Quantifying sources of bias in longitudinal data linkage studies of child abuse and neglect: measuring impact of outcome specification, linkage error, and partial cohort follow-up

Risk and Protective Factors for Parental Involvement and Early Indicators of School Achievement in Alaska

OBJECTIVES

Parental involvement can affect child school readiness, which in turn influences subsequent child learning outcomes. While social support, stress, caregiver psychological distress, and drinking could affect parental involvement, it is unknown whether and how these factors influence downstream child learning outcomes through parental involvement and child school readiness. This study tests those associations.

METHODS

Using de-identified data provided by the Alaska Longitudinal Child Abuse and Neglect Linkage project (N = 683), we use Structural Equation Modeling to assess direct and indirect effects of paths embedded in the proposed model.

RESULTS

This study found statistically significant indirect effects: (1) path linking stress faced by caregivers to child reading proficiency through caregiver psychological distress, parental involvement, and child school readiness, (2) path linking stress faced by caregivers to child reading proficiency through caregiver drinking, parental involvement, and child school readiness, and (3) path linking social support for caregivers to child reading proficiency through caregiver psychological distress, parental involvement, and child school readiness. Post-estimation showed that the sum of the magnitude of total effects of stress and the magnitude of total effects of support is significantly larger than either alone.

CONCLUSIONS FOR PRACTICE

Findings suggest that reducing caregiver stress and offering social support could not only benefit caregivers but learning outcomes of their children as well. For child learning outcomes, simultaneously reducing stress and offering social support for caregivers, rather than just one of them alone, is suggested. These results are important for children, particularly for those raised by caregivers experiencing psychological distress or drinking issues.

Leveraging Administrative Data to Better Understand and Address Child Maltreatment: A Scoping Review of Data Linkage Studies

BACKGROUND

This scoping review aimed to overview studies that used administrative data linkage in the context of child maltreatment to improve our understanding of the value that data linkage may confer for policy, practice, and research.

METHODS

We searched MEDLINE, Embase, PsycINFO, CINAHL, and ERIC electronic databases in June 2019 and May 2020 for studies that linked two or more datasets (at least one of which was administrative in nature) to study child maltreatment. We report findings with numerical and narrative summary.

RESULTS

We included 121 studies, mainly from the United States or Australia and published in the past decade. Data came primarily from social services and health sectors, and linkage processes and data quality were often not described in sufficient detail to align with current reporting guidelines. Most studies were descriptive in nature and research questions addressed fell under eight themes: descriptive epidemiology, risk factors, outcomes, intergenerational transmission, predictive modelling, intervention/service evaluation, multi-sector involvement, and methodological considerations/advancements.

CONCLUSIONS

Included studies demonstrated the wide variety of ways in which data linkage can contribute to the public health response to child maltreatment. However, how research using linked data can be translated into effective service development and monitoring, or targeting of interventions, is underexplored in terms of privacy protection, ethics and governance, data quality, and evidence of effectiveness.

Changes in Household Challenges and Subsequent Child Welfare Report

INTRODUCTION

Preventing child maltreatment and reducing adverse childhood experiences is critical for improving adult health. To inform prevention efforts, it is necessary to move beyond static risk models and instead model the dynamic changes in household challenges during the prebirth and early childhood periods. This study examined the effect of changes in the number of household challenges from prebirth (12 months before birth of a child) to early childhood (3 years after birth) period on the risk of a child maltreatment report by age 3 years.

METHODS

This retrospective cohort study linked data from the Alaska 2009-2011 Pregnancy Risk Assessment Monitoring System, its 3-year follow-up survey, and administrative records through 2019. Participants were 1,699 birthing parents. Latent class analyses identified prebirth and early childhood low- and high-challenge respondent groups on the basis of the level of reported household challenges. The authors then modeled the relationships between group transition membership and the risk of maltreatment using latent transition analysis. Analyses were conducted in 2021.

