The effects of prenatal exposure to opioids on early childhood development: context matters most

Abstract

Introduction:

The opioid crisis has brought an increasing focus on the long-term outcomes of children following prenatal opioid exposure. Evidence to date has been conflicting, which has caused confusion and concern amongst parents, caregivers, social service providers, medical providers and policy makers.

Methods:

This review systematically evaluated the highest quality studies relating prenatal exposure to opioids with early childhood developmental outcomes. It focused on developmental outcomes as measured by the Bayley Scales of Infant and Toddler Development, encompassing cognitive, motor, and psychosocial domains of child development.

Results:

Although several articles reported correlations between prenatal opioid exposure and poor early childhood developmental outcomes, these relationships were no longer statistically significant after adjusting for socio-environmental factors.

Conclusion:

Additional research is needed to determine the extent of any relationship of socio-environmental factors with early childhood development in children prenatally exposed to opioids. This review suggests that socio-environmental factors may be significantly related to poor early childhood outcomes in the presence of prenatal opioid exposure.

1. Introduction

The escalating opioid crisis in the United States is a significant public health challenge that harms both pregnant women and their babies. Diagnoses related to opioid use disorder (OUD) at delivery has increased by 131% between 2010 and 2017 [1]. The opioid epidemic is ever-evolving, with the rising use of fentanyl significantly contributing to an increase in drug-related overdose deaths. In 2020 and 2021, overdose deaths, especially those involving synthetic opioids reached record highs [2]. This escalating crisis is further complicated by the challenges pregnant women with substance use disorder face, including societal stigma and fear of legal repercussions, leading to barriers in accessing effective care [3–5]. These challenges, combined with the risk of inadequate prenatal care and the potential for social services involvement, compound the complexity of the crisis. Concern extends beyond maternal health to the potential impact on fetal and child development. The data regarding the impact of prenatal opioid exposure in the context of opioid use disorder and adverse developmental outcomes in children are conflicting and findings are often overstated in scientific discourse, policy discussions, and media representations [6].

With both rising numbers of pregnant woman on opioids who are untreated and treated for opioid use disorders and the intensity of the opioid crisis, it is crucial that we have a clear understanding of how prenatal opioid exposure affects early childhood development. The studies on prenatal opioid exposure’s impact on early childhood development demonstrate a wide range of findings. Reviews by Smith and Behnke [7], and Longo et al. [8] concluded that significant negative effects like impaired growth and behavioral issues evident in exposure studies, which identified abnormalities in newborn neurobehavior including impaired orientation, autonomic regulation, and muscle tone, as well as reports of deficits in attention processing and an increase in symptoms of attention-deficit/hyperactivity disorder and oppositional defiant disorder in exposed children. Conversely, reviews by Maguire et al. [9] and Camden et al. [10] present more varied conclusions, suggesting influences from environmental factors and potential for developmental recovery [9,10]. A review from Welton et al. [11] concluded early developmental delays with possible normalization by ages 2–3, indicating that the adverse impact of prenatal opioid exposure may not always lead to long-term developmental issues. These reviews collectively underscore the complexity and varied nature of prenatal opioid exposure to early childhood development.

Considering the conflicting literature, this review seeks to clarify the relationships of prenatal opioid exposure in the context of opioid use disorder with early childhood outcomes as measured by The Bayley Scales of Infant and Toddler Development (BSITD) [12]. The BSITD is a standardized assessment tool used to measure the developmental progress of infants and toddlers, including those impacted by prenatal opioid exposure. Understanding the role of prenatal opioid exposure in the context of opioid use disorder along with the myriad of other factors that are related to child development (e.g. as posited in the ecological model of child development), is crucial as we develop clinical guidelines, policy and potential interventions. This review aims to synthesize the current literature on examining the question of what are the developmental outcomes in children who are prenatally exposed to opioids in the context of opioid use disorder, highlight the limitations in existing research, and propose recommendations for future studies. While prenatal exposure to opioids in the context of acute or chronic pain would be interesting, it is outside this study’s scope.

2. Methods

2.1. Procedures

The medical librarian developed search strategies for the concepts of opioids, pregnancy, children, and developmental effects or disorders and had them peer reviewed by another medical librarian (see the Supplemental Table). The search strategies were created using a combination of subject headings and keywords and were used to search PubMed, Embase, PsycINFO, CINAHL Plus, and Web of Science from date of database inception to 1 June 2018, when all initial searches were completed, and then again to 1 December 2021, when all searches were updated and repeated. In PubMed and Embase, the search strategies of published Cochrane reviews were used to create the search hedges for children, pregnancy, and opioids and the Cochrane human filter was used. In PubMed, Embase, and Web of Science, search filters were used to remove non-journal article results such as books and newspaper articles and remove non-research articles such as editorials and meeting abstracts. In PsycINFO and CINAHL, a database-supplied filter was used to limit the search results to journal articles only. The searches in PubMed, Embase, PsycINFO, CINAHL, and Web of Science yielded a total of 17,904 citations. These citations were exported to Endnote and 5,552 duplicates were removed using the Endnote deduplication feature. This left a total of 12,352 unique citations found in all database searches. The complete search strategy can be found in the supplementary material.

