Prevalence of Births and Interactions with Child Protective Services of Children Born to Mothers Diagnosed with an Intellectual and/or Developmental Disability

Abstract

Objectives:

Concerns have been raised that parents with intellectual and/or developmental disabilities (IDD) interact with child protective services (CPS) at disproportionate rates than the general population as a result of bias and discrimination. However, there has been little empirical evidence to ascertain if these concerns are grounded. This study’s objectives were to identify (a) the prevalence and sociodemographic characteristics of children born to mothers diagnosed with IDD diagnoses, (b) how many of these children interact with CPS (reports and removals) and (c) when these CPS interactions are occurring.

Methods:

The dataset was comprised of linked administrative birth, hospital discharge, and CPS records for all children born in one U.S. state between 1999-2013 (N = 1,271,419). CPS records were available through the first quarter of 2018 and CPS reports and removals at the child’s first and fourth birthdays were identified. We conducted chi-square tests and multivariate survival Cox regression models.

Results:

A total of 567 children were identified as born to mothers with IDD diagnoses, which is 4.5 per 10,000 births. Of these children, 21.7% were the subject of a CPS report within one year and 35.8% within four years. In terms of removals, 6.5% experienced removals by one year and 8.6% by four years.

Conclusions for Practice:

This study provides population-based knowledge about how and when the children born to mothers diagnosed with IDD interact with CPS. These children have higher rates of CPS interactions than the general population, but these rates are not as high as previously reported.

Significance:

This is the first prospective population-based study utilizing medical and CPS records to examine the prevalence of births and CPS interactions of children born to mothers with IDD diagnoses. We found that children born to mothers with IDD diagnoses were reported to CPS and removed by CPS at higher rates than children born to mothers without such a diagnosis.

Introduction

Official child protective services (CPS) agencies received more than four million referrals in 2018 (US HHS, 2020). Concerns have been raised that children from certain populations interact with CPS at disproportionate rates. One of these populations is children born to parents with intellectual and/or developmental disabilities (IDD) (Azar, et al., 2012; Callow et al., 2011; Lightfoot & Hill, 2011). This fear of overrepresentation in CPS contacts is fortified when considering that 37 U.S. states include disability-related grounds for termination of parental rights, and that many statutes included vague definitions and obsolete language (Lightfoot et al., 2010). A prior population-based study identified 0.03% of births in California between 2000 and 2010 had a mother with an IDD diagnosis noted in the hospital discharge records and that the proportion of births with a maternal disability diagnosis doubled between 2000 and 2010 (Horner-Johnson et al., 2017).

An identified gap in the literature on parents with IDD is datasets that allow for comparisons of parents with IDD to the general population, which may be addressed with the use of administrative datasets (Llewellyn & Hindmarsh, 2015). To address this gap, a few studies have examined the interactions parents with IDD have with CPS in the United States using different administrative datasets. Some studies have utilized the national datasets of state CPS administrative data collected by the federal government, the Adoption and Foster Care Reporting System (AFCARS) and the National Child Abuse and Neglect Data System (NCANDS). Specifically, a study of CPS investigations using NCANDS compared parents identified with IDD to parents where this variable was not endorsed and found parents with IDD were more likely to have allegations substantiated with about a third experiencing this outcome (Slayter & Jensen, 2019).

Lightfoot and DeZelar (2016) used AFCARS to identify that 19% of children in foster care had parental disability indicated as a removal reason and that these children had longer stays in foster care than children without parental disability as a removal reason. Studies using NCANDS and AFCARS provide valuable overviews of CPS interactions with this population, yet important caveats include there is great variation in how states utilize and collect the variable of caregiver disability as a removal reason in AFCARS (DeZelar & Lightfoot, 2018) and the issue of missing data in NCANDS, including some states, like New Mexico and New York, not providing any information on the parent with IDD variable (US HHS, 2015).

To address some of these limitations, several studies have used linked administrative data, or data integrated from multiple sources, to investigate the experiences of parents with IDD with CPS. Specifically, these studies started with parents involved with CPS in Minnesota and examined their educational records as children to determine if parents have a disability (Lightfoot et al., 2017; LaLiberte et al., 2017). Of parents involved with CPS, one study found that parents identified with IDD in their childhood education records were more likely to have a CPS report investigated, have a child placed in foster care, and have their parental rights terminated compared to parents with other types of documented disability and no identified disability in their education records (LaLiberte et al., 2017).

