Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Dec 1.
Published in final edited form as: Child Abuse Negl. 2023 Oct 4;146:106473. doi: 10.1016/j.chiabu.2023.106473

Patterns of substance use among adolescents in and out of foster care: An analysis of linked health and child welfare administrative data

Sarah J Beal a,b, Mary V Greiner b,c, Robert T Ammerman a,b, Constance A Mara a,b, Katie Nause a, John Schulenberg e, Jennie G Noll d
PMCID: PMC10841651  NIHMSID: NIHMS1937161  PMID: 37801757

Abstract

Background:

Young adults with a history of foster care have higher risk for substance use disorders. Social systems can deliver substance use prevention to youth; however, the timing of intervention delivery and how needs differ for youth in foster care are unclear.

Objective:

To compare initiation and rates of substance use among adolescents in foster care to demographically similar adolescents never in foster care as identified by the healthcare system, and identify factors associated with increased substance use.

Participants and Setting:

Youth in foster care (n=2787, ages 10–20, inclusive) and demographically matched youth never in foster care (n=2787) were identified using linked child welfare and electronic health records from a single pediatric children’s hospital and county over a five-year period (2012–2017).

Methods:

All healthcare encounters were reviewed and coded for substance use by type (alcohol, tobacco, cannabis, other). Age of first reported or documented substance use was also captured. Demographic and child welfare information was extracted from administrative records. Survival and logistic regression models were estimated.

Results:

In adjusted models, youth in foster care initiated substance use at earlier ages (HR=2.50, p<.01) and had higher odds of engaging in use (AOR=1.54; p<.01) than youth never in care. By age 12, substance use initiation was more likely while youth were in foster care than when they were not in foster care (HR=1.47, p<.01). Placement stability and family care settings reduced odds of lifetime substance use.

Conclusions:

Foster care placement is associated with substance use. Screening may be important for prevention.

Keywords: adolescent, substance use, foster care, child welfare, prevention, health care

Adolescence is an ideal developmental period for detecting and deterring substance use and substance use disorders (SUD; Gil, Wagner, & Tubman, 2004; Walters & Urban, 2014). Substance use initiation often occurs during adolescence (Braciszewski & Stout, 2012; Gabrielli et al., 2016; Schulenberg et al., 2014; Schulenberg, Maslowsky, & Jager, 2018) and is common. Seven percent of 8th graders and 31% of 12th graders report using cannabis in the past year, and rates of alcohol use are even higher (17.2% and 46.5%, respectively; Johnston et al., 2022). Earlier initiation of substance use is associated with a higher frequency of SUD later in life (Han et al., 2019; Richmond-Rakerd et al., 2016; Volkow et al., 2021; Walters & Urban, 2014). For those reasons, education (U.S. Department of Education, 2023), social services (U.S. Department of Health and Human Services, 2023), and pediatric healthcare guidelines recommend substance use screening and prevention programming by age 12 (Bukstein, 2005; Levy et al., 2016; United Nations Office on Drugs and Crime, 2005) to identify youth at risk for health concerns associated with substance use, including SUD.

Social systems in place to provide substance use and SUD prevention to adolescents include the family, school, social services (e.g., child welfare), juvenile justice, and healthcare. Family system factors including family cohesion, parental supervision, parent-child communication broadly and about substance use specifically, and parent or caregiver use of consistent and contingent rewards to manage child behavior are all associated with reduced substance use among adolescents (Velleman et al., 2005). While these contextual family factors are important, few evidence-based programs that target only the family system have been found to be effective in substance use and SUD prevention, most notably the Strengthening Families Program (Kumpfer et al., 1996). Many more formalized programs have focused on school-based prevention (for a recent review, see Liu et al., 2023). A third service sector focused on substance use prevention is the healthcare system (US Preventive Services Task Force, 2020), where standardized screening, brief intervention, and referral to treatment has been identified as a promising strategy for adolescents. However, routine screening for substance use in pediatric healthcare settings has been universally low (Braciszewski et al., 2016). Young people who are disenfranchised or minoritized (e.g., those with child welfare involvement) are less likely to be engaged in the social systems where prevention is emphasized (e.g., families, schools), and are more likely to be interacting with systems where responding to SUD is the focus (e.g., child welfare, juvenile court; Smith et al., 2010; Terry et al., 2000; Tripodi & Bender, 2011) with some exceptions (e.g., Delaney et al., 2020). As a result, there may be gaps in delivering services that prevent substance use for young people who would benefit the most from those programs.

Leveraging adolescence to prevent SUD may be particularly important for adolescents experiencing maltreatment. Adults with a history of childhood maltreatment are at increased risk for SUD (Appleyard, Berlin, Rosanbalm, & Dodge, 2011; Cicchetti & Handley, 2019). This includes those exposed to sexual abuse (Kendler et al., 2000) and foster care (i.e., legal custody granted to a county or state child protective services entity with children removed from their family of origin and placed in out-of-home care; Braciszewski & Stout, 2012). As many as 21% of adults previously in foster care are diagnosed with SUD (Braciszewski & Stout, 2012) compared to 5% in the general population (White, O’Brien, White, Pecora, & Phillips, 2008). Recent statewide surveys of adolescents grades 6–12 who were previously or currently in foster care suggest that past 30-day substance use ranges between 4.2% (cigarettes) and 15.6 % (alcohol; McCurdy et al., 2023). Among adolescents in foster care who are 17, 49% reported having tried substances in their lifetime and 35% met criteria for SUD (Vaughn et al., 2007). This is substantially higher risk for SUD than reported for the general population.

Multiple factors likely contribute to increased risk for SUD among young people with foster care involvement. First, family history of substance use and SUD is common among families with child welfare involvement (Ghertner, Waters, Radel, & Crouse, 2018; Mowbray, Victor, Ryan, Moore, & Perron, 2017), and substance use and SUD among parents is associated with increased adolescent substance use (Rusby, Light, Crowley, & Westling, 2018) and SUD (Ali, Dean Jr, & Hedden, 2016; Yule, Wilens, Martelon, Rosenthal, & Biederman, 2018). Second, placement instability, which creates inconsistency in caregiver relationships, is associated with increased substance use (Stott, 2012). Finally, adolescents in foster care are often exposed to maltreatment and other adversity (Gabrielli, Jackson, & Brown, 2016), which contributes to sequelae of social, relational, and health challenges (Yoon et al., 2017), including strained parental relationships and higher rates of chronic physical and mental health symptoms and diagnoses (Kim & Leve, 2011). Several of these mechanisms, including family relationships (Yoon et al., 2017) and mental health concerns (Cicchetti & Handley, 2019; Gabrielli et al., 2016; Miettunen et al., 2014; Schulenberg, Patrick, Maslowsky, & Maggs, 2014; Sitnick, Shaw, & Hyde, 2014) are associated with increased risk for substance use. These issues are further exacerbated by the increased use of psychotropic medications for children in foster care compared to the general population (Allen et al., 2012; Bowden et al., 2022; Davis et al., 2021; Simmel et al., 2021), including in the sample used for this study (Tan et al., 2023). While there is no evidence of a causal association between psychotropic medication use and risk for substance use disorder in adolescence, some psychotropic medications are diverted for illicit use (Kontu et al., 2022). Thus, there is good reason to consider opportunities for substance use prevention among adolescents in foster care.

While family and school systems are disrupted for children in foster care, which may lead to less opportunity for prevention services, increased healthcare access may provide an effective avenue for prevention. Federal legislation promotes healthcare use for young people in foster care (Center for Health Care Strategies, 2010), which further justifies considering healthcare encounters as opportunities for delivering substance use and SUD preventive interventions. However, it remains unclear whether a specific focus on substance use and SUD prevention for young people in foster care is needed or, alternatively, whether patterns of substance use, and in turn opportunities for prevention, are similar for all young people, which would not require any unique tailoring of prevention efforts to the foster care population. Understanding whether differences in substance use patterns occur is important, because there is precedence to direct and guide screening for other health concerns specifically for foster youth (American Academy of Pediatrics Task Force on Health Care for Children in Foster Care, 2005), and similar strategies could be applied to substance use. Further, there may be protective factors associated with foster care involvement that could be leveraged for prevention if some young people in foster care are at greater risk than others.

