Factors Associated with Child Removal Among American Indian and Alaska Native People in an Alcohol Intervention Study

Introduction

American Indian and Alaska Native (AI/AN) adults are less likely than Non-Hispanic Whites to use alcohol in their lifetime, equally as likely to have had five or more drinks on one occasion in the past month, and more likely to have a diagnosis related to alcohol dependence (Center for Behavioral Health Statistics and Quality, 2018; Cunningham et al., 2016; Grant et al., 2015). AI/AN communities experience high rates of alcohol-related negative consequences and child out-of-home placements compared to Non-Hispanic White communities (Baldwin et al., 2000; Barlow et al., 2012; Carter, 2011; Delker et al., 2016; Grinnell Davis et al., 2022).

Previous research among the general population has indicated that parental alcohol and other substance use increases the risk that a child will be removed from their parent’s care (Child Welfare Information Gateway, 2014; English et al., 2015; Fusco, 2015; Myhra, 2011; Myhra & Wieling, 2014; Ondersma, 2002). Factors associated with child removal include being of a lower socio-economic status, lacking social supports, and having a parental history of early life interpersonal trauma, such as physical, emotional, and/or sexual abuse (Cohen et al., 2008; English et al., 2015; Fusco, 2015; Lindsey, 1991; Thieman & Dail, 1997). Furthermore, a history of previous out-of-home placements and parental separation in their own childhood has been linked to families’ prolonged involvement in the child welfare system (Buisman et al., 2020; English et al., 2015; Foster et al., 2015; Marshall et al., 2011; Mertz & Andersen, 2017; Thieman & Dail, 1997; Thornberry et al., 2013).

Racial and ethnic minority children have disproportionately higher rates of out-of-home placements than Non-Hispanic Whites (Puzzanchera et al., 2022; The Annie E. Casey Foundation, 2022). In 2018, AI/AN children were 2.6 times more likely to experience a foster care placement than the general population (National Indian Child Welfare Association, 2019; U.S. Department of Health & Human Services, 2020). While the U.S. AI/AN child population was 1.1% in 2020 they comprised 2.6% of the child population in an out-of-home placement (Puzzanchera et al., 2022; The Annie E. Casey Foundation, 2022). Factors for child removal among AI/AN families are similar to the general population (Carter, 2009a; 2009b; 2010; 2011; Fox, 2004; Galos, 2017; Lujan et al., 1989). However, AI/AN communities are affected by historical losses and traumas that are a consequence of racist and colonialist policies, such as a history of government boarding schools that prohibited Native languages and culture and indoctrinated children into Euro-American practices and values (Beardall & Edwards, 2021; Crofoot & Harris, 2012; Maxwell, 2014; Running Bear et al., 2018; Whitbeck et al., 2004). Assimilationist policies continued into the 1970s and included widespread adoption of AI/AN children into non-Native families (Crofoot & Harris, 2012; Evans-Campbell & Campbell, 2011; Jacobs, 2013). The underlying structures that gave rise to these policies remain largely unchanged, as indicated by the disproportionately high number of AI/AN children removed from their parents’ care compared to non-Hispanic Whites (National Indian Child Welfare Association, 2019; Puzzanchera et al., 2022; U.S. Department of Health & Human Services, 2020). The removal of AI/AN children from AI/AN communities disrupts both the affected families, as well as intergenerational transmission of cultural knowledge, values, and practices. While expanding the legacy of historical trauma, these experiences also increase the likelihood of these individuals developing negative health outcomes, such as post-traumatic stress disorder and substance use disorders (Bensen, 2001; Brave Heart et al., 2011; Evans-Campbell, 2008; Evans-Campbell et al., 2012; Gone et al., 2019; Haag, 2013).

Despite the history of disruption, family is central to many AI/AN peoples, who comprise 574 sovereign Nations with diverse cultures and societies (Getches et al., 1998; National Conference of State Legislatures, 2020; Red Horse, 1980). Efforts by Tribal leaders and AI/AN advocates to reinforce sovereignty by preserving AI/AN families resulted in the passing of the Indian Child Welfare Act of 1978 (ICWA; Fletcher, 2009). The ICWA mandates that states actively support AI/AN parents and prioritize out-of-home placements, when necessary, to AI/AN homes that reflect the family’s unique cultural values to maintain the child’s connection to their culture, family, and community (P.L. 95-608).

The goal of this secondary data analysis was to describe the prevalence and associations of 326 AI/AN adult participants enrolled in a multisite randomized clinical trial of a culturally-tailored contingency management intervention for alcohol dependence (McDonell et al., 2021) who reported ever having a child removed from their care, in part, because of their alcohol use (our definition of alcohol-related child removal—our primary outcome). Consistent with the literature, we hypothesized that alcohol-related child removal would be associated with the following factors: the participant’s own experience with parental separation when they were children (our definition of parental separation in childhood), higher levels of historical loss and trauma, and early life experiences with interpersonal trauma.

Materials and Methods

Instruments and procedures

Outcome Variable.

