Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Oct 22.
Published in final edited form as: Child Maltreat. 2024 Oct 7;30(4):707–717. doi: 10.1177/10775595241289894

Prenatal Substance Exposure and Infant Discharge Placement: Results From the ACT NOW Study

Tre D Gissandaner 1, Jordan A Gette 2, Kristin J Perry 3,4, Alainna Wen 5, Timothy Regan 6,7, Lauren M Mutignani 8, Dustin E Sarver 9, Crystal S Lim 10, Robert D Annett 11
PMCID: PMC11973236  NIHMSID: NIHMS2045628  PMID: 39374518

Abstract

It is critical that researchers gather evidence of factors that identify infants at risk of out-of-home placement based on types of substance exposures and demographic characteristics. This study applied a validated medical record data extraction tool on data derived from a multi-site (N = 30) pediatric clinical trials network (ISPCTN) study of Neonatal Opioid Withdrawal (ACT NOW study). Participants included 1808 birthing parent-infant dyads with documented NOWS scoring or prenatal opioid exposure. Non-Hispanic White pregnant persons comprised the largest proportion of the sample (69.8%), followed by Non-Hispanic Black (11.6%), Non-Hispanic Multiracial and Other race (8.5%), and Hispanic (6.2%). Most notably, infant prenatal substance exposure across alcohol, cocaine, meth/amphetamine, and opioids, had the lowest possibility of discharging to parent(s). Additionally, latent class analysis identified distinct classes of substance use during pregnancy that were associated with different probabilities of discharging to parent(s). Specifically, less than half of infants (47%–49%) in the Poly-use and Meth/amphetamine classes were discharged to their parent(s). Severity of infant withdrawal symptoms influenced placement decisions within the Poly-use and Prescription Opioid classes. Findings can inform standard practices for increasing support for pregnant persons and substance-exposed infants including identification, subsequent referrals, communication with Child Protective Services, and plans of safe care.

Keywords: infant, parents, neonatal abstinence syndrome, child protective services, latent class analysis, patient discharge

Neonatal Abstinence Syndrome (NAS), or its equivalent specifically caused by prenatal opioid exposure, Neonatal Opioid Withdrawal Syndrome (NOWS), are on the rise in the United States. Newborns diagnosed with NAS rose from 1.5 per 1000 hospital births in 2004 to 7.3 in 2017 (Hirai et al., 2021; Winkelman et al., 2018), an almost fivefold increase fueled by the U.S. opioid epidemic. Prenatal substance exposure, and opioid exposure specifically, is often a predictor of infant involvement with Child Protective Services (CPS). Of these newborns with NAS, those referred to CPS similarly rose from 4.7% in 2004 to 9.2% in 2014 (Lynch et al., 2018), a twofold increase. Laws and/or policies in 42 states require healthcare providers to report prenatal substance exposure to CPS for assessment of child abuse and neglect, and CPS in 33 of those states are required to develop plans of safe care to address the needs of these substance exposed infants (U.S. Department of Health and Human Services [HHS], 2020).

Given the focus of CPS is to protect children from abuse and neglect, decisions to refer cases to CPS for evaluations of parental capacity are influenced by a wide variety of factors including cumulative child safety risk, hospital- and/or state-wide policies, and an abundance of caution in instances when the duty to report is unclear (Murphy-Oikonen, 2021). Identifying infants at risk for negative outcomes due to prenatal substance exposure or out of home placement is key to implementing early intervention services that can begin during pregnancy or soon after delivery. NAS is a constellation of withdrawal symptoms associated with negative outcomes such as longer hospital stays and increased likelihood of readmission (Patrick et al., 2012). Therefore, it is critical that children with NAS receive early intervention, as fewer than half of eligible infants with NAS are enrolled in early intervention services (Peacock-Chamber et al., 2019).

Discharge placement decisions for infants prenatally exposed to substances can be based on factors like the type of substance exposure. For example, infants with prenatal cocaine, meth/amphetamine, or opioid exposure were less likely to be discharged to their parents, compared to other types of substance exposure (Austin et al., 2022; Canfield et al., 2017; Gissandaner et al., 2024; Prindle et al., 2018). However, research in this area has not accounted for the impact of contextual variables (e.g., race and ethnicity, rurality, and infant receipt of pharmacologic treatment for withdrawal symptoms) on decision-making regarding discharge placement, and how these interplay with prenatal substance exposures.

For instance, Black pregnant persons are more likely to be screened for illicit substance use (Kerker et al., 2004) and have their infants screened, regardless of whether they meet screening criteria (Ellsworth et al., 2010). As it relates to child placement and CPS involvement, among those screening positive for substances during pregnancy, placement with caregivers is comparable or higher for racial and ethnic minority parents compared to White parents (Putnam-Honstein et al., 2016; Rebbe et al., 2019). However, Black pregnant persons are more likely to be referred to welfare services than their White counterparts, highlighting inconsistent treatment of pregnant persons using substances as a function of racial identity (Office of National Drug Control Policy, 2021). Additionally, methamphetamine use tends to be more common in rural areas compared to urban areas (Gfroerer et al., 2007). Rural communities also tend to lack resources to provide services to parents struggling with substance use disorders, especially those involved with the child welfare system (Clary et al., 2020). These contextual variables habitually introduce bias when evaluating for prenatal substance exposure and parental capacity. A major key to reducing health disparities is understanding the differences of these influences (e.g., race/ethnicity) across contexts (e.g., U.S. geographic areas) (Wright et al., 2022).