We placed 12,352 remaining unique references into Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia, available at www.covidence.org) for screening. Two raters, researchers knowledgeable about the research on prenatal exposure to opioids and other substances first independently screened titles and abstracts and then full-text references for eligibility criteria. A third reviewer aided in resolution of conflicts. We checked references of included studies for eligibility for inclusion (see Figure 1 footnote for eligibility criteria). A review of the extent of bias of these studies was also conducted independently by two researchers using the JBI Critical Appraisal Tools (available at https://jbi.global/critical-appraisal-tools). There were only two discrepancies on all ratings of all publications, and these were resolved in discussion.

Figure 1.

PRISMA flow diagr-am for systematic review: Effects of prenatal opioid exposure in the context of opioid use disorder. (1) Original research published in a peer reviewed English language journal (2) Prospective data (3) Human studies (4) Exposure during pregnancy to opioids (5) Measured child outcomes using standardized methods, clinical examination or academic assessments of children from 3 months through 18 years old. (3 months of age was chosen due to ability to assess neurodevelopmental outcomes at this time.) (6) Examiners were masked to children’s opioid exposure status (7) Included a comparison group (8) All years Exclusion (1) Samples with large proportion of co-exposures (2) Unmasked assessment of child outcomes (3) Retrospective samples (4) Animal studies (5) Acute analgesic exposure (e.g. during labor and delivery).

A visualization of the methods process following the PRISMA reporting guideline is available in Figure 1.

2.2. Data extraction and synthesis

We created a word document table in which relevant information for each study meeting criteria was entered (n = 7 studies). We piloted the data abstraction in this form independently on two studies not included in the final set of studies. The following information was abstracted: author, year of publication, study location, study design, type and number of participants, control or comparison group, type of opioid exposure, data collection measures, covariates included in regression analysis (included in final model or not) and socio-environmental factors, outcomes documented concluding that prenatal exposure leads to adverse child outcomes measures (bivariate analysis), outcomes documented concluding that socio-environmental factors lead to adverse child outcome measures (bivariate analysis) and outcomes documented concluding that prenatal exposure leads to adverse child outcomes when controlling for socio-environmental factors (multivariate analyses) (effect size, confidence intervals).

For each study, two researchers independently extracted data and the discrepancies identified were resolved through discussion and a third reviewer made the final decision. These data were then synthesized with basic counts of variables and outcomes in the tables.

2.3. Measures: the bayley scales of infant and toddler development

The Bayley Scales of Infant and Toddler Development (BSITD) is a standardized assessment tool to evaluate the developmental progress of infants and toddlers from birth to approximately 3½ years [13]. The BSITD has had several revisions to reflect evolving research and clinical practices in child development. The first edition, BSID, developed in 1969, primarily examined mental and motor development and included a Behavior Rating Scale for qualitative assessment of child behavior during testing [14]. In 1993, the second edition, BSID-II, was released, which updated the age norms and items, and expanded the Behavior Rating Scale to incorporate both social-emotional and adaptive behaviors [14,15]. Significant changes were made in the 2006 third edition, Bayley-III. This version added new scales, to cover five key domains: Cognitive, Language (with subtests for receptive and expressive communication), Motor (including fine and gross motor subtests), Social-Emotional, and Adaptive Behavior, and aimed for expanded cultural sensitivity. The latest version, Bayley-4, was released in 2019 and continues to examine the same five domains and provides revised norms [16]. The BSITD is commonly used as a tool for healthcare professionals for the assessment of a child’s developmental progress, identification of developmental delays or concerns, and planning of appropriate interventions or support [15]. The application of the BSITD highlights its role as a key standard in the field of early childhood development. Based on its widespread use and appropriateness for this population, this review selected the BSITD as our outcome measure.

3. Results

This review yielded five overlapping areas of interest in the literature: definition of prenatal exposure, comparison group, simple opioid exposure effects, covariates and errors of inference. The way in which prenatal exposure to opioids was defined in each study varied as did the way in which the comparison group/control group was defined. To understand the relationship between prenatal opioid exposure and child development, the studies utilized different statistical methods including univariate analyses to assess the simple opioid exposure effects as well as multivariate analyses to elucidate the effects of prenatal opioid exposure and other covariates. Lastly, there is the risk of errors on inference that exists in small scale retrospective studies such as those studies included in this review.

3.1. The publications

This review ultimately included seven articles which were published between 1989 and 2021 (see Table 1). They included five studies from the United States and one each from Finland and New Zealand. Timing of maternal recruitment varied and none of the studies required recruitment by the 30^th week EGA. None of the seven studies reported or accounted for the number of prenatal care visits the women attended. The seven studies utilized either the BSID, the BSID-II, or the Bayley-III and age at testing varied from 4 to 36 months.