This study leverages child welfare administrative records linked to electronic health record (EHR) data for youth who experienced foster care between ages 10 and 20, demographically matched to a comparison sample of youth never in foster care to examine the onset of substance use and correlates associated with use as detected by the pediatric healthcare system. EHRs capture information about lifetime substance use, timing of initiation, and correlates (e.g., mental health concerns, family history of SUD) known to increase risk for early substance use initiation and SUD in adulthood. With these data, our multidisciplinary team sought to answer two questions. First, do pediatric healthcare systems detect differences in substance use for youth in foster care compared to the general population? Specifically, this study tested the hypothesis that healthcare systems will detect higher risk for substance use for youth in foster care than for youth not in foster care. Second, are there risk and protective factors related to substance use for young people in and out of foster care captured by pediatric healthcare systems? The effects of mental health diagnoses and family history of substance use were hypothesized to increase substance use across both groups of adolescents. Among youth in foster care, placement instability, placement into congregate care, and longer lengths of time in foster care were all expected to be associated with increased substance use. In this study, substance use was assessed two ways: age of initiation and lifetime use of substances (any use, alcohol, cannabis, tobacco, and multiple substances). Lifetime substance use and age of initiation were the primary focus because earlier initiation of substance use is associated with increased SUD risk in adulthood (Volkow et al., 2021) and substance use patterns can vary by type of substance used (Johnston et al., 2022) , so lifetime prevalence estimates are more stable in adolescence.

Methods

Study design and data sources

This retrospective study used EHR data extracted and linked to child welfare administrative records from the local child welfare agency, as described in Beal et al. (2022). Records were linked with high accuracy (Dexheimer et al., 2019) to ensure all adolescents in foster care between 2012 and 2017 who were between the ages of 10 and 20, inclusive, were represented. Data scientists then extracted records for the comparison sample, using inclusion criteria outlined below. From the pool of potential comparison records, data scientists developed algorithms to select a matched sample, where each selected comparison participant matched a foster care participant in date of birth (within 6 months), gender, race, and ethnicity. Researchers were provided with a limited dataset that contained both foster care and comparison participants’ data, using an honest broker protocol. The institutional review board at the academic medical center and the prosecutor’s office approved this study with a waiver of consent.

Participants and setting

This study took place in a Midwestern urban metro city and surrounding suburban neighborhoods where an established partnership exists between the large freestanding children’s hospital and the county-administered child welfare agency. Eligible participants (n = 2812) were all youth aged 10–20 years (M = 15.30 years, SD = 2.71) in foster care (i.e., in the protective custody of the county child welfare agency and placed into kinship care, non-relative foster care, congregate care, or independent living placement settings) in the participating county between 2012 and 2017. To be included, child welfare records had to have a corresponding EHR at the pediatric medical center where the study took place. Youth in foster care with no EHR (n = 25, 0.90%) were excluded, resulting in a study sample of N = 2787 youth in foster care. Participants were primarily Black/African American (n = 1522; 54.6%), White/European American (n = 1014; 36.4%) or multi-racial (n = 108; 3.9%). Only 2% of participants were identified as Hispanic. Race and ethnicity information was unknown for 4.6% (n = 129) of the sample. A remaining 0.5% were identified as a race other than white, black, or multi-racial (e.g., American Indian and Alaska Native [0.1%], Asian [0.3%], Native Hawaiian and Other Pacific Islander [0.1%]). Approximately half of participants were identified as female (n = 1375; 49%). The EHR did not have non-binary gender identities captured at the time these data were collected. Youth in foster care were matched one-to-one on age, gender, race, and ethnicity to youth with no history of foster care placement (n = 2787) who had at least one primary care visit between 2012 and 2017 at the same pediatric medical center. All participants (youth in foster care, youth never in foster care) were enrolled in Medicaid. The academic medical center where the study occurred is the only provider for specialty pediatric care that accepts Medicaid in the region and operates primary care clinics and provides behavioral health services for Medicaid-insured children and adolescents across 26 sites in the region where the academic medical center is located. More than 90% of Medicaid billing for the county of study is for services delivered at the academic medical center. The academic medical center also includes the only pediatric inpatient psychiatric treatment facility in the region, and contracts with the county of study to deliver healthcare services to all youth when they enter foster care or experience a placement change and reside within a 90-mile radius of the medical center.

Measures

Substance Use.

Multiple indicators of substance use from structured (i.e., encounter diagnoses, lab results, structured clinical data collection) and unstructured EHR data fields (i.e., clinical notes) were available to identify two main outcomes in this paper: 1) age of substance use initiation, and 2) lifetime substance use, which was classified into four categories (i.e., alcohol, tobacco, cannabis, or any use [alcohol, tobacco, cannabis, and other illicit substances, including opioids] over the study period). Diagnostic codes indicating current use or history of substance use were used to indicate lifetime use (see Supplemental Table 1). Age of onset of substance use was coded as the youth-reported age of onset in structured or clinical note data at the encounter in which substance use was first observed. The variable indicating lifetime use was coded 0 (no use) or 1 (use); use of multiple substances was coded 0 (use of 0 substances), 1 (use of 1 substance), or 2 (use of >1 substance over the observation period). The procedures for synthesizing substance use data across multiple structured and unstructured fields in the EHR have been previously described (Beal et al., 2022; Ni et al., 2021), where multiple coders classified clinical note data (κ > .90) and evidence of substance use derived from either structured data (e.g., diagnostic codes) or unstructured data (e.g., clinical notes) for a given encounter was used to indicate the presence of substance use at that encounter. Coders reliably classified clinical note data as indicating use and the age youth reported initiating substance use for the first time. Diagnostic codes indicating current use or history of substance use supplemented clinical note date to indicate lifetime use.

Foster Care Status.

Child welfare administrative records included dates of entry into and exit from foster care over the study period. Those dates were used to identify lifetime foster care history (i.e., ever in foster care during the study period, coded as 0 [never in foster care] or 1 [ever in foster care]) as well as whether youth were in foster care at the time of each healthcare encounter where a substance use screening occurred (i.e., foster care status concurrent to substance use screening, coded as 0 [not in foster care] or 1 [in foster care]). The effect of lifetime foster care history should be interpreted as the effect for young people who, at any point between the ages of 10 and 18, experienced foster care. In contrast, the effect of foster care status concurrent to screening should be interpreted as the effect of being in foster care at the time a young person is asked about substance use. These constructs help to differentiate substance use risk for the population that has experienced foster care from the risk of substance use while young people are in foster care.

Family History of Substance Use.

Two data sources were available to provide information about family history of substance use. In the EHR, family history was indicated by encounter diagnoses (e.g., fetal alcohol spectrum disorder) or clinical notes and coded as 0 (no family history of substance use) or 1 (family history positive for substance use) in a manner consistent with the coding of substance use data described above (see also Ni et al., 2021). In the child welfare administrative record, family history of substance use (0 = no family history of substance use, 1 = family history positive for substance use) was determined using reasons for removal from family of origin. Child welfare data was the only indicator of family history of substance use for 11% of foster youth.

Other Covariates from the EHR.

The number of mental health and chronic medical conditions were determined using the patient medical history in the EHR. Mental health conditions were coded using DSM-5 diagnostic criteria, and chronic medical conditions were coded by specialty (e.g., neurology, cardiology, endocrinology, etc.). ICD-9/10 codes were used to determine complex chronic condition classification in this sample, using previously published algorithms (Feudtner, Feinstein, Zhong, Hall, & Dai, 2014). To account for the co-occurrence of other risk behaviors which may be predictive of substance use (e.g., Baskin-Sommers & Sommers, 2006; Connell et al., 2009; Meader et al., 2016), sexual risk behavior was included as a covariate in lifetime substance use models, coded as 0 (none) or 1 (any), and was indicated by sexual debut before the age of 16, diagnosis of sexually transmitted infection, diagnosis of pregnancy, multiple sexual partners, or unprotected sex. See Beal et al. (2022) for complete details on data coding for these variables.

Other Child Welfare Covariates.

For models predicting substance use among those ever in foster care, child welfare administrative records were used to capture age of first entry into foster care. Length of time in care (i.e., the number of days in child welfare custody) was a time-invariant count variable of the lifetime number of days spent in foster care prior to the onset of substance use, or until the end of the study period if substance use was not initiated. Number of changes in placement (e.g., moving from a foster home to a kinship home) was a time-varying count variable determined by number of placement changes prior to the study period and number of placement changes observed over the study period. Type of placement was coded as family setting (i.e., licensed foster caregiver, kinship caregiver; 0) or non-family setting (i.e., group home, independent living, residential care; 1), reflecting differences in substance use for youth in family vs. non-family settings (Benedict et al., 1996; Keller et al., 2010; Sakai et al., 2011; Vaughn et al., 2007). Placement type was identified as the type of placement a young person was in at the time of substance use screening. Finally, reasons for placement as documented in structured data entered by caseworkers were used to identify exposure to physical abuse (i.e., physical abuse of the child, exposure to domestic violence; 0 = no documented exposure, 1 = documented exposure), exposure to sexual abuse (0 = no documented exposure, 1 = documented exposure), exposure to neglect (e.g., neglect, dependency, inadequate housing, relinquishment, incarceration of parent; 0 = no documented exposure, 1 = documented exposure), and history of child behavior problems (e.g., delinquency, unruly/status offender, parent inability to manage child behaviors, drug abuse of child; 0 = no documented history, 1 = documented history).