Preliminary work from our group revealed two distinct subgroups of substance-exposed newborns and how this related to discharge placement within Mississippi (Gissandaner et al., 2024). We identified (1) a Low-withdrawal Risk class characterized by prenatal substance exposure with low risk for NAS, accompanying non-pharmacologic infant intervention, with majority Black birthing parents and (2) a High-withdrawal Risk class characterized by high risk of NAS and infant pharmacologic intervention and predominantly White birthing parents (Gissandaner et al., 2024). We found no differences in discharge placement between these two subgroups. We also examined whether specific substances, race and ethnicity, rurality, and infant pharmacologic treatment were related to greater risk of infants’ out of home placement (Gissandaner et al., 2024). Here, we demonstrated that meth/amphetamine was the only substance type associated with greater risk of out of home placement, above and beyond other factors like NAS severity during hospitalization.

It is critical to understand the broad-level influences of parental substance use typology, race/ethnicity, and rural versus urban residencies, as these variables have a large influence on child maltreatment (Kepple, 2017; Maguire-Jack et al., 2021). However, less research has examined the influence of these variables on CPS placement decisions in a geographically diverse sample. The current study sought to expand upon our previous findings using a nationally representative sample to determine relations between substance use and demographic characteristics as they relate to infant placement outcomes among pregnant persons using substances. Aim one examined whether specific substance exposures, infant receipt of pharmacologic treatment, rurality, or birthing parent’s race and ethnicity were uniquely related to the infant’s discharge placement decisions. Aim two utilized latent class analysis (LCA) to characterize heterogeneity of prenatal substance exposure typology within a multi-state sample. Aim three determined how latent class moderates relations between infant receipt of pharmacologic treatment, rurality, and birthing parent’s race and ethnicity and infant’s discharge decisions, extending our preliminary findings.

Methods

Study Design and Participants

This cross-sectional study utilized a validated medical record data extraction tool to extract data from the medical records of 1808 birthing parent-infant dyads. Dyads met eligibility criteria of infants being at least 36 weeks gestation, and NOWS scoring in the first 120 hours of life and/or prenatal opioid exposure documented in the medical record (Devlin et al., 2023; Young et al., 2021). Opioid exposure documented in the medical record was defined as any of the following: (1) birthing parent history of prenatal opioid use; (2) record of positive birthing parent toxicology screen; (3) record of positive infant toxicology screen.

These data were obtained from a NICHD DASH data request (https://doi.org/10.57982/ye7z-n835). Medical record data derived from a multi-site (N = 30) pediatric clinical trials network (ISPCTN) study of NOWS (Advancing Clinical Trials in Neonatal Opioid Withdrawal Current Experience Study) in which data came from infants born at or admitted to one of the participating hospitals from July 1, 2016, to June 30, 2017. Participating hospitals were in 18 states across all regions of the U.S. (Young et al., 2021). Non-Hispanic White pregnant persons comprised the largest proportion of the sample (69.8%), followed by Non-Hispanic Black (11.6%), Non-Hispanic Multiracial and Other race (8.5%), Hispanic (6.2%), and unknown race and ethnicity (3.9%). Complete birthing parent and infant characteristics are displayed in Table 1. Further information regarding study procedures and participants have been published elsewhere (Devlin et al., 2023; Young et al., 2021).

Table 1.

Descriptive Statistics for Demographic and Birthing Parent and Infant Treatment Variables for the Overall Sample and According to Latent Class Membership.

Demographic and birthing parent and infant treatment variables
Overalla (N = 1808) % or M(SD) Poly-use class (n = 355) % or M(SE) Meth/amphetamine class (n = 204) % or M(SE) Cannabis class (n = 275) % or M(SE) Prescription opioid class (n = 974) % or M(SE) X2 Class contrasts
Birthweight (kg)b 3.07 (0.49) 2.87 (0.03) 3.06 (0.04) 3.04 (0.04) 3.16 (0.02) 53.23*** 4 > 1, 2, 3
APGAR 5c 8.75 (0.75) 8.72 (0.06) 8.69 (0.06) 8.70 (0.06) 8.79 (0.03) 2.64 1 = 2 = 3 = 4
Sexd Female (48.3) Male (51.7) Female (46.3) Male (53.7) Female (49.5) Male (50.5) Female (51.1) Male (48.9) Female (48.0) Male (52.0) 0.51 1 = 2 = 3 = 4
Non-hispanic black 11.6 8.2 4.5 23.8 11.8 29.76*** 3 > 1, 2, 4
Non-hispanic white 69.8 78.5 65.0 59.6 75.7 24.12*** 1 = 4 > 2, 3
Non-hispanic multiracial and other 8.5 5.7 21.7 8.4 7.4 18.38*** 2 > 1, 3,4
Hispanic 6.2 7.6 8.9 8.2 5.1 3.94 1 = 2 = 3 = 4
Rurality (RUCA code) 2.51 (2.64) 2.22 (0.16) 2.14 (.20) 2.84 (0.22) 2.60 (0.10) 9.21* 3 = 4 > 1, 2
Pharmacologic treatment Yes (38.6) Yes (65.8) Yes (20.8) Yes (26.6) Yes (35.8) 95.91*** 1 > 2, 3, 4
No (61.4) No (34.2) No (79.2) No (73.4) No (64.2)
MOUDe Yes (54.7) Yes (61.1) Yes (16.3) Yes (33.7) Yes (65.5) 175.69*** 4 = 1 >2, 3
No (45.2) No (38.9) No (83.7) No (66.3) No (34.5)
Open in a new tab

Abbreviations: P, parent exposure; I, infant positive toxicology screen; NOS, not otherwise specified; MOUD, prenatal exposure to opioid medication as treatment for opioid use disorder; RUCA, rural-urban commuting area.

*

p < .05;

***

p < .001.

a

n = 71 (3.9%) unknown race and ethnicity.

b

n = 3 missing data on birthweight.

c

n = 33 missing data on APGAR 5.

d

n = 1 sex identified as “ambiguous” and not included in analyses.

e

n = 163 missing data on MOUD.