Table 1.

Studies: samples, location of the study, risk of bias and child measures examined.

						Child Outcome Measures
Publication						Bayley Scale of Infant Development (BSID)
Author	Year Published	Opioid	Maternal Age (mean years)	Sample Size	Location and Risk of Bias	Version	Components	Age of Testing (months)
Bakhireva et al.	2019	MTD or BUP	Exposed: 28.3 Non: 26.8	Enrolled children: 184 Completed: 78 (42 exposed; 36 non)	USA (Albuquerque) Risk of Bias= moderate	Bayley-III	(1) Motor (2) Cognitive (3) Language	5–8 months
Bernstein & Hans	1994	MTD	Exposed: 27.1 Non: 25.3	Enrolled Children: 89 (42 exposed; 47 non) Completed: 69 (28 exposed; 41 non)	USA (Chicago) Risk of Bias= high	BSID	(1) Mental	24 months
Hans	1989	MTD	Exposed: 27.3 Non: 25.4	Enrolled children: 89 (42 exposed; 47 non) Completed: 74 (30 exposed; 44 non)	USA (Chicago) Risk of Bias= high	BSID	(1) Psychomotor (2) Mental	24 months ±2 weeks
Hans & Jeremy	2001	MTD	Exposed: 27.1 Non: 25.8	Enrolled children: 89 (42 exposed; 47 non) Completed: 78 (33 exposed; 45 non)	USA (Chicago) Risk of Bias= high	BSID	(1) Mental (2) Psychomotor	4, 8, 12, 18, 24 months
Messinger et al.	2004	Opiates (per maternal admission or positive meconium)	Exposed: 30.35 Non: 27.67	Enrolled: 1388 Completed: 1227 (98 exposed; 1129 non)	USA (Providence, RI; Miami, FL; Memphis, TN, Detroit, MI) Risk of Bias= moderate	BSID-II	(1) Mental (2) Psychomotor	1, 2, 3 years
Salo et al.	2009	BUP	Exposed (parental care):28.29 Exposed nonparental care: 33.93 Non: 29.08	Enrolled mothers: 28 Completed: 34 (21 exposed; 13 non)	Finland Risk of Bias= high	Bayley-III	(1) Cognitive (2) Language (3) Socio-Emotional	3 years ±3 months
Levine	2021	MTD	Exposed: 29 year 9 month Non: 31 years 6 months	Enrolled children: 210 (100 exposed; 110 non) Completed: 200 (92 exposed; 108 non)	New ZealandRisk of Bias= moderate	BSID-II	(1) PDI (2) MDI	2 years

Note: BSID: Refers to the original Bayley Scales of Infant Development; BSID-II: The second edition of the Bayley Scales of Infant Development; Bayley-III: The third edition of the Bayley Scales of Infant and Toddler Development.; Bayley-4: The fourth and most recent edition of the Bayley Scales of Infant and Toddler Development.

Of note, 3/7 studies [17–19] examined the same data for children exposed prenatally to methadone. Women were recruited between 1978 and 1982. The same control group was also examined in all three studies. However, the final samples in the three studies varied slightly based on some missing data of the outcome measures.

All studies demonstrated enough information to assess for bias in most of the predefined categories. Overall, (see Table 1), the studies collectively had a moderate to high risk of bias driven by issues such as lack of clarity in eligibility criteria, lack of clarity in consecutive inclusion of participants, confounding factors not fully considered and addressed. Based on the heterogeneity of the methods, metrics and outcomes as well as the generally poor data quality, a meta-analysis was not possible. As such, a narrative synthesis was conducted.

3.2. Definition of prenatal exposure

All studies included in this review focused on mother/baby dyads where there had been prenatal exposure to opioids (see Table 2). Recruitment and inclusion criteria varied between studies. The type of opioid exposure was different. The majority of the studies specified that methadone was the primary substance (n = 4), one study recruited women whose primary substance was buprenorphine, one study recruited women using either methadone or buprenorphine or both, and one study did not specify type of opioid. The source of the opioids also varied among the studies. Five studies recruited from treatment programs where women were prescribed methadone, one study included women in a maintenance program or with illicit opioid use and one study did not specify if the opioid exposure was illicit or part of a treatment program. Of the studies that included women receiving medication to treat opioid use disorder (MOUD) programs, only one study specified that 62% of the women were engaged in MOUD at the time of pregnancy, while 14% began MOUD during the first trimester. Three studies noted that some women were on MOUD at the time of pregnancy while others started during pregnancy. They did not report how many women were on MOUD versus how many started during pregnancy or when during the course of pregnancy medication was initiated. Two studies did not specify when MOUD was started in relation to the pregnancy. Women in all seven studies were not excluded for use of other substances and/or use of other opioids while receiving MOUD.

Table 2.

Type of opioid exposure examined.