RESULTS

Households transitioning from a high-challenge-prebirth status to a low-challenge-early-childhood status had a lower predicted risk for child services report than households remaining in the high-challenges group. Transitioning from low- to high-challenges status predicted the highest risk for child services report than that of all other groups.

CONCLUSIONS

To reduce the risk of child maltreatment and subsequent adverse childhood experiences, healthcare providers should screen parents for the presence of household challenges during both pregnancy and early childhood and connect patients to resources targeted at reducing those challenges and providing continuous familial support.

Pre-Birth Household Challenges Predict Future Child’s School Readiness and Academic Achievement

Early developmental success and school readiness strongly influence future skill development, occupational opportunities, and health. Therefore, it is critical to identify and address early determinants of school readiness for supporting children’s overall well-being and success. In this retrospective cohort study, we examined the effects of pre-birth household challenges, such as homelessness or experiences of intimate partner violence, on children’s early school readiness. We linked data from the Alaska 2009–2011 Pregnancy Risk Assessment Monitoring System (PRAMS) to administrative and education records through 2019. Education records included kindergarten developmental scores, third grade reading assessments, and attendance records. Generalized linear models with Quasi-Poisson distributions for each outcome of interest examined the predictive value of pre-birth household challenges on the risks of not meeting school readiness expectations. We found that experiencing higher numbers of pre-birth household challenges was related to higher risk of the child not meeting developmental and reading proficiency and having chronic absenteeism. These results suggest that it is imperative support systems for pregnant persons and their families be introduced as soon as possible in pre-natal care routines to address current pre-birth household stressors and prevent future challenges. Such early prevention efforts are needed to ensure the best possible developmental start for children.

Research using population-based administration data integrated with longitudinal data in child protection settings: A systematic review

INTRODUCTION

Over the past decade there has been a marked growth in the use of linked population administrative data for child protection research. This is the first systematic review of studies to report on research design and statistical methods used where population-based administrative data is integrated with longitudinal data in child protection settings.

METHODS

The systematic review was conducted according to Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) statement. The electronic databases Medline (Ovid), PsycINFO, Embase, ERIC, and CINAHL were systematically searched in November 2019 to identify all the relevant studies. The protocol for this review was registered and published with Open Science Framework (Registration DOI: 10.17605/OSF.IO/96PX8).

RESULTS

The review identified 30 studies reporting on child maltreatment, mental health, drug and alcohol abuse and education. The quality of almost all studies was strong, however the studies rated poorly on the reporting of data linkage methods. The statistical analysis methods described failed to take into account mediating factors which may have an indirect effect on the outcomes of interest and there was lack of utilisation of multi-level analysis.

CONCLUSION

We recommend reporting of data linkage processes through following recommended and standardised data linkage processes, which can be achieved through greater co-ordination among data providers and researchers.

Prebirth Household Challenges To Predict Adverse Childhood Experiences Score by Age 3

OBJECTIVES

With this study, we seek to understand the relationship between prebirth household challenges and the child's adverse childhood experiences (ACEs) score by age 3 in a statewide-representative birth cohort to inform primary prevention strategies.

METHODS

We used a longitudinally linked data set from the Alaska 2009-2011 Pregnancy Risk Assessment Monitoring System, its 3-year follow-up survey, and multiple administrative data sources. Using this linked data set, we predicted an expanded ACEs score by age 3 using maternal reported prebirth household challenges.

RESULTS

The number of household challenges reported during the 12 months before or during pregnancy predicted ACEs score in a graded, dose-response manner. On average, reporting 4+ prebirth household challenges was associated with an ACEs score 4.1 times that of those reporting 0 challenges. Homelessness was associated with the greatest increase in ACEs score (relative rate ratio = 3.0). Prebirth household challenges that were independently associated with an elevated ACEs score in our final model included problems paying bills, someone close to the mother having a drinking and/or drug problem, homelessness, mother or husband or partner being in jail, husband or partner losing job, separation or divorce, and being checked or treated for anxiety or depression.