Infant Discharge Placement

The data collection form documented where the infants were discharged or transferred. Options included home with a (1) parent(s); (2) relative; or (3) foster parent; or transfer to (4) another hospital for initiation of pharmacologic treatment; or (5) outpatient treatment center or another hospital for weaning of pharmacologic treatment; and (6) other. Given the ultimate placement of infants that were in the “transferred” or “other” categories could not be determined (i.e., an infant transferred to another hospital could have been discharged home with parent(s) from that hospital), discharge placement was coded into a categorical variable: 1 = “home with parent; ”2 = “home with relative; ”and 3 = “foster parent.” with transfers coded as missing. This categorical variable was then dummy coded for primary analyses.

Substance Exposure

Birthing parental substance exposure was determined by positive toxicology screening and/or history of substance use during the second and third trimester of the pregnancy as documented in the medical record. Infant substance exposure was determined by positive toxicology screen after delivery. This information was obtained from birthing parent and infant medical records. Substances were combined into the following categories: (1) cocaine; (2) meth/amphetamines; (3) cannabis; (4) prescription opioids (buprenorphine, hydrocodone, hydromorphone, methadone, and oxycodone); (5) non-prescription opioids (heroin and fentanyl); (6) opioids not otherwise specified (NOS); (7) alcohol; (8) nicotine; (9) depressants/sedatives (barbiturates, gabapentin, and benzodiazepine); 10) selective serotonin reuptake inhibitors (SSRIs); and 11) other. Each substance category was then coded as 0 = “no exposure by history/screening” and 1 = “exposure by history/screening.”

Pharmacologic Treatment for NOWS, Race and Ethnicity, and Rurality

One item determined whether the infant received pharmacologic treatment (Yes or No); this was treated as a binary variable in analyses. Two items determined birthing parent’s race and ethnicity indicated in the birthing parent’s medical chart. Rural-Urban Commuting Area (RUCA) codes were used to determine rurality of participants. RUCA codes ranged from 1 = Metropolitan area core to 10 = Rural areas (U.S. Department of Agriculture, 2020).

Covariates

Infant birth weight, sex, five-minute Apgar score, and prenatal exposure to opioid medication as treatment for opioid use disorder (MOUD) were obtained from the data collection form. Covariates were selected a priori based on previously demonstrated relations with bias in pediatric settings (Ellsworth et al., 2010; Kerker et al., 2004; Palusci & Botash, 2021) and associations with withdrawal severity (Bakhireva et al., 2019; Devlin et al., 2023; Gandhi et al., 2021). All multivariate models adjusted for covariates.

Analysis

Analyses were performed using SAS version 8.3 and MPlus version 7.31. First, a series of binary logistic models assessed whether birthing parent and/or infant exposure to the various substances (0 = No, 1 = Yes), demographic variables (i.e., race and ethnicity, rurality), or infant receipt of pharmacologic treatment were associated with differing odds of discharge placement (i.e., to parent(s), relative(s), or foster care). Next, multivariate models were conducted for each of the three outcomes to determine if relations were maintained when all substances, demographics, and pharmacologic treatment were entered simultaneously.

Class analyses used maximum likelihood with robust standard errors and 5000 random starts. To assess model fit, Akaike’s Information Criterion (AIC), Bayesian Information Criterion (BIC), and sample-size adjusted BIC (BICadj) were used. Following optimal class solution determination, the Bolck, Croon, and Haegenaars method (Asparouhov & Muthén, 2021; Bolck et al., 2004) assessed differences between classes on the indicators. Next, using the training weights derived from the class analysis (i.e., latent variables that account for measurement error in the class indicators), class membership was used as a moderator to predict infant placement decisions (Asparouhov & Muthén, 2021). We report odds ratios and 95% confidence intervals. Significance level was set at p < .05.

Due to the exploratory nature of this study, the impact of multiple testing on Type I error was adjusted for by calculating False Discovery Rates (FDR; Benjamini & Hochberg, 1995) via the Multiple Testing Correction application (Menyhart et al., 2021). Corrected FDR values were calculated based on the a priori statistical significance cutoff (.05), rank order of the observed p-values, and the number of model estimates. The p-value for each estimate was then compared to this FDR value, and if the observed p-value was less than or equal to the FDR value, the estimate retained its statistical significance. Additionally, throughout the results we report and interpret the adjusted odds ratios (aOR) and 95% confidence intervals (CI). If an estimate is determined to not retain its statistical significance after false discovery correction, this indicates the observed confidence interval for that estimate likely includes 1 (which signifies no statistically significant differences).

Results

Associations with Placement Outcomes

Supplemental Table 1 shows bivariate associations. When placed in a multivariate model (including birthing parent and infant substance exposure variables, pharmacologic treatment, race and ethnicity, rurality, and covariates), infant exposure to meth/amphetamines, opioids (NOS, prescription, and non-prescription opioids), cocaine, and receipt of pharmacologic treatment were associated with lower odds (aORs ranging from 0.22–0.67, 95% CIs: 0.13, 0.98; Figure 1) of discharging to parent(s) compared to discharging to relative(s) or foster parent(s). Parent exposure to meth/amphetamines, non-prescription opioids, cocaine, and alcohol were also related to lower odds (aORs ranging from 0.39–0.57, 95% CIs: 0.24, 0.88; Figure 1) of discharging to parent(s). Parent exposure to SSRIs were related to greater odds (aOR = 1.82, 95% CI: 1.07, 3.10) of discharging to parent(s). Infants of Black birthing parents had greater odds (aOR = 1.76, 95% CI: 1.04, 2.99) and infants of Multiracial and Other race birthing parents had lower odds (aOR = 0.52, 95% CI: 0.29, 0.91) of discharging to parent(s) as compared to White birthing parents. After correcting for false discovery, parent exposure to alcohol, cocaine, non-prescription opioids, infant exposure to meth/amphetamines, and receipt of pharmacologic treatment retained statistical significance. Infant exposure to prescription opioids and opioids NOS, parental exposure to SSRIs, and racial differences were no longer significant.