Publication	Type of Opioid	Medication to treat Opioid Use Disorder (MOUD)
Bakhireva et al. 2019	Methadone or Buprenorphine With or without exposure to other opioids	MOUD Unspecified start of MOUD
Bernstein & Hans, 1994	Methadone With or without exposure to other opioids	MOUD throughout pregnancy or during pregnancy
Hans 1989	Methadone With or without exposure to other opioids	MOUD throughout pregnancy or during pregnancy
Hans & Jeremy, 2001	Methadone With or without exposure to other opioids	MOUD throughout pregnancy or during pregnancy
Messinger et al. 2004	‘Opiates’ by maternal admission or meconium metabolite	Unknown
Salo et al. 2009	Positive urine drug screening at birth for Buprenorphine (illicit or MAT)	MOUD or illicit Unspecified MOUD start
Levine 2021	Methadone With or without exposure to other opioids	MOUD 62% at pregnancy 14% during 1st trimester

3.3. Comparison group

In order to study the effects of prenatal opioid exposure on early childhood developmental outcomes, studies compared the exposure group to a control/comparison group(s). There was great heterogeneity between how control/comparison group(s) were selected. Only one study [20] had a matched comparison cohort. They matched based on race/ethnicity, gender of infant and gestational age of infant.

In order to isolate the effect of prenatal opioid exposure, all studies took into consideration potential covariates including, but not limited to tobacco use, estimated gestation age at delivery, race, birth weight, maternal age, maternal education, socio-economic status, co-occurring mental health and prenatal care. Of the seven studies, two did not address, report, or account for tobacco use, three did not assess or report estimated gestation age at birth or co-occurring mental health diagnoses. One study did not consider race. Every study considered the potential impact of co-occurring substance use, birth weight, maternal education, and socio-economic status on child development, as measured by the BSITD.

3.4. Simple opioid exposure effects

As described in Table 4, all seven studies conducted hypothesis testing to examine the association of prenatal opioid exposure with child developmental outcomes as measured by the BSITD. Of the seven studies, five found that prenatal exposure was associated with poor BSITD scores, three studies utilized the original version of the BSID, two studies used the second edition (BSID-II) and two studies employed the third edition (Bayley-III) which was published in 2006. Two studies found no difference in scores between the prenatally exposed group and the control group. Five of the studies found an association between prenatal opioid exposure and poorer BSITD outcomes. These five studies included all three versions the BSITD and spanned ages 4 to 36 months.

Table 4.

Definition of Tobacco from Studies.

Publication
Author	Year Published	Tobacco Definition
Bakhireva et al.	2019	Tobacco use among participants was measured through a nicotine metabolites panel which included nicotine, cotinine, 3-hydroxycotinine, nornicotine, and anabasine. The frequency of these measurements was not explicitly stated, but it is mentioned that these panels were administered as part of the study design, (p. 70 bottom right paragraph). Reported as any tobacco use (treated for NOWS vs. not treated for NOWS) tobacco was excluded from the control group
Bernstein & Hans	1994	Tobacco use, specifically, isn’t operationalized or defined as a direct variable of study. The research focuses on the broader context of prenatal drug exposure, primarily to methadone, and its relationship with developmental outcomes in children at the age of 2 years. See limitations and screenshot below.
Hans	1989	This study acknowledges that women in methadone maintenance programs frequently use additional substances, but tobacco is not explicitly mentioned. The study doesn’t operationalize tobacco use separately but considers the overall impact of drug use on maternal health and prenatal development. Maternal drug use status was determined through self-reporting on the University of Washington Pregnancy and Health Questionnaire and by repeated urine toxicology screening during pregnancy.
Hans & Jeremy	2001	Tobacco use was operationalized into three categories: abstaining from tobacco, smoking less than a pack a day, and smoking at least a pack a day. These categories were used to classify the tobacco exposure level of the participants.
Messinger et al.	2004	Tobacco use was operationalized and quantified in terms of the number of cigarettes consumed daily. The study reported the average number of cigarettes smoked daily for different groups (opiate exposed vs. not opiate exposed), providing both the mean and standard deviation for these figures.
Salo et al.	2009	Tobacco use was operationalized through self-reports of substance use by the mothers, whose main drug of choice was buprenorphine. The data indicate that out of 21 mothers in the study, 20 (95.2%) reported using tobacco.
Levine	2021	Tobacco use during pregnancy was operationalized and quantified by categorizing the amount of smoking into three categories:  (1) No smoking (abstinence)  (2) Smoking 1–9 cigarettes per day  (3) Smoking ≥10 cigarettes per day

3.4.1. Mental development index (MDI)

The Mental Development Index (MDI) is a standardized measure used in the BSITD to assess cognitive development in children. The original BSID and the BSID-II produce this index [16]. All five studies that utilized the BSID or the BSID-II evaluated MDI scores at 24 months. At this time point, four of the five studies found that MDI scores were not related to prenatal opioid exposure [17–20], while one did find an association [21]. Two studies also evaluated MDI scores at other time points with one finding an association at 12 months [20] and one finding an association at 18 months [19].

