Predictors of child protective service contact between birth and age five: An examination of California's 2002 birth cohort

https://doi.org/10.1016/j.childyouth.2011.04.006Get rights and content

Abstract

This study utilizes population-level birth data to describe those children who may be at greatest risk of maltreatment during the first five years of life. Based on a unique dataset constructed by linking California's administrative child welfare data to statewide vital birth records, a cohort study design was employed to track reports of maltreatment involving children born in 2002. Twelve variables captured in the birth record were selected for analysis. Generalized Linear Models were used to estimate adjusted risk ratios (RR) for each independent variable. Predicted probabilities of CPS contact were computed based on the count of risk factors present at birth. Results suggest that many of the associations previously observed between birth variables and subsequent maltreatment have sustained value in foretelling which children will be reported to CPS beyond infancy. Of the 531,035 children born in California in 2002, 14% (74,182) were reported for possible maltreatment before the age of five. Eleven of the twelve birth variables examined presented as significant predictors of contact with child protective services.

Research Highlights

► This article presents findings from a prospective birth cohort analysis. ► We estimate a child’s risk of being reported for maltreatment by age 5. ► Findings demonstrate that children at-risk of maltreatment can be targeted using birth data. ► We discuss using birth record data to construct an epidemiologic risk assessment tool.

Introduction

The maltreatment of children, including neglect and various forms of physical, sexual, and emotional abuse, endures as a societal problem of significant scope. In 2008, referrals involving approximately 6 million children believed to have been harmed or at risk of harm were made to child protective service agencies (CPS) in the United States (U.S. Department of Health and Human Services, 2010). Of these children, an estimated 3.7 million were included in an investigation and over 700,000 (10.6 per 1000) were subsequently deemed victims of maltreatment. Among maltreated children, the nation's youngest children were disproportionately represented: 33% of all victims were under four years of age; the highest rate of maltreatment was observed during the first year of life (21.7 per 1000).

The profound vulnerability of these youngest children cannot be understated. Not only are they at greatest risk of death from maltreatment (Schnitzer and Ewigman, 2005, U.S. Department of Health and Human Services, 2010), but research suggests that the negative developmental consequences of non-fatal maltreatment manifest in multiple domains of later life functioning (Currie and Widom, 2010, Felitti et al., 1998, Glaser, 2000, Springer et al., 2007), with outcomes often observed to be most severe and intractable for children with an onset of abuse or neglect during the first few years of life (English et al., 2005a, English et al., 2005b, Kotch et al., 2008, Wulczyn et al., 2005).

The high rates at which these youngest children are reported to child protective service agencies suggest that a renewed focus on primary prevention through various early intervention activities prenatally and shortly after birth may prove particularly impactful. Successful efforts have the potential to result in large and rather immediate declines in the number of children subjected to abuse and neglect. This paper explores how population-level birth data may be employed to identify those children who are at greatest risk of maltreatment.

Prior studies have examined the relationship between birth indicator variables and a verified or substantiated case of infant maltreatment (Wu et al., 2003), maltreatment among children who received home visiting services (Murphey & Braner, 2000), as well as infant entries to foster care following a maltreatment substantiation (Needell & Barth, 1998). A body of literature linking child welfare and birth records also arises from Europe, where linkages between administrative data sources are more common (Murphy et al., 1981, Sidebotham and Heron, 2006, Spencer et al., 2006). Additionally, ecological studies examining associations between child maltreatment risk and perinatal characteristics at the community level have also been conducted (Lee and Goerge, 1999, Zhou et al., 2006).

From these separate empirical streams, a stable list of child and family characteristics at birth has emerged for their association with subsequent maltreatment. Pregnancy variables with prior predictive value include low birth weight (Murphy et al., 1981, Needell and Barth, 1998, Spencer et al., 2006, Wu et al., 2003), a birth abnormality (Murphy et al., 1981, Needell and Barth, 1998), late or an absence of prenatal care (Murphey and Braner, 2000, Murphy et al., 1981, Needell and Barth, 1998, Wu et al., 2003, Zhou et al., 2006), and a prior pregnancy termination (Parrish and Gessner, 2010, Wu et al., 2003).

Strong associations have also been observed between a child's report to child protective services and several sociodemographic variables captured on the birth record including maternal race/ethnicity (Lee and Goerge, 1999, Needell and Barth, 1998), young maternal age at birth (Lee and Goerge, 1999, Murphey and Braner, 2000, Murphy et al., 1981, Needell and Barth, 1998, Sidebotham and Heron, 2006, Wu et al., 2003, Zhou et al., 2006), low levels of maternal education (Murphey and Braner, 2000, Needell and Barth, 1998, Sidebotham and Heron, 2006, Wu et al., 2003), single parent status (Murphey and Braner, 2000, Murphy et al., 1981, Needell and Barth, 1998, Wu et al., 2003, Zhou et al., 2006), the number of children born to the mother (Lee and Goerge, 1999, Murphey and Braner, 2000, Needell and Barth, 1998, Wu et al., 2003, Zhou et al., 2006), and the receipt of public insurance (Lee and Goerge, 1999, Murphey and Braner, 2000, Murphy et al., 1981, Needell and Barth, 1998, Wu et al., 2003, Zhou et al., 2006).

