Maternal Mental Health Disorders and Reports to Child Protective Services: A Birth Cohort Study

Background. Existing literature has documented a strong relationship between parental mental illness and child maltreatment, but little is known about the prevalence of mental illness among childbearing women. In the present study, linked administrative records were used to identify the prevalence of maternal mental health (MH) disorders documented at birth and determine the associated likelihood of maltreatment reports during infancy. Materials and Methods. Vital records for California’s 2006 birth cohort were linked to hospital discharge and Child Protective Services (CPS) records. The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) billing codes from the mother’s delivery hospitalization were used to determine diagnosed maternal MH disorders for 551,232 infants born in 2006, and reports of alleged maltreatment were documented from CPS records. Vital birth records were used to control for sociodemographic factors. Finally, the associated risk of reported maltreatment during the first year of life was examined using generalized linear models. Results. Among infants in this statewide birth cohort, 2.8% were born to a mother with a documented MH disorder, of which 41.3% had documented maternal substance abuse issues versus less than 0.5% of infants born to mothers without a diagnosed MH disorder. Further, 34.6% of infants born to mothers with a MH disorder were reported to CPS within one year, and a majority of those reports were made within the first month of life (77.2%). In contrast, among children born to mothers without a MH disorder, 4.4% were reported to CPS during infancy. After controlling for sociodemographic factors, the rate of CPS reports during infancy for infants born to mothers with a MH disorder but no substance use disorder was 2.6 times that of infants born to mothers without a MH disorder (95% CI = 2.47, 2.73). Among infants born to mothers with MH and substance use disorders, the rate of CPS reports during infancy was 5.69 times that of infants born to mothers without a MH disorder (95% CI = 5.51, 5.87). Conclusions. Administrative records provide a method for identifying infants born to mothers with MH disorders, enabling researchers to track rates over time and generate population-level data to inform policy development and improve service delivery.


Introduction
The movement to deinstitutionalize psychiatric care in the 1970s led to notable increases in pregnancy and childbirth among women with diagnosed mental health (MH) disorders [1]. Whereas many mothers with MH disorders give birth to healthy infants and demonstrate appropriate parenting practices, existing research indicates that children of mothers with a history of mental illness are at heightened risk of adverse health and wellbeing outcomes, including birth abnormalities [2], low birth weight [3], preterm delivery [4][5][6], insecure parent-child bonding [7], and the transmission of mental illness [8]. Empirical literature has also documented a strong relationship between parental mental illness and child abuse and neglect [8], particularly among children living with a mother with mental illness [9,10] and within economically disadvantaged families [9].
Given the broad range of functional impairments and severity in symptoms across MH disorders [11][12][13], the effect of these disorders on parenting capacity and child safety almost certainly varies. One study found that Child Protective Services (CPS) reports were more likely for children of mothers with mental illness than for children born to mothers without mental illness, regardless of the type of diagnosis [12]. Another study found that mothers living with depression were less likely to use recommended infant care practices if they also had been diagnosed with a personality disorder [14]. Because researchers attempting to account for this heterogeneity have categorized MH disorders in different ways, corroborating findings across studies is difficult [13][14][15]. As such, prevalence estimates for specific mental disorders among mothers are not well-documented, and the relationships between the various categories of MH disorders and children's outcomes, including involvement with CPS, remain largely unknown.
The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes offer a standardized method for tracking population-level changes in diagnosis rates over time and have previously been used to document the prevalence of maternal MH and substance-related disorders [16,17] and their association with subsequent maltreatment and CPS involvement [12,15,18]. No studies, however, have assessed the occurrence of a CPS report regarding infants born to mothers with a MH disorder known to medical professionals at the time of delivery, which is a critical engagement and intervention point. In the current study, linked administrative records from California's 2006 birth cohort were used to determine the following: (1) the prevalence of infants born to mothers with a MH disorder documented in their medical records at the time of delivery, stratified by disorder type; (2) the cumulative percentage of infants reported to CPS during the first 12 months of life among those born to mothers with a MH disorder; and (3) the risk of a report to CPS in those infants born to mothers with a diagnosed MH disorder versus other infants after adjusting for co-documented substance exposure and other factors.