Demographic Characteristics.

Demographic data extracted from the EHR included dates of birth and encounter dates to calculate age in years at each encounter, sex (0 = male [50.7%], 1 = female [49.3%]), and race and ethnicity (White [36.4%], Black or African American [54.6%], American Indian and Alaska Native [0.1%], Asian [0.3%], Hispanic/Latino [<0.1%], Middle Eastern [0%], Multi-racial [3.9%], Native Hawaiian and Other Pacific Islander [0.1%], Hispanic [2.0%], Other [3.8%]) was recoded as non-Hispanic white (0, 34.5%) or people of color (POC; 1, 65.5%) to better understand the impact of marginalized race and ethnicity on screening and results. To ensure consistency in demographic characteristics for youth in and out of foster care, demographic data from the EHR was used exclusively, and demographic data from the child welfare record was not used in analyses.

Data analysis

Outcome: Age of Substance Use Initiation

To answer our first question, whether pediatric healthcare systems detect differences in substance use initiation for youth known to be at higher risk compared to the general population, descriptive data about occurrences of substance use screening and detection of positive screening results were examined using unconditional survival curves. Then cox proportional hazard regression models with robust standard error estimation with the stcox function in Stata version 18 (StataCorp, 2023) were conducted. Age was the time scale, left-truncated at age 10 to specify when participants were eligible to be “at risk” for substance use and were censored (i.e., no longer at risk) at their event onset age (or the end of the study timeframe, i.e., 20.99 years old). Data were coded to reflect whether each participant ever initiated substance use during the study timeframe and the age at which substance use was initiated, collected from the EHR. To address our second question, whether there are risk and protective factors related to substance use for young people in and out of foster care captured by pediatric healthcare systems, individual characteristics from the child welfare record (i.e., lifetime foster care status, foster care status at the time of screening, family history of substance use) and EHR (i.e., gender, POC status, complex chronic condition classification count, number of chronic conditions, number of mental health diagnoses, family history of substance use) were included as predictors (e.g., foster care status) or covariates (e.g., gender) in the cox proportional hazard model. Then, a separate cox proportional hazard model was estimated among those with a foster care history only (N = 2787) to examine child welfare characteristics associated with use (which were missing for young people never in foster care). Those models included predictors and covariates taken from the child welfare record (i.e., family history of substance use, number of previous placements, length of time in foster care) and EHR (i.e., gender, POC, complex chronic condition classification count, number of chronic conditions, number of mental health diagnoses, family history of substance use). Predictors were only included if they were observed before or concurrent with the onset of substance use, and this was determined using the dates of clinical observations. Placement type varied over time and was not applicable when screening was not concurrent with foster care involvement; for that reason, it was not included in the estimate of proportional hazard models.

For both models, where the proportional hazard assumption could not be met for a particular predictor, time-varying effects of that predictor were included in the model. Time-varying effects should be interpreted as the change in the hazard rate associated with a specific variable (e.g., foster care placement) with increasing age. For example, the effect of having ever been in foster care could increase the risk for substance use for the entire population exposed to foster care in the same manner, proportional across all ages, in which case no time-varying effect of lifetime foster care involvement would be needed. Alternatively, if the hazard rate cannot be assumed to be proportional across all ages (e.g., there is a stronger effect of ever having been in foster care when a young person is 17 than when a young person is 11), the time-varying effect of a particular variable would be significant. Multicollinearity was evaluated for both models as well, with a variance inflation factor (VIF) < 5 indicating acceptably low correlations among predictor variables included in each model.

Outcome: Lifetime Substance Use

To test both research questions with our lifetime substance use outcome, factors in administrative data associated with substance use (any and by type) for youth ever in foster care and youth never in foster care (N = 5574) were included as predictors of lifetime substance use in longitudinal multilevel logistic models estimated using lmer and glmer packages in R (R Core Team, 2019) via the integrated development environment, RStudio (Allaire, 2012). Those models included age (i.e., our time scale, centered at age 13) and nonlinear effects (slope and quadratic terms) to account for known increased probability of lifetime substance use with increasing age, which begins to plateau in late adolescence. Multilevel models were used for these analyses to allow for estimates of time-varying (e.g., age, length of time in foster care) and time invariant (e.g., ever being in foster care, gender, race, ethnicity) effects as predictors of lifetime substance use. All data were generated from a single healthcare system; therefore, no additional clustering needed to be accounted for.

Alpha was set to 0.05 for both sets of models (time-to-event and multilevel logistic). To address missing data in models estimating survival and continuous outcomes, maximum likelihood estimation with robust standard errors was used; weighted least squares estimation with robust standard errors was used for categorical outcomes. These methods allow for data estimation when predictor variables are present, even when outcome data are missing (Enders, 2011).

Results

Bivariate statistics for youth ever and never in foster care

Descriptive statistics for the sample of youth ever in foster care and youth never in foster care are provided in Table 1. Of all study participants, 23% (n = 1,279) had no documented screening for substance use (Supplemental Table 2). Youth ever in foster care had a significantly higher rate of screening compared to youth never in foster care (χ2 ([1, 4843] = 279.88, p<.01). Of screened youth, 36% (n = 1,283) had ≥1 positive substance use screening result, with increasing percentages of positive screening results with age and foster care history. Fewer young people were screened for alcohol than for tobacco and cannabis. A significantly higher rate of documentation of any substance use (χ2 ([1, 5697] = 81.06, p < .001), alcohol use (χ2 ([1, 4780] = 74.26, p < .001), tobacco use (χ2 (1, 5570] = 33.16, p < .001), and cannabis use (χ2 ([1, 4918] = 222.77, p < .001) was found in unstructured compared to structured data (Figure 1). For youth who were never screened, substance use variables were assigned a missing value.

Table 1.

Demographic characteristics of study participants by group.

Ever in Foster Care Never in Foster Care
Total sample N 2787 2787
Mean age at first encounter (SD) 12.28 (3.60) 12.18 (3.81)
Female N (%) 1375 (49.3%) 1375 (49.3%)
Persons of Color N (%) 1825 (65.5%) 1738 (63.0%)
Mean number of mental health diagnoses (SD) 1.50 (1.95) 0.67 (1.25)
Mean number of chronic conditions (SD) 0..68 (1.08) 0.99 (1.28)
Mean number of complex chronic conditions (SD) 0.20 (0.60) 0.27 (0.71)
Family history of substance use N (%) 1372 (49.2%) 406 (14.6%)
Exposure to physical abuse N (%) 723 (26.0%) --
Exposure to sexual abuse N (%) 178 (6.4%) --
Exposure to neglect N (%) 2088 (74.9%) --
History of child behavior problems N (%) 510 (18.3%) --
Open in a new tab

Figure 1.

Figure 1.

Open in a new tab

Percentage of youth ever experiencing a positive substance use screening result documented in the electronic health record, distinguishing source as structured or unstructured data from any encounter occurring between 2012 and 2017. Significant differences using p-values less than .001 from χ2 results are indicated as ***.

Age of Substance Use Initiation

Question 1: Differences in substance use for youth in and out of foster care.

Unconditional survival curves indicated that youth ever in foster care engaged in substance use at earlier ages and higher percentages overall than youth never in foster care (Table 2). A cox proportional hazard model estimated for all youth (VIF = 1.8) indicated that ever being in foster care increased the risk of substance use 2.5 times at age 10 (holding all other variables constant; Hazard Ratio (HR) = 2.50, p < .001, 95% CI = 1.74, 3.59). This effect varied across time and declined approximately 18% each year, such that, by age 14, there is no significant difference in risk of substance use between youth ever in foster care and those never in foster care. Placement in foster care concurrent to the report of substance use was not significantly associated with risk of substance use at age 10 or 11 (HR = 0.81, p = .22, 95% CI = 0.60, 1.13); however, a significant time-varying effect of foster care status concurrent to the report of substance use indicated that by age 12, those placed in foster care concurrent to the report of substance use had a 47% increased risk of engaging in substance use compared to youth never in foster care, and that risk continued to increase with age.

Table 2.

Survival model predicting substance use initiation by age, across ages 10–20 years, inclusive.