CONCLUSIONS

The accumulation and certain prebirth household challenges are strongly associated with the accumulation of childhood ACEs. Addressing and reducing household challenges during the prebirth period may serve as a primary point of ACEs prevention. Many evidence-based, multidisciplinary intervention strategies can and should be implemented in the prebirth period to strengthen the household unit before the introduction of a new child.

Patterns of Risk and Protective Factors Among Alaska Children: Association With Maternal and Child Well-Being

This study used population-representative data to examine associations of risk and protective factor patterns among Alaska Native/American Indian (AN/AI; N = 592) and non-Native (N = 1,018) children with maternal and child outcomes at age 3 years. Among AN/AI children, a high risk/moderate protection class was associated with child developmental risk and mothers being less likely to feel comfortable asking for help or knowing where to go for parenting information compared to a low socioeconomic status/high protection class. Among non-Native children, a moderate risk/high protection class was associated with child developmental risk and mothers being less likely to feel comfortable asking for help compared to a low risk/high protection class. Results provide insight on the intersection of risk and protective factors among Alaska families.

Measuring the Incidence of Child Maltreatment Using Linked Data: A Two-State Comparison

INTRODUCTION

Measuring and comparing the incidence of child maltreatment is challenging. Linkage of statewide birth cohorts with Child Protective Services reports to study incident child maltreatment over the life course are becoming more common. This study compares the reported incidence between 2 states derived from population-based administrative data linkages.

METHODS

Linked births (2009-2011) with Child Protective Services records (2009-2015) and deaths in each state were used to compare the cumulative incidence of a Child Protective Services report before age 7 years. Given differences in population race structure and documented disparities of race groups in Child Protective Services data, variation was adjusted for using direct standardization. Unadjusted cumulative incidence, race cumulative incidence, and race-adjusted cumulative incidence were compared. Analyses were completed in 2018.

RESULTS

Before age 7 years, 26.0% of Alaskan children and 19.0% of Californian children were reported to Child Protective Services (RR=1.37, p<0.001). Aside from Asian/Pacific Islanders, the cumulative incidence between states was similar for each race. The race-adjusted cumulative incidence indicated that children born in Alaska were 1.10 times as likely to experience a report before age 7 years compared with children in California.

CONCLUSIONS

Much of the difference in risk for child maltreatment observed between Alaska and California is most likely due to variation in the population structure by race as opposed to modifiable factors. Standardization is a simple method to adjust for population structure differences. This study contributes to the growing body of knowledge regarding the use of linked administrative data to study maltreatment and provides insights into considerations for making comparisons or conducting cross-jurisdictional analyses based on commonly aligned data sets.

Heterogeneity in Risk and Protection Among Alaska Native/American Indian and Non-Native Children

Currently, little is known about patterns of co-occurring risk and protective factors among young children. Understanding variations in co-occurring risk and protective factors among children in Alaska is important as experiences of collective trauma may contribute to differences in the intersection of risk and protective factors between Alaska Native/American Indian (AN/AI) and non-Native children. Using data from the Alaska Longitudinal Child Abuse and Neglect Linkage (ALCANLink) project, a linkage of the 2009-2011 Alaska Pregnancy Risk Assessment Monitoring System survey and administrative data sources, and the 2012-2014 Childhood Understanding Behaviors Survey, we conducted latent class analysis to identify classes of AN/AI (N = 593) and non-Native (N = 1018) children in terms of seven risk factors (poverty, maternal depression, maternal binge drinking, parental incarceration, intimate partner violence exposure, other violence exposure, child maltreatment) and four protective factors (father figure involvement, reading by adults, family meals, peer interactions) experienced prior to age 3 years. We identified two classes among AN/AI children: (1) high risk-moderate protection (29.1%) and (2) low socioeconomic status-high protection (70.9%). We identified two classes among non-Native children: (1) moderate risk-high protection (32.9%) and (2) low risk-high protection (67.1%). A test of invariance revealed that risk and protective factor probabilities differed significantly for corresponding classes of AN/AI and non-Native children. Overall, results demonstrate heterogeneity within and between AN/AI and non-Native children in early experiences of risk and protection and suggest that interventions will be more effective if tailored to the experiences and developmental needs of specific groups of Alaska children.