These studies are similar to those using NCANDS and AFCARS in that they start with populations of families involved with CPS and then identify, of those involved with CPS, which have a parent with IDD. A missing gap in the literature is starting with a population of mothers with IDD who have become parents and, prospectively, identifying which of the mothers with IDD have involvement with CPS and when the CPS involvement occurs. This approach differs from previous research in regards to the comparison group. Specifically, previous research compared parents with IDD within the larger population of parents involved with CPS. By starting with mothers with IDD and following them prospectively from the birth of the child, knowledge can be generated about the frequency, risk factors, and timing mothers with IDD have regarding CPS involvement and compare them with other mothers with IDD and the general population. This study takes this prospective approach to add new knowledge to the literature.

Therefore, using a unique set of integrated public health and CPS data from Washington State, the aim of this study was to investigate if, how, and when children born to mothers with IDD interact with CPS. Specifically, the research questions were: 1) What is the prevalence and sociodemographic characteristics of children born to mothers with IDD diagnoses? 2) How many children born to mothers diagnosed with IDD interact with CPS (reports and removals)? and 3) When are these CPS interactions occurring?

Methods

Data

This prospective cohort study utilized a linked dataset of birth record, birth hospital discharge, and CPS administrative data for all children born in Washington State from 1999 through 2013, the most recently available years of data (N = 1,271,419). The datasets were linked by an individual outside of the analysis team using a sequential deterministic linkage methodology using personal identifiable information, such as names and birthdates, and is described further in Herman et al. (1997). After linkages were completed, the dataset was de-identified and provided to the researchers for analysis. The CPS records through the first quarter of 2018 are included. This study is part of a larger investigation regarding CPS involvement and child injury/health risks, which has been approved by the Washington State Institutional Review Board.

Variables

Our primary outcomes of interest were the interactions with CPS, specifically 1) any report made to CPS and 2) a report to CPS followed by a child removed by CPS. These outcomes were chosen because reports are the entry point to CPS involvement and a removal indicates an escalation of the case requiring court involvement (Gelles, 2020). Binary variables, using state administrative CPS data, were constructed for each outcome for the timeframes within one year and within four years, with reports that occurred prior to a child’s birth included in the first-year variable. Time to these events was constructed in days. Four years is the furthest time point available in the data for the oldest cohort of children (born in 2013). Additionally, the youngest children have the highest rates of maltreatment victimization according to nationwide CPS data (US HHS, 2020).

Children were identified as being born to mothers with IDD through the use of medical standardized diagnostic codes, the International Classification of Disease, Ninth Revision (ICD-9) from the birth hospital discharge records. ICD-9 codes have been used successfully to identify individuals with IDD in a variety of previous studies (Chang et al., 2014; Horner-Johnson et al., 2017; Kancherla et al., 2012; Mitra et al., 2018; Parish et al., 2015). Mothers were identified as having an IDD diagnosis if the following diagnoses were included in their medical record at the time of the child’s birth, the same list used by Mitra et al. (2018): 317 (mild intellectual disability), 318 (moderate intellectual disability), 319 (unspecified intellectual disability), 759.83 (Fragile X syndrome), 759.81 (Prader-Willi syndrome), 758 (Down syndrome), 330.8 (Rett syndrome), 277.2 (Lesch Nyhan), 758.31 (Cri du chat), 299 (pervasive developmental disorder), 759.5 (Tuberous Sclerosis), 760.71 (fetal alcohol syndrome), 343 (cerebral palsy), 330 (cerebral degeneration), with the addition of 741 (spina bifida), and 907 (traumatic brain injury).

Covariates were included in the analysis from the child’s birth record, all of which have been shown to be predictors of CPS contact within the first five years of life (Parish et al., 2011; Putnam-Hornstein & Needell, 2011). Maternal race included the categories of Asian/Pacific Islander, Black, Hispanic, Native American, White, and other/unknown. Maternal age at birth was divided into four categories for the descriptive analysis (19 years and younger, 20 to 24 years, 25-29 years, and 30 or more years) and as a binary variable with a cutoff at 20 years old to identify if the mother was a teenager for the inferential analysis. Birth payment method was constructed as a binary variable as either public (Medicaid) or private funds. The timing of prenatal care initiation was a categorical variable that included each trimester or no prenatal care sought for the descriptive analyses and as a binary variable indicating no or late (third trimester) prenatal care for the inferential analyses. We included a binary parity variable indicating whether this was the mother’s first child or not. Low birth weight (less than 2500 grams) was included as a binary variable. The year the child was born was included as an interval variable.

