Substance use in pregnancy is a growing public health concern. In 2019, 5.8% of U.S. pregnant mothers reported using illicit drugs, 9.5% reported using alcohol, and 9.6% reported using tobacco products (Substance Abuse and Mental Health Services Administration, 2020). Infants with prenatal substance exposure (PSE) are at increased risk of child maltreatment reporting and foster care placement (Rebbe et al., 2019a). Moreover, mental health conditions including substance use disorder are the leading cause of pregnancy-related deaths in the U.S. (Trost et al., 2022). One way to improve outcomes for infants with PSE and mothers with substance use in pregnancy is through prompt identification of the exposure and connection to services that enhance protective factors and support early development (Bada et al., 2012).

To aid in early identification and postnatal support for this population, the Child Abuse Prevention and Treatment Act (CAPTA), a primary U.S. child welfare law, added a mandate in 2003 that hospital personnel notify child protective services (CPS) of infants “born with and identified as being affected by illegal substance abuse or withdrawal symptoms resulting from prenatal drug exposure”. In 2010, CAPTA added fetal alcohol spectrum disorder (FASD) and in 2016, removed the term “illegal” (Lloyd et al., 2019). Despite inclusion in a child welfare law, these provisions do not establish a federal definition of child abuse or neglect that includes PSE. That said, CAPTA’s implications include that state child welfare agencies are tasked with 1) ensuring that hospitals identify infants with PSE and 2) coordinating multisystem responses to identified infants and mothers through a Plan of Safe Care (called a “Family Care Plan” in Connecticut [CT], the setting for the current study, and used for the remainder of this manuscript). Accurate, early, and equitable identification of infants with PSE and their mothers, coupled with non-stigmatized, non-punitive intervention is central to any public health response, including with this population.

Before an infant is born, the primary means for identifying substance use in pregnancy include self-report (either spontaneously or in response to screening questions), other report/medical record review, or toxicology test (Behnke & Smith, 2013). After delivery, infant withdrawal symptoms may indicate exposure, but are often corroborated with toxicology (Behnke & Smith, 2013). Each approach to detection has limitations. Objective methods such as toxicology and withdrawal symptoms differ in accuracy depending on substance type, while self-report may be unreliable, particularly if the person perceives risk of criminal or civil penalties (Ondersma et al., 2019). Additionally, thirty years of research has documented disproportionality in hospital providers’ detection of, and response to, substance use in pregnancy (e.g., Chasnoff et al., 1990). More recently, Rebbe et al. (2019a, 2019b) linked hospital birth records with child welfare records in Washington State to understand rates of diagnosed prenatal substance exposure and the association between diagnosis and child welfare reporting. Native American infants were diagnosed at four times the rate of white infants and Black infants at 1.5 times the rate of white infants. These rates further varied by substance type. Given these challenges, providers and policymakers continue to seek effective, reliable, and equitable means of identifying prenatal substance exposure.

In March 2019, Connecticut implemented a statewide CAPTA response that relies on hospitals to identify infants exposed in utero to a range of legal and illegal substances (including alcohol), infants showing withdrawal symptoms, or those diagnosed with FASD. Upon identification, the Department of Children and Families requires providers to submit a “notification” via an online submission portal that alerts the Department of the infant’s birth but withholds any identifying information to prevent CPS surveillance (hereafter, the policy as implemented in Connecticut is referred to as “CT CAPTA” and the process created under CT CAPTA is referred to as a “CAPTA notification”). This approach aims to address certain barriers to accurately measuring population prevalence of substance use in pregnancy including those related to punitive responses by the child welfare system.

Of relevance to the current study, CT CAPTA does not clarify how the identification of PSE is made. Across the state, providers use a variety of methods to identify these infants including, in order of reported frequency, maternal urine toxicology, infant urine toxicology, maternal self-report, indication in the medical record, infant meconium analysis, neonatal withdrawal symptoms, behavioral disturbance, and growth/developmental disturbance (Hasan, 2019). Thus, the notification approach remains vulnerable to most common challenges to accurate identification. Although beyond the scope of this paper, but perhaps of interest to the reader, the CAPTA notification form also asks three questions to providers pertaining to safety concerns and maternal substance use: (a) Has the child tested positive because of maternal substance misuse? (b) Is there a concern that mother’s substance use will impact parental functioning? and (c) Does the family present with suspicions of abuse or neglect? If a provider answers affirmatively to any of these questions or indicates that no Family Care Plan is in place, the individual making the notification is instructed to use a separate system to report suspected maltreatment. If the provider answers no to these items and the dyad has a Family Care Plan, the system does not prompt the user to make a CPS report and the anonymous information is submitted to the state.