3.4.2. Psychomotor developmental index (PDI)

The Psychomotor Developmental Index (PDI), is a component of the BSITD that evaluates motor skills, conceptual, and psychological aspects [22]. The original BSID and the BSID-II assessed a standardized measure for motor development. Of the seven studies, four evaluated the relationship between prenatal opioid exposure and PDI while one study that utilized the BSITD, did not include PDI as an outcome measure [17]. All four studies that included the PDI found that prenatal opioid exposure was associated with lower PDI scores at 24 months [18–21]. Two studies examined their samples at additional time points and found no relationship at 4, 8, 12, 18, 36 months [19,20] or PDI mean across age (4–24 months [19] and 12–36 months [20]).

3.4.3. Infant behavior record (IBR)

The Infant Behavior Record (IBR) is a component of the BSITD, used to assess a range of infant behaviors during interactions with an examiner, providing insights into the infant’s developmental status [15]. The original BSID had a third section, the Infant Behavior Record (IBR), where the examiner would rate a series of behaviors witnessed during the interaction. The three studies that utilized the BSID as their outcome measure, included some individual and some combinations of the items from the IBR (of note, all three studies examined the same sample) [17–19]. One study did not specify what items were evaluated but found that tension, gross motor coordination, and fine motor coordination at 24 months were related to prenatal opioid exposure [18]. This same study found that activity level and attention span at 24 months were not related to prenatal opioid exposure [18]. The other two studies found that attention was not related to prenatal opioid exposure at 4, 8, 18, 24 months or across the age range, but found an association at 12 months [17,19]. Social resiliency was also not related to prenatal exposure at 24 months [17]. Coordination was not related to prenatal exposure at 4, 8, 12 or 18 months, but was related at 24 months and across ages [19]. Activity level was not related to prenatal exposure at 8, 12, 18, 24 months or across ages, but was associated at 4 months [19].

3.4.4. Behavior rating scale (BRS)

The Behavior Rating Scale (BRS) serves as a specialized tool for assessing the behaviors of infants and young children during developmental evaluations. This scale examines key areas including orientation/engagement, emotional regulation, motor quality, and attention/arousal, offering crucial insights into developmental progress [23] The BSID-II added the BRS, which incorporates 30 items that evaluate the behavior of the child during the exam. It has four main domains: 1) orientation/engagement, 2) emotional regulation, 3) motor quality, 4) attention/arousal (for 1- to 5-month olds) [24]. Of the two studies that utilized the BSID-II, only one included analysis of the BRS [20]. Prenatal opioid exposure was associated with BRS at 24 months, however, no association was found at 12 months, 36 months, or across all ages [20].

3.4.5. Bayley-III

The Bayley Scales of Infant and Toddler Development, Third edition (Bayley-III) evaluates five domains which include adaptive behavior, cognitive, language, motor, and social-emotional [25]. The social-emotional domain is derived via a questionnaire that is filled about by caregivers of the infants and was excluded from this paper. Two studies utilized the Bayley-III. The first study administered the Bayley-III between 5 and 8 months and found no difference between the prenatal exposed group and the control group in the cognitive, language and motor domains [26]. The second study found that at 3 years of age, the cognitive and language domains were significantly negatively associated with group status (prenatal exposure vs. control) [22].

3.5. Covariates: socio-environmental factors evaluated

To further interpret the relationship between prenatal opioid exposure and childhood outcomes, four of the seven studies examined other covariates including socio-environmental risk factors [19–22]. After controlling for these variables and risk factors, prenatal opioid exposure no longer had a statistically significant relationship with BSITD scores. The studies controlled for a variety of maternal, infant, and socio-environmental factors. The socio-economic factors that were significant were birth weight (two studies) [19,20] social-environmental risk [19], maternal care (vs. foster care, two studies) [19,22] poverty status [20], Peabody Picture Vocabulary Test-Revised Vocabulary [20], HOME Scores [20], gestational age (two studies) [21,22], caregiver age [22], self-efficacy [22], breastfed [21], and non-punitive parenting [21]. Of note, all four studies also examined the influence of other illicit psychoactive substances and tobacco (see Table 4 for operationalized definitions of tobacco) and found that none of them were significantly related to BSID scores [19–22].

3.6. Errors of inference

Retrospective studies and studies utilizing convenience samples with small sample sizes that fail to control for conducting multiple tests of significance are prone to errors of inference, notably, both Type I and Type II errors. On the one hand, multiple inferential tests at a per comparison error rate of .05 will yield rejections of the hypothesis of no differences between groups when in fact that is not the case were the two populations actually compared – a Type I error. Conversely, it is also the case that a small sample size will not allow for the detection of a true difference because of inadequate power for the test of significance – a Type II error. So it is likely that some effects that are reported to be ‘significant’ in the reviewed studies are spurious, and it is likewise possible for the studies to have missed detection of a true difference between the groups due to low power.