Variables Hazard Ratio SE 95% CI
All youth (N = 5574)
Main effect
 History of foster care 2.50** 0.46 1.74 3.59
 Foster care status concurrent to screening 0.81 0.14 0.58 1.13
 Family history of substance use 2.28** 0.13 2.03 2.56
 Number of mental health diagnoses 1.42** 0.05 1.33 1.51
 Number of chronic conditions 1.04 0.03 0.99 1.09
 Number of complex chronic conditions 0.92 0.04 0.84 1.01
 Gender 0.64** 0.09 0.48 0.84
 POC status 0.67** 0.10 0.50 0.90
Time-varying effect
 History of foster care 0.82** 0.02 0.78 0.87
 Foster care status concurrent to screening 1.35** 0.04 1.27 1.43
 Number of mental health diagnoses 0.97** 0.01 0.96 0.98
 Gender 1.08** 0.02 1.03 1.12
 POC status 1.11** 0.03 1.06 1.16
Youth Ever in Foster Care (N = 2787)
Main effect
 Foster care status concurrent to screening 0.90 0.15 0.64 1.25
 Family history of substance use 3.62** 0.71 2.47 5.32
 Number of mental health diagnoses 1.33** 0.05 1.24 1.42
 Number of chronic conditions 1.05 0.04 0.98 1.12
 Number of complex chronic conditions 0.97 0.07 0.85 1.12
 Gender 0.82** 0.05 0.73 0.93
 POC status 1.13 0.08 0.99 1.29
 Length of time in care 1.00** 0.00 1.00 1.00
 Number of placements 1.01 0.01 0.99 1.04
Time-Varying Effect
 Foster care status concurrent to screening 1.42** 0.04 1.34 1.50
 Family history of substance use 0.87** 0.03 0.82 0.93
 Number of mental health diagnoses 0.98* 0.01 0.97 1.00
 Length of time in care 1.00** 0.00 1.00 1.00
Open in a new tab
*

p < .05,

**

p < .01.

Note: SE = Standard Error; CI = Confidence Interval. Main Effects indicate effects of predictors on substance use, with hazard ratios greater than 1 indicating increased risk and hazard ratios less than 1 indicating decreased risk. Time-varying effects indicate effects of predictors that changed over time (age), such that effects were stronger with older ages when values are greater than 1 and effects are decreased with age when values are less than 1.

Question 2: Risk and protective factors.

Across all adolescents (with and without a history of foster care), females had a 36% lower risk of engaging in substance use than males at age 10 (HR = 0.64, p = .002, 95% CI = 0.48, 0.84), with this effect varying over time such that females were at significantly higher risk of substance use than males by age 19 (Table 2). Young people of color had a 33% lower risk of engaging in substance use at age 10 compared to white, non-Hispanic youth (HR = 0.67, p = .009, 95% CI = 0.50, 0.90), which also varied over time, where young people of color were at significantly higher risk for substance use by age 15 than their white, non-Hispanic peers. At age 10, every additional mental health diagnosis increased the risk of engaging in substance use by 42% (HR = 1.42, p < .001, 95% CI = 1.33, 1.51), although this risk significantly decreased 2.7% per year. Those with a family history of substance use were more than 2 times more likely to engage in substance use (HR = 2.28, p < .001, 95% CI = 2.03, 2.56) than those without, and this effect did not change with age. Chronic conditions were not significantly associated with risk of substance use.

Among youth ever in foster care, a cox proportional hazard model (VIF = 2.79) indicated that after accounting for currently being in foster care (HR = 0.90, p = 0.53, 95% CI = 0.64, 1.25), other child welfare characteristics were unassociated with risk of substance use initiation among young people who ever experience foster care (Table 2). Consistent with the model for all adolescents, the effect of current foster care status varied significantly across age, such that those in foster care at age 12 were 80% more likely to engage in substance use than young people with a lifetime history of substance use who were not in foster care at age 12, and this effect continued to increase with age.

Lifetime substance use

Question 1: Differences in substance use for youth in and out of foster care.

In multilevel logistic regression models (Table 3) examining lifetime odds of any substance use, alcohol use, tobacco use, and cannabis use, a nonlinear change with age was identified, such that odds of substance use increased with age (slope) and plateaued by 20 (quadratic). Lifetime substance use was 1.5 times higher for youth ever in foster care compared to youth never in foster care (95% CI = 1.39–1.71). Significantly higher odds were also observed for alcohol use (OR = 1.12; 95% CI = 1.01–1.24), tobacco use (OR = 2.29; 95% CI = 2.08–2.53), and cannabis use (OR = 1.72; 95% CI = 1.56–1.90). In contrast, placement in foster care concurrent to a substance use screen was only associated with a significantly increased odds for lifetime use of alcohol (OR = 1.12; 95% CI = 1.02–1.23).

Table 3.

Multilevel logistic regression models predicting lifetime substance use among adolescents ages 10–20 years of age, inclusive.

Variable Any Substance Use Alcohol Use Tobacco Use Cannabis Use
OR Lower CI Upper CI OR Lower CI Upper CI OR Lower CI Upper CI OR Lower CI Upper CI
All youth (N = 5574)
Ever in foster care 1.54** 1.39 1.71 1.12* 1.01 1.24 2.29** 2.08 2.53 1.72** 1.56 1.90
Foster care status concurrent to screening 1.02 0.92 1.13 1.12* 1.02 1.23 0.96 0.88 1.06 1.07 0.97 1.17
Family history of substance use 2.57** 2.34 2.82 2.00** 1.83 2.18 2.01** 1.85 2.19 2.05** 1.89 2.23
Number of mental health diagnoses 1.51** 1.47 1.55 1.36** 1.34 1.39 1.39** 1.36 1.42 1.22** 1.20 1.25
Number of chronic conditions 0.97* 0.93 0.99 0.88** 0.85 0.91 1.08** 1.05 1.11 0.93** 0.90 0.96
Number of complex chronic conditions 0.92** 0.87 0.97 0.97 0.91 1.02 0.92** 0.88 0.97 0.84** 0.80 0.90
Sexual risk behaviors 1.31** 1.20 1.44 1.20** 1.10 1.31 1.27** 1.16 1.38 1.33** 1.22 1.45
Gender 0.72** 0.65 0.78 1.15** 1.05 1.26 0.62** 0.57 0.68 0.66** 0.61 0.72
POC status 1.59** 1.44 1.76 1.02 0.93 1.12 1.03 0.94 1.13 1.47** 1.34 1.61
Slope (Age) 1.44** 1.41 1.47 1.47** 1.43 1.51 1.33** 1.31 1.36 1.53** 1.49 1.57
Quadratic(Age*Age) 0.97** 0.96 0.97 0.97** 0.96 0.97 0.97** 0.97 0.98 0.95** 0.95 0.96
R2 0.43 0.37 0.34 0.38
Youth Ever in Foster Care (N = 2787)
Family history of substance use 1.91** 1.65 2.20 1.93** 1.69 2.21 1.46** 1.28 1.66 1.67** 1.46 1.90
Number of mental health diagnoses 1.49** 1.43 1.56 1.39** 1.35 1.44 1.32** 1.28 1.36 1.21** 1.17 1.24
Number of chronic conditions 0.98 0.92 1.04 0.95 0.89 1.00 1.06* 1.01 1.12 0.85** 0.80 0.90
Number of complex chronic conditions 0.80** 0.71 0.89 0.92 0.82 1.01 0.90* 0.82 0.99 0.87** 0.79 0.97
Sexual risk behaviors 1.30** 1.09 1.56 1.02 0.87 1.19 1.16 1.00 1.35 1.11 0.95 1.29
Age at first time in foster care 0.99 0.97 1.01 1.00 0.99 1.02 0.97** 0.96 0.99 0.99 0.97 1.01
Lifetime number of placements 1.03* 1.00 1.05 1.01 0.99 1.03 1.02 1.00 1.04 1.07** 1.05 1.10
Placement in nonfamily settings 1.07* 1.01 1.13 0.99 0.94 1.03 1.12** 1.07 1.17 1.13** 1.07 1.18
Exposure to physical abuse 0.79** 0.67 0.94 0.79** 0.67 0.93 0.84* 0.72 0.98 0.99 0.85 1.16
Exposure to sexual abuse 0.66** 0.51 0.87 1.49** 1.15 1.92 0.65** 0.51 0.83 1.02 0.79 1.31
Exposure to neglect 1.04 0.88 1.23 1.04 0.89 1.21 0.88 0.76 1.02 1.83** 1.58 2.13
History of child 1.80** 1.53 2.13 1.14 1.00 1.31 1.68** 1.47 1.93 1.38** 1.20 1.58
behavior problems Gender 0.57** 0.49 0.66 0.90 0.78 1.03 0.64** 0.56 0.73 0.54** 0.47 0.62
POC status 0.96 0.81 1.13 0.73** 0.62 0.84 0.75** 0.65 0.86 1.01 0.87 1.17
Slope (Age) 1.44** 1.40 1.49 1.35** 1.31 1.40 1.37** 1.33 1.41 1.43** 1.39 1.48
R2 0.46 0.36 0.33 0.38
Open in a new tab
*

p < .05,

**

p < .01.

