Early parent‐child interaction and home environments of children exposed prenatally to opioids: A comparison of biological mothers and out‐of‐home caregivers

Abstract

Children born to mothers with opioid use disorder (OUD) are at increased risk of maltreatment and out‐of‐home care (OOHC) placement. This study examines the parent‐child interaction quality and home environments of 92 New Zealand children with prenatal opioid exposure (OE) and 106 non‐opioid‐exposed (NE) children. Experiences for those in maternal care versus OOHC were of particular interest. Biological mothers completed a lifestyle interview during late pregnancy/at birth. At 18 months, parent‐child interaction observations, maternal/primary caregiver interviews, and the Home Observation for Measurement of the Environment were completed during a home visit. At age 4.5, children underwent developmental assessment. By 18 months, 20% of OE children were in OOHC. Mothers with OUD who were younger, less cooperative, and had increased polysubstance use during pregnancy were more likely to have lost custody of their child. OE children in their mother's care experienced less positive parenting and lower‐quality home environments than NE children. OE children in OOHC had similarly resourced environments to NE children, yet experienced lower levels of parental warmth and responsiveness. Early parenting predicted child cognition, language, and behavior 3 years later, underscoring the critical importance of supporting the parenting and psychosocial needs of OE children's parents/caregivers to improve long‐term outcomes.

1. INTRODUCTION

Opioid misuse and dependence is a global health problem, accounting for more fatalities and years of “healthy” life lost to disability than any other drug (UNODC, ²⁰²³). Worldwide, an estimated 60 million people engage in the non‐medical use of opioids annually and, although the majority (∼75%) of opiate (e.g., heroin) users are men, women make up almost 50% of those engaging in pharmaceutical opioid misuse (UNODC, ²⁰²⁴).

Relevance and Key Findings

• By age 18 months, close to one in five children born to mothers with OUD were living in out‐of‐home care (OOHC).

• Significant maternal predictors of OOHC placement were younger maternal age at childbirth, greater maternal polysubstance use/abuse, and an uncooperative personality style.

• Children with prenatal opioid exposure (OE) experienced less positive parenting behaviors than comparison children regardless of caregiver type (biological mother vs. alternative), and children with prenatal OE who were raised by their biological mothers also experienced poorer quality/less enriched home environments, impacting later developmental outcomes.

Over the past two decades alongside the most recent opioid epidemic, there has also been a stark increase in the prevalence of opioid use during pregnancy in the United States. For example, a large medical insurance database study reported that over 14% of mothers obtained opioids for medical use at some point during their pregnancy (Bateman et al., 2014). It is also estimated that around one‐fifth of the prescription opioids used by women during pregnancy were obtained via someone other than a healthcare provider (Ko et al., 2020). Alarmingly, between 1999 and 2014, the national prevalence of opioid use disorder (OUD) in pregnancy increased by 333% (Haight et al., 2018). By 2017, there were 8.2 per 1000 cases of maternal opioid‐related diagnoses at hospital delivery (Hirai et al., 2021). Consequently, there has been a sharp increase in the number of infants with neonatal opioid withdrawal syndrome (Hirai et al., 2021; Patrick et al., ²⁰¹⁵; Winkelman et al., ²⁰¹⁸), with these infants requiring high levels of medical care during an often prolonged hospitalization to assess and manage infant withdrawal.

Women with an OUD often have a history of trauma and childhood adversity, higher rates of psychiatric comorbidity, and are more likely to be in violent partner relationships. They also tend to be less well‐educated and subject to greater financial and material hardship than parents without a substance use disorder (Davie‐Gray et al., 2013; Oei et al., ²⁰⁰⁹; Saia et al., ²⁰¹⁶; Whiteman et al., ²⁰¹⁴). Such factors are likely to pose serious challenges for parents in providing their children with home environments that are nurturing and enriching. They also raise significant concerns regarding their capacity for healthy parenting (Dawe et al., 2003; Paris et al., ²⁰²³; Suchman & Luthar, ^{2000, 2001}). There is evidence to support these concerns, with parental substance use disorder being associated with an elevated risk of child maltreatment and in turn, child placement in alternative family or foster care (Lean et al., 2013; Taplin & Mattick, ²⁰¹³). The adoption of punitive child welfare policies that criminalize prenatal substance use in some countries may further limit therapeutic service engagement and increase child risks (Maclean et al., 2022; Roberts et al., ²⁰²⁴). Thus, not unexpectedly, the opioid epidemic has been accompanied by an increase in child welfare reports and foster care placements (Edwards et al., 2023; Quast, ²⁰¹⁸; Radel et al., ²⁰¹⁸).

Statement of Relevance

Children with prenatal opioid exposure (OE) are at increased risk of experiencing maltreatment and out‐of‐home care. Our findings showed that although OE children raised in their biological mothers' care experienced the poorest quality or least enriched home environments, all OE children, regardless of caregiver type, experienced greater parent‐child emotional disconnection and lower parental warmth/responsiveness compared to non‐OE children. These early parenting experiences predicted children's later cognition, language, and behavioral adjustment, reinforcing the critical need to provide targeted support for children and families affected by maternal opioid use disorder.

The transition from hospital and the early childhood years is an important window of opportunity to support new mothers with an OUD during their transition to parenting. Understanding their parenting capabilities, including strengths and support needs, is important for the implementation of effective intervention strategies to nurture mother‐infant emotional connection and improve infant safety in the context of maternal recovery (Bosk et al., 2019; Flykt et al., ²⁰²¹), to improve infant health and development. Given the serious impacts of child removal on early attachments and relational health, information about the parenting quality and caregiving experiences of children raised by their biological mothers relative to alternative providers is needed to help better inform child welfare and placement decision‐making.

