Punitive legal responses to prenatal drug use in the United States: A survey of state policies and systematic review of their public health impacts

Legal review

We updated previous reviews of policies governing prenatal drug use, documented heterogeneity, and incorporated legislation adopted after 2016.^3,17,34 The timeframe was selected to extend existing legal reviews given recent policy shifts.^3,35 Specifically in 2016, the federal child abuse law—the Child Abuse Prevention and Treatment Act (CAPTA)—was amended to require that states institute care planning and notification policies for infants identified as affected by prenatal substance exposure.^36-38

Our inclusion criteria restricted searches to policies that explicitly allow criminal charges for prenatal drug use, authorize involuntary commitment, or impose civil penalties with custody-related repercussions (Table 1). We excluded polices that do not target pregnant people directly, such as mandatory reporting policies which require health and social service providers to report prenatal drug use to state authorities. We similarly excluded policies that target the romantic partners of pregnant people, or that only apply to businesses or other entities like social service agencies.^34,39 For criminal policies, we specifically searched for enacted statutes but also considered states to have adopted criminalization based on judicial interpretation—here defined as a state supreme court decision. Policies associated with custody loss included those that classify prenatal drug use as civil child abuse, or that specify prenatal drug use as grounds for child abuse substantiation or parental rights termination. We excluded policies that mentioned parental drug use but did not clearly specify prenatal use as a criterion for enforcing penalties. For each policy, we additionally identified whether evidence of symptomology or harm was required (versus exposure only), whether exceptions for drug treatment were included, or whether other special circumstances were noted.

We obtained legal data from two sources. First, we identified legal citations from existing datasets including the Alcohol Policy Information System (APIS), the Guttmacher Institute Prenatal Substance Use Policies dataset, the Child Welfare Information Gateway State Statute Database and previously published studies.^3,35,40,41 From this corpus, we identified the most common terms related to prenatal substance use, then used terms to search LexisNexis for additional legislation (Supplemental Table 1). Because our focus was on current policies, we did not necessarily capture all repealed policies, but noted them when identified. After deduplication, we reviewed the text of collected statutes for inclusion. Lastly we abstracted relevant information based on themes identified from prior literature³⁴ and questions that arose during the corresponding literature review.

Literature review

For the literature review, we identified studies estimating associations between punitive prenatal drug policies and health-related outcomes, published January 2000 to December 2022. This window was selected to include the period when punitive policy enactment began to escalate in response to increasing prenatal substance use, thus allowing for empirical investigations.^3,26,42,43 Studies with any research design were included if they reported a quantitative association between a punitive policy and any potentially relevant health outcome, including in the context of mixed methods designs. Other than specifying quantitative research, we did not impose other restrictions on study design. Commentaries, dissertations, and other non-peer reviewed articles were excluded, as were studies that did not include a U.S. population.

To identify studies meeting these criteria, we searched EBSCOhost, Embase, ProQuest, PubMed, Scopus, and Web of Science using prespecified search terms (Supplemental Table 1) and exported results to Covidence for review. After deduplication, studies were screened on title and abstract, then relevant papers went through full text review. For each included study, we extracted article features, methods, and results.

We assessed potential validity threats using prespecified criteria based on methodological considerations relevant to policy evaluation (Supplemental Table 2).^44-51 Focal areas included: study design and timeframe, exposure and outcome definitions, adjustment for time-varying confounders and co-occurring policies, and methods to account for baseline and time-varying differences between locations and over time, if applicable. For study design, we identified whether analyses were cross-sectional or longitudinal by level of analysis, and if a comparison group was included. For example, studies were considered to have a longitudinal comparison group design if they utilized a difference-in-differences, event-study, or synthetic control approach.⁵² Strengths and weaknesses across these domains were described narratively.

Each phase of the review processes was completed by two independent reviewers with disagreements resolved by discussion and consensus. Legal review content was presented to a third reviewer with expertise in prenatal substance use policy for additional validation. Initial legal and literature searches began in January 2022 and findings are current as of December 2022.

Legal survey

The legal search identified 567 unique citations (Figure 1). In total, 131 statutes met criteria and were included. By 2022, approximately half of states had at least one punitive prenatal drug policy (n=25, Figure 2). The most common were those designating prenatal drug use as civil child abuse (n=23). Explicit criminalization (n=2), involuntary commitment (n=5) and parental rights termination (n=6) were less common. Geographic variation was noted, with criminalization limited to Southern states and involuntary commitment prevalent in the Midwest.