Note: OR = Odds Ratio, CI = 95% Confidence Interval. In models with all youth, odds of lifetime substance use increased with age (slope) and plateaued by age 20 (quadratic). In models with foster youth, odds of lifetime substance use increased and did not plateau.

With respect to lifetime use (Table 4), 64% of young people (n= 2281) reported no lifetime use of any substance, 16% (n= 570) reported lifetime use of only one substance, 11% (n= 392) reported lifetime use of two substances, and 9% (n= 321) reported lifetime use of three substances (i.e., alcohol, tobacco, and cannabis use). Use of multiple substances increased with age (slope) and plateaued in late adolescence (quadratic). A foster care history was associated with an increase in the number of substances used (B = 0.46, SE = 0.02, p < .01), while placement in foster care concurrent to substance use screening was associated with a decrease in the number of substances used (B = −0.07, SE = 0.01, p < .01).

Table 4.

Multilevel regression models predicting lifetime use of multiple substances.

Variable B SE
All Youth (N = 5574)
Intercept 0.22** 0.03
Slope (Age) 0.15** 0.01
Quadratic (Age*Age) −0.02** 0.00
History of foster care 0.46** 0.02
Foster care status concurrent to screening −0.07** 0.01
Family history of substance use 0.38** 0.02
Number of mental health diagnoses 0.17** 0.00
Number of chronic conditions 0.02* 0.01
Number of complex chronic conditions 0.09** 0.01
Sexual risk behaviors 0.07** 0.01
Gender 0.02 0.02
POC status 0.13** 0.02
R2 0.83
Youth Ever in Foster Care (N = 2787)
Intercept 0.68** 0.05
Slope (Age) −0.12** 0.01
Quadratic −0.01** 0.00
Family history of substance use 0.34** 0.03
Number of mental health diagnoses 0.13** 0.01
Number of chronic conditions 0.02 0.01
Number of complex chronic conditions −0.14** 0.02
Sexual risk behaviors 0.03 0.02
Age at first time in foster care 0.00 0.00
Lifetime number of placements 0.00 0.00
Exposure to physical abuse −0.20** 0.03
Exposure to sexual abuse 0.10 0.06
Exposure to neglect 0.08* 0.03
History of child behavior problems 0.26** 0.03
Placement in non-family settings 0.00 0.01
Gender 0.04 0.03
POC status −0.01 0.03
R2 0.83
Open in a new tab
*

p < .05

**

p < .01.

Note: SE = Standard Error. Use of multiple substances increased with age (slope) and plateaued in late adolescence (quadratic).

Question 2: Risk and protective factors.

In models estimated for adolescents with and without a history of foster care, a family history of substance use and mental health diagnoses increased odds of lifetime substance use (Table 3) and an increase in the number of substances used (Table 4). Additionally, among youth ever in foster care, the odds of lifetime substance use was significantly higher with increased number of placements (OR = 1.03; 95% CI = 1.00–1.05) and placement in non-family settings (OR = 1.07; 95% CI = 1.01–1.13; Table 3). A similar pattern was observed for lifetime cannabis use, with higher odds for increased placements (OR = 1.07; 95% CI = 1.05–1.10) and placement in non-family settings (OR = 1.13; 95% CI = 1.07–1.18). Earlier age of entry into foster care was associated with a significantly lower odds (0.97; 95% CI = 0.96–0.99) of tobacco use, while non-family placement was associated with a higher odds (1.12; 95% CI = 1.07–1.17) of tobacco use. Number of placements, placement in non-family settings, and age of entry into foster care were not associated with use of more than one substance (Table 4).

Discussion

Adolescents and young adults with a history of exposure to maltreatment, including those in foster care, are at elevated risk for early initiation of substance use and SUD (e.g., Capusan et al., 2021; Cicchetti & Handley, 2019; Tonmyr et al., 2010). Healthcare systems may be well-positioned to detect and intervene to reduce risk for SUD (Ghitza & Tai, 2014; Ozechowski et al., 2016) if there is adequate information collected and documented to inform evidence-based care delivery (Ghitza et al., 2013). The potential to leverage healthcare systems for substance use prevention and harm reduction is promising, given that children in foster care are required to receive healthcare services (Center for Health Care Strategies, 2010) and that other social systems often leveraged for substance use prevention are disrupted when children enter foster care. This study examined substance use initiation and lifetime use of alcohol, tobacco, and cannabis for youth ever in foster care and youth never in foster care, using linked EHR and child welfare administrative data. Risk and protective factors, including foster care status and other covariates (e.g., mental health, family history of substance use, placement history), were examined.

Substance use for adolescents in and out of foster care.

The current study’s findings highlighted the critical importance of distinguishing between ever having been in foster care and being in foster care concurrent to substance use initiation, with implications for prevention and intervention across multiple social service sectors. As expected, youth ever in foster care initiated substance use at earlier ages, had higher rates of substance use, and used more substances than youth never in foster care, consistent with prior research (Gabrielli et al., 2016; Traube, James, Zhang, & Landsverk, 2012). However, by age 14, we found no difference between youth ever and never in foster care after accounting for other covariates. This suggests that lifetime history of foster care may be important for early initiation of substance use (i.e., before the age of 14), but that there is not a universal risk of increased substance use incurred for all young people with a history of foster care. In contrast, being in foster care at the time of screening was initially not associated with substance use risk. As youth got older, current foster care status significantly increased risk, such that by age 12 foster care status concurrent to substance use screening was associated with increased risk for substance use initiation compared to young people never in foster care and youth with a history of foster care who were not in care at age 12. Additionally, and consistent with the prior literature (Schulenberg et al., 2017), lifetime odds of substance use increased with age and plateaued in late adolescence. However, the plateau in odds of lifetime substance use observed for all youth was not observed when the population of youth ever in foster care was examined separately, suggesting that risk for substance use among those ever in foster care continues beyond adolescence. This is consistent with studies of young people leaving foster care (e.g., Courtney et al., 2011). Foster care status concurrent to screening also contributed to higher odds of lifetime alcohol use but was protective for lifetime use of multiple substances. Together, these findings indicate that it is advantageous to consider a young person’s history of foster care along with current foster care status, particularly in identifying programs and other services that might interact with young people who are most likely to benefit from substance use screening and intervention. Foster care clinics may be an important target for universal substance use screening and intervention, for example, to ensure services are delivered to those currently in foster care to reduce SUD risk. Given that 5.9% of the US population experiences foster care at least once (Wildeman & Emanuel, 2014), it is critical that all professionals and partners who work with young people in foster care, including healthcare and social service systems, recognize the need to need to monitor and emphasize prevention of substance use with this population.

Risk and protective factors.

Several important and modifiable protective factors were identified that were associated with delayed initiation and reduced use of substances for young people in foster care. First, for adolescents between the ages of 10 and 12, youth currently and ever in foster care experienced similar risk for substance use initiation. This may suggest that for this age group, delivering interventions to reduce risk of substance use initiation is important for those currently in foster care and those receiving services to prevent entry/re-entry into foster care, to ensure that all young adolescents benefit from preventive interventions. Additionally, placement stability was associated with a reduced odds of lifetime substance use and use of more than one substance, as was placement in family-type settings (i.e., with foster or kinship caregivers). Shorter duration of time in foster care was also associated with a slight decline in lifetime substance use risk. Given the focus of new policies to address many of these issues, including promoting family-based placement settings, placement stability, and permanency (Family First Prevention Services Act, 2017), future research focused on the downstream impact of those policies on substance use for young people in foster care is warranted. Additional areas of opportunity for future research highlighted by this study include examining the effect of standardized screening practices, to assess the effect of screening occurrences on the detection of substance use differences for young people in and out of foster care, along with studying additional modifiable protective factors that contribute to lower risk of substance use initiation and/or delays in initiation while young people are in foster care. The findings from this study also point to opportunities to better understand how mental health diagnoses and treatment might impact risk for substance use while young people are in foster care, along with the role of child behavior problems in contributing to substance use risk. Finally, our findings identified an initial protective effect of minoritized race and ethnicity on reducing risk of substance use, consistent with other studies in adolescence (Chen & Jacobson, 2012; Griffin et al., 2019). However, this protective effect diminished with increasing age, such that by age 15, young people of color were just as likely to initiate substance use as young people who were non-Hispanic white. A better understanding of this protective effect and why it diminishes with age for young people in this study is needed.