To date, a handful of observational studies based on very small samples (n = 7–36) have assessed the early parenting behaviors of mothers with an OUD who are engaged in medication‐assisted treatment (MAT) using direct observation of parent‐child interaction. Sarfi et al. (2011) compared the quality of mother‐infant interactions of mothers treated with methadone or buprenorphine during pregnancy (n = 36) and mothers without an OUD (n = 35) at infant age 6 months. Mothers in the MAT group showed less warmth/positivity, sensitivity/responsiveness, and cognitive stimulation as well as increased intrusiveness and detachment than comparison mothers. Similar findings were also reported at age 12 months and 4 years, suggesting parenting challenges, even for mothers in treatment, tend to generally persist (Konijnenberg et al., 2016). In contrast, Bernstein and Hans (1994) assessed the parenting behaviors of a small sample (n = 25) of mothers treated with methadone during pregnancy and a demographically matched control group of mothers (n = 38). No differences were found between the two groups in overall positive parenting behaviors (warmth, sensitivity, cognitive scaffolding) at ages 4 or 12 months. However, the mothers in methadone treatment demonstrated fewer of these positive parenting behaviors with their children at 24 months of age than the comparison mothers. Bauman and Dougherty (1983) also examined the parenting behaviors of 15 mothers treated with methadone compared to a matched control group of 15 mothers without an OUD. They found that mothers who were treated with methadone showed significantly higher observed levels of aversive behavior (e.g., disapproval, threats, provocations, commands) towards their preschool children.

Finally, only one study has examined the parenting experiences of children born to women who used illicit buprenorphine during pregnancy. These children were either raised by their biological mothers (n = 7) or foster parents (n = 14). They found that compared to a group of control mothers (n = 13), both mothers with an OUD and foster mothers had lower observed levels of sensitivity and non‐hostility on the Emotional Availability Scales (Salo et al., 2009). Mothers with an OUD also had lower levels of these behaviors than the foster mothers in this small sample. There were no between‐group differences in structuring or intrusive behavior.

Collectively, these studies suggest an overall trend toward less positive and more negative parenting behaviors among mothers with an OUD compared to mothers without a substance use disorder. While these studies do indicate ongoing parenting challenges for this vulnerable group, several important research gaps remain. First, the results of these studies are limited by their inclusion of small and unrepresentative samples. In addition, with one exception (Salo et al., 2009), the parenting experiences of a significant proportion of prenatally opioid exposure (OE) children, those raised in foster or out‐of‐home care (OOHC), have been largely overlooked. Further, almost no information exists on the quality of the home environments of these children, regardless of caregiver.

Finally, and even more importantly, whilst studies have measured the parenting of women with an OUD it is still unclear how this influences child outcomes. This is important given mounting evidence suggesting children with prenatal OE are at an increased risk of experiencing a range of developmental problems or impairments (Jaekel et al., 2021; Konijnenberg & Melinder, ²⁰¹⁵; Lee, Bora, et al., ²⁰²⁰; Lee, Pritchard, et al. (2020); Lee et al., ²⁰¹⁹; Oei et al., ²⁰¹⁷). The ideal research design to address these limitations is a longitudinal study that can track and trace children who may be otherwise unmonitored or become lost in the child welfare system, and one that includes both parent‐report measures of parenting and independent observer ratings of the parenting and quality of the child's actual home environment. Addressing these gaps will assist with the implementation of early family intervention programs for this population that focus on enhancing caregiving in the safest and most suitable environment for children's optimal development. As such, the aims of the study were to:

1. Describe the early care placements of children born to women with an OUD from birth to age 18 months.

2. Identify maternal factors associated with placement in OOHC care by age 18 months.

3. Examine the quality of early parent‐child interactions and the home environments of children with prenatal OE raised by (a) their biological mothers or (b) in OOHC, relative to (c) a comparison control group of non‐opioid‐exposed (NE) children and their parents.

4. Examine the extent to which these postnatal family factors contribute to children's later neurodevelopmental outcomes at age 4.5 years.

2. METHODS

2.1. Participants

Participants included two groups of children who were enrolled in a prospective longitudinal study of the effects of prenatal OE. Each child's biological mother was recruited in their third pregnancy trimester or at birth, with all study children born in Christchurch, New Zealand, between December 2002 and 2008. The groups are described in detail below, with sample characteristics shown in Table 1.

TABLE 1.

Characteristics of the sample at birth.

Measure	Methadone Group (N = 92)	Comparison Group (N = 106)	t/χ2	p
Maternal ethnicity
% NZ/Other European	76.1	81.1
% Māori	23.9	10.4
% Pasifika	.0	1.9
% Asian/African/Other	.0	6.6	13.39	.004
Maternal social background
% low SES	94.6	26.4	91.18	<.001
% no educational qualification	81.5	19.8	72.64	<.001
% single parent	50.0	9.4	37.98	<.001
% young mother (<21 years)	3.3	5.7	0.22	0.64
Mean (SD) maternal age	29.82 (5.24)	31.65 (5.43)	−2.41	.008
Maternal prenatal substance use
% smoked any cigarettes	92.4	16.0	111.92	<.001
% any alcohol use	19.6	19.8	.00	1.00
% any cannabis use	52.2	0.9	66.69	<.001
% any benzodiazepines use	54.3	0.9	70.69	<.001
% any stimulant use	25.0	0.9	24.55	<.001
% any additional opioid use	26.1	.0	–	–
Mean (SD) polysubstance use score	3.64 (2.04)	0.58 (1.04)	13.00	<.001
Maternal mental health
Mean (SD) depression score at term	12.23 (6.54)	5.06 (4.71)	8.74	<.001
% Any psyc. treatment in pregnancy	57.6	16.0	35.45	<.001
Maternal personality (TCI)
Mean (SD) Novelty seeking	12.06 (4.07)	8.09 (3.71)	7.15	<.001
Mean (SD) Harm avoidance	11.74 (5.33)	8.16 (4.86)	4.91	<.001
Mean (SD) Reward dependence	8.88 (3.23)	10.87 (2.91)	−4.55	<.001
Mean (SD) Persistence	2.01 (1.55)	3.17 (1.50)	−5.31	<.001
Mean (SD) Self‐directedness	13.74 (5.42)	21.49 (3.34)	−11.86	<.001
Mean (SD) Cooperativeness	18.77 (4.81)	5.05 (4.14)	−6.24	<.001
Mean (SD) Self‐transcendence	5.05 (4.14)	4.59 (3.67)	0.85	0.20
Infant Characteristics
% Male sex	58.7	47.2	2.18	0.14
% < 33 weeks GA	12.0	7.5	0.65	0.42
Mean (SD) Gestational age, weeks	38.78 (1.69)	39.21 (1.70)	−1.81	.04
Mean (SD) Birth weight, grams	3052.12 (464.45)	3418.29 (584.29)	−4.91	<.001
Mean (SD) Birth head circ., cm	33.82 (1.51)	34.70 (1.47)	−4.12	<.001
% Pharmacological treatment for NAS/NOWS	88.0	.0	–	–