This problem can be compounded by low reliability associated with the predictor variables. For example, tobacco use was variously defined in the seven studies (see Table 4), with several studies [19,21,27,28] using crude classification of the observations [29] a measurement approach known to lower reliability and hence validity (as the upper-bound of a validity coefficient is the index of reliability of a measure known to be measured with error).

4. Discussion

Based upon a review of eligible studies, this review found that after adjusting for socio-environmental factors, there was no association between prenatal opioid exposure and child developmental outcomes as measured by the BSITD.

4.1. Use of BSITD to measure aspects of child development

There is no universal way to measure child development and in fact there are multiple domains to consider. Limitations also include limited generalizability of measures and outcomes, as well as challenges inherent in measuring socio-environmental factors accurately. This review does not serve to compare different methods of assessing development, but instead chose to evaluate the results of the utilization of one common development assessments, the BSITD.

The BSITD is widely recognized as the standard measure for early childhood development and evaluating infants and toddlers’ cognitive, language, and motor skills. It serves as a critical tool in understanding at-risk populations, especially children exposed prenatally to opioids [16]. While the BSITD is a gold standard measure, each edition – from the original to the Bayley-4—reflects revisions in the field of child development, and due to the nature of the revisions, it is difficult to compare scores across all versions [30,31].

The review emphasizes that despite the differences in BSITD versions, studies on the effects of prenatal opioid exposure on early childhood development consistently report no significant impacts. This review included three of the four versions. No studies utilizing the fourth edition met the inclusion criteria. Initially, this review set out to examine if there was a difference in results between BSITD versions as some versions have been found to have significantly higher scores for certain domains [31]. However, the results of this review did not warrant further investigation since the effect of prenatal exposure did not remain significant in any of the final models. This review underscores the importance of recognizing the nuanced differences between the various versions of the BSITD. Future research should continue to consider these variations when using the BSITD to measure child developmental outcomes and also strive to understand the continued contradictory findings in the literature by examining the different domains of child development under consideration.

4.2. Prenatal exposure effect

In crude (unadjusted) analyses, this review found inconsistent associations between prenatal opioid exposure and poorer child development. However, when these associations were found, they did not persist in any models adjusted for confounders (including demographics and other socio-environmental factors).

4.3. Implications for future work and recommendations

This review has several implications for addressing the challenges of prenatal opioid exposure in the context of opioid use disorder. In research, there is a clear indication for more studies that not only investigate the direct effects of prenatal opioid exposure on early childhood development but also comprehensively measure and control for socio-environmental factors linked to child development. Developing methodologies that can accurately distinguish the impacts of opioid exposure from other variables will be crucial in these studies. Future-forward studies would benefit from the use of longitudinal designs, diverse populations, and innovative statistical techniques to better isolate the effects of prenatal opioid exposure from those of confounding variables. Future studies should examine the extent to which prenatal exposure to opioids in the context of acute and chronic pain have similar challenges and opportunities when determining child outcomes.

4.3.1. Need for clearcut definition of opioid exposure

Future studies need to carefully define prenatal opioid exposure. As seen in this review, there is no universal definition of prenatal opioid exposure. While prenatal opioid exposure does is commonly used to signify any fetal exposure to any opioids, it is also important to recognize the opioids are common during labor and delivery and during certain prenatal procedures. Therefore, clearly defining groups with ‘prenatal opioid exposure’ by outlining the type of opioid (e.g. fentanyl, methadone, oxycodone, buprenorphine, heroin, fentanyl etc.), the source of the opioids (illicit, prescribed, MOUD program, etc.), duration, quantity, frequency of use, and timing of use will help new research clearly evaluate the role of prenatal opioid exposure on early childhood development.

Buprenorphine and methadone are commonly recognized by Substance Abuse and Mental Health Services Administration (SAMHSA) and ACOG as the most effective and safest way to manage a pregnant woman with OUD [32]. The majority of the included studies in this review recruited from MOUD programs. However, even if engaged in treatment programs, there often is co-occurring opioid use, a return to use, and/or overdose. Studies need to distinguish between prescription use, illicit use and MOUD, and distinguish when there is co-occurring use. Women with an OUD receiving integrated care during pregnancy have better birth outcomes than women not receiving care [33]. These better birth outcomes may also affect child development, so it is crucial that we carefully define the population under investigation.

Future studies need to carefully consider which population they intend to study, choose a methodology that allows them to follow and capture the prenatal opioid exposure and explicitly define the group for dissemination. Factors to consider are type of opioid used, co-occurring substance use, route of use, length of use and dose. While this review consistently found no adjusted association between prenatal opioid exposure and child outcomes as measured by the BSID, the larger body of literature today is inconsistent perhaps attributed to the varied definitions of ‘prenatal opioid use.’ Studies are comparing heterogeneous populations and/or combining very different populations under a catch all of ‘prenatal opioid use.’

