1 delivery to the same woman during the study period, we used generalized estimating equations to adjust for clustering.³⁶ We generated crude RRs and RRs adjusted for sociodemographic, health, service, and perinatal characteristics associated with IDD status and with risk of newborn discharge to child protective services.^14,17,–20 Because poverty, isolation, and mental illness increase the risk of child custody loss,^14,17,–20 we tested for interactions between IDD status and maternal age (<25 vs ≥25 years old), income (quintiles 1–2 vs 3–5), comorbid mental illness (any mental illness or substance use disorder vs none), and social assistance (receiving vs not) and then conducted analyses stratified on these variables.

Among women with IDDs only, to identify risk factors for newborn discharge to child protective services, we first described the characteristics of women who did and did not experience the outcome, as above. We generated crude RRs for the associations between each covariate and the outcome, followed by RRs adjusted in separate blocks: sociodemographic, health, service, and perinatal characteristics.³² The final multivariable model included only covariates that had P < .20.

In additional analyses, we estimated and compared the risk of sustained maternal-infant separation in women with versus without IDDs and identified risk factors for sustained maternal-infant separation among women with IDDs. Analyses used SAS 9.4 (SAS Institute, Inc, Cary, NC).

There were 3845 live births to women with IDDs and 379 834 to women without IDDs. Women with IDDs were more likely than those without IDDs to be <20 years of age and to live in low-income neighborhoods and rural areas. They were more likely to have chronic medical conditions, mental illness, and substance use disorders and to receive social assistance. Their infants were more likely to be born preterm and to have neonatal morbidity and congenital anomalies (Table 1).

The rate of discharge to child protective services directly from the birth hospitalization in newborns of women with IDDs was 5.7% vs 0.2% in newborns of women without IDDs (RR 32.80; 95% CI 27.94–38.50). Risk remained statistically significant after adjusting for baseline sociodemographic, health, service, and perinatal characteristics (adjusted relative risk [aRR] 8.10; 95% CI 6.51–10.09) (Fig 1). Interactions between IDD status and age (P = .0002), income (P = .002), comorbid mental illness (P = .0001), and social assistance (P = .002) were statistically significant. In stratified analyses, IDD status had a weaker impact on outcome risk in women who lived in low-income neighborhoods, had comorbid mental illness, and received social assistance (also Fig 1).

In women with IDDs, those with a newborn discharged to child protective services were more likely than those without to be multiparous and to live in low-income neighborhoods but were less likely to live in rural areas. They were more likely to have chronic medical conditions, mental illness, and substance use disorders, to receive social assistance, to have no ultrasound by 20 weeks, and to have <4 prenatal care visits by 36 weeks. Their infants were less likely to be born by cesarean delivery but were more likely to have neonatal morbidity or congenital anomalies (Table 2).

In crude analyses, neighborhood income quintile, chronic medical conditions, psychotic and nonpsychotic mental illness, substance use disorders, social assistance receipt, failure to receive an ultrasound by 20 weeks, receipt of <4 prenatal care visits by 36 weeks, neonatal morbidity, and congenital anomalies were associated with risk of discharge to child protective services among newborns of women with IDDs. In the final multivariable model, psychotic disorders (aRR 2.58; 95% CI 1.90–3.50), social assistance receipt (aRR 2.55; 95% CI 1.87–3.47), failure to receive an ultrasound by 20 weeks (aRR 1.76; 95% CI 1.32–2.34), and receipt of <4 prenatal visits by 36 weeks (aRR 1.71; 95% CI 1.05–2.78) remained statistically significant (Table 3).

The rate of sustained maternal-infant separation in women with IDDs was 4.9% compared with 0.1% in those without IDDs (RR 37.40; 95% CI 31.41–44.52) (ie, 85.9% vs 74.8% of those initially discharged to child protective services). This risk remained statistically significant in adjusted analyses (aRR 8.83; 95% CI 6.99–11.17). Among women with IDDs, risk factors for sustained maternal-infant separation were psychotic disorders (aRR 1.71; 95% CI 1.03–2.85), social assistance receipt (aRR 2.96; 95% CI 2.10–4.17), failure to receive an ultrasound by 20 weeks (aRR 1.64; 95% CI 1.21–2.23), and neonatal morbidity (aRR 1.73; 95% CI 1.05–2.86) (Table 4).

In this large, population-based study, 1 in 20 newborns of women with IDDs were discharged to child protective services immediately after the birth hospitalization. This was >30 times the rate in newborns of women without IDDs. Even after taking into account characteristics that could be used to explain the difference, maternal IDD status remained associated with an eightfold increased risk for newborn discharge to child protective services. Among women with IDDs, risk factors for newborn discharge to child protective services were maternal comorbid psychotic disorders, social assistance receipt, and indicators of inadequate prenatal care. Women with IDDs in these high-risk groups may benefit from improved support before and during pregnancy to prevent maternal-newborn separations, when possible. On the other hand, the finding that 19 out of 20 newborns of women with IDDs were discharged from the hospital with their mothers suggests that, in many cases, women with IDDs are given the opportunity to parent and should be given appropriate supports to optimize maternal-newborn bonding, breastfeeding, and early infant care.