Methadone Treatment Group: The first group comprised a consecutive series of 100 children who were born to women with an OUD who were prescribed methadone as part of their MAT (83% recruitment rate). Methadone was the only approved MAT for use during pregnancy in New Zealand at the time of recruitment. The buprenorphine mono‐product is not registered for use in New Zealand, although buprenorphine with naloxone is now approved and funded. Pregnant women are provided priority access to state‐funded MAT and encouraged to enroll as early as possible, meaning untreated OUD is extremely rare among pregnant women in New Zealand. As such, our recruited sample was highly representative of all pregnant mothers with an OUD in the Canterbury region at the time. The mean daily methadone dose during the third trimester for these women was 64.9 mg (12.5–195 mg). All of these mothers delivered their infants at Christchurch Women's Hospital, the largest maternity care provider in the Canterbury region. Full details of recruitment procedures are reported in (Davie‐Gray et al., 2013). At age 18 months, 92 parent‐child dyads participated in the study (92% retention). Of those mothers who retained custody of their infant at age 18 months (N = 74), the majority (92%) were still engaged in MAT with methadone. Reasons for attrition to 18 months included infant death (n = 3), declined participation (n = 4), and severe child neurodevelopmental impairment (n = 1).

Comparison Group: The second group comprised 110 comparison children whose mothers were randomly selected from the delivery schedule of Christchurch Women's Hospital, based on the expected due dates of children in the OE group (65% recruitment rate). Comparison of the socioeconomic profile of this group of families with regional census data showed that they were representative of families residing in Canterbury during the recruitment period. At age 18 months, 108 comparison parent‐child dyads participated in the study (98% retention). Reasons for attrition included declined participation (n = 1) and relocated overseas (n = 1). A further two dyads were excluded from the current study due to severe child neurodevelopmental impairment (n = 2).

2.2. Procedures

At birth, a senior research nurse completed a structured psychosocial interview with all study biological mothers to collect information about their life histories, social background, prenatal substance use/dependence, mental health, and personality. Self‐reported pregnancy substance use was then cross‐checked with random maternal urine screening results across pregnancy and infant meconium analysis for drug metabolites shortly after birth (see “Measures” for further details). Then, around age 18 months, a home visit was conducted by a child clinical psychologist or doctoral‐level students. This home visit included a direct observation of parent‐child interaction, assessment of the quality of the home environment using the Home Observation for Measurement of the Environment (HOME) inventory, and a semi‐structured interview with the child's primary caregiver. Then, at age 4.5 years all children completed neurodevelopmental assessments at the University of Canterbury Child Development Centre. This was administered by a clinical psychologist and postgraduate‐level students who were blinded to group status where possible. All procedures were approved by the Canterbury Regional Ethics Committee (18‐month follow‐up; Ref CTB/04/07112) and the Upper South B Regional Ethics Committee (4.5‐year follow‐up; Ref URB/07/10/042). Written informed consent was obtained from the participating parent or the child's social worker for those children in OOHC.

2.3. Measures

2.3.1. Caregiver status at 18 months

When children were 18 months of age, their primary custodial caregiver was recorded as either their biological mother (with or without cohabiting fathers) or an alternative caregiver in OOHC. Alternative caregivers included kinship or related carers such as biological fathers (sole paternal care, where the child had been removed from the mother's custody), other extended family members, and non‐related foster parents. Previous primary caregiver changes from birth to 18 months were also recorded.

2.3.2. Parenting experiences at 18 months

Parenting behaviors

Parents and their children were video‐recorded during a 5‐minute parent‐child free‐play observation in the child's home. Parents and children were given a standard box of toys and invited to play with whatever they liked. The parent was told to support their child's play as they felt appropriate. Toys included a doll, tea set, toy phone, rattles, puzzles, a book, cars, and toy animals. Parenting behaviors were subsequently assessed using a coding schedule adapted from the 24 and 36‐month NICHD Early Child Care Study (NICHD Early Child Care Research Network, ¹⁹⁹⁹). Mother's behaviors assessed during parent‐child interaction observations using this schedule have shown to be stable over time (r = 0.39 from 6 to 15 months and r = 0.48 from 24 to 36 months) and to be significantly positively correlated with other parenting dimensions assessed in a naturalistic context, such as the HOME inventory (Caldwell & Bradley, 1984; NICHD Early Child Care Research Network, ¹⁹⁹⁹). For the current study, each of the following parenting behaviors was rated on a 5‐point Likert scale (1 = none, 5 = pervasive/strong):

Warmth: the extent of parental warmth towards the child, including both the degree of physical and verbal affection (e.g., smiling, warm tone of voice).

Emotional support: the parent's responsiveness and involvement, including their responses to the child's needs and requests, the provision of encouragements to engage in an activity, and the extent of their enthusiasm regarding child capacities and effort.

Cognitive support: the extent to which the parent facilitated and scaffolded child efforts and cognitive ability verbally or with practical scaffolding support. Cognitive support is a measure of how contingent and appropriate the parent's behavior was for the child's level and quality of play.

Passivity: a lack of containment behavior. A passive parent was defined as not actively engaged in their child's play and not paying attention to the child's attempts to involve them in a game (e.g., physically distant, silent).

Criticism: the extent to which the parent showed negative regard for the child, including any disapproval, irritation, negative tone of voice, tense facial expressions, and/or impatience.

Intrusiveness: the degree to which the parent lacked respect for the child's autonomy and acted in an overly demanding and controlling manner, interfering verbally or physically, cutting across the child's behavior or imposing their own play agenda.

Dyadic synchrony: this scale assessed the extent of mutual synchronicity and emotional connection between the child and parent/caregiver, with higher scores reflecting a stronger sense of relatedness and mutual engagement between the dyad; the pair seems “in tune” with each other and relaxed, evidenced by harmonious and synchronous interactions.

Coding procedures and reliability: Parenting behaviors were coded by trained, blinded raters, with inter‐observer checks conducted for 20% of participating dyads. Inter‐rater reliability was assessed using the intra‐class correlation coefficient minute by minute, with correlations ranging from 0.78 (0.28–0.94) for Intrusiveness to 0.94 (0.80–0.98) for Emotional Support.

Parent‐child interaction data were available for 91/92 OE and 100/106 comparison children but lost video data could not be recovered following severe earthquake damage to our research facility.