Figure 1.

Figure 1.

Open in a new tab

Adjusted Odds of Caregiver at Discharge on Substance Exposure, Demographic Factors, and Pharmacologic Treatmenta. Abbreviations: P, parent exposure; I, infant positive toxicology screen; NOS, not otherwise specified; MOUD, prenatal exposure to opioid medication as treatment for opioid use disorder; Rx, prescription; Non-rx, non-prescription; Meth, meth/amphetamine; RUCA, rural-urban commuting area. aModels adjusted for infant sex, birthweight, MOUD, and APGAR 5 score. Significant findings displayed. *These estimates retained their statistical significance following false discovery correction for multiple testing.

Regarding discharging to relative(s), infant exposure to non-prescription and opioids NOS were associated with greater odds (aORs ranging from 2.59–2.61, 95% CIs: 1.01, 6.79; Figure 1) of relative placement compared to placement with parent(s) or foster parent(s). Birthing parent exposure to cocaine and non-prescription opioids (aORs ranging from 1.94–2.02, 95% CIs: 1.04, 3.61; Figure 1) were also related to greater odds of placement with relative(s) versus placement with parent(s) or foster parent(s). Parent exposure to SSRIs were related to lower odds (aOR = 0.22, 95% CI: 0.08, 0.64) of discharging to relative(s). Infants of Black birthing parents also had lower odds (aOR = 0.44, 95% CI: 0.21, 0.93) of discharging to relative(s) compared to White birthing parents. From this model, only infant exposure to opioids NOS maintained statistical significance after correction.

Lastly, infant receipt of pharmacologic treatment, and prescription opioid and meth/amphetamine exposure were related to greater odds (aORs ranging from 1.71–3.61, 95% CIs: 1.15, 6.46; Figure 1) of placement with foster parent(s) compared to placement with parent(s) or relative(s). Additionally, parent exposure to meth/amphetamines, depressant/sedatives, non-prescription opioids, cocaine, and alcohol were related to greater odds of placement with foster parent(s) compared to placement with parent(s) or relative(s) (aORs ranging from 1.65–2.08, 95% CIs: 1.03, 3.69; Figure 1). Parent exposure to depressant/sedatives, alcohol, and meth/amphetamines did not remain significantly associated with infant placement with foster parent(s) after false discovery correction. Rurality was not associated with any placement outcomes. See Supplemental Table 2 for all model estimates.

Latent Class Solutions

Examining substance use variables, comparison of class solutions specifying 1–8 classes found that a 4-class solution maximized fit across AIC, BIC, BICadj, and entropy. The first class, Poly-use (n = 355), was defined by increased substance exposure across multiple substances, most notably, cocaine and opioids. The Meth/amphetamine class (n = 204) had high rates of meth/amphetamine exposure but lower rates of other substance exposure. Similarly, the Cannabis class (n = 275) had high rates of exposure to cannabis but not to other substances. The Prescription Opioids class (n = 974) evinced high prevalence of prescription opioids, with low endorsement of other substances including non-prescription opioids (Figure 2 and Supplemental Table 3).

Figure 2.

Figure 2.

Open in a new tab

Proportion of Infant and Parent Substance Exposures by Class Membership. Abbreviations: Class 1, poly-use class; Class 2, meth/amphetamine class; Class 3, cannabis class; Class 4, prescription opioid class; I, infant toxicology screen; P, parent exposure; NOS, not otherwise specified; Rx, prescription.

Moderation by Class

Several significant differences emerged in placement outcomes according to class membership (Table 2). Among the Poly-use class, Hispanic birthing parents had greater odds (aOR = 6.99, 95% CI: 1.53, 31.99) of infants discharging to parent(s) than discharging to relative(s) or foster parent(s) compared to White birthing parents, but this did not hold after false discovery correction. Infant receipt of pharmacologic treatment was related to lower odds (aOR = 0.28, 95% CI: 0.13, 0.56) of infants discharging to parent(s) than discharging to relative(s) or foster parent(s) compared to infants not receiving pharmacologic treatment. The overall probability of discharging to parent(s) versus discharging to relative(s) or foster parent(s) was .47.

Table 2.

Adjusted Odds of Caregiver at Discharge on Substance Exposure, Demographic Factors, and Pharmacologic Treatment Stratefied by Class.a

Poly-use class
 Meth/amphetamine class
 Cannabis class
 Prescription opioid
aOR 95% CI aOR 95% CI aOR 95% CI aOR 95% CI
Race and ethnicity (ref = non-hispanic white)
 Non-hispanic black 1.98 0.59, 6.59 20.57 0.14, 3041.13 3.57 0.99, 12.87 0.83 0.31,2.19
 Non-hispanic other 0.52 0.14, 1.96 0.24 0.07, 0.80 2.42 0.09, 64.53 2.20 0.10, 46.47
 Hispanic 6.99 1.53, 31.99 0.73 0.16, 3.27 8.84 0.21, 376.33 0.64 0.21, 1.99
 Rurality (RUCA code) 1.03 0.87, 1.22 1.05 0.87, 1.26 0.98 0.85, 1.14 1.04 0.92, 1.18
 Pharmacologic treatment 0.28 * 0.13, 0.56 1.87 0.54, 6.44 0.98 0.30, 3.22 0.29 * 0.14, 0.57
 Infant sex (ref = male) 0.58 0.29, 1.11 0.53 0.22, 1.30 1.28 0.51, 3.22 1.65 0.85, 3.23
 Birthweight 1.37 0.69, 2.73 1.80 0.66, 4.88 1.55 0.55, 4.41 0.62 0.33, 1.14
 APGAR 5 score 0.46 * 0.27, 0.77 0.62 0.25, 1.57 1.25 0.79, 2.00 1.09 0.66, 1.81
 MOUD 1.97 1.00, 3.89 1.63 0.41, 6.53 1.44 0.48, 4.36 2.25 1.05, 4.83
Open in a new tab

Abbreviations: MOUD, prenatal exposure to opioid medication as treatment for opioid use disorder; RUCA, rural-urban commuting area.

a

Due to issues with cell sizes and model convergence, these analyses were limited to comparison between discharging to parent(s) v. discharging with relative(s) or foster parent(s).

