Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2022 Apr 20;31(7):1452–1467. doi: 10.1002/hec.4518

Prenatal substance use policies and newborn health

Angélica Meinhofer 1,✉, Allison Witman 2, Johanna Catherine Maclean 3, Yuhua Bao 1
PMCID: PMC9177792  NIHMSID: NIHMS1802916  PMID: 35445500

Abstract

We study the effect of punitive and priority treatment policies relating to illicit substance use during pregnancy on the rate of neonatal drug withdrawal syndrome, low birth weight, low gestational age, and prenatal care use. Punitive policies criminalize prenatal substance use, or define prenatal substance exposure as child maltreatment in child welfare statutes or as grounds for termination of parental rights. Priority treatment policies are supportive and grant pregnant women priority access to substance use disorder treatment programs. Our empirical strategy relies on administrative data from 2008 to 2018 and a difference‐in‐differences framework that exploits the staggered implementation of these policies. We find that neonatal drug withdrawal syndrome increases by 10%–18% following the implementation of a punitive policy. This growth is accompanied by modest reductions in prenatal care, which may reflect deterrence from healthcare utilization. In contrast, priority treatment policies are associated with small reductions in low gestational age (2%) and low birth weight (2%), along with increases in prenatal care use. Taken together, our findings suggest that punitive approaches may be associated with unintended adverse pregnancy outcomes, and that supportive approaches may be more effective for improving perinatal health.

Keywords: infant health, maternal substance use disorder, prenatal substance use policies

1. INTRODUCTION

Illicit substance use during pregnancy has increased substantially over the past decade in the United States. In 2018, an estimated 8.5% of pregnant women ages 15–44 used illicit substances in the past month, reflecting a 70% increase from 2010 levels (Center for Behavioral Health Statistics and Quality, 2018). Substance use during pregnancy is of public health concern because of potential adverse maternal and newborn health outcomes. Observational studies suggest that in utero exposure to substances is associated with higher rates of perinatal mortality, birth defects, low birth weight, low gestational age, and neonatal drug withdrawal syndrome (NDWS; Behnke et al., 2013). NDWS is primarily caused by opioid exposure in utero, but can also be caused by exposure to other substances. Withdrawal symptoms in newborns often include diarrhea, vomiting, difficulty sleeping, respiratory distress, and seizures (Desai et al., 2015; Huybrechts et al., 2018; Maeda et al., 2014; Patrick et al., 2012).

US states began to adopt a diverse set of prenatal substance use policies (PSUPs) in the 1980s, when the crack cocaine epidemic created public concern about the adverse effects of prenatal exposure to illicit substances (Thomas et al., 2018). Some PSUPs are supportive and focus on providing access to substance use disorder (SUD) treatment and related supports, for instance, by creating or funding targeted SUD treatment programs specifically for pregnant women or by granting pregnant women priority access to SUD treatment programs (Angelotta et al., 2016; Christian, 2004; Figdor & Kaeser, 1998; Miranda et al., 2015). Other PSUPs are punitive, criminalizing illicit prenatal substance use or defining prenatal substance exposure as child abuse or neglect in child welfare statutes. When defined as such, prenatal substance exposure could result in Child Protective Services (CPS) involvement, foster care placement, and termination of parental rights. Prenatal substance exposure in these definitions generally implies that healthcare providers must inform CPS when a newborn exhibits signs of prenatal substance exposure. Regardless of whether a state defines prenatal substance exposure as child abuse or neglect, amendments to the Child Abuse Prevention and Treatment Act (CAPTA) in 2003 mandated states to implement statutes, codes, programs, or procedures requiring or encouraging healthcare providers to notify CPS or other authorities of substance‐exposed newborns (Administration for Children and Families, 2017).

The ongoing opioid crisis and its effects on pregnant women and newborns have refocused national attention on PSUP implementation in the United States. As states seek to address the growing incidence of NDWS (Patrick et al., 2012; Winkleman et al., 2018), foster care placement for parental drug use (Meinhofer & Angleró‐Díaz, 2019), as well as overall healthcare costs in the state budget, elucidating the question of PSUP effectiveness is a priority policy question. We study the effect of punitive and priority treatment PSUPs on neonatal drug withdrawal syndrome, low birth weight, very low birth weight, low gestational age, and prenatal care. In an extension, we consider heterogeneity by socioeconomic characteristics. We employ a difference‐in‐differences framework and analyze 2008–2018 data from the Healthcare Cost and Utilization Project (HCUP) in 46 states and from restricted National Vital Statistics System Natality Files (NVSS) in 51 states. Our study provides policymakers with comprehensive and timely information regarding evidence‐based approaches for improving pregnancy outcomes in perinatal populations affected by substance use disorders.