Data Sources
The analytic dataset was constructed by linking three administrative data sources from California: (1) vital birth records for all infants born in 2006; (2) maternal hospital discharge records corresponding to those births; and (3) CPS records for allegations of abuse or neglect made within 12 months of birth. The authors probabilistically linked birth records to CPS records using available unique (i.e., maternal Social Security number) and non-unique (e.g., parent and child first names or parent and child dates of birth) personal identifiers. Among infants reported to CPS and born in 2006, 92% were successfully linked to a birth record. In a separate linkage, 97% of maternal hospital discharge records were probabilistically linked to birth records as part of the ongoing California Vital Stats to Care Delivery Project [19]. The two linked files were then de-identified and merged using the unique birth-record number common to both files. The analysis includes an estimated 95% of all live births in California during 2006, excluding births that occurred in military hospitals and at home. These linked data were previously used to evaluate CPS reporting among mothers with a history of documented substance abuse [20]. This study received human subject approvals from both state (CA HHS #09-12-51) and university (UCB #2010-01-592) institutional review boards and was additionally reviewed by all departments from which data were obtained.

Dependent Variable
The dependent variable was defined as a first report of alleged maltreatment made to CPS during the first 12 months of life. All first reports of maltreatment were included, irrespective of whether the allegation was investigated or substantiated by CPS. Existing research has documented that infants reported to CPS are at heightened risk of negative outcomes, including re-reporting to CPS, placement into foster care, and critical injury and fatality, regardless of whether or not the allegations were substantiated [21,22]. The decision to include all allegations arose from an interest in clinical and community decisions to report infants born to mothers with MH disorders, as opposed to the type of response to those reports by the CPS system.

Independent Variables
Maternal MH disorders were coded based on ICD-9-CM diagnostic codes [23]. Up to 25 codes were documented in the medical records of delivery hospitalizations. Each birth was categorized based on the presence of an MH disorder documented at the time of birth (i.e., no MH disorder or MH disorder). Subsequently, births with a MH disorder were grouped into four non-mutually exclusive diagnostic categories used in previous research: (1) psychotic disorders; (2) mood disorders; (3) anxiety disorders; and (4) MH Disorder at Delivery [12,15,17]. Each category was dichotomously coded, and mothers with more than one diagnosis fell into multiple categories. Mental disorders at delivery is an ICD diagnostic code (648.4) used to document any mental disorder complicating pregnancy, childbirth, or the puerperium [16]. We excluded disorders occurring in childhood, personality disorders, and mental disorders attributable to medical conditions in an effort to focus on the relationship between a mother's experience of a MH disorder leading up to childbirth and CPS involvement for the infant. Our decision was informed by parameters established in earlier studies that used the same data sources from California [16,17]. A full list of included codes and the corresponding number of births for each are included as Appendix A.

Covariates
To characterize mothers diagnosed with a MH disorder and isolate the association between a diagnosed maternal MH disorder and CPS reports during infancy, several covariates available in the birth records were examined. Covariates included maternal race/ethnicity (White, Black, Hispanic, or other/missing) and maternal age at first-ever birth in years (≤19, 20-24, 25-29, and ≥30). The decision to use maternal age at first-ever birth aligns with research indicating that the timing of a first birth is an important proxy for future outcomes of a mother and her children [24,25]. Insurance type was coded dichotomously (private or public). California allows for retroactive enrollment in the state's public system if there is no insurance coverage at the time of birth. Additional covariates included paternity establishment (established or missing), the timing of the initiation of prenatal care (first trimester, second trimester, and third trimester/no care), and parity (first birth or non-first birth). Finally, all births were coded based on the presence or absence of an ICD-9-CM code in the mother's or infant's hospital discharge records indicating maternal substance use or prenatal substance exposure (no or yes). A list of ICD-9-CM substance exposure codes is also included in Appendix A.

Analysis
Covariate distributions were assessed using χ 2 tests to determine statistically significant variations (p < 0.05) based on a diagnosed MH disorder. Distributions of covariates were also assessed for specific MH diagnostic categories. The relationship between a diagnosed maternal MH disorder and CPS reporting was descriptively presented as the cumulative percentage of infants reported to CPS by month throughout the first year of life. To isolate the increased risk of CPS reporting associated with a diagnosed MH, a series of generalized linear models were estimated based on a Poisson distribution, log link, and robust standard error adjustments to account for binary outcomes [26,27]. Estimates were reported as risk ratios (RR) with 95% confidence intervals (95% CI) and presented for any MH disorder, as well as specific categories of diagnosed MH disorders. Model 1 shows the unadjusted risk of being reported for an infant born to a mother with a MH disorder. Model 2 reflects the main effects of MH disorders and includes interactions for co-diagnosed substance use or substance exposure at birth. Model 3 additionally controls for other sociodemographic and pregnancy-related covariates. All analyses were conducted using Stata SE, version 13 (StataCorp LLC, College Station, TX, USA).