*

These estimates retained their statistical significance following false discovery correction for multiple testing.

For the Meth/amphetamine class, Multiracial and Other race birthing parents had lower odds (aOR = 0.23, 95% CI: 0.07, 0.76) of infants discharging to parent(s) than discharging to relative(s) or foster parent(s) compared to White parents. This finding did not retain its significance following false discovery correction. The overall probability of discharging to parent(s) versus discharging to relative(s) or foster parent(s) was .49. Among the Cannabis class, no association between predictor variables and discharge to parent(s) emerged. The overall probability of discharging to parent(s) was .80. Finally, within the Prescription Opioids class, infant receipt of pharmacologic treatment was related to lower odds (aOR = 0.31, 95% CI: 0.16, 0.61) of discharging to parent(s) than discharging to relative(s) or foster parent(s) compared to infants that did not receive pharmacologic treatment. The overall probability of discharging to parent(s) for this class was .91.

Discussion

This large, geographically diverse, retrospective chart review found that cocaine, meth/amphetamine, alcohol, and opioid exposure were uniquely related to decreased odds of discharging to the birthing parent(s) (Aim one). Interestingly, this pattern of findings differs from our group’s previous work (Gissandaner et al., 2024), which found that only methamphetamine use predicted infant discharge placement. Although, this previous work was limited by the small sample size (N = 136) and site-specific context of Mississippi (Gissandaner et al., 2024). Other recent work has found patterns of effects vary by substance type and differ from both our groups’ studies (Prindle et al., 2018). These disparate findings within the literature may be due in part to the variability in perceptions and behaviors, at both the individual and cultural level, related to various substances. Future work may want to examine how individual and group perceptions influence relations between substance type and CPS reporting/infant placement decisions.

The pattern found in the current study is likely due in part to negative perceptions about these substances, particularly methamphetamine (Carlson et al., 2012; Messina et al., 2014; Zernike, 2005), which is also associated with greater risk for child maltreatment (Carrillo et al., 2022; Rommel et al., 2015). There was no clear pattern differentiating infant toxicology screening from parent report/screening regarding placement outcomes. These findings paint a complicated picture of how substance use influences infant placement decisions: specific patterns of use (e.g., under medical supervision), recency of use, and severity of infant withdrawal may be more strongly considered in placement decisions than history of exposure alone. Overall, Aim one results indicated that cocaine, meth/amphetamine, alcohol, and opioid use were all associated in some ways with decreased likelihood of infants discharging to their birthing parent(s) while other substance types (e.g., cannabis, nicotine, etc.) were not.

The LCA yielded four classes, largely driven by single substances, and relations between predictors and placement differed within classes (Aim two). At discharge, most infants in the Prescription Opioid (91%) and Cannabis classes (80%) were discharged to parent(s), compared to less than half of infants in the Meth/amphetamine and Poly-use classes. Although it did not hold up to false discovery correction, prenatal exposure to opioid medication as treatment for opioid use disorder (MOUD) was indicative of greater odds of parental placement for the Prescription Opioid class (Table 2). While we regard this finding as exploratory, this potentially suggests that a record of MOUD may be protective against child removal from the birthing parent for infants with primarily prescription opioid exposure.

Infant pharmacologic treatment was associated with lowered odds of discharging to parent(s) for the Poly-use and Opioid classes (Aim three). Thus, severity of infant withdrawal appears to be a strong predictor of negative determinations of parental capacity, particularly for subgroups of infants with these exposure profiles. Given the low likelihood of infant withdrawal from meth/amphetamines and cannabis, it is reasonable that pharmacological intervention was unrelated to placement in these classes. This low likelihood of withdrawal from meth/amphetamine is particularly notable considering infants in this class had the overall lowest probability of discharging to their parent(s), providing further evidence that infant safety risk factors and negative perceptions of meth/amphetamine use may influence placement decision more than withdrawal symptoms (Gissandaner et al., 2024).

Regarding birthing parent race and ethnicity and placement outcomes, our results align with recent work which has found that Black and Hispanic infants diagnosed with prenatal substance exposure are removed from their birthing parent(s) at comparable or lower rates to White infants diagnosed with prenatal substance exposure (Putnam-Hornstein et al., 2016; Rebbe et al., 2019). Given studies that document racial disparities at other points in the CPS decision making process (e.g., Black pregnant persons more likely to be screened for illicit substance use regardless of meeting criteria for screening; Ellsworth et al., 2010), it is important that future work continue to parse apart how race, ethnicity, and other demographic characteristics influence relations between prenatal substance use and CPS placement decisions. Finally, it should also be noted that the present sample lacked heterogeneity in terms of birthing parent race and ethnicity (i.e., 70% of birthing parents were White), therefore, these differences could be due to restricted variability. Consequently, there does not appear to be a clear pattern of disparity at this time. However, more diverse samples in the future will provide knowledge about risk profiles for out-of-home placement that are more generalizable for different racial and ethnic groups.