Alcohol-related child removal was self-reported. The variable was assessed with a single question only gathered at baseline asking, “Have your kids ever been removed from your care, in part, because of your alcohol use?” Participants selected from four choices: “yes,” “no,” “N/A,” and “decline to answer.” Those who answered “N/A” or “decline to answer” were excluded from the analysis.

Other Variables.

We explored nine hypothesized independent variables, collected at baseline and hypothsized to and with a theories be associated with alcohol-related child removal, as follows:

Parental Separation in Childhood. The Early Life and Chronic Stress Scale (Jiang et al., 2008) was used to explore participants’ exposure to parental separation in childhood and interpersonal trauma before the age of 18. This scale’s parental separation in childhood domain has four binary yes/no variables that include whether the participant lived with others because of problems in the family, whether they went to boarding/reform school, whether they had been in foster care, and whether they had been adopted. The domain was analyzed as separate binary parental separation in childhood individual event types.

Interpersonal Trauma. The Early Life and Chronic Stress Scale also contains an interpersonal trauma domain with six questions. This domain was converted into a binary variable consistent with how the parental separation in childhood domain was converted (Jiang et al., 2008).

Historical Loss and Trauma. The Historical Loss and Trauma Scale (Whitbeck et al., 2004) has two domains to assess the frequency of thoughts and feelings related to historical loss and historical trauma, each containing 12 questions. A Likert scale was used for each domain, which was then converted to a binary response for each question. Consistent with published scoring of the measure, answers were coded “1” if they were greater than the median response, and “0” if they were equal to or less than the median response within the present study sample. All questions were added together to create a continuous composite score from 0 to 12 for both domains (Whitbeck et al., 2004). A higher historical loss score indicates more frequent contemplation of traumatic historical events that happened in their community. A higher historical trauma score indicates a higher level of negative emotional responses to those losses.

Demographic Factors. Parent’s age, sex, and education were included as control variables to increase the precision of our estimates (Table 1). Age in years was analyzed as a continuous variable. Biological sex was analyzed categorically as male or female. High school equivalent education was operationalized categorically as “12 or more years” or “less than 12 years.”

OPEN IN VIEWER

Table 1.

Descriptive Statistics of the Total Sample as well as Participants Who Have and Have Not Had an Alcohol-Related Child Removal at Baseline.^{symbol a,b}

	Total Sample (N= 326)	Alcohol-Related Child Removal (n = 132)	No Alcohol-Related Child Removal (n = 194)
Total participants		40.5	59.5
Age	42.4(11.0)	41.2(10.3)	43.3(11.3)
Female sex	50.0	65.9	39.2
Education (≥12 years)	81.2	74.2	86.0
The early life and chronic stress scale
Parental separation in childhood, by event type
Lived with others because of problems in the family	63.3	68.2	59.9
Attended boarding school	28.0	32.6	24.9
Were in foster care	27.1	28.0	26.4
Were adopted	9.2	13.6	6.2
Interpersonal trauma	73.5	26.7	26.4
Historical loss and trauma measures (0 to 12)
Historical loss	5.3(4.1)	6(4.3)	4.8(3.9)
Historical trauma	4.4(3.6)	4.1(3.6)	4.6(3.7)
Ethyl glucuronide negative (EtG<150 ng/mL)	57.1	41.6 symbol c	58.4 symbol d

a

All values indicate %/Mean(SD). “%” represents percent of “yes” answers out of total responses unless otherwise stated.

b

No row was missing more than 5 participants, unless otherwise noted.

c

Sample includes 77 participants.

d

Sample includes 108 participants.

Intervention Outcomes.

The primary outcome of the intervention study was urine alcohol biomarker ethyl glucuronide (EtG) level. Alcohol abstinence was defined as EtG<150 ng/mL, a threshold that can detect alcohol use for two to five days depending on the level of alcohol consumed and individual factors (Lowe et al., 2015; McDonell et al., 2015; 2017). Urine samples were collected twice a week for 12 weeks to compare the effect of contingency management intervention on urine EtG levels between the treatment and control groups.

Results

Of the 326 participants, 40.5% (n = 132) met criteria for alcohol-related child removal (i.e., participant had a child removed from their care due, in part, to alcohol use; Table 1). These individuals were predominantly female (65.9%). Seventy-four percent of the sample had 12 or more years of education. In the total sample, 74.5% were separated from their parents as children (i.e., parental separation in childhood) at least once (M = 1.3, SD = 1.0). In the subsample of participants who had a child removed from their care, approximately 80.0% experienced their own parental separation in childhood. In the subsample of parents who had not had a child removed from their care, 71.5% had personally experienced a separation from their parents. Seventeen percent of the overall sample experienced neither a child being removed from their care nor their own parental separation in childhood.

In the unadjusted associations, participants who were adopted had higher odds of being separated from their child (OR = 2.3; 95% CI = 1.1,5.1; p = .03). Those who reported higher levels of reflection on historical loss also had greater odds of an alcohol-related child removal (OR = 1.1; 95% CI = 1.0.1.1; p = .02). All other independent variables did not have statistically significant, unadjusted associations.