2. BACKGROUND

2.1. Conceptual framework

The impact of punitive PSUPs is theoretically ambiguous. Stigma and fear of punishment may deter pregnant women from using illicit substances, which would lead to improved newborn health. However, among pregnant women unwilling or unable to discontinue using substances, an unintended consequence could be deterrence from seeking prenatal care and SUD treatment, which would lead to worse newborn health. Pregnant women may also substitute illicit substances with legal substances (i.e., tobacco, alcohol), although the net effects on newborn health are uncertain and will depend on the intensity of use and toxicity between substituted substances. Priority treatment PSUPs may improve newborn health through greater SUD treatment and prenatal care utilization, but the effect on NDWS, which primarily results from prenatal exposure to opioids, is unclear. In particular, methadone and buprenorphine, the gold standard medications used for treating opioid use disorders (OUD) in pregnancy, are also opioids and may induce NDWS despite improving other pregnancy outcomes (Goodman et al., 2019). Moreover, priority treatment PSUPs may be less impactful in the context of prenatal opioid use and NDWS because federal requirements on opioid treatment programs, which provide SUD treatment for persons with opioid use disorder, already mandate that pregnant women receive priority access to SUD treatment at opioid treatment programs. Nevertheless, priority treatment PSUPs may improve pregnancy outcomes for women seeking treatment at other types of SUD treatment programs.

2.2. Prenatal substance use policies

U.S. states began to adopt a diverse set of punitive and supportive PSUPs during the crack cocaine epidemic of the 1980s. Punitive PSUPs attempt to deter prenatal substance use through potential sanctions on pregnant women. For example, in 2014 Tennessee enacted a law criminalizing narcotic drug use among pregnant women. The law specified prenatal substance use as a criminal act, amending the state's initial policy which defined a viable fetus as a victim and deemed any criminal acts against it a criminal offense. The statute expired under a sunset clause in 2016 (TN Code § 39‐13‐107, 2014). Many other states have policies that punish pregnant women for substance use, such as mandatory civil commitment to SUD treatment, or that define prenatal substance exposure as child abuse or neglect. When defined as such, prenatal substance exposure could result in CPS involvement, foster care placement, and termination of parental rights. Some states with punitive policies may require that any reports of prenatal substance exposure to CPS trigger a referral for an addiction assessment, for SUD treatment if recommended, as well as a plan of safe care for the substance exposed newborn. As mandated reporters of child abuse and neglect, healthcare providers in states with punitive PSUPs are generally required to inform CPS of newborns with prenatal substance exposure. Some states have implemented punitive PSUPs that are arguably of smaller scope in that prenatal substance exposure is not specifically defined as child abuse or neglect but might result in some negative consequences. For instance, in Idaho prenatal substance use may result in inclusion in a Child Protection Central Registry for a minimum of 10 years and in New Mexico, an allegation of child abuse or neglect can be substantiated if a newborn has been prenatally exposed to substances.

In contrast to punitive PSUPs, supportive PSUPs are rehabilitative and seek to provide SUD treatment services and other related assistance to pregnant women, and prevent discrimination. Some states have statutes that give pregnant and postpartum women priority access to SUD treatment programs, or that create targeted SUD treatment programs specifically for pregnant and postpartum women. The comprehensiveness of supportive PSUPs also varies across states. For instance, different states specify that treatment priority PSUP requirements apply exclusively to opioid treatment programs, which is less comprehensive because it is already federally mandated that opioid treatment programs give priority SUD treatment access to pregnant women.

2.3. Previous literature

Despite considerable debate regarding the various approaches for addressing illicit substance use during pregnancy, empirical evidence on PSUP effectiveness is scarce. Previous studies primarily focus on the characterization and historical evolution of these policies, documenting that PSUPs have become more punitive over time (Angelotta et al., 2016; Christian, 2004; Figdor & Kaeser, 1998; Jarlenski et al., 2018; Miranda et al., 2015; Thomas et al., 2018). Other studies are qualitative and document pregnant women's reactions to PSUPs (Jessup et al., 2003; Roberts & Nuru‐Jeter, 2010; Roberts & Pies, 2011; Stone, 2015).