Black and Hispanic pregnant persons with opioid use are less likely to receive MOUD than White pregnant persons (Henkhaus et al., 2022), possibly exacerbated by declining numbers of treatment centers in Black communities (Cummings et al., 2016). Further, Black pregnant persons that use substances are more likely to be referred to welfare services and less likely to be reunited with removed children compared to their White counterparts (National Office on Drug Control Policy, 2021). Assessing and challenging professionals’ beliefs about who uses substances and who would most benefit from substance use treatment may improve outcomes. Increasing access to care for pregnant persons, particularly those in minoritized groups, is a critical need as foster care placements in the U.S. continue to rise, more than half of which are in relation to substance use referrals (Patrick et al., 2019).

Several limitations warrant discussion. Secondary use of this data did not permit access of geographical information; therefore, we were unable to account for state-level variations in policies designating substance use during pregnancy as child abuse (HHS, 2020). This may have led to overestimation of discharge odds. Though, similar findings were demonstrated from our previous examination in a state where substance use during pregnancy is not considered child abuse (e.g., Mississippi; Gissandaner et al., 2024), suggesting the present study results still yield important implications at the national level. We were only able to assess substance exposure with binary (yes/no) responses. Understanding frequency and quantity of substance use and provider oversight for prescription substances would add nuanced understanding of placement decision-making. Notably, we were unable to assess indicators such as level of familial and social support for the birthing parent, previous history of CPS involvement, paternal substance use, and household income that may influence discharge outcomes (Canfield et al., 2017). Further, data were not available regarding birthing parent involvement in discharge decision making. A proportion of birthing parents may have elected to have the infant discharged to relatives or foster care, not because of substance use or other indicators included here.

Despite these limitations, this work informs several future directions for research and practice. These data are cross-sectional but highlight the need for longitudinal work to assess substance use prior to, during, and after pregnancy to monitor how placement decisions are made long-term. For example, some infants may have temporarily been placed with a relative or foster parent while the birthing parent engaged in additional treatment before the infant was returned to their care. Additionally, a more comprehensive approach is needed to examine how different layers of the socio-ecological environment (including state-level policies, paternal substance use) uniquely and synergistically relate to outcomes (Bronfenbrenner, 1994). Qualitative work could also inform our understanding of placement decisions. Interviews with medical and social work professionals may identify greater nuance to infant discharge decision making that retrospective chart review studies cannot determine. Qualitative methodology may also point to the need for enhancing provider education or awareness of discharge disparities and bias. Ultimately, this study provides information on who substance exposed infants are discharged to after birth and factors predicting discharge decisions to which caregiver. Future research should follow these infants, collecting developmental and psychosocial outcome measures, to determine which discharge placement(s) leads to the most favorable short- and long-term outcomes.

Conclusions

The results of the present investigation add to our understanding of substance use and exposure among pregnant persons and infants and findings are important for informing early prevention and intervention efforts. Specifically, we found evidence that substance type (i.e., cocaine, meth/amphetamine, opioids, and alcohol) and severity of infant withdrawal symptoms (measured by need for pharmacological treatment after birth) acted as risk factors and decreased the likelihood of infants being discharged to their birthing parent(s) in both multivariate models and LCA. Exploratory LCA results also revealed that, for birthing parent(s) in the Prescription Opioid class, parent engagement in treatment, specifically MOUD, may function as a protective factor and increase likelihood of being discharged with birthing parent(s). There is much work still needed to best serve this population and understand mechanisms guiding relations between substance use and outcomes as well as increasing support for pregnant persons with current substance use or a history of use.

Supplementary Material

supplemental table 2
supplemental table 1
supplemental table 3

Acknowledgements

Data from this study was made available to the authors following a data request submitted to the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) Data and Specimen Hub portal: Snowden, Jessica (2020). Advancing Clinical Trials in Neonatal Opioid Withdrawal Syndrome (ACT NOW) Current Experience: Infant Exposure and Treatment (Version 1) [dataset]. NICHD Data and Specimen Hub. https://doi.org/10.57982/ye7z-n835. No agency had a role in the preparation of the manuscript or in the decision to submit the manuscript for publication. The opinions and assertions expressed herein are those of the author(s) and do not reflect the official policy or position of the Uniformed Services University of the Health Sciences or the Department of Defense.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Research reported in this article was supported by: The Institutional Development Award States Pediatric Clinical Trials Network funded by the National Institutes of Health’s Office of the Director, as part of the Environmental Influences on Child Health Outcomes Program; the Neonatal Research Network, supported by the National Institutes of Health, the Eunice Kennedy Shriver National Institute of Child Health and Human Development, and the National Center for Advancing Translational Sciences (Principal Investigator: Jessica N. Snowden, MD). A postdoctoral fellowship by the National Institute of Mental Health supported Dr. Tre Gissandaner (award number T32 MH016434). Postdoctoral fellowships by the National Institute on Drug Abuse supported Dr. Kristin Perry (award number T32 DA017629) and Dr. Timothy Regan (award number T32 DA007292).

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Supplemental Material

Supplemental material for this article is available online.