Home Observation for Measurement of the Environment (HOME)

The infant‐toddler version of the HOME inventory (Caldwell & Bradley, 1984) was used to provide an objective measure of the quality and quantity of stimulation and support available to the study children in their homes at the age of 18 months. The HOME is comprised of 45 items, with six subscales (emotional and verbal responsivity, acceptance of suboptimal behavior and avoidance of restriction and punishment, organization of the physical and temporal environment, provision of appropriate play materials, parental involvement with the child, and opportunities for variety in daily stimulation). Items are scored (1) yes, or (0) no, and then summed. Higher scores indicate a more stimulating and supportive caregiving environment (Totsika & Sylva, 2004). The HOME inventory has good psychometric properties, with inter‐observer reliability consistently reported as 0.80 or more (Totsika & Sylva, 2004). There was also sufficient internal consistency for the current sample (α = 0.74).

Verbal aggression and physical punishment

The parent‐reported Parent‐Child Conflict Tactics Scale (CTS–PC) was used to assess parental psychological and physical punitive parenting/abuse (Straus et al., 1998). Parents responded to statements on a 7‐point scale, from never (0) to more than 21 times in the last year (6) based on how often they had used various discipline techniques/aggressive actions towards their child. Psychological aggression and physical punishment/assault items were used to examine the frequency of harsh discipline experienced by each child in the past year by their current caregiver and/or their caregiver's partner (if applicable). There was sufficient internal consistency for the psychological aggression items (α = 0.77). Cronbach's alpha for the physical punishment items was low, which is typical when overall reported levels of punishment are low.

Early childhood education

Enrolment in early childhood education (ECE) was based on parent/caregiver reports of child enrolment status (yes/no) and the total hours of ECE per week.

Background television

Overall indirect exposure to television was recorded as the total number of hours that parents/caregivers reported that the television was on in the home on average each day.

2.4. Maternal predictors of out‐of‐home care by 18 months

2.4.1. Maternal social background information

Information on each mother's age, partner status, ethnicity, educational attainment level, and socioeconomic status (SES; Elley & Irving, ²⁰⁰³) was gathered in a comprehensive maternal psychosocial interview conducted during late pregnancy/at birth by a senior research nurse specialist. A social risk composite was also created that was a summative index of the following risk indicators: (1) early motherhood (< 21 years), (2) single parent (not married/cohabiting), (3) ethnic minority (Māori, Pacific Islander, Asian or African), (4) maternal educational underachievement (did not complete high school), and (5) low SES (unemployed and semi/unskilled work).

2.4.2. Maternal personality

Maternal personality was measured using The Temperament and Character Inventory (TCI‐125; Cloninger et al., ^{1993, 1994}). The TCI‐125 assesses four temperament (Novelty Seeking, Persistence, Harm Avoidance and Reward Dependence) and three character dimensions (Self‐Directedness, Cooperativeness and Self‐Transcendence). Mothers rated whether they considered a statement as true or false of themselves, for example, “I often try new things just for fun and thrills, even if most people think it is a waste of time”. Items are reverse‐coded where appropriate and are summed to provide a score on each scale. In the current study, Cronbach's alpha for the individual temperament and character scales ranged from .93 to .99.

2.4.3. Maternal mental health

Maternal depression around the time of birth was measured using the Edinburgh Postnatal Depression Scale (EPDS; Cox et al., ¹⁹⁸⁷). The EPDS is a 10‐item questionnaire regarding depressive symptomology experienced over the past 2 weeks. Item responses ranged between (0) often, (1) sometimes, (2) hardly ever, and (3) never. Cronbach's alpha for the current sample was .96 A cut‐off score of ≥13 has adequate sensitivity (79%) and specificity (85%) for identifying depression (Cox et al., 1996). Mothers in the current study also self‐reported whether they were receiving or had sought any treatment for a psychiatric illness or mental health concerns during their pregnancy.

2.4.4. Maternal prenatal polysubstance use

The extent of maternal prenatal polysubstance use was measured using (a) maternal report of the frequency and amount of a range of substances used during each pregnancy trimester, (b) random maternal urine sampling across pregnancy, and (c) meconium samples obtained from 81% of OE infants after birth. In the current analysis, a summative index that included both the number of different substances used (cigarettes, alcohol, cannabis, benzodiazepines, stimulants) as well as the level of use (e.g., heavy vs. light use) was calculated (α = .71) to indicate overall prenatal polysubstance use (see Lee, Pritchard, et al. (2020) for further details).

2.5. Child developmental outcomes at age 4.5 years

2.5.1. Cognitive ability

A short form of the Wechsler Preschool and Primary Scales of Intelligence—Revised (WPPSI‐R) was used to assess children's IQ/general cognition (Wechsler, 1989). Total IQ scores were derived from two performance (Block Design, Picture Completion) and two verbal (Arithmetic and Comprehension) subtests, that have excellent internal consistency (α = .92–.96). Scores from this short form also correlate highly (r = .92) with full‐scale IQ scores (Lobello, 1991).

2.5.2. Language

Children's language skills were assessed using the Clinical Evaluation of Language Fundamentals—Preschool (Wiig et al., 1992). Standardized total language scores were derived, based on children's performance across six internally consistent (α = .76–.91) receptive and expressive language subtests.

2.5.3. Behavioral adjustment

Children's overall behavioral adjustment was assessed using the valid and reliable Strengths and Difficulties Questionnaire (SDQ; Goodman, ²⁰⁰¹). Parents rated their children's behavior on the 25‐item questionnaire, and scores from the emotional, conduct, inattention/hyperactivity, and peer problems subscales were summed to provide a total difficulties score, that had adequate internal consistency for the current sample (α = .71).

2.6. Data analysis

Data were analyzed in four steps using SPSS version 28. First, we examined rates of OOHC at age 18 months using chi‐square tests of independence. Predictors of OOHC placement at 18 months for children born to mothers with an OUD were then identified. This involved examining bivariate associations between maternal psychosocial characteristics and OOHC status (see Table S1). Variables that were associated with OOHC were then entered into a logistic regression model using forward and backwards methods. Variables that did not make a significant independent contribution to or weakened the predictive ability of the model were removed to ensure parsimony.