Only a handful of quantitative studies consider the question of PSUP effectiveness. Two recent studies find that punitive PSUPs are negatively associated with the proportion of pregnant women entering SUD treatment (Atkins & Durrance, 2020; Kozhimannil et al., 2019). Another set of studies finds that punitive PSUPs are associated with increases in foster care placement for parental substance use (Atkins & Durrance, 2021; Sanmartin et al., 2019, 2020). Three studies use a two‐way fixed‐effects (TWFE) design to estimate the impact of punitive PSUPs on NDWS with mixed findings (Atkins & Durrance, 2020; Faherty et al., 2019, 2022). Using individual‐level data from the 2003–2014 HCUP State Inpatient Databases in eight states, one study finds increases in NDWS after the implementation of punitive PSUPs (Faherty et al., 2019). Atkins and Durrance (2020) use state‐level data from the 2000–2014 HCUPNet in 37 states and finds no significant association between PSUPs and NDWS. Both studies, however, rely on limited variation from five changing punitive PSUPs. Using a sample of Medicaid beneficiaries from 39 states, Faherty et al. (2022) find no statistically significant association between PSUPs and NDWS for Medicaid births, but only observe four punitive PSUP changes. A related literature considers the impact of prenatal alcohol use policies on newborn health and finds that policies classifying alcohol use during pregnancy as child abuse or neglect are associated with adverse birth outcomes (Roberts et al., 2019; Subbaraman et al., 2018).

2.4. Contributions

Our study contributes to the nascent PSUP literature by generating new evidence about health implications for newborns, and by addressing some of the most pressing limitations. First, previous studies of the impact of PSUPs on newborn health have exclusively focused on NDWS (Atkins & Durrance, 2020; Faherty et al., 2019, 2022). While NDWS is an important measure of newborn health directly affected by prenatal opioid exposure and one considered in our study, it is nonspecific in that it can be potentially induced by any type of opioid, including illegal opioid use, prescription opioid misuse, or appropriate use of prescription opioids. Methadone and buprenorphine, the gold standard medications used for treating opioid use disorder in pregnancy, are also prescription opioids and may induce NDWS despite improving other pregnancy outcomes (Goodman et al., 2019). In sum, studies that rely on NDWS alone cannot assess whether newborn health overall improves post‐policy. Our study is the first to consider the impact of PSUPs on other key measures of newborn health, including low birth weight, very low birth weight, and low gestational age. These outcomes are major risk factors for infant mortality and long‐term morbidity and may be influenced by maternal use of other illegal substances, not only opioids (Behnke et al., 2013).

Second, there is limited empirical evidence regarding the mechanisms through which PSUPs may influence pregnancy outcomes. Only two PSUP studies have evaluated this question by considering maternal SUD treatment utilization (Atkins & Durrance, 2020; Kozhimannil et al., 2019). We are the first to consider the impact of PSUPs on prenatal care utilization, another potential mechanism reflecting maternal behaviors that can affect newborn health. Third, previous studies primarily focus on punitive policies. Our study evaluates the impact of both, punitive and priority treatment PSUPs. Fourth, previous PSUP studies of NDWS analyze data up to 2014 and estimate treatment effects using variation from a limited number of switching states (four to five states with switching punitive PSUPs). We rely on data up to 2018 capturing all three waves of the opioid crisis (Maclean et al., 2020) and a larger number of switching states, which considerably increases the number of PSUP adoptions and thus, generalizability. Fifth, researchers have noted inconsistencies in PSUP statues and effective dates used across previous studies (Reddy & Schiff, 2022). We make efforts to reconcile these mismatches, and generate a PSUP database with up‐to‐date policy dates based on data from the Guttmacher Institute, the Children's Bureau, other researchers, direct communication with state child welfare agencies, and our own original legal analysis. Moreover, we test the sensitivity of findings to changes in PSUP selection criteria based on the scope of the policies.

Finally, we contribute to the literature by assessing whether PSUP effectiveness varies across subpopulations. There are important differences across socioeconomic characteristics in the prevalence of prenatal opioid use and subsequent NDWS, including among pregnant women with low income, receiving Medicaid, and living in rural areas (Desai et al., 2015; Leech et al., 2020; Patrick et al., 2012, 2019; Shaw et al., 2015; Villapiano et al., 2018; Winkelman et al., 2018). Thus, establishing the extent to which PSUPs may exacerbate or mitigate existing disparities in substance use‐related outcomes is important for promoting population health equity.