Analysis

For our descriptive analysis to address the first question regarding the sociodemographic characteristics of children born to mothers diagnosed with IDD, we first examined the distributions of our covariates between infants born to mothers with IDD diagnoses and those who were not, using chi-square tests to identify differences by maternal IDD diagnosis status. Similarly, to address the second research question of how many children born to mothers diagnosed with IDD experience CPS reports and removals, we tested the distributions of the CPS outcomes at both one year and four years for differences by maternal IDD diagnosis status utilizing chi-square tests.

To answer our third research question of when do children born to mothers diagnosed with IDD experience these CPS interactions, we used survival analysis to investigate the time from birth to CPS reports and removals (Singer & Willett, 2003). First, we plotted Kaplan-Meier estimates for both outcomes, examining the survival probability curves by the mother’s IDD diagnosis status. Second, we used Cox proportional hazards regression to test multivariate models including our birth record covariates. The model is as follows:

$$h\left(t\right)=h_0\left(t\right) \text{exp} [{\beta }_1 \text{IDD diagnosis}+ {\beta }_2 \text{maternal race}+ {\beta }_3 \text{maternal age}+ {\beta }_4 \text{public birth payment}+ {\beta }_5 \text{no/late prenatal care}+ {\beta }_6 \text{first born}+ {\beta }_7 \text{low birth weight + }{\beta }_8 \mathrm{b}\mathrm{i}\mathrm{r}\mathrm{t}\mathrm{h}\mathrm{ }\mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r} ]$$

where t represents the time to the CPS outcome and h(t) represents the hazard function. All analyses were performed using R version 3.4.1 (R Core Team, 2017).

Results

Descriptive Statistics

Of the 1,271,419 births in Washington State between 1999 and 2013, 567 were born to mothers with IDD diagnoses (4.5 per 10,000 births or 0.045%). Table 1 presents the percentages by diagnosis. The most frequent diagnoses were cerebral palsy (39.7%) and spina bifida (26.3%).

Table 1.

Percentage of Children Born to Mothers with IDD Diagnosis by ICD-9 Code

	N = 567
Diagnosis (ICD Code)	%
Cerebral Palsy (343)	39.7
Spina Bifida (756.17)	26.3
Traumatic Brain Injury (907)	9.9
Unspecified Intellectual Disability (319) a	6.9
Pervasive Developmental Disorder (299)	4.9
Mild Intellectual Disability (317) a	3.2
Down Syndrome (758) a	2.5
Fetal Alcohol Syndrome (760.71)	2.1
Tuberous Sclerosis (759.5)	1.6
Fragile X Syndrome (759.83)	1.2
Cerebral Degeneration (330)	0.9
Moderate Intellectual Disability (318) a	0.5
Lesch Nyhan (277.2)	0.2
Prader-Willi Syndrome (759.81)	0.2
Cri du chat (758.31)	0.0
Rett Syndrome (330.8)	0.0

Note: IDD = Intellectual disability diagnoses

Table 2 displays the distributions of the covariates by maternal IDD diagnosis. More infants born to mothers with IDD diagnoses were white (X² = 71.60, p < .001), were their mother’s first-born child (X² = 33.78, p < .001), and had low birth weight (X² = 30.85, p < .001) compared to infants whose mothers did not have an IDD diagnosis. The births of more than half (50.1 %) of the infants born to mothers with IDD diagnoses were paid for using public health insurance compared to 35% of births for mothers without an IDD diagnosis (X² = 54.84, p < .001).

Table 2.