The third step involved comparing the early parenting behaviors and home environment experiences of children in different caregiving placement contexts at age 18 months. Comparisons were made between three groups of children, including (1) prenatally OE children being raised by their biological mother, (2) prenatally OE children in OOHC being raised by an alternative caregiver, that is in foster or kinship (relative) care, and (3) non‐OE comparison children being raised by their biological parents. Measures of early parenting spanned three measurement sources: (a) direct observational measures of parental behavior during a free‐play parent‐child interaction, (b) examiner ratings of the child's home environment on the HOME scale, and (c) parent‐reported caregiving variables including parental use of verbal aggression/physical punishment, children's television exposure, and ECE/preschool enrolment. One‐way analysis of variance (ANOVA; Welch where variances were unequal) was conducted for continuous and normally distributed variables, Kruskal–Wallace tests for continuous and non‐normally distributed variables, and chi‐square tests of independence for categorical variables. Post‐hoc t‐tests were also conducted. Effect sizes were assessed using eta squared (η ²) for one‐way ANOVA and Kruskal–Wallace tests, Cramer's V for chi‐square tests, and Cohen's d for t‐tests. Common conventions were used to determine small, medium and large effect sizes. These were .02, .06 and .14 for eta squared, .07, .21, and .35 for Cramer's V, and .2, .5, and .8 for Cohen's d.

Finally, linear regression analyses were used to examine associations between measures of early parenting and the home environment with child developmental outcomes at age 4.5, controlling for caregiver group status at 18 months, gestational age at birth, maternal social risk score at term, and maternal prenatal polysubstance use.

3. RESULTS

3.1. Postnatal caregiving circumstances of children exposed to opioids during pregnancy

All comparison children were still in the custodial care of their biological mothers and living in either a two‐parent or single‐parent household. In contrast, whilst four out of five (80.4%, n = 74) OE children were still in the legal and residential custody of their biological mothers, around a fifth (19.6%, n = 18) had been placed in OOHC. These alternative care placements included (1) sole custody by the child's biological father (n = 1), (2) another family member (e.g., aunt or grandmother, n = 7), or (3) non‐relative foster parent/s or guardians (n = 10). The total number of primary caregiver changes between birth and 18 months for these children ranged from one (n = 11) to six (n = 1). Most children in OOHC were placed before age 3 months (n = 11), with five of these infants discharged from hospital to an alternative caregiver shortly after birth. An additional two children had been in OOHC but had subsequently returned to their mothers’ custody by the 18‐month research home visit.

3.2. Predictors of out‐of‐home care

Table S1 summarizes the maternal psychosocial characteristics and prenatal substance exposures assessed across pregnancy at birth for children being cared for by their biological mothers’ and those in OOHC at age 18 months. The independent effects of those factors on the likelihood of OE children entering OOHC are summarized in Table 2.

TABLE 2.

Logistic regression predicting the likelihood of entering into out‐of‐home care by age 18 months.

Variable	B (SE)	p	Odds ratio (95% CI)
Maternal age (years)	−.29 (.09)	.001	.75 (.63–.89)
Maternal Cooperativeness score (TCI)	−.16 (.08)	.04	.85 (.73–1.00)
EPDS score ≥13 at term/late pregnancy	1.50 (.81)	.06	4.47 (.92–21.80)
Any psych. treatment in pregnancy	−1.23 (.75)	.10	.29 (.07–1.28)
Polysubstance use score	.37 (.18)	.04	1.44 (1.01–2.05)
Constant	8.06 (3.32)	.02	3156.49

Note: Polysubstance use score sums of the total number and extent of cigarettes, alcohol, cannabis, benzodiazepines, and stimulants used during pregnancy.

The final logistic regression model was statistically significant χ ²(5) = 39.23, p < .001 and explained 55% (Nagelkerke R² ) of the variance in OOHC status, with the model correctly classifying 89% of children in OOHC. Key independent predictors of OOHC placement at age 18 months included maternal age (OR = .75 [.63–.89]), maternal cooperativeness score on the TCI (OR = .85 [.73–1.00]), and the extent of maternal polysubstance use/abuse during pregnancy (OR = 1.44 [1.01–2.05]). Specifically, younger mothers who were characterized by a less cooperative personality style, and who were engaged in increased polysubstance use during pregnancy were more likely to have lost residential custody of their child. Maternal depression (OR = 4.5 [.92–21.80]) and maternal psychiatric treatment during pregnancy (OR = 3.4 [.07–1.28]) approached significance, suggesting a trend for biological mothers with higher levels of depressive symptoms and those who were not receiving any psychiatric treatment during pregnancy to have been more likely to lose residential custody of their child by age 18 months.

3.3. Observed parenting behavior of mothers treated with methadone, alternative caregivers, and control group parents

We next examined the observed parent‐child interaction experiences of (a) OE children being raised by their biological mother, (b) OE children raised in OOHC by alternative (foster or kinship) caregivers, and (c) comparison children being raised by their biological parents, using the home‐based, blinded free‐play interaction data collected at age 18 months. Almost all custodial biological parents included in this analysis were the child's mother. The only exceptions were one OE and three comparison children who were observed playing with their biological fathers given family co‐parenting and shared childcare practices.

Table 3 describes the parenting behavior of the three groups of study children's parents/primary caregivers and overall ANOVA results. Post‐hoc tests showed the following:

1. Significant between‐group differences were observed in the parenting behaviors of the biological mothers of OE and comparison children. Specifically, at child age 18 months, the biological mothers in the OE group were less warm (p = .008, d = .40 [.07–.68]), provided their child less emotional support (p = .001, d = .47 [.16–.77]), cognitive support (p = .001, d = .47 [.17–.78]) and were more intrusive/critical (p = .03, OR = 3.0 [1.1–8.5]) when playing with their child than comparison parents. The quality of interactions between these mothers and their toddlers was also less synchronous and in tune relative to comparison group dyads (p < .001, d = .51 [.21–.82]).

2. A similar pattern of results was found when comparing the parenting behavior of OE children's alternative caregivers and comparison children's parents, with significant differences in parental warmth (p = .006, d = .67 [.14–1.19]), emotional support (p < .001, d = .85 [.32–1.37]), and cognitive support [p = .02, d = .56 [.04–1.08]. They also showed higher rates of passivity (OR = 2.5 [.7–9.0]), but this difference was non‐significant, likely reflecting low statistical power. Lastly, parent‐child interactions between alternative caregivers and their OE children were less synchronous and in tune than comparison group dyads (p < .001, d = .83 [.31–1.36]).

3. Finally, examination of the general parenting profiles of OE children's biological parents and alternative caregivers showed trends towards alternative caregivers being characterized by lower levels of warmth (p = .07, d = .028 [−.25–.81]) and emotional support (p = .06, d = .32 [−.21–.85]) than biological mothers in the OE group. In addition, the alternative caregivers had double the odds of exhibiting passive behavior (p = .23, OR = 2.2 [.6–8.3]), although this was non‐significant, which, as mentioned above, likely reflects a statistical power issue.