References

  1. Asparouhov T, & Muthén B (2021). Advances in Bayesian model fit evaluation for structural equation models. Structural Equation Modeling: A Multidisciplinary Journal, 28(1), 1–14. 10.1080/10705511.2020.1764360 [DOI] [Google Scholar]
  2. Austin AE, Gest C, Atkeson A, Berkoff MC, Puls HT, & Shanahan ME (2022). Prenatal substance exposure and child maltreatment: A systematic review. Child Maltreatment, 27(2), 290–315. 10.1177/1077559521990116 [DOI] [PubMed] [Google Scholar]
  3. Bakhireva LN, Holbrook BD, Shrestha S, Leyva Y, Ashley M, Cano S, Lowe J, Stephen JM, & Leeman L (2019). Association between prenatal opioid exposure, neonatal opioid withdrawal syndrome, and neurodevelopmental and behavioral outcomes at 5–8 months of age. Early Human Development, 128(2019), 69–76. 10.1016/j.earlhumdev.2018.10.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Benjamini Y, & Hochberg Y (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B, 57(1), 289–300. 10.1111/j.2517-6161.1995.tb02031.x [DOI] [Google Scholar]
  5. Bolck A, Croon M, & Hagenaars J (2004). Estimating latent structure models with categorical variables: One-step versus three-step estimators. Political Analysis, 12(1), 3–27. 10.1093/pan/mph001 [DOI] [Google Scholar]
  6. Bronfenbrenner U (1994). Ecological models of human development. International Encyclopedia of Education, 3(2), 37–43. [Google Scholar]
  7. Canfield M, Radcliffe P, Marlow S, Boreham M, & Gilchrist G (2017). Maternal substance use and child protection: A rapid evidence assessment of factors associated with loss of child care. Child Abuse & Neglect, 70(2017), 11–27. 10.1016/j.chiabu.2017.05.005 [DOI] [PubMed] [Google Scholar]
  8. Carlson BE, Williams LR, & Shafer MS (2012). Methamphetamine-involved parents in the child welfare system: Are they more challenging than other substance-involved parents? Journal of Public Child Welfare, 6(3), 280–295. 10.1080/15548732.2012.683361 [DOI] [Google Scholar]
  9. Carrillo Beck R, Szlapinski J, Pacheco N, Sabri Laghaei S, Isard R, Oudshoorn A, & Marshall CA (2022). Violence and victimisation in the lives of persons experiencing homelessness who use methamphetamine: A scoping review. Health and Social Care in the Community, 30(5), 1619–1636. 10.1111/hsc.13716 [DOI] [PubMed] [Google Scholar]
  10. Clary E, Ribar C, Weigensberg E, Radel L, & Madden E (2020). Challenges in providing substance use disorder treatment to child welfare clients in rural communities. US Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation. [Google Scholar]
  11. Cummings JR, Wen H, & Ko M (2016). Decline in public substance use disorder treatment centers most serious in counties with high shares of black residents. Health Affairs, 35(6), 1036–1044. 10.1377/hlthaff.2015.1630 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Devlin LA, Hu Z, Ounpraseuth S, Simon AE, Annett RD, Das A, Fuller JF, Higgins RD, Merhar SL, Smith PB, Crawford MM, Cottrell LE, Czynski AJ, Newman S, Paul DA, Sánchez PJ, Semmens EO, Smith MC, Whalen BL, & Young LW (2023). The influence of mediators on the relationship between antenatal opioid agonist exposure and the severity of neonatal opioid withdrawal syndrome. Maternal and Child Health Journal, 27(6), 1030–1042. 10.1007/s10995-022-03521-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Ellsworth MA, Stevens TP, & D’Angio CT (2010). Infant race affects application of clinical guidelines when screening for drugs of abuse in newborns. Pediatrics, 125(6), e1379–e1385. 10.1542/peds.2008-3525 [DOI] [PubMed] [Google Scholar]
  14. Gandhi KD, Schak KM, Vande JL, Shekunov J, Lynch BA, Rummans TA, Geske JR, Pease ER, Limbeck MG, Desai J, Croarkin PE, & Romanowicz M (2021). The negative impact of maternal perinatal opioid use on neonatal outcomes. The Primary Care Companion for CNS Disorders, 23(3), 20m02737. 10.4088/PCC.20m02737 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gfroerer JC, Larson SL, & Colliver JD (2007). Drug use patterns and trends in rural communities. The Journal of Rural Health, 23(Suppl), 10–15. 10.1111/j.1748-0361.2007.00118.x [DOI] [PubMed] [Google Scholar]
  16. Gissandaner TD, Wen A, Gette JA, Perry KJ, Mutignani LM, Regan T, Malloch L, Tucker LC, White CB, Fry TB, Lim CS, & Annett RD (2024). Considerations and determinants of discharge decisions among prenatal substance exposed infants. Child Maltreatment, 29(2), 246–258. 10.1177/10775595231161996 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Henkhaus LE, Buntin MB, Henderson SC, Lai P, & Patrick SW (2022). Disparities in receipt of medications for opioid use disorder among pregnant women. Substance Abuse, 43(1), 508–513. 10.1080/08897077.2021.1949664 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hirai AH, Ko JY, Owens PL, Stocks C, & Patrick SW (2021). Neonatal abstinence syndrome and maternal opioid-related diagnoses in the US, 2010–2017. JAMA, 325(2), 146–155. 10.1001/jama.2020.24991 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kepple NJ (2017). The complex nature of parental substance use: Examining past year and prior use behaviors as correlates of child maltreatment frequency. Substance Use & Misuse, 52(6), 811–821. 10.1080/10826084.2016.1253747 [DOI] [PubMed] [Google Scholar]
  20. Kerker BD, Horwitz SM, & Leventhal JM (2004). Patients’ characteristics and providers’ attitudes: Predictors of screening pregnant women for illicit substance use. Child Abuse & Neglect, 28(2), 209–223. 10.1016/j.chiabu.2003.07.004 [DOI] [PubMed] [Google Scholar]