Preconception and Prenatal Predictors of Early Experiences of Risk and Protection Among Alaska Children

OBJECTIVES

Our objective was to identify preconception and prenatal predictors of early experiences of co-occurring risk and protective factors to help target prevention efforts to the highest-need families prior to the birth of the child.

METHODS

Data were from the Alaska Longitudinal Child Abuse and Neglect Linkage project and the 2012-2014 Alaska Child Understanding Behaviors Survey. We used latent class analysis and Vermunt's three-step approach to examine predictors of latent classes of risk and protective factors among Alaska children.

RESULTS

Among children of Alaska Native/American Indian mothers, financial (OR 2.02, 95% CI 1.04, 3.90) and partner stress (OR 2.06, 95% CI 1.02, 4.10) prior to childbirth, maternal education < 12 years (OR 2.29, 95% CI 1.05, 4.96), and maternal substance use (OR 2.52, 95% CI 1.30, 4.89) were associated with a higher likelihood of membership in a high risk/moderate protection class as compared to a low socioeconomic status/high protection class. Among children of non-Native mothers, partner stress prior to childbirth (OR 3.92, 95% CI 1.08, 14.19), maternal education < 12 years (OR 2.69, 95% CI 1.24, 5.81), maternal substance use (OR 2.69, 95% CI 1.24, 5.81), younger maternal age (OR 0.87, 95% CI 0.80, 0.95), and a greater number of children (OR 1.62, 95% CI 1.09, 2.41) were associated with a higher likelihood of membership in a moderate risk/high protection class as compared to a low risk/moderate protection class.

CONCLUSIONS

Results can inform eligibility criteria for prenatal home visiting programs and prenatal screening in Alaska to ensure prevention programming and referrals are directed to families most in need of additional support.

Trajectories of child protective services contact among Alaska Native/American Indian and non-Native children

BACKGROUND

Contact with child protective services (CPS) functions as an independent marker of child vulnerability. Alaska children are an important population for understanding patterns of CPS contact given high rates of contact overall and among specific demographic groups.

OBJECTIVE

We aimed to identify longitudinal trajectory classes of CPS contact among Alaska Native/American Indian (AN/AI) and non-Native children and examine preconception and prenatal risk factors associated with identified classes.

PARTICIPANTS AND SETTING

We used data from the Alaska Longitudinal Child Abuse and Neglect Linkage (ALCANLink) project, a linkage of 2009-2011 Alaska Pregnancy Risk Assessment Monitoring System (PRAMS) births with administrative data including CPS records.

METHODS

We conducted growth mixture modeling to identify trajectory classes of CPS contact from birth to age five years. We used Vermunt's three-step approach to examine associations with preconception and prenatal risk factors.

RESULTS

Among AN/AI children, we identified three classes: 1) no/low CPS contact (75.4%); 2) continuous CPS contact (19.6%), and 3) early, decreasing CPS contact (5.0%). Among non-Native children, we identified four classes: 1) no CPS contact (81.3%); 2) low, increasing CPS contact (9.5%); 3) early, rapid decline CPS contact (5.8%); and 4) high, decreasing CPS contact (3.3%). Maternal substance use had the largest impact on probabilities of class membership, increasing the probability of membership in classes characterized by CPS contact, among both AN/AI and non-Native children.

CONCLUSIONS

Results reveal heterogeneity in longitudinal patterns CPS contact across early childhood among Alaska children and identify maternal substance use as an important target for primary prevention.