Characteristics of Births with a Maternal Mental Health Disorder
As shown in Table 1, among the 551,232 infants born in 2006 and linked to a maternal hospital discharge record, 2.8% (n = 15,516) were born to mothers with a documented MH disorder in their medical records. Overall, the majority (93.5%) of mothers with a MH disorder received the general diagnosis of 648.4 or "Mental disorders complicating pregnancy, childbirth, or the puerperium" (herein referred to as MH Disorder at Delivery). Births with a maternal MH disorder were associated with younger maternal age at first birth, a higher percentage of White mothers (49.2% vs. 27.0%), more coverage by public health insurance (56.3% vs. 48.2%), and higher rates of missing paternity on the birth certificate (27.5% vs. 8.7%). The percentage of mothers who received prenatal care during the first trimester was lower among mothers with a MH disorder (70.7% vs. 85.7%). Notably, 41.3% of infants with mothers who had a MH disorder had documented maternal substance abuse versus less than 0.5% of infants born to mothers without a diagnosed MH disorder. Differences emerged among mothers with specific diagnoses of psychotic, mood, or anxiety disorders at birth compared with the diagnosis of MH Disorder at Delivery. Mothers with at least one of these specific diagnoses tended to be older at the time of their first birth than mothers with a MH Disorder at Delivery diagnosis, with roughly 40% aged 30 years or older. Among mothers with diagnosed psychotic disorders, Black mothers were disproportionately represented. Nearly a quarter of mothers with a psychotic disorder were Black (23.4%) versus only 5.4% of mothers without a MH disorder. White mothers were overrepresented among those classified with mood and anxiety disorders. Public insurance covered 80.7% of births in which there was a diagnosed psychotic disorder, versus only 40.5% and 29.9% of those with mood and anxiety disorders, respectively. Nearly 50% of births to mothers with psychotic disorders were missing paternity, and one third had started prenatal care only after the first trimester or not at all. Underscoring the distinct profiles of mothers with specific MH disorder diagnoses versus a general diagnosis at delivery is the percentage of mothers with co-diagnosed substance exposure, which ranged from 5.4% (anxiety disorder) to 24.3% (psychotic disorder) for specific diagnoses compared to 44.0% for a MH Disorder at Delivery. of mothers with a psychotic disorder were Black (23.4%) versus only 5.4% of mothers without a MH disorder. White mothers were overrepresented among those classified with mood and anxiety disorders. Public insurance covered 80.7% of births in which there was a diagnosed psychotic disorder, versus only 40.5% and 29.9% of those with mood and anxiety disorders, respectively. Nearly 50% of births to mothers with psychotic disorders were missing paternity, and one third had started prenatal care only after the first trimester or not at all. Underscoring the distinct profiles of mothers with specific MH disorder diagnoses versus a general diagnosis at delivery is the percentage of mothers with co-diagnosed substance exposure, which ranged from 5.4% (anxiety disorder) to 24.3% (psychotic disorder) for specific diagnoses compared to 44.0% for a MH Disorder at Delivery. Figure 1 depicts the cumulative percentage of infants reported to CPS during the first 12 months of life based on the diagnosis and type of maternal MH disorder. Overall, more than one third (34.6%) of infants born to mothers with a MH disorder were reported before one year of age, and a majority of those initial reports to CPS were made within the infant's first month of life (77.2%). In contrast, among children born to mothers without a MH disorder, 4.4% were reported to CPS during the first year of life. Among children born to mothers with a documented psychotic disorder, 68.5% were reported to CPS during infancy. However, the cumulative percentage reported to CPS was much lower among mother-child dyads in which anxiety or mood disorders were diagnosed (17.8% and 9.2%, respectively), yet these values were still twice as high as the percentage of those without any documented MH disorder (4.4%). Among children born to mothers with a documented psychotic disorder, 68.5% were reported to CPS during infancy. However, the cumulative percentage reported to CPS was much lower among mother-child dyads in which anxiety or mood disorders were diagnosed (17.8% and 9.2%, respectively), yet these values were still twice as high as the percentage of those without any documented MH disorder (4.4%). Table 2 presents findings from three different models examining the unadjusted and adjusted association between maternal MH disorders (with and without co-documented substance exposure) and the likelihood of a maltreatment report to CPS during infancy. As shown in Model 1, children born to mothers with any MH disorder were reported to CPS at nearly 8 times the rate of those whose mothers were without a MH disorder (RR: 7.82; 95% CI: 7.63, 8.02). The increased risk of reporting was most pronounced among births with a specific diagnosis of a psychotic disorder (RR: 13.10; 95% CI: 12.09, 14.19), although significant differences (p < 0.001) also emerged for births in which the specific diagnosis was for a mood or anxiety disorder. Model 1 displays the risk of being reported during infancy among children born to mothers with a medically documented MH condition compared with children whose mother was not documented to have a MH condition. Model 2 first compares children born to mothers with a MH diagnosis but no substance abuse diagnosis relative to children whose mothers did not have a MH condition and second shows the risk for children born to mothers with a comorbid MH and substance abuse condition documented relative to children born to moms with no MH condition. Model 3 builds upon Model 2 with adjustments for other risk factors present at birth.