4.3.2. Selection of appropriate control/comparison group

This review highlights the importance of selecting an appropriate control/comparison group. Randomization integral to study design would be ideal to maximize the potential to balance potential confounders across exposure groups but is not likely in studies such as these, given numerous considerations including ethics, costs, feasibility and resource requirements. Matching is a potential strength in retrospective studies, but the critical variables on which matching should take place are frequently unknown. Moreover, matching may end up discarding cases from one or both groups, narrowing the sample differences that actually exist in the populations of interest. It should be understood in the context of matching that equating two groups on some characteristics (say, age and marital status) is not matching – matching takes places on a case-by-case basis, not on a group basis as would seem to have occurred in studies in Table 3. Attempting to somehow equate groups by the use of covariates is likewise fraught with problems, as, again, the choice of covariates is critical, and numerous covariates will reduce power. Propensity scoring may be a useful alternative in this case.

Table 3.

Outcomes, Covariates and Significant Effects.

Publication	Bayley Version	Prenatal Opioid Exposure Effect Found	Covariates	Prenatal Opioid Exposure Effect Remained Significant	Significant Covariates
Bakhireva et al. 2019 [26]	Bayley-III	No	N/A	N/A	N/A
Bernstein & Hans, 1994 [17]	BSID	No	N/A	N/A	N/A
Hans 1989 [18]	BSID	Yes	N/A	N/A	N/A
Hans & Jeremy, 2001 [19]	BSID	Yes	opioid exposure marijuana exposure tobacco exposure cocaine exposure birth weight social-environmental risk	No	birth weight (PDI) social-environmental risk (MDI)
Messinger et al. 2004 [20]	BSID-II	Yes	infant age infant birth weight infant HIV exposure maternal care (biological mother) maternal education maternal age poverty status PPVT-R vocab SES HOME ethnicity tobacco exposure geographical site opiate exposure marijuana exposure alcohol exposure	No	infant birth weight (MDI, PDI, BRS) maternal care (biological mother) (MDI, PDI) poverty status (BRS) PPVT-R vocab (MDI, BRS) HOME (PDI, BRS)
Levine 2021 [21]	BSID-II	Yes	other licit/illicit substances maternal education gestational age breastfed parental stress non -punitive parenting parental involvement	No	gestational age (PDI, MDI) breastfed (PDI) non -punitive parenting (PDI)
Salo et al. 2009 [22]	Bayley-III	Yes	gestational age birth weight birth height number of placements caregiver age caregiver education sensitivity structuring nonhostility nonintrusiveness self-efficacy exposure status parental vs. foster care	No	gestational age (cognition, language) caregiver age (language) self-efficacy (language) parental vs. foster care (language)

Note. Salo et al. (2009) contains an apparent typo when it states on p. 253 ‘For the language Scale, gestational age along with maternal age and perceived Maternal Self-Efficacy accounted for significant variance; in addition, group status (drug exposed versus nonexposed) remained significant.’ where group status (parental versus foster care) should have been significant and group status (drug exposed versus nonexposed) should have been nonsignificant to be consistent with the findings reported in Table 4 on page 253.

Future research should carefully consider different methodologies with sufficient statistical power in order to appropriately investigate the effects of prenatal opioid exposure. Given the multifactorial nature of child development, thoughtful consideration is needed regarding whether it is appropriate to have a single comparison group or multiple comparison groups. Depending on how prenatal opioid exposure is operationalized, selection of an appropriate comparison group(s) will differ across future studies. However, ensuring that the comparison group(s) are appropriately selected and well defined will allow future research to elucidate the effect on child development.

4.3.3. Inclusion of appropriate covariates

This review highlights the need to adjust for appropriate confounders in the final analysis, as all four studies that included a multivariate analysis found that prenatal opioid exposure was not related to child development. Instead, it was a mix of other socio-environmental and health variables that were more associated with child development.

When considering child development outcomes it is critical that there is inclusion of known co-occurring factors. Tobacco exposure during pregnancy is linked to decreased language functions and negative impacts on cognition and motor abilities in children [34,35]. Gestational age at delivery also plays a pivotal role, with preterm babies showing varied developmental outcomes [36]. Furthermore, higher maternal education levels correlate with better cognitive outcomes in the general population of children [37,38], and higher socioeconomic status is linked to improved cognitive and social development [39]. Additionally, maternal mental health significantly influences using crude classification of the observationscognitive develo pment in infants [40].

Additionally, this review found that a combination of socio-environmental and other factors were predictive of BSITD scores. These factors included: birth weight, gestational age, socio-environmental risk, caregiver (biological mom vs. foster care/other caregiver), caregiver age, caregiver self-efficacy, poverty status, maternal vocabulary, HOME inventory, breastfed, and non-punitive parenting. While some of these factors are nonmodifiable, others have great potential for intervention. Further research is needed to better understand the relationship of these factors with child development in children prenatally exposed to opioids, identify other factors not highlighted here, and help guide interventions for clinical practice.