Distribution of Characteristics of Children Born in WA 1999-2013 by Maternal IDD Diagnosis Status

	All Births		No IDD Diagnosis		IDD Diagnosis
	N = 1,271,419		N = 1,270,852		N = 567
	n	%	n	%	n	%	X2
Race/Ethnicity							71.60***
White	840,356	66.1	839,917	66.1	439	77.4
Hispanic	203,385	16.0	203,333	16.0	52	9.2
Other/Unknown	18,644	1.5	18,632	1.5	12	2.1
Asian/Pacific Islander	120,485	9.5	120,472	9.5	13	2.3
Black	59,762	4.7	59,736	4.7	26	4.6
Native American	28,787	2.3	28,762	2.3	25	4.4
Maternal Age							8.17*
< 19 years	102,940	8.1	102,882	8.1	58	10.2
20-24 years	295,316	23.2	295,167	23.2	149	26.3
25-29 years	361,060	28.4	360,904	28.4	156	27.5
30+ years	511,562	40.3	511,358	40.3	204	36.0
Insurance Type							54.84***
Public	444,914	35.0	444,630	35.0	284	50.1
Private	826,505	65.0	826,222	65.0	283	49.9
Prenatal Care							3.65
1st trimester	666,769	52.4	666,480	52.4	289	51.0
2nd trimester	403,985	31.8	403,813	31.8	172	30.3
3rd trimester	70,231	5.5	70,194	5.5	37	6.5
None	130,434	10.3	130,365	10.3	69	12.2
Parity							33.78***
First birth	519,097	40.8	518,797	40.8	300	52.9
Non-first birth	752,322	59.2	752,055	59.2	267	47.1
Infant Birth Weight							30.85***
Low (<2500 g)	82714	6.5	82,644	6.5	70	12.3
Normal (2500+)	1,188,705	93.5	1,188,208	93.5	497	87.7

Note: missing values were not included in calculations of percentages; IDD = Intellectual disability diagnoses

^***p < .001,

^**p < .01,

^*p < .05

Table 3 presents the distributions of CPS interactions by maternal IDD diagnosis status. Children born to mothers with IDD diagnoses were more likely to have contact with CPS at both time intervals. Specifically, of infants born to mothers with IDD diagnoses, 21.7% were the subject of a CPS report within one year and 35.8% within four years. These are both higher than infants born to mothers without an IDD diagnosis at 6.3% (one year (X² = 284.9, p < .001)) and 9.9% (four years (X² = 248.0, p < .001)), respectively. In regards to removals by CPS, 6.5% of infants born to mothers with IDD diagnoses experienced removals by one year and 8.6% by four years. This is in comparison to 0.9% (one year (X² = 202.5, p < .001)) and 1.8% (four years (X² = 148.2, p < .001)) of infants born to mothers without an IDD diagnosis.

Table 3.

Distribution of CPS Interactions by Maternal IDD Diagnosis Status

	All Births	No IDD Diagnosis	IDD Diagnosis
	N = 1,271,419	n = 1,270,912	n = 567
	%	%	%	X2
CPS Report Within 1 Year				284.9***
Yes	5.5	5.5	21.7
No	94.5	94.5	78.3
CPS Report Within 4 Years				248.0***
Yes	13.3	13.3	35.8
No	86.7	86.7	64.2
Removal Within 1 Year				202.5***
Yes	0.9	0.9	6.5
No	99.1	99.1	93.5
Removal Within 4 Years				148.2***
Yes	1.8	1.8	8.6
No	98.2	98.2	91.4

Notes:

^***p < .001; IDD = Intellectual disability diagnoses; CPS = child protective services

Survival Analysis

Figure 1 presents the Kaplan-Meier survival estimates of both CPS reports and reports and removals by maternal IDD diagnosis status. Both outcomes were statistically significant (p< .0001). As illustrated in these curves, right after birth is the time period with the greatest number of children being reported or reported and removed. The figure also demonstrates that more than 64% and 91% of these infants do not experience CPS reports or CPS reports and removals, respectively, within their first four years of life. Some children also experienced reports during the mother’s pregnancy, which is why the survival curve starts prior to Day 0. Per policy, these reports were recorded as information only.

Figure 1.