3. DATA AND METHODS

3.1. Hospitalization data

Healthcare Cost and Utilization Project is sponsored by the Agency for Healthcare Research and Quality (AHRQ) and includes the largest collection of hospital data in the U.S., with all‐payer, encounter‐level information. We analyzed the HCUP FastStats Neonatal Abstinence Syndrome data (Patrick et al., 2019), which are based on the State Inpatient Databases and report NDWS‐related inpatient measures. Data are available at the state‐year level in 46 states during years 2008–2018, for the population overall and for select subpopulations. Subpopulations of interest included expected payer (Medicaid and private insurance), community‐level income quartiles based on the median household income of the patient's Zip code of residence (Q1, Q2, Q3, and Q4, with Q1 representing lowest income communities), and urbanicity (large central, large fringe, medium and small metropolitan, and rural; HCUP FastStats, 2019).

Outcomes include the rate of NDWS per 1000 newborn hospitalizations. Agency for Healthcare Research and Quality generates NDWS diagnoses with ICD‐9 code 779.5 for 2008–2015 and ICD‐10 code P96.1 for 2016–2017 (Patrick et al., 2019). Consistent with previous NDWS studies, possible iatrogenic NDWS cases are excluded (HCUP FastStats, 2019; Patrick et al., 2012). NDWS is missing for approximately 6% of state‐year observations. As this issue is concentrated in a small number of states, we remove two states (Alaska and Delaware) especially subject to missing data. Although Delaware is not consistently available in HCUP, Delaware's Department of Health (DOH) shared 2008–2018 individual‐level hospital inpatient data directly with us. Using DOH data, we successfully replicated the same state‐year outcomes as in HCUP during the years in which both datasets were available, and incorporate them to our HCUP sample for a total of 46 states.

3.2. Birth certificate data

We analyze restricted use 2008–2019 NVSS Natality Files, which provide demographic and health information for nearly all U.S. births and are based on data abstracted from birth certificates. We identify singleton hospital births and examine the proportion of births with low birth weight (<2500 g), very low birth weight (<1500 g), low gestational age (<37 weeks), and for which the mother received any prenatal care. In analyses of prenatal care, we drop three states (Alaska, Hawaii, and Rhode Island) with poor data reporting for this specific variable during our study period. The unit of analysis is the state‐year‐quarter of conception in 51 states for conception years 2008–2018, constructed following previous studies (Meinhofer et al., 2021b).

3.3. Prenatal substance use policies

We review PSUP statutes and effective dates from various sources, including the Guttmacher Institute's PSUP statute database (Guttmacher Institute, 2019), the Children's Bureau (Administration for Children and Families, 2021), and other studies and reports (Atkins & Durrance, 2020; Faherty et al., 2019; Miranda et al., 2015; Thomas et al., 2018). We identify some inconsistencies in PSUP definitions and effective dates across sources, as previously documented by other researchers (Reddy & Schiff, 2022). We investigate and reconcile these mismatches through original legal research, by contacting state child welfare agencies, 1 by reading available documentation in state websites and other official sources, and through consensus across sources. See Table A1 for details.

We define punitive and priority treatment PSUPs as follows. In states with punitive PSUPs, prenatal substance use is either criminalized, defined as child abuse or neglect, grounds for termination of parental rights, used as evidence to substantiate reports of child abuse or neglect, or may result in being placed in a CPS registry. In some state statutes, exposing a child to controlled substances is defined as abuse or neglect but there is no explicit mention of a fetus. Since a fetus is not a child in most states, these policies are not included in our study for those cases. In states with priority treatment PSUPs, some or all types of SUD treatment programs are required to provide pregnant women with priority access to SUD treatment. During our sample period, punitive PSUPs became effective in 13 states and were repealed in two states; priority treatment PSUPs became effective in 12 states and were repealed in three states (see Table A1). Some states eventually adopted both types of PSUPs, with 17 states ever implementing both types of policies at some point as of 2018. However, no state adopted both policies at the same time during our study period and there are at least four quarters between each policy change (within a same state) that occurred during our study period. Nevertheless, as some states adopted both policies, one potential issue might be if there is insufficient independent variation in PSUP indicators to identify the effect of both punitive and priority treatment policies on outcomes. We explore this potential issue with diagnostic tests by calculating variance inflation factors (VIF) for our two PSUP indicators and other variables using the HCUP FastStats sample, the smaller of the two samples we analyze. We find that VIF values are 1.19 for punitive PSUPs and 1.20 for priority treatment PSUPs, well below the conventional threshold of 5 that would indicate problematic collinearity (Kutner et al., 2004; Sheather, 2009). Thus, while some states eventually adopted both PSUPs, this policy adoption rollout does not appear to lead to concerning multicollinearity and we have sufficient variation to identify the effects of both PSUPs. Figure A1 characterizes time trends in punitive and treatment priority PSUP.