  21. Lynch S, Sherman L, Snyder SM, & Mattson M (2018). Trends in infants reported to child welfare with neonatal abstinence syndrome (NAS). Children and Youth Services Review, 86(2018), 135–141. 10.1016/j.childyouth.2018.01.035 [DOI] [Google Scholar]
  22. Maguire-Jack K, Jespersen B, Korbin JE, & Spilsbury JC (2021). Rural child maltreatment: A scoping literature review. Trauma, Violence, & Abuse, 22(5), 1316–1325. 10.1177/1524838020915592 [DOI] [PubMed] [Google Scholar]
  23. Menyhart O, Weltz B, & Gy}orffy B (2021). MultipleTesting. com: A tool for life science researchers for multiple hypothesis testing correction. PLoS One, 16(6), Article e0245824. 10.1371/journal.pone.0245824 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Messina N, Jeter K, Marinelli-Casey P, West K, & Rawson R (2014). Children exposed to methamphetamine use and manufacture. Child Abuse & Neglect, 38(11), 1872–1883. 10.1016/j.chiabu.2006.06.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Murphy-Oikonen J (2021). Neonatal abstinence syndrome: Decisions concerning infant safety. Journal of Public Child Welfare, 15(3), 384–411. 10.1080/15548732.2020.1733731 [DOI] [Google Scholar]
  26. Office of National Drug Control Policy. (2021). Substance use disorder in pregnancy: Improving outcomes for families. Retrieved from. https://www.whitehouse.gov/wp-content/uploads/2021/10/ONDCP_Report-Substance-Use-Disorder-and-Pregnancy.pdf
  27. Palusci VJ, & Botash AS (2021). Race and bias in child maltreatment diagnosis and reporting. Pediatrics, 148(1), Article e2020049625. 10.1542/peds.2020-049625 [DOI] [PubMed] [Google Scholar]
  28. Patrick SW, Frank RG, McNeer E, & Stein BD (2019). Improving the child welfare system to respond to the needs of substance-exposed infants. Hospital Pediatrics, 9(8), 651–654. 10.1542/hpeds.2019-0106 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Patrick SW, Schumacher RE, Benneyworth BD, Krans EE, McAllister JM, & Davis MM (2012). Neonatal abstinence syndrome and associated health care expenditures: United States, 2000–2009. JAMA, 307(18), 1934–1940. 10.1001/jama.2012.3951 [DOI] [PubMed] [Google Scholar]
  30. Peacock-Chambers E, Leyenaar JK, Foss S, Feinberg E, Wilson D, Friedmann PD, Visintainer P, & Singh R (2019). Early intervention referral and enrollment among infants with neonatal abstinence syndrome. Journal of Developmental and Behavioral Pediatrics: JDBP, 40(6), 441–450. 10.1097/DBP.0000000000000679 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Prindle JJ, Hammond I, & Putnam-Hornstein E (2018). Prenatal substance exposure diagnosed at birth and infant involvement with child protective services. Child Abuse & Neglect, 76(2018), 75–83. 10.1016/j.chiabu.2017.10.002 [DOI] [PubMed] [Google Scholar]
  32. Putnam-Hornstein E, Prindle JJ, & Leventhal JM (2016). Prenatal substance exposure and reporting of child maltreatment by race and ethnicity. Pediatrics, 138(3), Article e20161273. 10.1542/peds.2016-1273 [DOI] [PubMed] [Google Scholar]
  33. Rebbe R, Mienko JA, Brown E, & Rowhani-Rahbar A (2019). Child protection reports and removals of infants diagnosed with prenatal substance exposure. Child Abuse & Neglect, 88(2019), 28–36. 10.1016/j.chiabu.2018.11.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Rommel N, Rohleder NH, Wagenpfeil S, Haertel-Petri R, & Kesting MR (2015). Evaluation of methamphetamine-associated socioeconomic status and addictive behaviors, and their impact on oral health. Addictive Behaviors, 50(2015), 182–187. 10.1016/j.addbeh.2015.06.040 [DOI] [PubMed] [Google Scholar]
  35. U.S. Department of Agriculture. (2020). Rural-urban commuting areas codes. Retrieved from. https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes.aspx
  36. U.S. Department of Health and Human Services, Administration for Children and Families, Children’s Bureau. (2020). Parental substance use as child abuse. Retrieved from. https://www.childwelfare.gov/topics/systemwide/laws-policies/statutes/parentalsubstanceuse/
  37. Winkelman TN, Villapiano N, Kozhimannil KB, Davis MM, & Patrick SW (2018). Incidence and costs of neonatal abstinence syndrome among infants with medicaid: 2004–2014. Pediatrics, 141(4), Article e20173520. 10.1542/peds.2017-3520 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Wright JL, Davis WS, Joseph MM, Ellison AM, Heard-Garris NJ, & Johnson TL, AAP Board Committee on Equity. (2022). Eliminating race-based medicine. Pediatrics, 150(1), Article e2022057998. 10.1542/peds.2022-057998 [DOI] [PubMed] [Google Scholar]
  39. Young LW, Hu Z, Annett RD, Das A, Fuller JF, Higgins RD, Lester BM, Merhar SL, Simon AE, Ounpraseuth S, Smith PB, Crawford MM, Atz AM, Cottrell LE, Czynski AJ, Newman S, Paul DA, Sánchez PJ, & Semmens EO, Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network and the Nih Environmental Influences on Child Health Outcomes Echo Program Institutional Development Awards States Pediatric Clinical Trials Network. (2021). Site-level variation in the characteristics and care of infants with neonatal opioid withdrawal. Pediatrics, 147(1), Article e2020008839. 10.1542/peds.2020-008839 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Zernike K (2005, July 11). A drug scourge creates its own form of orphan. The New York Times. https://www.nytimes.com/2005/07/11/us/a-drug-scourge-creates-its-own-form-of-orphan.html [Google Scholar]

Associated Data

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

Supplementary Materials

supplemental table 2
NIHMS2045628-supplement-supplemental_table_2.docx (18.2KB, docx)
supplemental table 1
NIHMS2045628-supplement-supplemental_table_1.docx (21.1KB, docx)
supplemental table 3
NIHMS2045628-supplement-supplemental_table_3.docx (18KB, docx)

RESOURCES