A secondary finding of this study was the evidence indicating that healthcare systems are screening young people for substance use generally, and that youth in foster care are more likely to have screening occurrences and results documented. Further, screening results for all young people were most often documented in unstructured clinical notes, indicating a lack of standardization of screening practices. Early detection of substance use through standardized and routine screening is critically important to support the prevention of substance use disorder. These findings highlight opportunities for improvement in screening and documentation to ensure that all young people have the opportunity to receive prevention and early intervention to reduce SUD. Studies have repeatedly demonstrated the association between early initiation of substance use and lifelong risk for SUD (Grant & Dawson, 1997), as well as the utility of evidence-based screening, brief intervention, and referral to treatment to identify substance use (Mitchell et al., 2016; Ozechowski et al., 2016), reduce substance use and prevent problematic use (e.g., Hammock et al., 2020), and support access to treatment for SUD (Scott et al., 2018). Among adolescents, research also indicates that screening and intervention may be effective in preventing or delaying initiation of substance use (Ozechowski et al., 2016). Healthcare and other social service systems (e.g., community behavioral health services, schools, children’s services) implementing universal screening for substance use among adolescents should consider screening at younger ages (i.e., prior to age 12; Han et al., 2019; Richmond-Rakerd et al., 2016; Volkow et al., 2021; Walters & Urban, 2014) and also delivering prevention and intervention to reduce substance use (Ozechowski et al., 2016), in order to address the concerns about substance use highlighted in the current study.

This study’s findings should be interpreted within the context of several limitations. Retrospective EHR and child welfare administrative data were used, which are not collected for research purposes, creating limitations in our understanding of important factors contributing to substance use (e.g., detailed reasons for placement). There is mixed evidence that adolescents will disclose substance use in healthcare settings. Some studies report rates of disclosure in clinical settings that match anonymous self-report data from research studies (Levy et al., 2004; Williams & Nowatzki, 2005). Others have identified under-reporting of substance use among adolescents screened in clinical settings compared to self-report in research studies (Gryczynski et al., 2019). Therefore, data from the EHR reflects substance use detected and documented by healthcare providers and acknowledged by patients in pediatric settings, the time-period where prevention is critical for curtailing later SUD. Additionally, this study deployed deterministic matching strategies and reviewed both structured and unstructured data to ensure the highest quality and rigor possible; however, substance use may still have been missed (either by the research team or because information was never collected). Disparities in screening practices also introduce bias, making findings primarily generalizable to those who underwent screening. Findings may have differed substantially if only structured data fields were used. Efforts to standardize screening and documentation (Mitchell et al., 2016) and leverage natural language processing to access unstructured substance use data without relying on laborious chart review techniques (Ni et al., 2021) may aid in alleviating these challenges. Additionally, we found that adolescents ever in foster care were significantly more likely to receive screening compared to adolescents never in foster care. As a result, this study’s findings reflect differences among youth with and without a history of foster care who received substance use screening. Further, this study only included data from one EHR system and one child welfare agency, representing one geographic region. The vast majority (99.1%) of children in foster care had EHR documentation, and the academic medical center where the study occurred includes a specialized foster care clinic that provides consultation care to all youth in this study, increasing the likelihood that a child in foster care would be screened for substance use, and that complete data for this population is available, which may explain higher rates of screening among foster youth. To maximize the likelihood that complete data was available for the comparison sample, it was selected from those enrolled in Medicaid and receiving primary care with the same healthcare system, which is the safety net Medicaid provider for the region for both primary and specialty care. However, differences in screening rates remained. The increased screening may in part be a result of biased assumptions about substance use among young people in foster care (Pryce et al., 2017); it may also reflect that state-mandated healthcare visits when young people enter foster care recommend substance use screening as part of the evaluation, which resulted in screening occurrences on a more consistent basis in our healthcare system than has been reported in other primary care sites (American Academy of Pediatrics Task Force on Health Care for Children in Foster Care, 2005; Camenga, 2023; Levy et al., 2016). Regardless of cause, it is likely that the estimates for elevated substance use for young people in foster care are influenced by the increased rates of screening. Third, EHR and other clinically generated data cannot always distinguish timing of onset of substance use from the timing of detection. In this case, age of initiation, reported by youth, was available for the majority of youth screened. EHR data are also limited in that some forms of substance use (e.g., injection drug use, inhalants) are less well documented. Finally, individual and child welfare characteristics, including exposure to maltreatment, known child behavior problems, juvenile justice involvement, and family history of substance use was limited to what was available in the EHRs or child welfare records, and reflects what was observed or captured by the clinician or caseworker. It is likely that rates of maltreatment and family history of substance use are both under-reported, and more detail regarding child behavior problems noted in the child welfare record (e.g., juvenile justice involvement) would have been useful.

Conclusions.

This study makes three important contributions to our understanding of the relations between child welfare involvement and substance use during adolescence. First, findings indicate that substance use screening is inconsistently occurring and documented in the EHR, screening occurrences differ for youth ever and never in foster care, and substance use initiation is detected at younger ages (i.e., before age 12). Together, these findings suggest that young people would benefit from healthcare and other social service systems (e.g., schools, child welfare, juvenile justice, family-based services) implementing standardized and consistent screening before age 12 to ensure substance use and SUD prevention is universally available for all young people who have had or will have foster care involvement. Given that ages 12–13 are recommended for initiating confidential screening (Levy et al., 2016), substance use screening before age 12 is a departure from current recommendations. Second, among youth ever in foster care, findings suggest that substance use initiation more frequently occurs while adolescents ages 12 and older are in foster care, and this risk continues to increase with older ages. These findings strongly suggest that foster care is not protective against substance use by age 12; therefore, social service systems serving young people while they are in foster care need to deliver interventions to reduce SUD risk. Finally, these results indicate that supporting placement stability, placement in family-based settings, and shorter duration of foster care placement may also help to reduce substance use. Social service systems across domains (e.g., healthcare, education) could benefit from partnering with child welfare to support these protective factors and further reduce SUD risk. These findings represent the full population of youth in foster care in a single county with a matched comparison sample and confirmed child welfare status, which increases generalizability and rigor over previous studies (Keller, Blakeslee, Lemon, & Courtney, 2010; Shin, 2004; Smith, Chamberlain, & Eddy, 2010; Vaughn, Ollie, McMillen, Scott Jr, & Munson, 2007). Given that early initiation of substance use and use of multiple substances are both established risk factors for later SUD (Gil et al., 2004; King & Chassin, 2007), and that youth in foster care who use substances are at increased risk for placement instability and placement in non-family settings (e.g., group homes, independent living), effective programs to screen for and prevent or reduce substance use may have long-term positive impact to improve health and wellbeing.

Supplementary Material

1

Highlights:

  • Foster youth disclose past and current substance use to healthcare providers

  • Foster youth initiate substance use by age 12 more frequently than non-foster youth

  • Providers detect more substance use and use of multiple substances for foster youth

  • Healthcare providers could deliver substance use prevention to youth in foster care

Disclosure of Interest:

The project described was supported in part by the National Center for Advancing Translational Sciences of the National Institutes of Health, under Award Number 5UL1TR001425-02 and the National Institute of Drug Abuse, under Award Number 1 K01 DA041620-01A1. The content is solely the responsibility of the authors and does not necessarily represent the official views of NIDA or the National Institutes of Health. All authors declare that they have no conflicts of interest. Clinical trials registration was not required for this study. The data used in this paper are available with established data use and transfer agreements from the healthcare and child welfare systems that supplied the data. Please contact the corresponding author to establish those agreements. The authors thank the community and clinical partners involved in the gathering and analysis of this data, and specifically acknowledge Dr. Schulenberg† for his significant contribution to all aspects of this study, including multiple drafts of this paper, prior to his passing in February 2023.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Ahrens KR, Garrison MM, & Courtney ME (2014). Health outcomes in young adults from foster care and economically diverse backgrounds. Pediatrics, 134(6), 1067–1074. doi: 10.1542/peds.2014-1150 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ali MM, Dean D Jr, & Hedden SL (2016). The relationship between parental mental illness and/or substance use disorder on adolescent substance use disorder: Results from a nationally representative survey. Addictive behaviors, 59, 35–41. [DOI] [PubMed] [Google Scholar]
  3. Allaire J (2012). RStudio: integrated development environment for R. Boston, MA, 770, 394. [Google Scholar]
  4. Appleyard K, Berlin LJ, Rosanbalm KD, & Dodge KA (2011). Preventing early child maltreatment: Implications from a longitudinal study of maternal abuse history, substance use problems, and offspring victimization. Prevention Science, 12(2), 139–149. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Allen KD, Pires SA, Mahadevan R, Chazin S, & Baruchin A (2012). Improving Outcomes for Children in Child Welfare: A Medicaid Managed Care Toolkit. Center for Health Care Strategies. https://www.chcs.org/media/Child_Welfare_Quality_Improvement_Collaborative_Toolkit.pdf [Google Scholar]
  6. American Academy of Pediatrics Task Force on Health Care for Children in Foster Care. (2005). Fostering health: Health care for children and adolescents in foster care (2nd ed.). American Academy of Pediatrics. [Google Scholar]
  7. Baskin-Sommers A, & Sommers I (2006). The co-occurrence of substance use and high-risk behaviors. Journal of Adolescent Health, 38(5), 609–611. 10.1016/j.jadohealth.2005.07.010 [DOI] [PubMed] [Google Scholar]
  8. Beal SJ, Nause K, Ammerman RT, Hall ES, Mara CA, & Greiner MV (2022). Careful: An administrative child welfare and electronic health records linked dataset. Data in Brief, 44, 108507. 10.1016/j.dib.2022.108507 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Benedict MI, Zuravin S, & Stallings RY (1996). Adult functioning of children who lived in kin versus nonrelative family foster homes. Child Welfare, 75(5), 529–549. [PubMed] [Google Scholar]
  10. Bowden CF, Simmel C, Mendez A, Yu M, Neese-Todd S, & Crystal S (2022). The Complexity of Psychotropic Medication Prescription and Treating Trauma Among Youth in Foster Care: Perspectives from the Lived Experience. Administration and Policy in Mental Health and Mental Health Services Research, 49(5), 821–833. 10.1007/s10488-022-01203-4 [DOI] [PubMed] [Google Scholar]
  11. Bukstein OG (2005). Practice Parameter for the Assessment and Treatment of Children and Adolescents With Substance Use Disorders. Journal of the American Academy of Child & Adolescent Psychiatry, 44(6), 609–621. 10.1097/01.chi.0000159135.33706.37 [DOI] [PubMed] [Google Scholar]
  12. Camenga D (2023). Substance Use Screening Practices among Primary Care Pediatricians in the US: Results from the 2021 AAP Periodic Survey. https://www.aap.org/en/research/pas-abstracts/substance-use-screening-practices-among-primary-care-pediatricians-in-the-us-results-from-the-2021-aap-periodic-survey/
  13. Capusan AJ, Gustafsson PA, Kuja-Halkola R, Igelström K, Mayo LM, & Heilig M (2021). Reexamining the link between childhood maltreatment and substance use disorder: A prospective, genetically informative study. Molecular Psychiatry, 26(7), Article 7. 10.1038/s41380-021-01071-8 [DOI] [PubMed] [Google Scholar]
  14. Center for Health Care Strategies. (2010, November 2). Health Screening and Assessment for Children and Youth Entering Foster Care: State Requirements and Opportunities. Center for Health Care Strategies. https://www.chcs.org/resource/health-screening-and-assessment-for-children-and-youth-entering-foster-care-state-requirements-and-opportunities/ [Google Scholar]
  15. Chen P, & Jacobson KC (2012). Developmental Trajectories of Substance Use From Early Adolescence to Young Adulthood: Gender and Racial/Ethnic Differences. Journal of Adolescent Health, 50(2), 154–163. 10.1016/j.jadohealth.2011.05.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Cicchetti D, & Handley ED (2019). Child maltreatment and the development of substance use and disorder. Neurobiology of Stress, 10, 100144. 10.1016/j.ynstr.2018.100144 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Connell CM, Gilreath TD, & Hansen NB (2009). A multiprocess latent class analysis of the co-occurrence of substance use and sexual risk behavior among adolescents. J Stud Alcohol Drugs, 70(6), 943–951. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Courtney ME, Dworsky A, Hook J, Brown A, Cary C, Love K, Vorhies V, Lee J, Raap M, Cusick G, Keller T, Havlicek J, Perez A, Terao S, & Bost N (2011). Midwest Evaluation of the Adult Functioning of Former Foster Youth – Chapin Hall. https://www.chapinhall.org/research/midwest-evaluation-of-the-adult-functioning-of-former-foster-youth/ [Google Scholar]
  19. Davis DW, Lohr WD, Feygin Y, Creel L, Jawad K, Jones VF, Williams PG, Le J, Trace M, & Pasquenza N (2021). High-level psychotropic polypharmacy: A retrospective comparison of children in foster care to their peers on Medicaid. BMC Psychiatry, 21(1), 303. 10.1186/s12888-021-03309-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Delaney D, Balestrieri SG, Bassett SS, & Stein LAR (2020). A Brief Screen to Detect Cannabis Use Disorder Among Incarcerated Youth. Journal of Child & Adolescent Substance Abuse, 29(3), 189–196. 10.1080/1067828X.2021.1943587 [DOI] [Google Scholar]
  21. Dexheimer JW, Beal SJ, Divekar P, Hall ES, Patel V, & Greiner MV (2019). Automated patient linking for electronic health record and child welfare databases. Journal of Technology in Human Services, 37(4), 286–292. 10.1080/15228835.2019.1578327 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Enders CK (2011). Analyzing longitudinal data with missing values. Rehabilitation Psychology, 56(4), 267. 10.1037/a0025579 [DOI] [PubMed] [Google Scholar]
  23. Family First Prevention Services Act, H.R.253, 115th Cong. (2017). https://www.congress.gov/bill/115th-congress/house-bill/253/text?q=%7B%22search%22%3A%5B%22family+first+prevention+services+act%22%5D% 7D&r=1%29
  24. Ghitza UE, Gore-Langton RE, Lindblad R, Shide D, Subramaniam G, & Tai B (2013). Common data elements for substance use disorders in electronic health records: The NIDA Clinical Trials Network experience. Addiction, 108(1), 3–8. [DOI] [PubMed] [Google Scholar]
  25. Ghitza UE, & Tai B (2014). Challenges and opportunities for integrating preventive substance-use-care services in primary care through the Affordable Care Act. Journal of Health Care for the Poor and Underserved, 25(1 0), 36. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Grant BF, & Dawson DA (1997). Age at onset of alcohol use and its association with DSM-IV alcohol abuse and dependence: Results from the National Longitudinal Alcohol Epidemiologic Survey. Journal of Substance Abuse, 9, 103–110. [DOI] [PubMed] [Google Scholar]
  27. Griffin KW, Lowe SR, Botvin C, & Acevedo BP (2019). Patterns of adolescent tobacco and alcohol use as predictors of illicit and prescription drug abuse in minority young adults. Journal of Prevention & Intervention in the Community, 47(3), 228–242. 10.1080/10852352.2019.1603672 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Gryczynski J, Mitchell SG, Schwartz RP, Kelly SM, Dušek K, Monico L, O’Grady KE, Brown BS, Oros M, & Hosler C (2019). Disclosure of Adolescent Substance Use in Primary Care: Comparison of Routine Clinical Screening and Anonymous Research Interviews. Journal of Adolescent Health, 64(4), 541–543. 10.1016/j.jadohealth.2018.10.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Hammock K, Velasquez MM, Alwan H, & von Sternberg K (2020). Alcohol Screening, Brief Intervention, and Referral to Treatment (SBIRT) for Girls and Women. Alcohol Research : Current Reviews, 40(2), 07. 10.35946/arcr.v40.2.07 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Han B, Compton WM, Blanco C, & Jones CM (2019). Time since first cannabis use and 12-month prevalence of cannabis use disorder among youth and emerging adults in the United States. Addiction, 114(4), 698–707. 10.1111/add.14511 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Johnston LD, Meich RA, O’Malley PM, Bachman JG, Schulenberg JE, & Patrick ME (2022). Monitoring the future national survey results on drug use, 1975–2000. National Institute on Drug Abuse. chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://deepblue.lib.umich.edu/bitstream/handle/2027.42/171751/mtf-overview2021.pdf?sequence=1&isAllowed=y [Google Scholar]