In this study, we utilize 12 birth record variables to predict a child's risk of being reported for maltreatment before the age of five. Although variables were chosen based on their previously demonstrated value in predicting maltreatment during infancy, to the best of our knowledge this is the first study to extend this method of population-based birth record linkage and inquiry to all children reported for maltreatment 1) through the age of five, 2) regardless of report disposition, and 3) inclusive of reports that were screened out and therefore not investigated.

This study serves as a preliminary examination of birth variables that predict which children will be reported for maltreatment by age five. The potential usefulness of using birth data to construct an epidemiologic risk-assessment tool was argued by Wu and colleagues who were able to identify 50% of all substantiated infant maltreatment cases from 13% of Florida's 1996 birth cohort, using just three variables from the birth record (2003). Other researchers have found that many risk factors measured in the neonatal period continue to predict CPS referrals through the fourth year of life (Kotch, Browne, Dufort, Winsor, & Catellier, 1999). In this paper, we seek to both describe a recent birth cohort of children who were reported for maltreatment during early childhood and explore the utility of birth variables for predicting a CPS referral throughout the years of peak child maltreatment vulnerability.

Section snippets

Data

This study utilizes a unique dataset constructed by linking California's administrative child welfare data to statewide vital birth records. Child welfare records were extracted from the state's Child Welfare Services Case Management System (CWS/CMS)—the statewide database for tracking children reported for possible abuse or neglect. CWS/CMS data are hosted at the Center for Social Services Research at the University of California at Berkeley through a longstanding interagency agreement with

Descriptive statistics

Table 1 presents the distribution of children who were and were not reported for maltreatment across variable levels, as well as the unadjusted risk of a referral. Of the full 531,035 children born alive in California in 2002, 13.9% (74,182) were reported for possible abuse or neglect before their fifth birthday, roughly consistent with earlier published maltreatment prevalence estimates from California (Magruder & Shaw, 2008). This rate of children who were reported should be considered a

Discussion

Prior research established that data universally collected at the time of birth and recorded in the birth record could be used to identify those infants most at risk of maltreatment. In this study, we find that many of the birth indicator variables previously observed to predict a report of maltreatment continue to emerge for a recent cohort of children born in California, beyond infancy and through the age of five, and inclusive of reports that are screened out or unfounded.

Overall, the

Research support

Support for this research was received from the Harry Frank Guggenheim Foundation, the Fahs-Beck Foundation, and the Center for Child and Youth Policy. Ongoing support for the California Performance Indicator's Project is provided by the California Department of Social Services and the Stuart Foundation.

Financial disclosure/conflict of interest

None.

Acknowledgments

The authors would like to acknowledge and thank colleagues at both the California Department of Social Services and the Center for Social Services Research for their assistance in preparing data files underlying this analysis. The authors would also like to thank Wendy Wiegmann for her assistance in editing this article.

References (45)

  • B. Needell et al.

    Infants entering foster care compared to other infants using birth status indicators

    Child Abuse & Neglect

    (1998)
  • J.W. Parrish et al.

    Infant maltreatment-related mortality in Alaska: correcting the count using birth certificates to predict mortality

    Child Abuse & Neglect

    (2010)
  • P. Sidebotham et al.

    Child maltreatment in the “children of the nineties”: A cohort study of risk factors

    Child Abuse & Neglect

    (2006)
  • K.W. Springer et al.

    Long-term physical and mental health consequences of childhood physical abuse: Results from a large population-based sample of men and women

    Child Abuse & Neglect

    (2007)
  • E. Chen et al.

    Socioeconomic differences in children's health: How and why do these relationships change with age?

    Psychological Bulletin

    (2002)
  • D.E. Clark

    Practical introduction to record linkage for injury research

    Injury Prevention

    (2004)
  • J. Currie et al.

    Long-term consequences of child abuse and neglect on adult economic well-being

    Child Maltreatment

    (2010)
  • H. Davies et al.

    When can odds ratios mislead?

    British Medical Journal

    (1998)
  • B. Drake

    Unraveling “unsubstantiated”

    Child Maltreatment

    (1996)
  • B. Drake et al.

    Substantiation and recidivism

    Child Maltreatment

    (2003)
  • I. Fellegi et al.

    A theory for record linkage

    Journal of the American Statistical Association

    (1969)
  • L. Franzini et al.

    Understanding the Hispanic paradox

    Ethnicity & Disease

    (2001)
  • Cited by (125)

    • Documenting Opportunity for Systematic Identification and Mitigation of Risk for Child Maltreatment

      2022, Journal of the American Academy of Child and Adolescent Psychiatry
    View all citing articles on Scopus
    View full text