Although CT CAPTA aims to leverage its notification approach to increase identification and reduce unnecessary CPS intervention, it is unknown whether population prevalence of substance use in pregnancy as detected by CAPTA notifications aligns with prevalence of self-reported substance use in pregnancy. To begin understanding the alignment between the mother-infant dyads identified via CAPTA notifications and population-level self-report data on substance use in pregnancy, we aimed to compare CAPTA notifications against a special supplement to the Pregnancy Risk Assessment Monitoring System (PRAMS) survey that measured retrospective self-reported substance use during pregnancy (hereafter, this specific survey as implemented in Connecticut is referred to as “CT PRAMS”; the national survey mechanism is referred to as “PRAMS”). Although retrospective self-report presents its own validity issues (Ramos et al., 2020), the fact that the CT PRAMS survey was a population-level public health instrument, administered outside of the hospital within two to six months of the infant’s birth, and with no real or perceived involvement by CPS, it offers a unique and important method for capturing population-level prevalence of substance use in pregnancy.

Current federal CAPTA policy requires states to identify infants born affected by prenatal substance exposure. In 2019, CT implemented a statewide CAPTA response that relies on individual hospital policy or practices to identify these infants. To begin understanding the alignment between the mother-infant dyads identified via CT CAPTA and population-level self-report data on substance use in pregnancy, we aimed to compare CAPTA notifications against a special supplement to the PRAMS survey that measured retrospective self-reported substance use during pregnancy.

Our study contributes three key findings to research on policy-driven identification of substance use in pregnancy: (1) hospital approaches to detection dramatically under-identify alcohol use in pregnancy compared to self-report, (2) racial disproportionality persists, but may be the result of under-identification of white and Hispanic mothers, rather than over-identification of Black mothers, (3) hospital approaches to detection identify more cannabis and polysubstance use versus self-report.

Regarding our first key finding, CT PRAMS documented alcohol consumption at over 800 times the rate compared to CAPTA notifications (8.1% of mothers vs. 0.1% of mothers). These findings are consistent with previous research documenting much lower rates of alcohol exposure identified in hospital records (Rebbe et al., 2019a, 2019b) compared to self-reported alcohol use in pregnancy (Gosdin et al., 2022). This observed under-reporting of low levels of alcohol exposure begs the question as to whether hospital personnel are identifying the use and not completing a notification, or whether the use goes undetected. Previous research documents discrepancies between provider responses to alcohol use compared to drug use, suggesting that the practice gap involves failure to document, not failure to detect. A survey of health care and legal system professionals in Iowa found that providers were significantly more likely to take several actions including consulting social work and reporting to CPS for drug use compared to alcohol use (Chasnoff et al., 2018).

The CAPTA notification and PRAMS datasets aligned more closely when three or fewer drinks per week categories were excluded from analysis via the PRAMS-4+ subsample, suggesting that providers submitted CAPTA notifications for heavier drinkers only. This may indicate that providers used the notification system in instances of suspected alcohol use disorder; perceiving the notification as an apparatus of child welfare or substance use concern rather than public health surveillance. Alternatively, it may point to the fact that routine toxicology tests in healthcare settings exclude alcohol and, unless a provider is alarmed by the level of alcohol use and orders a separate test, the use will go undetected. In the first scenario, providers are making an implicit risk assessment and acting based on that assessment. In the second, established practices systematically limit alcohol identification and response. Both scenarios, however, illuminate the need for continued emphasis on the importance of monitoring alcohol use in pregnancy and implementing non-punitive interventions to minimize harm to the mother and infant.

Our second key finding documented racial disproportionality among CAPTA notifications compared to self-reported substance use. According to CT PRAMS data, there were no differences in any substance use between race/ethnicity groups, and this did not change when low alcohol use was excluded. On the other hand, CAPTA notifications were made for twice as many Black mothers compared to White (10.4% vs. 5.2%) and more than twice as many compared to Hispanic mothers (4.6%). These findings align with a long history of research documenting racial disparities in hospital provider behaviors despite equivalent rates of substance use in pregnancy (Chasnoff et al., 1990; Patel et al., 2021; Perlman et al., 2021). Although we do not have data on each hospital’s approach to toxicology testing, previous research on hospital policies suggests that selective screening/testing is common practice in CT (Hasan, 2020) and other states (Rebbe et al., 2019b). Our findings suggest the current approach disproportionately identifies substance use in pregnancy across racial/ethnic identities.

The mechanisms that explain this disproportionality are unclear from the current study, although the different substances identified across each dataset may contribute. The PRAMS sample is too small to conduct a race/ethnicity-based subgroup analysis of self-reported substance use. Prior research does point to racial/ethnic differences in type, magnitude, and timing of substances used (Roberts & Nuru-Jeter, 2011). For example, Rebbe et al. (2019a), using a Washington state birth cohort, found that Native American and Black infants were significantly more likely to have cannabis exposure compared to white and Hispanic infants. In this study, Hispanic infants had the lowest rates of prenatal substance exposure of any type. In our study, the high rates of cannabis notifications (close to 80% of the notifications included single or polysubstance cannabis; Lloyd Sieger et al., 2021) may partly contribute to the observed race disparity. Like our discussion of under-identifying alcohol exposure, cannabis may be more readily identified due to systemic factors such as common toxicology test types used during prenatal care or practices among providers serving lower income mothers.