Survey evidence of the decline in child abuse in younger Canadian cohorts

Physical and sexual abuse in childhood is a worldwide phenomenon with potentially dramatic consequences of both a psychological and physical nature. Measures of primary prevention have been developed in some countries. In the USA, child protection services reports and research surveys indicate that child sexual abuse has been on the decline in recent decades. Results are less clear for physical and overall abuse. The aim of this study was to describe how childhood abuse has changed over the years in Canada through an analysis of the 2012 Canadian Community Health Survey: Mental Health Edition data. The sample comprised 22,775 respondents ages 20 and over who completed a child abuse questionnaire. Respondents born from 1983 to 1992 reported significantly less overall abuse, physical abuse, and sexual abuse than did older generations, with the exception of people born in 1942 or earlier. The decrease was observed among men and women and across all the regions of Canada.Conclusion: The results are encouraging in that they may have an impact on life expectancy, severity of various chronic disorders, and suicide in the population. They also support policies that have focused on improving the childhood environment in the 1990s. Results also underline the importance of using different kinds of data sources for evaluating child abuse. What is Known: • Physical and sexual abuse in childhood has been associated with lower life expectancy in connection with an array of chronic diseases, including mental disorders, and with suicide. • Measures of primary prevention have been developed in some countries, such as the USA and Canada. What is New: • Canadians born from 1983 to 1992 report significantly less overall abuse, physical abuse, and sexual abuse than older generations do. • These encouraging results support policies implemented in the 1990s focused on improving the childhood environment.

Assessment of Rates of Child Maltreatment in States With Medicaid Expansion vs States Without Medicaid Expansion

IMPORTANCE

Physical abuse and neglect affect a significant number of children in the United States. The 2014 Medicaid expansion, in which several states opted to expand their Medicaid programs, is associated with parental financial stability and access to mental health care.

OBJECTIVE

To determine whether Medicaid expansion is associated with changes in physical abuse and neglect rates.

DESIGN, SETTING, AND PARTICIPANTS

This ecological study used state-level National Child Abuse and Neglect Data Systems (NCANDS) data from January 1, 2010, through December 31, 2016, to compare the change in physical abuse and neglect rates in states that chose to expand Medicaid vs those that did not. All cases of physical abuse and neglect of children younger than 6 years during the study period that were referred to state-level Child Protective Services and screened in for further intervention after having met a maltreatment risk threshold were included. Cases with only documented sexual or emotional abuse were excluded. A difference-in-difference analysis was conducted from April 12, 2018, through March 26, 2019.

EXPOSURES

State-level Medicaid expansion status.

MAIN OUTCOMES AND MEASURES

Incidence rate of screened-in referrals for physical abuse or neglect per 100 000 children younger than 6 years per year by state.

RESULTS

Data were analyzed for 31 states and the District of Columbia that expanded Medicaid and 19 states that did not during the study period, with baseline neglect counts of 646 463 and 388 265, respectively. After Medicaid expansion, 422 fewer cases of neglect per 100 000 children younger than 6 years (95% CI, -753 to -91) were reported each year after adjusting for confounders for comparison of postexpansion and preexpansion rates in states that expanded Medicaid contrasting with the change during that time in nonexpansion states. From 2013 to 2016, Medicaid coverage for adults with dependent children increased a median 1.9% (interquartile range, 0.4% to 4.3%) in the states that did not expand Medicaid and 4.2% (interquartile range, 0.9% to 6.0%) in the states that did. No associations were found between Medicaid coverage or Medicaid eligibility criteria and physical abuse or neglect rates.

CONCLUSIONS AND RELEVANCE

Medicaid expansion was associated with a reduction in the reported child neglect rate, but not the physical abuse rate. These findings suggest that expanding Medicaid may help prevent child neglect.

Sociodemographic differences in linkage error: an examination of four large-scale datasets

BACKGROUND

Record linkage is an important tool for epidemiologists and health planners. Record linkage studies will generally contain some level of residual record linkage error, where individual records are either incorrectly marked as belonging to the same individual, or incorrectly marked as belonging to separate individuals. A key question is whether errors in linkage quality are distributed evenly throughout the population, or whether certain subgroups will exhibit higher rates of error. Previous investigations of this issue have typically compared linked and un-linked records, which can conflate bias caused by record linkage error, with bias caused by missing records (data capture errors).