Criminalization and involuntary commitment

South Carolina, Alabama, and Tennessee have each formally criminalized prenatal drug use. Tennessee’s policy has not been in effect since mid-2016 after expiring due to a sunset provision.^26,53 Tennessee is the only state to have enacted a criminal policy via the legislature; criminalization in South Carolina and Alabama occurred first via court decisions, then later amendments in Alabama. In Tennessee, the legislature passed Fetal Assault as an amendment to the state’s existing fetal homicide policy. The change allowed prosecutors to charge pregnant people with misdemeanor assault following diagnosis of NDWS or evidence of prenatal use of an illegal drug. In South Carolina and Alabama, state supreme courts interpreted the term ‘child’ to include fertilized eggs or fetuses, thereby allowing legal grounds for felony or misdemeanor child abuse prosecutions for any type of prenatal drug use. The Alabama policy was initially geared specifically towards methamphetamine.⁵⁴ Even though remaining states have not adopted policies explicitly authorizing criminalization, it should be noted that prenatal drug use is frequently charged under existing criminal statutes—for example pregnant people who experience stillbirth being charged with murder. Therefore, even in the absence of an overt criminal policy, individuals could be criminally prosecuted in other states and existing legal literature suggests that this practice is relatively common.²⁶

In addition to criminal approaches, five states (MN, ND, OK, SD, WI) have policies that allow pregnant people to be held in physical custody or involuntarily committed to treatment for reasons related to prenatal drug use. In North Dakota, the mandatory reporting statute grants the state authority to act under the state’s involuntary commitment policy during pregnancy. In the remaining 4 states, protection of an ‘unborn child’ is explicitly cited as grounds for detention directly within the language of the involuntary commitment statute.

Custody-related penalties

Custody-related penalties were the most prevalent punitive policy observed, with policies in 23 and 6 states clearly considering prenatal drug use sufficient grounds for child abuse substantiation or parental rights termination (FL, IL, KY, MO, ND, TX, Figure 2). Several other states had policies that could arguably include prenatal drug use under child abuse statutes, however the text was ambiguous. For example, in Kentucky, the definition of child abuse includes parental incapacity due to substance use disorder, but does not cite prenatal exposure as a condition for enforcement.

We identified significant heterogeneity across state policies. For example, parental rights termination in Illinois and Missouri (but not FL, ND, and TX), is authorized only after a substance-exposed infant was previously born to the same person. Similarly, some states require evidence of infant symptomatology or NDWS before applying penalties. In most states, however, neonatal exposure alone—typically established via toxicology—is sufficient to trigger penalties. State policies also vary in terms of the substances and circumstances covered under punitive provisions. Most policies (n=18) included language to exclude medical treatments or legally prescribed medications from prenatal drug use definitions (Supplemental Table 3). In some instances, such language could be interpreted to carve out allowances for MOUD, but only one state policy (TN) appears to explicitly state this exception.

Literature review

The literature search identified 241 unique citations, 119 of which underwent full text review, resulting in 16 included articles. Search strategy and inclusion decision details are provided in Table 1 and Figure 1.

Twelve studies grouped policies dichotomously, with eight of these defining exposure using a combined measure capturing presence or absence of any policy (Table 2). The most frequent were criminal and civil child abuse. Relatively fewer studies included civil commitment (k=2)^40,55 or parental rights termination (k=2) in operationalizations of punitive definitions.^41,56 Six studies examined the impacts of individual policies in primary^57,58 or sensitivity analyses,⁵⁶ often using single-state analyses.^53,59,60 Although most studies utilized the same exposure data source—the Guttmacher Institute Prenatal Substance Use Policy dataset—legal dates were sometimes conflicting, as has been noted previously.^41,41,61,62

Threats to validity

Threats to validity were concentrated in three areas, overall study design, control for time-varying confounding, and consideration of effect measure modification. First, in terms of study design, most included studies utilized a longitudinal approach with a comparison group (k=14), typically difference-in-differences implemented using two-way fixed effects or event study models. One early study employing treatment admissions data from 2012 was cross-sectional,⁶³ and another was primarily descriptive, presenting only unadjusted analyses.⁶⁴ Of the remaining longitudinal comparison group studies, all addressed common concerns in policy evaluation such as accounting for time trends (e.g., year or month fixed effects), including unit-specific effects (e.g., state fixed effects, generalized estimating equations, or by study design), and addressing serial correlation (e.g. clustered standard errors) where applicable. However, only some studies presented formal assumption testing analyses, such as parallel trends testing for difference-in-differences designs.⁶⁵ While most included studies utilized national administrative data, five^{40,53,55,59,60} used a convenience sample of states or conducted single-state analyses.