TABLE 3.

Parenting behaviors during free play observations age 18 months.

Parenting Behavior	Methadone Group: Biological Parent (N = 74)	Methadone Group: Alternative Caregivers (N = 17)	Non‐exposed Comparison Group (N = 100)	F/χ2	p	η 2 (95% CI)/ Cramer's V
Warmth, M (SD)	2.32 (.51) **	2.19 (.26) **	2.51 (.51)	5.05	.007	.05 (.00–.12)
Emotional support, M (SD)	2.40 (.63) **	2.21 (.37) ***	2.67 (.57)	10.70	<.001	.08 (.02–.15)
Cognitive support, M (SD)	2.68 (.58) **	2.64 (.43) *	2.94 (.55)	5.82	.004	.06 (.01–.13)
Any passivity, %	12.2	23.5	11.0	2.09	.351	.11
Any harshness, %	16.2 *	5.9	6.0	5.30	.071	.17
Dyadic synchrony, M (SD)	3.00 (.76) ***	2.76 (.66) ***	3.39 (.76)	8.57	<.001	.08 (.02–.16)

Note: F/χ² statistics, p‐values and effect sizes are shown for overall ANOVAs and Chi Square tests. For post‐hoc tests. Differences between the two methadone groups did not reach statistical significance. Smaller sample size for parent‐child interaction data due to earthquake damage to several video recordings (alternative care group n = 1, comparison group n = 6).

^*denotes significant difference to comparison group (* < .05, ** < .01, *** < .001).

3.4. Quality of the caregiving home environment

Table 4 describes the quality of each study child's home environment. Measures spanned HOME subscales, caregiver self‐reported punitive parenting, children's daily television exposure, and ECE enrolment. The key results were as follows:

1. In general, the early caregiving environments of OE children being raised by their biological parents were rated as being of lower quality than comparison children. Specifically, they were characterized by lower scores across all subscales of the HOME and, as such, the OE children had lower total HOME scores (p < .001, d = .90 [.58–1.21]). Biological parents of OE children reported higher median levels of verbal aggression (p = .02, r = .02) but not physical punishment than comparison parents. Children with prenatal opioid‐exposure who were raised by their biological parents were less likely to be enrolled in ECE than comparison children and those that were enrolled attended for fewer hours than comparison children (p = .02, r = .20). Finally, OE children were exposed to significantly more background television than comparison children (p < .001 d = .94 [.62–1.25]).

2. In contrast, the early caregiving environments of OE children raised by alternative caregivers were, on average, similar to those of comparison children. The only exception was for a lower score on opportunities for daily stimulation for children in alternative care compared to comparison children (p = .047, d = .67 [.17–1.18], and lower rates of ECE enrolment among the OE children raised by alternative caregivers (p < .04, OR = 2.89 [1.04–8.00]).

3. Finally, a comparison of the two groups of OE children showed that children being raised by their biological parents had lower‐quality early caregiving environments, on average, than OE children raised by alternative caregivers. Specifically, OE children raised by their biological parents had lower total HOME (p = .005, d = .69 [.16–1.21]), emotional and verbal responsivity (p = .04, d = .46 [.06–.98]), organization of the environment (p = .03, d = .52 [.01–1.04]), and parent involvement (p < .001, d = .85 [.32–1.38]) scores, and were exposed to more daytime television (p < .001 d = .84 [.31–1.37]) than OE children raised by alternative caregivers. There were no significant differences in parent/caregiver verbal aggression or physical punishment, although results favored the OE children raised by alternative caregivers. Rates of enrolment in ECE were similar.

TABLE 4.

Quality of the child‐rearing environments of prenatally methadone‐exposed and non‐exposed children at age 18 months.

Environmental Variable	Methadone Group: Maternal Care (N = 74)	Methadone Group: Alternative Care (N = 18)	Non‐Exposed Comparison (N = 106)	F /χ2/H	p	η 2 (95% CI)/ Cramers V
Supportive Parenting
Total HOME Score	33.88 (6.18) ***,††	37.99 (5.02)	38.45 (4.21)	15.26	<.001	.16 (.07–.24)
Emotional and Verbal Responsivity	8.08 (2.07) **,†	9.00 (1.64)	8.74 (1.80)	3.26	.041	.03 (.00–.08)
Acceptance of Child Behaviour	6.49 (1.50) **	6.83 (1.15)	6.98 (1.04)	2.94	.063	.03 (.00–.09)
Organisation of Environment	5.24 (.86) *,†	5.67 (.59)	5.47 (.75)	3.06	.049	.03 (.00–.09)
Provision of Play Materials	7.34 (1.63) ***	7.93 (1.16)	8.04 (.98)	5.34	.008	.06 (.01–.13)
Parent Involvement	3.13 (1.85) ***,†††	4.67 (1.57)	4.79 (1.39)	24.73	<.001	.20 (.11–.29)
Opportunities for Daily Stimulation	3.59 (1.22) ***	3.89 (1.28) *	4.43 (.70)	16.29	<.001	.14 (.06–.23)
Harsh Discipline
Verbal Aggression Score, median (range)	12 (0–138) *	3 (0–83)	8 (0–150)	5.56	.062	.022
Physical Punishment Score, median (range)	4 (0–40)	0 (0–40)	1 (0–60)	2.07	.356	.004
Other Experiences
Any Early Childhood Education (ECE), %	54.1 *	44.4 *	69.8	7.04	.030	.03
Hours ECE per week, median (range)	0 (0–36) *	0 (0–45)	6 (0–40)	7.64	.020	.026
Hours TV on per day	8.13 (5.04) ***,†††	4.13 (3.06)	4.13 (3.65)	21.18	<.001	.18 (.09–.27)

Note: Results presented as Mean (SD) unless otherwise specified. F/χ²/H statistics, p‐values and effect sizes are shown for overall ANOVAs and chi square tests.

For post‐hoc tests, ^*denotes significant difference to comparison group (^* < .05, ^** < .01, ^*** < .001);

^†denotes significant difference to alternative caregivers (^† < .05, ^†† < .01^{, †††} < .001).