Kaplan-Meier Estimates for CPS Interactions by Mothers’ IDD Diagnosis Status

Table 4 and Figure 1 present the multivariate results of the Cox regression models for the outcomes of CPS reports and reports and removals. In the first model examining only CPS reports, children born to mothers with intellectual disability diagnoses were reported at a rate 2.7 times that of children born to mothers with other IDD diagnoses (Odds Ratio (OR): 2.71, 95% Confidence Interval (CI): 2.36-3.11). Children whose births were paid for through public insurance (Medicaid) were reported at a rate three times as great as children whose births were paid through private means (OR: 3.07, CI: 3.03-3.10). Additionally, increased rates of reports to CPS were also identified for children born to teenaged mothers (OR: 2.55, CI: 2.51-2.58) and whose mothers who received late or no prenatal care (OR: 1.21, CI: 1.19-1.22).

Table 4.

Multivariate Cox Regression Models Results in Hazard Ratios

	CPS Report		Report & Removal
	HR	95% CI	HR	95% CI
IDD Diagnosis
Present	Ref.	Ref.	Ref.	Ref.
Not Present	2.71	[2.36, 3.11]	3.65	[2.76, 4.83]
Race/Ethnicity
White	Ref.	--	Ref.	--
Hispanic	0.56	[0.55, 0.56]	0.33	[0.32, 0.35]
Asian/Pacific Islander	0.48	[0.47, 0.49]	0.32	[0.30, 0.35]
Other/Unknown	0.95	[0.92, 0.99]	1.25	[1.13, 1.38]
Black	1.27	[1.25, 1.29]	1.13	[1.08, 1.18]
Native American	2.03	[1.99, 2.07]	3.10	[2.97, 3.23]
Maternal Age
20+ years	Ref.	--	Ref.	--
≤ 19 years	2.55	[2.51, 2.58]	2.47	[2.37, 2.56]
Insurance Type
Private	Ref.	--	Ref.	--
Public	3.07	[3.03, 3.10]	3.98	[3.85, 4.11]
Prenatal Care
1st/2nd Trimester	Ref.	--	Ref.	--
None/3rd Trimester	1.21	[1.19, 1.22]	1.68	[1.63, 1.73]
Parity
Non-first birth	Ref.	--	Ref.	--
First birth	0.57	[0.56, 0.57]	0.50	[0.48, 0.52]
Infant Birth Weight
Normal (2500+)	Ref.	--	Ref.	--
Low (<2500 g)	1.33	[1.31, 1.36]	1.89	[1.82, 1.97]
Birth Year	0.99	[0.98, 0.99]	1.01	[1.00, 1.01]

Notes: HR = hazard ratio; CI = confidence interval; Ref. = reference; IDD = Intellectual disability diagnoses; CPS = child protective services

For the model testing both CPS reports and removals and consistent with the model regarding CPS reports, children born to mothers with an intellectual disability diagnosis were removed at more than three times the rate of children born to mothers without an IDD diagnosis (OR: 3.65, CI: 2.76-4.83). Children whose birth was paid for using public funds were removed at almost four times the rate of children whose births were paid for through private means (OR: 3.98, CI: 3.85-4.11). Children who were born to mothers who were teenagers at the time of birth (OR: 2.47, CI: 2.37-2.56) and who received late or no prenatal care (OR: 1.68, CI: 1.63-1.73) also had increased risk of experiencing a CPS removal by the time they turned age four.

Discussion

This is the first prospective population-based study utilizing birth and CPS records to examine the prevalence of births and CPS interactions of children born to mothers with IDD diagnoses. We identified 0.045% of all births in Washington State during the study time period had a mother with an IDD diagnosis, which is in line with a California study that identified 0.024% of births between 2000 and 2010 had a mother with an IDD diagnosis (Horner-Johnson et al, 2017). Overall, our finding indicated that children born to mothers with IDD diagnoses had higher percentages of being reported to CPS (35.8% by age four) and removed by CPS than children born to mothers without such a diagnosis. Yet, the observed rate of 8.6% experiencing a removal within the first four years of life for children born to mothers with an IDD diagnosis is much lower than the previous estimates of removals which were cited at 40-60% based on international studies (McConnell & Llewellyn, 2002) and used by U.S.-based advocacy organizations such as the National Council on Disability (NCD) (2012).

The use of a prospective population-based dataset provides a more precise understanding of the prevalence of interactions with CPS because we are able to determine the entire population who was also at risk of interacting with CPS, in this case, the population of children born to mothers with diagnoses of IDD. Previous population-based research on the CPS interactions of parents with IDD has relied on datasets of children after involvement with CPS and compared them with other families interacting with CPS (LaLiberte et al., 2017; Lightfoot & DeZelar, 2016; Slayter & Jensen, 2019). Similar to the findings of those studies, we find that parents with IDD have increased contacts with CPS compared to parents with no documented disability.