Despite generally strong observational study designs, six^40,53,66-69 of the included longitudinal studies did not control for co-occurring policies, and two^53,67 did not control for any other theoretically relevant time-varying confounders. The co-occurring policies most frequently included in adjusted models were related to Medicaid (k=5),^{41,55,56,60,62} or were aimed at supporting pregnant people with drug use disorders—for example, policies mandating drug treatment access (k=5),^{57,58,60,62,67} or prohibiting treatment discrimination (n=2)^55,67 during pregnancy. Eight studies also addressed policies that require health care providers to drug test, or to report evidence of prenatal drug use to child welfare or social services agencies for investigation (i.e., mandatory testing and reporting).^{40,41,55,57-60,67} Five of these treated mandatory reporting policies as an additional exposure category distinct from other punitive policies,^{40,55,59,60,67} while three treated them as covariates.^41,57,58 Despite interdependencies between punitive mandatory reporting and civil child abuse policies, only one study considered joint effects.⁶⁰ Five studies controlled for other treatment-related factors such as numbers of substance use disorder treatment providers, facilities, or funding levels within a state.^{40,41,55,59,60} Analytically accounting for the potentially countervailing effects of supportive policies and substance use disorder treatment capacity is an important consideration given that many states maintain both punitive and supportive policies.^3,70 Several supportive-oriented policies are associated with increases in pregnant people’s use of treatment services and decreases in NDWS,^41,67 though not all prior research is consistent.⁷¹

Lastly, only four studies considered the differential impacts that punitive policies may have on low-income pregnant people and pregnant people of color.^41,62,67-69 Analyses of differential effects by race and ethnicity were included in three^62,68,69 of the five studies examining maltreatment or foster care outcomes, but only in one of five studies⁶⁷ on drug treatment. One additional study⁴¹ examined potential differential effects of punitive policies by income and Medicaid status on multiple pregnancy and birth-related outcomes.

Drug use disorder diagnosis and treatment and overdose

Drug use disorder diagnosis and treatment, including MOUD, were some of the most commonly assessed outcomes (Table 2).^{56,57,63,64,66,67} We did not identify any studies examining associations between punitive policies and measures of drug use directly, however studies in the treatment-related category hypothesized that inverse associations with punitive policies were indicative of reductions in treatment rather than reductions in drug use directly.

Two studies using national drug treatment admissions program data through 2014^57,67 reported that adoption of a criminal or civil child abuse policy was associated with an approximately 1-percentage-point decrease in the proportion of women who were pregnant upon entering drug treatment, suggesting treatment-inhibiting effects. Gressler et al.,⁶⁴ additionally found a lower prevalence of opioid use disorder (OUD) diagnoses among pregnant people in punitive compared to non-punitive states in unadjusted analyses (Table 3).

Among pregnant people diagnosed with OUD, two additional studies reported reduced use of MOUD generally,⁶³ methadone⁵⁶ and psychosocial drug treatment⁵⁶ (but not buprenorphine use), in punitive compared to non-punitive states. Choi et al.⁶⁶ further examined associations between Medicaid expansion and MOUD in both punitive (criminal or civil child abuse) and non-punitive states. They found that while expansion was associated with increases in MOUD broadly, increases were relatively smaller in punitive contexts (8.7 versus 5.6 percentage points), but not statistically significant in formal interaction testing. Lastly, Tabatabaeepour et al.,⁵⁶ analyzing commercial insurance claims, found that punitive policies overall were not associated with changes in fatal or non-fatal overdose among pregnant people with OUD. However, in some sensitivity analyses there was an imprecise relationship between criminalization and increasing non-fatal overdose prevalence, highlighting that further research is needed to better characterize the impacts of criminal policies specifically.