3.5. Associations between early parenting behaviors, home environment quality and child developmental outcomes at 4.5 years

Linear regression analyses were used to examine the extent to which variations in early parent‐child interaction quality and home environment factors helped to explain between‐group differences in cognition, language, and behavior at age 4.5 years of OE and comparison children (see Table S2 for descriptive statistics). After controlling for covariates (OOHC status, prenatal polysubstance exposure, gestational age, maternal social risk score), parental emotional support score during the free‐play task (β = .14, p = .046) and hours of background television (β = −.17, p = .02) reported at 18 months were significantly, independently associated with children's IQ score at age 4.5 years. Parental emotional support score (β = .17, p = .009) and overall HOME score (β = .16, p = .02) at 18 months significantly, independently predicted children's Total Language score at age 4.5 years, while children's Total Difficulties score on the SDQ at age 4.5 years was marginally associated with parental cognitive support (β = −.12, p = .08) measured at 18 months.

4. DISCUSSION

We examined the postnatal placement outcomes and parenting experiences of children born to mothers with an OUD who were treated with methadone during pregnancy, compared to a regionally representative control group of children. This study has numerous strengths that assisted in addressing some of the challenges of earlier research. These include its prospective longitudinal design, a relatively large, representative group of OE children, including those living with biological and alternative caregivers, as well as a regionally representative comparison group. We also utilized multiple measures of parenting and the home environment to capture the early parenting experiences of these children in a naturalistic context. Study findings are discussed below.

We found that by age 18 months, close to one in five children born to mothers with an OUD had been placed in OOHC. Alternative caregivers included biological fathers, other extended family members, and non‐relative foster caregivers. These data are in line with recent reports describing the increase in child welfare cases associated with parental opioid use in North America (Lynch et al., 2018; Quast, ²⁰¹⁸; Radel et al., ²⁰¹⁸). Although most of the biological mothers remained in MAT, unaddressed psychosocial challenges appeared to continue to create difficulties for these families. We found that vulnerable mothers with an OUD during pregnancy who gave birth at a younger age, had higher levels of polysubstance use/abuse during pregnancy, and were less cooperative were significantly less likely to retain care of their child. Untreated poor mental health was also associated with OOHC entry. These findings indicate the intensive wraparound support that is likely to be needed to assist many families affected by OUDs to ensure safe and nurturing environments, and reduce the rates of OOHC for OE children. Although we were unable to examine protective factors, for example, partner relationship quality and the availability of social support (Dawe et al., 2003) in mitigating OOHC risk, these would be important to include in intensive parenting support programs that simultaneously address mother's psychological functioning and trauma histories alongside important social contextual factors.

Examination of the parenting experiences of children who were still being raised by their biological mothers at age 18 months relative to those living with an alternative caregiver and the regionally representative comparison group showed that children born to mothers with an OUD experienced less positive parenting behaviors than the comparison group children, regardless of caregiver type. The biological mothers who were treated with methadone and the alternative caregivers were characterized by lower levels of warmth and emotional and cognitive support than comparison mothers. Further, these two groups showed less dyadic synchrony relative to the comparison group, indicative of lower levels of parent‐child reciprocity and emotional connection. The findings are consistent with previous research showing lower warmth, sensitivity, and scaffolding among mothers with an OUD, including those treated with methadone or buprenorphine (Konijnenberg et al., 2016; Salo et al., ²⁰⁰⁹; Sarfi et al., ²⁰¹¹).

In addition, similar to existing studies (Bauman & Dougherty, 1983; Konijnenberg et al., ²⁰¹⁶; Salo et al., ²⁰⁰⁹; Sarfi et al., ²⁰¹¹), we also found that the biological mothers of OE children exhibited higher rates of negative parenting behaviors compared to the other two groups. Specifically, mothers with an OUD who were treated with methadone during pregnancy demonstrated higher rates of criticism and intrusiveness when observed interacting with their children. In line with this observation, these mothers also reported higher levels of parental verbal aggression than alternative caregivers and comparison parents. However, there were no between‐group differences in self‐reported physical punishment use by parents/caregivers, with very few reporting any physical punishment at all. Corporal punishment was criminalized in New Zealand prior to the time these data were collected, and the results are likely reflecting a combination of actual lower population rates and/or underreporting. Further, those who may have been at the highest risk of experiencing physical punishment or maltreatment are likely those children who were already in OOHC at follow‐up.

Interestingly, our results showed a trend for the alternative caregivers group to demonstrate lower levels of warmth, emotional support, and dyadic synchrony than the group of biological mothers who were treated for an OUD. They also showed higher rates of passivity, suggesting less engagement with their child during free play. The ease and strength with which caregivers interact with their child may be influenced by several factors, including the length of time in which the child has been in their custody. Caregiver instability was common in our cohort of OE children, with more than a third of those in OOHC having had two or more primary caregivers in their first 18 months of life. As such, some of the alternative caregivers and children may have had less time to form a close connection than other dyads. Multiple foster care placements raise concerns about impacts on children's socioemotional wellbeing and this, alongside evidence that positive parenting skills by foster parents protect against further placement instability (Konijn et al., 2019), strongly suggests a need for new foster caregivers to receive interventions to learn supportive ways to interact with their child and build a reciprocal, positive caregiver‐child relationship.

In contrast, an examination of the home life experiences of the cohort showed that OE children raised in OOHC had similar mean HOME scores and caregiver‐reported rearing environment quality to the comparison group. Whereas, results from the HOME inventory showed that OE children in their biological mother's care were, on average, experiencing significantly poorer quality rearing environments than OE children in alternative care and comparison children. OE children living with their biological mothers had less parental involvement/engagement during day‐to‐day activities, and less exposure to a variety of books/toys, learning activities, and other care providers. These findings, as well as the greatly increased indirect television exposure among OE children living with their biological mothers compared to the other two groups, are suggestive of a less socially and physically enriched home environment, alongside material hardship and lack of parental social support.

It is important to note that maternal OUD is unlikely to be the only reason for the observed differences in parenting quality and home environments in this study, but more likely one of many factors that contribute to compromised parenting in the context of substance use disorder and associated cumulative risk (Hatzis et al., 2017; Suchman & Luthar, ²⁰⁰⁰). The mothers in treatment for an OUD were characterized by significantly increased social disadvantage at the time of birth than the comparison mothers, but as we did not have the same historical psychosocial data available for the alternative caregivers at 18 months, we could not account for these confounding social background risks in the analysis.