METHODS

Four large administrative datasets were individually de-duplicated, with results compared to an available 'gold-standard' benchmark, allowing us to avoid methodological issues with comparing linked and un-linked records. Results were compared by gender, age, geographic remoteness (major cities, regional or remote) and socioeconomic status.

RESULTS

Results varied between datasets, and by sociodemographic characteristic. The most consistent findings were worse linkage quality for younger individuals (seen in all four datasets) and worse linkage quality for those living in remote areas (seen in three of four datasets). The linkage quality within sociodemographic categories varied between datasets, with the associations with linkage error reversed across different datasets due to quirks of the specific data collection mechanisms and data sharing practices.

CONCLUSIONS

These results suggest caution should be taken both when linking younger individuals and those in remote areas, and when analysing linked data from these subgroups. Further research is required to determine the ramifications of worse linkage quality in these subpopulations on research outcomes.

Using time-to-event analysis to identify preconception and prenatal predictors of child protective services contact

We examined preconception and prenatal predictors of time to first child protective services (CPS) contact among Alaska children. Data were from the Alaska Longitudinal Child Abuse and Neglect Linkage (ALCANLink) project, a population-representative data source linking 2009-2011 Alaska Pregnancy Risk Assessment Monitoring System (PRAMS) data with administrative data sources through 2015. We examined the incidence CPS contact using the Kaplan-Meier method and predictors of CPS contact using Cox proportional hazards regression. Using data from the Alaska Permanent Fund Dividend and Child Death Review, we censored children who emigrated out-of-state or died during the study period. Significant predictors included low socioeconomic status (HR = 2.23, 95% CI 1.68, 2.96), maternal smoking during pregnancy (HR = 1.87, 95% CI 1.55, 2.24), unmarried maternal marital status (HR = 1.62, 95% CI 1.31, 1.99), urban residence (HR = 1.59, 95% CI 1.32, 1.92), lower maternal education (HR = 1.54, 95% CI 1.24, 1.92), maternal experience of intimate partner violence in the 12 months before childbirth(HR = 1.32, 95% CI 1.01, 1.74), Alaska Native/American Indian race (HR = 1.40, 95% CI 1.15, 1.71), a greater number of living children (HR = 1.20, 95% CI 1.13, 1.29), a greater number of stressful life eventsin the 12 months before childbirth (HR = 1.16, 95% CI 1.11, 1.21), and younger maternal age at childbirth (HR = 0.95, 95% CI 0.93, 0.97). Use of multiple linked data sources and time-to-event analysis methods adds to the growing literature regarding predictors of CPS contact. Results suggest that assessing for and addressing clinical, social, and environmental indicators during the prenatal period may aid prevention efforts in mitigating family need for involvement with CPS.

A Longitudinal Study of Child Maltreatment and Mental Health Predictors of Admission to Psychiatric Residential Treatment Facilities

The child welfare system is an access point for children's mental health services. Psychiatric residential treatment facilities (PRTFs) are the most restrictive, and most expensive setting for children to receive long-term care. Given the high rates of behavioral health concerns among maltreated children in out-of-home care, research is needed to examine the factors that predict entry in PRTFs among children investigated for maltreatment. This exploratory study used cross-sector administrative records linked across multiple systems, including child welfare records and Medicaid claims, from a single state over a five-year period (n = 105,982). Cox proportional hazards modeling was used to predict entry into a PRTF. After controlling for many factors, PRTF entry was predicted by diagnosis code indicating a trauma-related condition, antipsychotic medication prescriptions, and entry into lower levels of out-of-home care, supporting the view that youth are admitted to PRTFs largely due to clinical need. However, PRTF admission is also associated with characteristics of their experiences with the social service system, primarily foster care placement stability and permanency. Implications for practice and research are discussed.