Of note, the only substance category more frequently identified among CAPTA notifications compared to self-report in the PRAMS-4+ subsample was polysubstance use. One percent of mothers were included in CAPTA notifications involving polysubstance use compared to .3% of mothers in the PRAMS 4+ sub-sample. Despite differing prevalence rates, three substance types were involved in most polysubstance exposures: cannabis, MOUD, and other illegal drugs. The fact that CAPTA notifications identified higher rates of polysubstance use compared to CT PRAMS may reflect that mothers who used illegal drugs failed to self-disclose in the CT PRAMS survey due to inaccurate recall, stigma, or social desirability bias. Because hospitals would not have to rely on self-report to detect polysubstance use, hospital approaches likely identified a more accurate prevalence rate.

Each of these findings speak to the same broad set of challenges when implementing a policy like CAPTA: (1) engaging in public health surveillance of a stigmatized health status, (2) conducting this surveillance in taxed healthcare settings, and (3) leveraging an historically punitive system to manage this effort. The fact of racial/ethnic health disparities in the U.S. means that most health conditions, including stigmatized conditions, are more prevalent among populations “weathered” by racial discrimination and socioeconomic disadvantage (Forde et al., 2019). Compounding this is the long history of racism in healthcare settings (Green et al., 2021), as well as larger societal policies that result in racially disproportionate contact with punitive systems, including policing, courts, and child protection systems. It is within this context that frontline healthcare providers must gather information on substance use in pregnancy from mothers and then submit a deidentified notification to the state’s child welfare agency. Simply put, this is a tough ask.

What can policymakers and administrators do to support frontline providers in this effort? We make two recommendations. First, in order to ensure that the CAPTA notification system is used for public health surveillance, additional and ongoing training of frontline staff to differentiate between public health surveillance and child maltreatment risk assessment is needed. Achieving a public health response to substance use in pregnancy is undermined by a pervasive belief that certain types of drug use in the prenatal period is child abuse. While substance use in pregnancy may indicate, depending on perspective, a questionable health choice or a symptom of a behavioral health disorder, neither proposition is limited to people who use illicit drugs, as many long-term health and developmental consequences to children stem from alcohol or tobacco use in pregnancy (Behnke & Smith, 2013). And while untreated SUD is a risk for child maltreatment, it is not a condition of child maltreatment. Minimizing the social and legal consequences of disclosing substance use during pregnancy must be viewed as the most evidence-based strategy to increase self-report by pregnant individuals (Nguemeni Tiako & Sweeney, 2021).

Second, both the American College of Obstetricians and Gynecologists and American Academy of Pediatrics have long recommended universal substance use screening in pregnancy. Mental health concerns frequently co-occur with substance use and co-occurring disorders present much higher maltreatment risk than either substance use or mental health disorder alone (Deutsch et al., 2020). Therefore, screening for both substance use and mental health throughout women’s healthcare encounters, including during pregnancy and at delivery, is crucial. Combining verbal screening with calendar recall may provide more reliable assessment than any single tool on its own (Donohue et al., 2014; Ondersma et al., 2019). Given our findings that Black mothers are part of CAPTA notifications at different rates than White and Hispanic mothers, ensuring that universal screening achieves equitable identification of substance use will require concomitant and ongoing systems-level monitoring and revising of hospital practices (e.g., toxicology testing practices) that result in racial disproportionality (Green et al., 2021).

Several limitations to our study must be noted. One, CT PRAMS was based on self-reported information that may be subject to recall and/or reporting bias towards socially desirable answers. This undoubtedly resulted in the underestimation of substance use in pregnancy captured by CT PRAMS data and therefore impacted our ability to measure the extent of inaccurate identification in CAPTA. Two, heavier substance use in pregnancy is a low frequency event and thus more difficult to measure using a partial year of CT PRAMS data. Three, CAPTA notifications and CT PRAMS measured substance use in pregnancy differently. We attempted to align the substance categories, but different measurement approaches could have influenced our results. Four, missing race/ethnicity for 10% of CAPTA notification may not be missing at random. Finally, because the datasets used in this analysis were unlinked, we could not conduct standard statistical tests to compare rates across subsamples. We used a simplistic approach to compare rates based on confidence intervals, but future research is needed to address these limitations.

Despite these limitations, our study contributed to knowledge by documenting racial/ethnic disparities between self-reported substance use in pregnancy and hospital-identified substance use in pregnancy submitted to the State’s notification system. Although the notifications withheld personally identifying information for dyads without safety concerns, the observed racial/ethnic disparities raise concerns regarding provider approaches to identifying substance use in pregnancy. Additional effort is needed to support pregnant mothers’ disclosure of substance use status with care providers and providers utilization of the CAPTA notification system for public health surveillance.