Likewise, there are different contextual circumstances at age 18 months that would be important to examine for their role in determining parenting/environment quality to help inform interventions, including the mother's recovery status or addiction severity, ongoing use of additional substances, mental distress, parenting stress, and partner/social support, alongside financial and material hardship (Paris et al., 2023; Suchman & Luthar, ²⁰⁰¹). It would also be important for intervention planning to understand the psychosocial factors associated with the parenting of alternative caregivers in the OOHC group.

A better understanding of these psychosocial influences and how to address them to improve parenting quality and the early home life experiences of OE children in both maternal and OOHC will be important to minimize longer‐term developmental and caregiving risks. This is supported by our findings examining the extent to which the early parenting experiences of OE children were associated with later child developmental outcomes, including general cognitive functioning (IQ), language skills, and behavioral adjustment over and above prenatal substance exposure, maternal social background and OOHC status. Specifically, parental emotional support during free play was a significant, independent predictor of both higher IQ and language scores at age 4.5 years, while parental cognitive support was associated with better behavioral adjustment. Background television was also negatively associated with cognition while total HOME score was positively associated with language skills. These findings support the limited longitudinal research demonstrating the influence of positive parenting on development within this population (Konijnenberg et al., 2016) and further highlight the importance of responsive/sensitive and emotionally connected parenting in the context of an enriched environment to improve developmental outcomes for children subject to prenatal/early life stress (Conradt et al., 2018).

The results from the current research combined with those of earlier studies highlight the numerous parenting challenges of mothers with an OUD, and underscore three target areas for early intensive support of this high‐risk parenting group. First, detecting those mothers with an OUD who are at risk of losing custody of their child, and implementing effective interventions is essential to reduce potential child maltreatment and rates of entry into OOHC (Dawe et al., 2003). Key to this will be strengthening parent‐child reciprocity and connection (Bosk et al., 2019; Flykt et al., ²⁰²¹), and providing guidance for parents on the developmental needs of their child/children. Ongoing supports must continue to consider the psychosocial resources of these mothers that contribute to their parenting capacities, as well as the financial resources available for providing a caregiving environment that contains materials for children's learning and play and translates to a positive rearing experience for children (Bernstein & Hans, 1994). As found in the present study, both observed parenting behaviors and environment quality factors were important predictors of longer‐term child development, suggesting parenting‐specific and financial/material support would be appropriate.

Importantly, this study has also demonstrated that OE children who enter OOHC will likely require ongoing monitoring, surveillance, and caregiver support. While OE children in OOHC tended to have caregivers who were more engaged in instrumental care and had more resources available to them than OE children raised at home, the quality of the interactions with their caregivers appeared less optimal, with potential negative implications for longer‐term development.

This study has some limitations that are worth noting. First, although our sample was large relative to previous studies, the group of OE children in OOHC was relatively small. With larger samples, there is a need to assess differences in the parenting experiences of children residing in foster care versus kinship care and examine the role of caregiver stability and length of placement on the provision of parental sensitivity and dyadic synchrony.

In addition, while maternal sensitivity is a crucial element in fostering parent‐child synchrony (Hammer et al., 2019), the child's role in the dynamic and bi‐directional dyadic relationship should also be considered. Infants with prenatal OE are at increased risk of showing disorganized attachment behavior (Goodman et al., 2005), regulatory problems, and attention difficulties (Jaekel et al., 2021; Levine et al., ²⁰²¹) that may challenge a parent's own self‐regulation and ability to provide emotionally supportive parenting. Acknowledging and addressing negative feedback loops in families affected by maternal OUD may assist with further enhancing parental sensitivity and parent‐child reciprocity, with important implications for children's developmental outcomes (Goodman et al., 2005; Jaekel et al., ²⁰²⁴; Rasmussen et al., ²⁰¹⁶).

Finally, we note that since this cohort was recruited, certain practices have changed in relation to the treatment, services, and support offered to mothers with an OUD and their children. One notable change includes the ESC approach to infant care that is increasingly being used to support the in‐hospital management of neonatal withdrawal. The emphasis of the ESC approach on increasing family involvement in helping to manage infant symptoms instead of pharmacologic approaches (Grossman et al., 2017; Young et al., ²⁰²³) may help to enhance early mother‐infant emotional connection, which has been shown to improve longer‐term dyadic outcomes (Woodward et al., 2024). However, follow‐up studies are needed to fully evaluate the long‐term effects of ESC on mother‐infant relationships and overall developmental outcomes.

Parenting interventions and the support services offered to mothers with an OUD postnatally are also evolving to better meet the needs of these parents who often have associated financial difficulties, trauma and mental health challenges (Bosk et al., 2019; Paris et al., ^{2015, 2023}; Stulac et al., ²⁰²⁴). The current findings reinforce the critical need for these approaches and potential expansion, especially in the infancy and early childhood years. Mothers engaged in treatment, particularly residential treatment programs with integrated parenting interventions, have been found to show more sensitive caregiving than untreated mothers or those in the very early phases of recovery (Hatzis et al., 2017), highlighting the importance of family support in improving maternal and infant outcomes. However, for the benefits of parenting interventions to be realized consistently, access to these must be integrated into routine care.

5. CONCLUSION

To conclude, this study suggests that children who are born to mothers with an OUD with comorbid mental health problems and unaddressed substance misuse are at high risk of removal from their biological mother's care during infancy. Those who remain in maternal care are more likely to experience lower positive parenting behaviors, increased harsh parenting, lower parent‐child synchrony, and lower quality home environments characterized by low parent involvement and fewer child learning/play resources compared to NE comparison children. Those who are raised in OOHC tend to have similar access to books, toys and activities to NE children, but experience lower caregiver warmth, emotional and cognitive support, as well as poorer parent‐child synchrony than comparison children without prenatal OE. Importantly, these parenting experiences were significantly associated with child cognition, language, and behavior at age 4.5 years, highlighting the potential for early positive experiences to have long‐lasting impacts on child development. Findings underscore the critical need for ongoing family support for parents and children impacted by maternal OUD.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Supporting information

Lee, S. J. , Davie‐Gray, A., & Woodward, L. J. (2025). Early parent‐child interaction and home environments of children exposed prenatally to opioids: A comparison of biological mothers and out‐of‐home caregivers. Infant Mental Health Journal, 46, 343–358. 10.1002/imhj.22157

DATA AVAILABILITY STATEMENT

The data presented in this study are available on request from the corresponding author. Some restrictions may apply. The data are not publicly available due to privacy and ethical concerns.