Several studies have documented the overrepresentation of parents with IDDs in child protective services cases, with rates ranging from 26.1% to 86.9%.^12,–14,19 Most studies were restricted to high-risk samples from court documents^12,13,19 or social service agency records.¹⁴ In addition, although 1 study suggested that younger child age was associated with increased odds of court applications for guardianship orders,¹⁹ only 2 studies specifically examined custody loss directly after delivery.^15,16 The only other population-based study revealed that, among 53 565 women in Manitoba, Canada, of whom 69 had an IDD, IDD status was associated with 6.54 greater odds of having an infant taken into care at birth,¹⁶ consistent with our aRR of 8.10. A case series from the United Kingdom revealed that 40% of infants born to 20 parents with IDDs were removed from their care at birth,¹⁵ a rate considerably higher than our rate of 5.7%. However, this was a high-risk sample of parents receiving disability-related services. Our findings align with those of studies identifying subgroups of women with IDDs who were at particularly high risk for custody loss.^17,18,20 Two of these studies had a comparison group of women with IDDs without child protective services involvement.^17,20 Authors of a Canadian study of court documents of 1243 children of parents with IDDs found that low social support and mental illness were predictive of applying to child welfare court for guardianship orders.²⁰ A smaller Canadian study of social service agency records of 47 women with IDDs revealed that women with custody loss had higher rates of poverty and social isolation than those without custody loss.¹⁷ In an Australian study of court documents of 12 women with IDDs who were at risk for losing custody, 42% had comorbid mental illness.¹⁸ Our results build on this research by using population-based data to provide clear quantification of the burden of custody loss immediately postdelivery in women with IDDs and by introducing key risk factors for this outcome.

Several factors could be used to explain the higher rate of discharge to child protective services in newborns of women with IDDs compared with newborns of women without IDDs, including concerns about parenting ability related to memory, organizational skills, judgement, or social skills^7,37 or the attitudes and perceptions of child protective services workers.³ In our study, the association between IDD status and risk of discharge to child protective services remained after accounting for sociodemographic, health, service, and perinatal characteristics that could be used to explain the difference, suggesting some impact attributable to these unmeasured factors. It is notable that the impact of IDD status on child protective services involvement was not as great in high-risk groups defined by poverty, social isolation, and poor mental health. This suggests that IDD status has a smaller impact in women already experiencing social vulnerability. As shown previously,^17,18,20 a variety of factors likely interact to increase the risk of child protective services involvement. This explanation is further reflected in our analyses of risk factors for child protective services involvement among women with IDDs; psychotic disorder, social assistance receipt, and inadequate prenatal care are all indicative of marginalization and were the main predictors of newborn discharge to child protective services, suggesting the existence of high-risk subgroups. These high-risk subgroups are significant because they speak to potential targets for supports aimed at improving the parenting skills of women with IDDs.

Our work has limitations common to studies using administrative data. IDD status may have been misclassified if disability was not recorded in health care encounters and disability-related income supports were not used.²⁴ However, the prevalence of IDD in our cohort (0.8%)²⁶ is similar to previous reports.¹ Our conceptualization of IDD includes a broad range of disabilities that may be associated with variability in parenting ability. However, we did not stratify analyses by diagnosis because many individuals with IDDs have multiple diagnoses and, even within a diagnosis, there can be considerable range in parenting ability and support needs. In addition, our broad definition is consistent with disability legislation²⁵ and clinical definitions.² We did not know the original method of IDD or mental illness diagnosis, because only diagnostic codes are recorded in administrative data. However, the standard of care in Ontario for the diagnosis of these disorders includes the use of standardized measures.³⁸ We did not have data on social context, including maternal marital status or the IDD status of the father, which may influence child custody decisions. This should be addressed in future research, as interventions for parents with IDDs tend to be designed for mothers.³⁹ We also had no data on housing (independent, group home), familial support, or previous custody losses. We were unable to determine the subsequent custody status of newborns discharged to child protective services. To measure sustained maternal-infant separation, we used a proxy comprising different maternal and infant postal codes at 12 months. This does not provide information on the continuity of separation during that period or the reasons for different residences. In addition, albeit rarely, differences in postal code could reflect a failure to update health card information as opposed to separate residences. Finally, we had no data on apprehension at older ages. Even mothers whose newborns were not discharged to child protective services directly after the birth hospitalization may have eventually lost custody.^12,–14 

The immediate newborn period is important for breastfeeding and maternal-infant bonding, and separation of mothers and infants during this time is a risk factor for maternal suicide¹¹ and child developmental problems.¹⁰ Although there are cases in which custody loss among women with IDDs may be warranted, such as in instances of neglect or abuse, researchers suggest that many separations are preventable.^8,9 With our findings, we suggest that evidence-based interventions to avoid preventable maternal-infant separations at delivery and to support women who retain custody of their children after delivery are warranted, especially for women in the high-risk subgroups we identified. Previous research has revealed that behavior-based approaches to teach parenting skills are most effective at reducing risk for child removal, but these must be delivered in a manner that is tailored to a mother’s learning style.^8,9 Other factors, such as poverty and comorbid mental illness, would also need to be addressed by comprehensive intervention services to achieve optimal results from such interventions, and ultimately, good maternal and child outcomes.

aRR

adjusted relative risk

CI

confidence interval

ICES

Institute for Clinical Evaluative Sciences

IDD

intellectual and developmental disability

RR

relative risk

SD

standardized difference