Association between Maternal Mental Health Disorders and Maltreatment Reporting
Model 2 additionally examined the influence of co-documented substance exposure on CPS reporting. The combined effect of a maternal MH disorder and substance exposure was associated with a rate of CPS reporting 5 to 21 times that of other infants. For example, among mothers with a MH Disorder at Delivery diagnosis but no co-occurring substance abuse disorder, the RR was 3.16 (95% CI: 3.46, 3.84). The RR increased to 16.36 (95% CI: 15.54, 17.22) among mothers with a MH Disorder at Delivery diagnosis and comorbid substance use.
Finally, in Model 3, attempts to better isolate the relationship between maternal MH disorders and CPS reporting by adjusting for the covariates presented in Table 1 were made. After controlling for these covariates, the children of mothers with any maternal MH disorder but no a substance use disorder were still reported to CPS more during the first year of life at more than twice the rate of children born to mothers without a diagnosed MH disorder (RR: 2.60; 95% CI: 2.47, 2.73). Among infants born to mothers with both a maternal MH disorder and substance exposure, the risk remained even greater (RR: 5.69; 95% CI: 5.51, 5.87), albeit notably lower in the context of covariate adjustments.

Discussion
In the present study, linked administrative records were used to examine the relationship between maternal MH disorders documented at birth and CPS reporting during infancy. Several findings emerged that highlight the potential to not only use administrative records to monitor population-level dynamics, but also generate new knowledge concerning the relationship between maternal MH and CPS involvement.
First, this study provides a statewide estimate of the prevalence of infants born to mothers with MH disorders in the 2006 birth cohort. The percentage of MH disorders documented in the hospital delivery records of mothers in California increased from 1.7% in 1995 [17] to 2.8% a decade later, whereas the proportion of documented co-occurring MH and substance use disorders decreased. Additional research is needed to determine whether substance abuse has, indeed, decreased among mothers with a MH diagnosis or whether these findings reflect changes that occurred in documentation practices over the 10-year period. Whereas these estimates may be imperfect measures of the true prevalence of maternal MH disorders at birth, they underscore the potential for using hospital discharge data for population-level health surveillance. By focusing on mothers in which MH disorders were known to service providers at delivery, this analysis has implications for targeted prevention and early intervention services. Future studies would benefit from the inclusion of postpartum MH diagnoses in the data, which were not available for the present study.
Second, the present study produced a first-ever estimate of the occurrence of alleged maltreatment among infants born to mothers with MH disorders among both publicly and privately insured mothers in California. Statewide, one third of infants born to mothers with a MH disorder were reported to CPS by 12 months of age compared to fewer than one in every 20 infant born to a mother without a documented MH disorder. There were notable differences in the likelihood of CPS reporting by the specific MH disorder diagnosed. Overall, two thirds of infants born to mothers with a psychotic disorder were reported to CPS, which represents a rate more than 4 times that of infants born to mothers without a MH disorder after controlling for other risk factors at birth. Infants born to mothers with a mood disorder were more than twice as likely to have a CPS report, and infants born to mothers with an anxiety disorder experienced a small, but significant increase relative to infants of mothers without a MH disorder. Parental MH information is not always made available to child welfare social workers, and when provided, it is not systematically documented within CPS record systems when available. These findings provide useful estimates for understanding the scope of MH supports needed to service CPS-involved families. Further, the variations observed among CPS contact across the different types of MH disorders have implications for the development of child safety and risk-assessment tools and case-related decision making [28].
Third, for a majority of infants born to mothers with MH diagnoses who experienced a CPS report, the first report was made during the first month of life. This early reporting pattern indicates that those making a report to CPS were concerned about the mother's capacity to care for her infant shortly after delivery and responded proactively. One could argue that high rates of CPS reporting immediately following childbirth may be more indicative of preconceptions about mental illness than of actual risk to a child's safety. However, results from the present study also indicate that mothers with MH disorders who were not reported to CPS immediately following childbirth still had a comparatively higher occurrence of CPS reports by the infant's first birthday compared with infants born to mothers without a MH diagnosis. This finding supports earlier work suggesting that maternal mental illness may have a lasting impact on an infant's safety and wellbeing [14]. The decision to use administrative records for population surveillance undoubtedly requires ethical consideration; however, despite the risks, the present findings have the potential to enable better-informed decision making, communication, and collaboration between the MH and child welfare fields.