  32. Keller TE, Salazar AM, & Courtney ME (2010). Prevalence and Timing of Diagnosable Mental Health, Alcohol, and Substance Use Problems among Older Adolescents in the Child Welfare System. Child Youth Serv Rev, 32(4), 626–634. 10.1016/j.childyouth.2009.12.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Kendler KS, Bulik CM, Silberg J, Hettema JM, Myers J, & Prescott CA (2000). Childhood Sexual Abuse and Adult Psychiatric and Substance Use Disorders in Women: An Epidemiological and Cotwin Control Analysis. Archives of General Psychiatry, 57(10), 953–959. 10.1001/archpsyc.57.10.953 [DOI] [PubMed] [Google Scholar]
  34. Kontu M, Kantojärvi L, Hakko H, Riala K, & Riipinen P (2022). Misuse of prescribed psychotropic medication and drug crime offending: A follow-up case-control study of former adolescent psychiatric inpatients. Criminal Behaviour and Mental Health, 32(2), 124–137. 10.1002/cbm.2254 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Kumpfer KL, Molgaard V, & Spoth R (1996). The Strengthening Families Program for the prevention of delinquency and drug use. In Preventing childhood disorders, substance abuse, and delinquency (pp. 241–267). Sage Publications, Inc. 10.4135/9781483327679.n11 [DOI] [Google Scholar]
  36. Levy SJL, Williams JF, COMMITTEE ON SUBSTANCE USE AND PREVENTION, Ryan SA, Gonzalez PK, Patrick SW, Quigley J, Siqueira L, Smith VC, & Walker LR (2016). Substance Use Screening, Brief Intervention, and Referral to Treatment. Pediatrics, 138(1), e20161211. 10.1542/peds.2016-1211 [DOI] [PubMed] [Google Scholar]
  37. Liu X-Q, Guo Y-X, & Wang X (2023). Delivering substance use prevention interventions for adolescents in educational settings: A scoping review. World Journal of Psychiatry, 13(7), 409–422. 10.5498/wjp.v13.i7.409 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. McCurdy AL, Gower AL, Rider GN, Thomas D, Watson RJ, Eisenberg ME, & Russell ST (2023). Adolescent substance use at the intersections of foster care, sexual orientation and gender identity, racial/ethnic identity, and sex assigned at birth. Child Abuse & Neglect, 137, 106042. 10.1016/j.chiabu.2023.106042 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Meader N, King K, Moe-Byrne T, Wright K, Graham H, Petticrew M, Power C, White M, & Sowden AJ (2016). A systematic review on the clustering and co-occurrence of multiple risk behaviours. BMC Public Health, 16(1), 657. 10.1186/s12889-016-3373-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Mitchell SG, Schwartz RP, Kirk AS, Dusek K, Oros M, Hosler C, Gryczynski J, Barbosa C, Dunlap L, Lounsbury D, O’Grady KE, & Brown BS (2016). SBIRT Implementation for Adolescents in Urban Federally Qualified Health Centers. Journal of Substance Abuse Treatment, 60, 81–90. 10.1016/j.jsat.2015.06.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Ni Y, Bachtel A, Nause K, & Beal S (2021). Automated detection of substance use information from electronic health records for a pediatric population. Journal of the American Medical Informatics Association, 28(10), 2116–2127. 10.1093/jamia/ocab116 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Ozechowski TJ, Becker SJ, & Hogue A (2016). SBIRT-A: Adapting SBIRT to Maximize Developmental Fit for Adolescents in Primary Care. Journal of Substance Abuse Treatment, 62, 28–37. 10.1016/j.jsat.2015.10.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Pryce J, Napolitano L, & Samuels GM (2017). Transition to Adulthood of Former Foster Youth: Multilevel Challenges to the Help-Seeking Process. Emerging Adulthood, 5(5), 311–321. 10.1177/2167696816685231 [DOI] [Google Scholar]
  44. Richmond-Rakerd LS, Slutske WS, Lynskey MT, Agrawal A, Madden PAF, Bucholz KK, Heath AC, Statham DJ, & Martin NG (2016). Age at first use and later substance use disorder: Shared genetic and environmental pathways for nicotine, alcohol, and cannabis. Journal of Abnormal Psychology, 125(7), 946–959. 10.1037/abn0000191 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Sakai C, Lin H, & Flores G (2011). Health outcomes and family services in kinship care: Analysis of a national sample of children in the child welfare system. Arch Pediatr Adolesc Med, 165(2), 159–165. 10.1001/archpediatrics.2010.277 [DOI] [PubMed] [Google Scholar]
  46. Schulenberg J, Johnston L, O’Malley P, Bachman J, Miech R, & Patrick M (2017). Monitoring the Future national survey results on drug use, 1975–2016: Volume II, college students and adults ages 19–55.
  47. Scott CK, Grella CE, Dennis ML, & Nicholson L (2018). Linking Individuals with Substance Use Disorders (SUDs) in Primary Care to SUD Treatment: The Recovery Management Checkups– Primary Care (RMC-PC) Pilot Study. The Journal of Behavioral Health Services & Research, 45(2), 160–173. 10.1007/s11414-017-9576-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Simmel C, Bowden CF, Neese-Todd S, Hyde J, & Crystal S (2021). Antipsychotic treatment for youth in foster care: Perspectives on improving youths’ experiences in providing informed consent. American Journal of Orthopsychiatry, 91(2), 258–270. 10.1037/ort0000532 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Smith DK, Chamberlain P, & Eddy JM (2010). Preliminary Support for Multidimensional Treatment Foster Care in Reducing Substance Use in Delinquent Boys. Journal of Child & Adolescent Substance Abuse, 19(4), 343–358. 10.1080/1067828X.2010.511986 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. StataCorp. (2023). Stata Statistical Software: Release 18. [StataCorp LLC.]. [Google Scholar]
  51. Tan C, Greiner MV, Nause K, Shahabuddin Z, & Beal SJ (2023). Mental Health Diagnoses, Health Care Utilization, and Placement Stability on Antipsychotic Prescribing Among Foster Care Youth. Academic Pediatrics, 23(3), 675–680. 10.1016/j.acap.2022.08.005 [DOI] [PubMed] [Google Scholar]
  52. Terry YM, VanderWaal CJ, McBride DC, & Van Buren H (2000). Provision of drug treatment services in the juvenile justice system: A system reform. The Journal of Behavioral Health Services & Research, 27(2), 194–214. 10.1007/BF02287313 [DOI] [PubMed] [Google Scholar]
  53. Tonmyr L, Thornton T, Draca J, & Wekerle C (2010). A Review of Childhood Maltreatment and Adolescent Substance Use Relationship. Current Psychiatry Reviews, 6(3), 223–234. 10.2174/157340010791792581 [DOI] [Google Scholar]
  54. Tripodi SJ, & Bender K (2011). Substance abuse treatment for juvenile offenders: A review of quasi-experimental and experimental research. Journal of Criminal Justice, 39(3), 246–252. 10.1016/j.jcrimjus.2011.02.007 [DOI] [Google Scholar]
  55. United Nations Office on Drugs and Crime (Ed.). (2005). Schools: School-based education for drug abuse prevention. United Nations Publications. https://www.unodc.org/pdf/youthnet/handbook_school_english.pdf [Google Scholar]
  56. U.S. Department of Education. (2023). Preventing and Reducing Youth and Young Adult Substance Misuse: Schools, Students, Families. https://www.ed.gov/opioids/
  57. U.S. Department of Health and Human Services. (2023). Impact of Substance Use on the Child Welfare System. Child Welfare Information Gateway. https://www.childwelfare.gov/topics/systemwide/bhw/impact-substance/ [Google Scholar]
  58. US Preventive Services Task Force. (2020). Screening for Unhealthy Drug Use: US Preventive Services Task Force Recommendation Statement. JAMA, 323(22), 2301–2309. 10.1001/jama.2020.8020 [DOI] [PubMed] [Google Scholar]
  59. Vaughn MG, Ollie MT, McMillen JC, Scott L Jr, & Munson M (2007). Substance use and abuse among older youth in foster care. Addictive Behaviors, 32(9), 1929–1935. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Velleman RDB, Templeton LJ, & Copello AG (2005). The role of the family in preventing and intervening with substance use and misuse: A comprehensive review of family interventions, with a focus on young people. Drug and Alcohol Review, 24(2), 93–109. 10.1080/09595230500167478 [DOI] [PubMed] [Google Scholar]
  61. Volkow ND, Han B, Einstein EB, & Compton WM (2021). Prevalence of Substance Use Disorders by Time Since First Substance Use Among Young People in the US. JAMA Pediatrics, 175(6), 640–643. 10.1001/jamapediatrics.2020.6981 [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Walters GD, & Urban H (2014). Age of Substance Use Onset as a Predictor of Early Adult Substance Dependence and Offending in Male and Female Delinquents: Simple and Mediated Moderation. Journal of Drug Issues. 10.1177/0022042614542513 [DOI] [Google Scholar]
  63. Wildeman C, & Emanuel N (2014). Cumulative Risks of Foster Care Placement by Age 18 for U.S. Children, 2000–2011. PLOS ONE, 9(3), e92785. 10.1371/journal.pone.0092785 [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Yoon S, Kobulsky JM, Yoon D, & Kim W (2017). Developmental pathways from child maltreatment to adolescent substance use: The roles of posttraumatic stress symptoms and mother-child relationships. Children and Youth Services Review, 82, 271–279. 10.1016/j.childyouth.2017.09.035 [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Yule AM, Wilens TE, Martelon M, Rosenthal L, & Biederman J (2018). Does exposure to parental substance use disorders increase offspring risk for a substance use disorder? A longitudinal follow-up study into young adulthood. Drug and Alcohol Dependence, 186, 154–158. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1937161-supplement-1.docx (47.8KB, docx)

RESOURCES