The results from our multivariate survival analysis identified additional predictors of CPS involvement beyond mothers with an IDD diagnosis. The finding that births paid for with public health insurance, young maternal age, and no prenatal care increase the likelihood of being reported to CPS is consistent with other studies of predictors of CPS contact (Parish et al., 2011; Putnam-Hornstein & Needell, 2011). This finding supports a call for targeted prevention efforts at or just after birth utilizing the risk factors identified at the time of birth across subpopulations.

Prior research has cited a systemic, attitudinal issue that is not related to the availability of supports for families with parental diagnoses of IDD as the reason for the higher rate of interaction with CPS (Callow et al., 2011). Our findings indicate that this conclusion warrants further investigation. While we demonstrate that the removal rates are higher than the general population, we cannot identify how those decisions are made with our data. Future research would benefit from another administrative data source, namely the state’s IDD administrative services, a resource explored but not available to us for this study. This would provide more insight into the services that a family may be receiving. Additionally, more information is needed in regards to whether the observed level of CPS involvement is in relation to the needs of the families. Therefore, future research that examined the case notes of files to understand the precise allegations, the source of the CPS report, the needs of these families, and more objective measures of child maltreatment, such as medical encounters, could provide further insights on the needs and responses for this population.

In a review of the evidence, Azar & Read (2009) concluded that increasing the capacity of CPS systems to intervene appropriately with parents with IDD, including the use of developmentally sensitive intervention techniques and improved linkages with other systems outside of CPS, would result in keeping families intact. Research from Europe and Australia supports the importance of individualized supports, including modified parenting education programs, which can develop the skills individuals with IDD may need to parent successfully (Wilson et al., 2014). Something that is not clear from our results is what supports the families of the 64% of infants born to mothers with IDD diagnoses are offered and receive, as they are not coming to the attention of CPS within the child’s first four years of life.

Limitations

While this study provides new, population-based knowledge utilizing linked administrative data regarding the prevalence and interactions of children born to mothers with IDD diagnoses, there are a number of limitations to consider. First, our use of ICD-9 codes likely underestimates the true population of children born to mothers with IDD diagnoses because this would be a secondary diagnosis to those related to the birth. Similarly, our data regarding the mothers is limited to the hospitalization related to the birth not allowing us to check previous or subsequent hospitalizations for IDD diagnoses or providing any information about the IDD status of the fathers. Future research should explore if there are variations by hospitals, geographical regions, and health care providers in the recording of these codes.

Second, there is not a gold standard of medical diagnoses for IDD and our results only reflect the specific ICD-9 code diagnoses that we included in our analysis. While other studies have included psychiatric disability diagnoses (Llewellyn et al., 2003), we have chosen to focus on developmental and intellectual disabilities and not include another category, following the lead of Mitra et al. (2018). Further, as identified in the present study and the California study of IDD births (Horner-Johnson et al, 2017), compared to the general population, mothers with diagnoses of IDD on the birth records is a small percentage of the overall population. Thus, this must be taken into consideration when examining our results.

Third, as with all administrative data, we could not capture the full complexity of CPS cases and the myriad factors that mandated reporters and CPS workers consider when making decisions to file a CPS report or to remove a child from his/her parents. Lastly, these records are from a single U.S. state. Other states may have different policies and procedures related to parents diagnosed with IDD that may influence how their families interact with CPS. Future research should explore how other states and CPS agencies interact with parents diagnosed with IDD, including comparisons between states.

Conclusion

This study identified that 0.045% of births in the population were to mothers with IDD diagnoses. Of these children, 21.7% were the subject of a CPS report within their first year of life and 35.8% were reported by their fourth birthdays, both higher than for the general population. The percentage of these children being removed by CPS was also higher than the general population with 6.5% removed by their first birthday and 8.6% by age four. Children born to mothers with an intellectual disability diagnosis were at higher risk of both outcomes than children of mothers with the broader IDD diagnosis. These findings are important in understanding how this subpopulation interacts with CPS, which can inform practice, policy, and training.

Figure 2.

Cox Regression Results for CPS Reports and Removals