4.3.4. Prenatal recruitment

None of the studies included in this review required prenatal recruitment before 30 weeks estimated gestational age. Enrollment before 30 weeks gestation is needed to allow time for MOUD to be effective in treating OUD before birth. It is also needed for adequate prospective data collection to measure the amount, timing and frequency of exposure to opioids and other factors. To understand the effect of prenatal opioid exposure on child development, future research needs to have an accurate picture of prenatal exposure. While this can be done in a retrospective manor, future research should consider earlier recruitment in order to help understand the nuances of type, route, dose of opioids and how this influences child development. Likewise, prenatal recruitment would also allow for a better understanding of the other factors that are affecting pregnancy and fetal development such as prenatal care, tobacco use during pregnancy, and other substance use during pregnancy.

5. Significance

Regarding public policy, the findings highlight the importance of policies grounded in a comprehensive understanding of the spectrum of maternal opioid use and opioid use disorders. This encompasses a range of behaviors from opioid use without a disorder, opioid use for acute or chronic pain, misuse of prescription medications, untreated opioid use disorder and treated opioid use disorder. Assumptions about the risks associated with any prenatal opioid exposure have led to hasty decisions, such as the involvement of child protective services and the separation of infants from their mothers. To this end, effective policies require a thorough consideration of the diverse impacts of these substances, including factors that impact child development and maternal health conditions to provide better and more effective interventions. Given the important overarching role that Socio-economic status plays in child development, supportive polices to lift families out of poverty are imperative.

6. Conclusion

Based on the findings of this review, there is not enough evidence to suggest that prenatal exposure to opioids negatively impacts early childhood development. In addition, this review supports a growing body of evidence that socio-environmental factors may play a more substantial role in children exposed to opioids in utero. Child development is multifaceted. In response to the growing opioid crisis, it is imperative that we better understand the relationship between prenatal opioid exposure and child development. The needed recognition to focus on socio-environmental factor support of the family over the prenatal exposure status of the child has wide ranging implications for education systems, community health resources, and long-term economic outcomes.

7. Expert opinion

Our field continues to be challenged to learn from our shared history and change the narrative that is intertwined with drug policy and sociocultural views of women, pregnancy and children. In 2001 Frank et al. [41], provided a compelling review of the literature summarizing prenatal exposure to cocaine and crack that that provided a road map for future studies that included: (1) define appropriate comparison group or groups so that the questions being posed about outcomes of prenatal substance exposure can objectively be answered; (2) recruit participants prospectively during ideally early pregnancy rather than rely on postnatal records or retrospective data that has more biases; (3) use masked assessment to reduce bias and (4) exclude participants exposed in utero to other substances. In 2018 Jones et al. [6], affirmed that contemporary researchers should apply these lessons to the study of neonatal abstinence syndrome and also extended this essential list of methods to include modern statistical approaches such as adjustment for multiple comparisons, causal inference assessment, and propensity scores, changes in how statistical significance is determined should be incorporated into study design and analysis. Based on the findings from this systematic review, researchers of prenatal opioid exposure can no longer ethically avoid such calls for advancing science. Identifying developmental problems in children and adolescents prenatally exposed to opioids needs to focus on causal factors. As yet, studies have not shown a simple linear cause-and-effect relationship between either NAS diagnosis or prenatal opioid exposure and compromised developmental outcomes.

While the future is challenging to predict, the trajectory of the science needs to arc toward a focus on the multiple factors that can help mitigate risk and support healthy outcomes for children and families. For example, while traditionally mechanistic studies examine the primary prenatal exposure of interest or a ‘stressor,’ future studies to consider the cumulative effect of multiple substance exposures and prenatal psychosocial stress on fetal programming of stress reactivity and regulation. A focus too needs to include paternal contributions to long-term outcomes of children. There is a need for researchers to recognize the powerful responsibility of data interpretation for policy makers. If socio-environmental factors outweigh prenatal exposure status with regard to contribution to long-term outcomes of children then federal incentives are needed for states to provide robust early intervention and in home services for all children with socio-environmental risks in order to enhance infant maternal bonding and support healthy infant development. It may be advantageous to utilize funding from the National Opioid Settlement [42] to support such recommendations.

In summary, pregnant and postpartum individuals with OUD and their children need support and treatment and not punishment. As a research and practice community we should strive to advocate for and provide dyads with the highest level of evidence-based treatment and support to promote healthy outcomes.

Article highlights.

• There is inadequate evidence to support the assertion that prenatal opioid exposure is associated with detrimental early childhood developmental outcomes.

• There is evidence to support the assertion that, in particular, socio-environmental factors (for example, home environment, socio-economic status) may be associated with adverse child developmental outcomes in prenatally exposed children.

• Further studies should clearly define opioid exposure groups, include appropriate non-exposed comparator groups, and sufficiently measure potential confounders of child developmental outcomes (including socio-environmental factors). Such actions will advance the evaluation of the role of prenatal opioid exposure on early childhood development.

• There are major health, social, and legal ramifications that result from overattributing harms to the child from prenatal opioid exposure. Focus needs to be given to approaches for improving the resilience factors that support healthy child development for children with prenatal opioid exposure.