Pregnancy and birth-related outcomes and prenatal care utilization

Three studies^41,53,60 examined associations of punitive policies with pregnancy and birth-related outcomes (Table 3). Each found that punitive policy adoption adversely affected measures of prenatal care, but results were inconclusive for other outcomes. First, using 2008–2019 U.S. birth data, Meinhofer et al.,⁴¹ found that adoption of any punitive policy (criminalization, civil child abuse, parental rights termination and maltreatment registry requirements—policies requiring individuals investigated/convicted of child abuse to be place on state registries) was associated with decreased utilization of any prenatal care. Austin et al.⁶⁰ likewise found that child abuse policies alone, and child abuse policies adopted in conjunction with mandatory reporting, were both associated with delays in prenatal care initiation. Lastly, Boone and McMichael,⁵³ focusing on the Tennessee Fetal Assault statute, identified decreases in the county-level prevalence of any prenatal care following enactment, relative to counties in control states without criminalization.

Despite the generally consistent finding that punitive policies adversely impact uptake of prenatal care, results were more mixed for other birth outcomes. Boone and McMichael found that the Tennessee Fetal Assault statute adversely affected mean Apgar score, stillbirth, and infant mortality but not mean gestational age.⁵³ Similarly, Meinhofer et al.⁴¹ reported that punitive policies were associated with imprecise increases in low-birthweight and very low-birthweight and decreases in gestational age.

Neonatal drug withdrawal syndrome (NDWS)

The four studies that examined relationship of punitive policies with prevalence or individual odds of NDWS also suggested null or harmful results (Table 3).^40,41,55,57 It is important to note that in the context of NDWS, increases in prevalence following policy adoption are challenging to interpret given that NWDS can also be caused by MOUDs like methadone—the historical gold standard of care for treating OUD among pregnant people.⁸

In hospital discharge data from 37 states 2000-2016, Atkins and Durrance⁵⁷ identified a 10%–14% increase in NDWS prevalence in states adopting a civil child abuse policy, relative to comparators. Using the same data source, but restricting to the 2008-2018 period, Meinhofer et al.,⁴¹ found that punitive policies were associated with greater NDWS prevalence at approximately the same magnitude. Further, the authors reported that punitive policies specifically increased NDWS among births covered by Medicaid and births to lower-income people.

Two studies assessing NDWS at the individual-level also reported inconsistent findings. Faherty et al.,⁴⁰ tested whether the enactment of any policy related to criminal or civil child abuse or civil commitment, was associated with changes in odds of NDWS. Across the eight included states, these policies as a group predicted a 17-51% relative increase in the odds of NDWS 1-year after enactment. However, given that none of these eight states appear to have authorized criminalization or civil commitment during the study timeframe, these findings may be limited to enactment of civil child abuse statutes. In a second study by Faherty et al.,⁵⁵ evaluating NDWS within a sample of births to people covered by Medicaid, the authors found that any punitive policy (criminal child abuse, civil child abuse, involuntary commitment) adoption was associated with small imprecise relative increase in NDWS odds 1-year (OR: 1.130 [95% CI: 0.880, 1.480]) and 2-years (OR: 1.050 [95% CI: 0.810, 1.370]) post-policy.

Child maltreatment and foster care

Five studies that used national maltreatment or foster care records reported null or adverse associations with punitive policies (Table 3).^{58,59,62,68,69} First, Maclean et al.⁶² identified increases in total (19%) and substantiated (33%) reported infant maltreatment prevalence in states adopting a criminal or civil child abuse, or parental rights termination policy, relative to control states. Associations were especially pronounced among Black infants, suggesting differentially harmful effects by race and ethnicity. In contrast to these overall results, however, an additional study reported inconclusive findings for states implementing civil child abuse policies alone, or in combination with criminalization.⁵⁹ Bandara et al.⁵⁹ identified a potentially beneficial association—depending on interpretation—between civil child abuse policy adoption and reported infant maltreatment prevalence in a single state (UT) compared to a selection of comparison states (AK, CA, KY, MA, ME, MI, MT, PA). Although reductions in maltreatment prevalence were not observed in Utah in the first three years following enactment, decreases in maltreatment were noted four years post-enactment, relative to controls. In contrast, the study also reported that the coadoption of civil and criminal child abuse in Alabama was not associated with meaningful changes in maltreatment. As noted by the authors, however, reductions in observed maltreatment could reflect a chilling effect on reporting, rather than reductions in maltreatment incidents directly, making it challenging to interpret findings without broader context.