In our adjusted multivariable model, those participants who reported attending a boarding school as children were 93% more likely (OR = 1.9; 95% CI = 1.1.3.4; p = .03) to have experienced removal of a child from their care as an adult, compared with those who did not attend a boarding school. Other associations in two control variables (i.e., parent sex and education level) were significant as described in Table 2.

OPEN IN VIEWER

Table 2.

Associations Between Predictors and Alcohol-Related Child Removal. (n = 319) ^a .

	Adjusted OR [95% CI]
Parental separation in childhood, by event type
Lived with others because of problems in the family	1.32[0.74.2.36]
Attended boarding school	1.93[1.10,3.38]*
Were in foster care	0.85[0.46.1.55]
Were adopted	2.24[0.94.5.33]
Interpersonal trauma	0.68[0.38.1.24]
Historical loss and trauma measures
Historical loss	1.05[0.98.1.13]
Historical trauma	1.00[0.92.1.08]
Female sex	3.32[2.02,5.47]***
Age	0.98[0.95.1.00]
Education (≥ 12 years)	0.52[0.30,0.96]*

After fitting a logistic mixed effects model, participants who had a child removed from their care did not have a statistically significant different response to treatment (odds of submitting an alcohol negative urine test), relative to those who did not have their child separated from their care (OR = 1.2; 95% CI = 0.5.2.9; p = .68; Supplemental Table 1).

Discussion

In this sample of 326 AI/AN adults enrolled in a clinical intervention study for alcohol use, nearly 41% of participants reported ever having a child removed from their care due, in part, to alcohol use. In addition, 75% of participants in the study reported being separated from their parents as children. The adjusted multivariable model showed alcohol-related child removal was more likely if participants had attended a boarding school. These findings substantiate the intergenerational impact of boarding schools on AI/AN families, building on prior research that demonstrated boarding school effects on future mental and physical health, familial relationships, parenting behaviors, and alcohol and substance use (Colmant, 2000; Evans-Campbell et al., 2012; Running Bear et al., 2018; Zephier Olson & Dombrowski, 2020). Congruent with other studies, participants who were male and had less than 12 years of education had a higher likelihood of child removed from their care in the primary model (Crawford & Bradley, 2016; Hunter & Flores, 2021). There was no relationship between having a child removed from the participant and the effectiveness of the contingency management intervention on alcohol abstinence.

This study adds to the relatively small body of literature examining the risk factors associated with AI/AN child removal and is unique for its focus on examining those variables in alcohol use disorder treatment (Carter, 2009a, 2009b, 2010, 2011; Grinnell Davis et al., 2022; Lujan et al., 1989). Addressing these risk factors requires efforts on multiple levels. Practitioners may consider assessing the parent’s history of out-of-home placement types (e.g., attending a boarding school) to understand current or future child welfare involvement. Moreover, the parent’s service providers and their child’s wellbeing team may also consider a Systems of Care approach that requires intensive collaboration between each other, the parent, and the parent’s support network to achieve a reunification permanency plan (Cook & Kilmer, 2010; Simpson et al., 2001; Stroul & Friedman, 1986; U.S. Department of Health and Human Services, 2010). Evidence suggests that a Systems of Care approach has had positive outcomes among minoritized families, including geographically distinct AI/AN communities (Cross et al., 2000; Dettlaff & Rycraft, 2022; Nebelkopf & King, 2003). Additionally, the inclusion of family supports and resources, such as mental health services, parenting skills, and other evidence-based interventions, should be integrated into alcohol use treatment programs in AI/AN communities.

While beyond the scope of this study, addressing social determinants of health (e.g., poverty, discrimination) that underlie disproportionately high levels of alcohol use disorder and child removal are also necessary through social advocacy. The American Public Health Association recently declared structural racism as a public health concern (American Public Health Association, 2020). This declaration has implications for the child welfare system which maintains structures that disproportionality effect minority families, particularly AI/AN people (Crofoot & Harris, 2012; Dettlaff et al., 2011; Donald et al., 2003; Fox, 2004).

There are limitations to our research that should be noted. Although the research included a diverse sample of AI/AN adults from urban and rural settings, we may not be able to generalize our findings to other Tribal communities. In addition to common limitations of cross-sectional research, the outcome was based upon a single lifetime binary question of whether the participant ever had a child removed from their care, restricting the amount of information available for a holistic understanding apart from alcohol use. The participant’s self-reported answers to the outcome question may be another limitation to the outcome variable. The participant’s child removal experience may reflect the parent’s perception of the incident and not the reason documented by the child welfare administration. In addition, our study did not control for some potential unmeasured confounding variables such as mental health or non-problematic substance use. Strengths of our study include three geographically and culturally diverse study sites, the use of measures that were validated among AI/AN communities, our assessment of cultural and historical domains that may impact AI/AN alcohol-related child removal, and our large sample (N = 326) of AI/AN parents seeking alcohol treatment. Future research should focus on studying the intergenerational out-of-home placements among AI/AN adults with alcohol use disorder and seek to develop novel methods for keeping AI/AN families together.

Supplemental Material