Finally, the present study highlights the risk associated with co-occurring MH and substance abuse diagnoses. Previous literature has documented an association among child maltreatment, parental mental illness [8], and substance abuse [29,30]. Findings from this study indicate that comorbid substance use amplifies the likelihood of CPS contact among mothers with MH diagnoses and that substance use may impact mothers differently based on their specific MH diagnosis type, even after controlling for covariates. This finding suggests that whereas a MH diagnosis may indicate a heightened level of risk for an infant, information about maternal substance use will provide critical information for understanding the risk level of child maltreatment. Mothers who are known to have co-occurring MH and substance abuse diagnoses are in need of enhanced services to prevent CPS involvement. Findings from the present study highlight the importance of collaboration among obstetrics, pediatric, MH, drug treatment, and child welfare service providers.
There are at least four limitations to this study. First, ICD codes are an imperfect measure of maternal mental illness, as all administrative data analyses are limited by errors in data entry [31], suggesting that ICD codes undoubtedly provide an undercount of the true occurrence of maternal MH and substance use disorders [31]. Yet, these codes offer a standardized tracking method that can be used to evaluate changes over time and are available for full birth cohorts in the hospital setting. Future research would benefit from the inclusion of ICD codes documented by outpatient service providers. Second, there may be surveillance bias in screening, diagnosing, and documenting individuals who receive diagnoses and are reported to CPS; for example, one earlier study found that sociodemographic risk factors increased the likelihood that a new mother's MH and substance use symptoms and diagnoses would be documented [16]. Third, whereas the MH Disorder at Delivery code fell within the parameters of MH problems defined for this study and was included in earlier California analyses [16,17], this code appears to be used as a catchall diagnostic category and may, therefore, document MH disorders that would more appropriately be classified as psychotic, mood, or anxiety disorders. Future research regarding maternal MH disorders would benefit from a more in-depth study of the use and significance of the mental disorders at delivery category. Fourth, the year and geographic location in which the data were collected limit the generalizability of this study's findings.

Conclusions
Infants born to mothers with a diagnosed MH disorder at the time of birth have a heightened likelihood of experiencing early CPS involvement, particularly in infants born to mothers with psychotic disorders or with a comorbid substance use disorder. CPS reports are most likely to occur during the child's first month of life, making this a critical window for intervention. Early identification of infants born to mothers with MH disorders and co-occurring substance abuse should be viewed as an opportunity to provide parents with integrated support services early to prevent the occurrence of child maltreatment. Administrative records provide a cost-effective method for developing population-level data to inform policy development and improve service delivery.
Acknowledgments: Kechen Zhou assisted with the preparation of the data files, and John Prindle provided technical assistance in model development. Although the findings reported and conclusions drawn from these data are solely those of the authors and should not be considered to reflect those of any agency of the California government, this analysis would not be possible without the partnership of the California Department of Social Services and the county child welfare departments, reflecting their ongoing commitment to data-driven program and policy development. Supported by grants from First 5 Los Angeles and the Maternal and Child Health Research Program, Maternal and Child Health Bureau (Title V, Social Security Act), Health Resources and Services Administration, and Department of Health and Human Services (R40MC25689. Ongoing infrastructure support for the linkage and analysis of administrative data is provided to the Children's Data Network by First 5 Los Angeles and the Conrad N. Hilton Foundation.
Author Contributions: All authors have participated in the concept and design; analysis and interpretation of data; drafting or revising of the manuscript, and have approved the manuscript as submitted. Authors Ivy Hammond and Andrea Lane Eastman prepared the first draft of the manuscript. John Leventhal provided technical assistance with model development and content expertise. Emily Putnam-Hornstein is the principal investigator and was responsible for the extract and preparation of the data, conceptualization of the statistical models, and contributed to the preparation of the article.

Conflicts of Interest:
The authors declare no conflicts of interest.