For foster care, Sanmartin et al,⁶⁸ found increases in the prevalence of substance use-related foster care entries in states that newly adopted a criminal or civil child abuse policy between 2005 and 2016. They also reported that while the increase was observed for both non-Hispanic white and Black children, impacts were concentrated among non-Hispanic white infants. These findings were similar to those of Atkins and Durrance,⁵⁸ who likewise identified an approximately 10% increase in foster care entries among infants removed for parental drug use-related reasons in punitive states. In a second study, Sanmartin et al.⁶⁹ also found that among infants in foster care for reasons related to parental substance use, lower parental reunification was observed among infants in non-punitive states. Further, while infants experienced reduced parental reunification in punitive contexts overall (Hazzard Ratio [HR]: 0.87; 95% confidence interval, 0.86-0.88), non-Hispanic white infants were more likely to be reunited, suggesting more profound adverse impacts for Black families.

Limitations

This review had limitations. First, we imposed multiple restrictions to manage the complexity of the review but acknowledge that these decisions imply tradeoffs. For example, we excluded legal mechanisms that did not directly impose penalties on pregnant people themselves, including mandatory reporting policies which typically target providers.³⁴ The evaluation of mandatory reporting policies, which are fundamentally punitive in nature,⁸ is an active area of ongoing research, but one that was outside the scope of this initial review. We similarly excluded polices aimed at supporting pregnant people with drug use disorder, another important area for further investigation. A related issue is that this review focused solely on policies on the books (as enacted), without tracking enforcement or “on the ground” impact. Relatedly, we did not conduct a systematic search for case citations, only for legislation. Our search was similarly not designed to reflect settings where prenatal drug use could be considered legally relevant in the absence of a specific statute, including in states that have adopted other fetal personhood measures.⁷³

Several scoping-related limitations also apply to the systematic review. For example, we may have missed important studies published after our search window was closed, and that were conducted in other settings. Prenatal substance use is not limited to the U.S., and evidence indicates that multiple other countries including Australia, Canada, Norway, South Africa and the U.K. have adopted or are considering policies that can result in punitive consequences for pregnant people who use drugs.^74-78 Given that these policies likely operate similarly across contexts, we would expect punitive policies to also exert adverse health consequences in these settings. Additional research is needed to explore this possibility. Triangulating findings in an international perspective could also strengthen the broader evidence base by leveraging population registries with more comprehensive dyadic data to address observed shortcomings of U.S.-based studies.⁷⁹ Regardless of setting, as legislative and epidemiologic contexts evolve, additional research is warranted. Emerging data suggests that the prevalence of prenatal substance use and related outcomes like overdose increased in many places during the COVID-19 pandemic.^80-83 How these trends may have been affected by policy environments is unclear.

Lastly, another limitation is that we did not conduct a meta-analysis given differences in variable definitions and few studies in some outcome categories. However, as previously noted, while this methodological diversity complicates generalizations about the potentially harmful effects of specific punitive policies, the consistency of null or adverse associations across relatively dissimilar studies increases our confidence that these policies do not yield benefits.

Conclusions

Our findings suggest that nearly half of states have adopted policies imposing penalties on prenatal drug use. This means that a significant proportion of pregnant people who use drugs in the U.S. face potential legal implications. Moreover, the existing evidence base does not support this policy approach. While we found relatively consistent evidence that punitive approaches do not improve health, there was not enough evidence to definitively conclude that policies increase harms—at least for most exposure-outcome combinations. An important factor likely contributing to this uncertainly is that most studies assessed the effects of combinations of policy exposures, without disaggregating specific policies in the analysis. Additional research is therefore required to better characterize the pathways through which punitive legal approaches may result in unintended harmful consequences, particularly for explicit criminalization, parental rights termination, and involuntary commitment, which remain understudied. Further, given the strong possibility that punitive policies engender adverse health consequences, future work must investigate potential differential effects among families known to be unequally and unfairly targeted by criminal-legal system enforcement.