In robustness checks, we re‐define punitive and priority treatment PSUPs using variation from “stronger” policies. In states with “stronger” punitive PSUPs, prenatal substance use is either criminalized, defined as child abuse or neglect, or grounds for termination of parental rights. In states with “stronger” priority treatment PSUPs, other SUD treatment programs besides opioid treatment programs, are required to provide pregnant women with priority access to SUD treatment. We also test the robustness of findings to other potential definitions of PSUPs. While CAPTA federally mandates that states implement statutes, codes, programs, or procedures requiring or encouraging healthcare providers to notify CPS or other authorities of substance‐exposed newborns since 2003 (Administration for Children and Families, 2017), well before our sample period, many states have implemented their own statutes or administrative codes in addition to the federal law requiring healthcare provider reporting of substance‐exposed newborns. Additionally, a small number of states (Indiana, Texas, and Minnesota) implemented policies that created or funded targeted SUD treatment programs for pregnant and postpartum women during our sample period. We test the robustness of findings to controlling for these other state‐specific reporting and targeted SUD treatment funding PSUPs.

3.4. Control variables

We account for several control variables that may be correlated with PSUPs and birth outcomes to reduce potential omitted variable bias (see Tables 1 and 2). All models control for state‐year differences in the unemployment rate to account for economic factors, in Medicaid income thresholds for pregnant and postpartum women to account for changes in health insurance affecting healthcare utilization (i.e., prenatal care, SUD treatment), and for the proportion of Black and Hispanic newborns. In analyses of NVSS, we are also able to account for maternal age and newborn sex as these variables are included in the data.

TABLE 1.

Summary statistics in HCUP FastStats

Measures Mean Standard deviation N
Outcome variables
Rate of neonatal drug withdrawal syndrome (per 1000 newborn hospitalizations)
Overall 5.10 4.77 490
Payer
Medicaid 9.28 9.18 473
Private insurance 1.26 0.92 473
Self‐pay 5.92 6.03 473
Urbanicity
Large center metropolitan 4.20 3.94 382
Large fringe metropolitan 4.72 4.51 403
Medium metropolitan 6.07 5.92 451
Small metropolitan 6.68 7.32 445
Rural 6.40 6.23 463
Community level income
Quartile 1 (lowest) 6.75 6.83 478
Quartile 2 5.61 5.46 478
 Quartile 3 4.50 3.84 479
Quartile 4 (highest) 2.97 2.63 479
Policy variables (PSUPs)
Punitive 0.48 0.50 490
Priority SUD treatment 0.33 0.47 490
Control variables
Unemployment rate 6.73 2.31 490
Medicaid income thresholds for pregnant women 2.19 0.47 490
Proportion of black birth 0.15 0.09 490
Proportion of hispanic births 0.24 0.16 490
Open in a new tab

Note: The outcome variable was obtained from HCUP FastStats, 2008–2018. The unemployment rate was obtained from the Bureau of Labor Statistics, Medicaid thresholds for pregnant women were obtained from the Kaiser Family Foundation, and the proportion of Black and Hispanic births was obtained from the National Vital Statistics System. See Table A1 for PSUP data sources. The unit of analysis is a state‐year. Statistics are weighted by the number of births in a state‐year.

Abbreviations: PSUP, prenatal substance use policies; SUD, substance use disorders.

TABLE 2.

Summary statistics in NVSS Natality Files

Measures Mean Standard deviation N
Outcome variables
Low gestational age 0.081 0.010 2244
Low birth weight 0.064 0.010 2244
Very low birth weight 0.011 0.002 2244
Prenatal care 0.984 0.013 2112
Policy variables (PSUPs)
Punitive 0.480 0.500 2244
Priority SUD treatment 0.336 0.472 2244
Control variables
Unemployment rate 6.72 2.31 2244
Medicaid income thresholds for pregnant women 2.17 0.48 2244
Proportion of black births 0.16 0.09 2244
Proportion of hispanic births 0.24 0.16 2244
Proportion male 0.51 0.00 2244
Maternal age 28.17 1.06 2244
Open in a new tab

Note: Outcome variables were obtained from the NVSS Natality Files, 2008–2019. The unemployment rate was obtained from the Bureau of Labor Statistics and Medicaid thresholds for pregnant women were obtained from the Kaiser Family Foundation. The proportion of Black births, Hispanic births, males, and mean maternal age was obtained from the National Vital Statistics System. See Table A1 for PSUP data sources. The unit of analysis is a state‐year‐quarter of conception. Statistics are weighted by the number of births in a state‐year.

Abbreviations: PSUP, prenatal substance use policies; SUD, substance use disorders.

In robustness checks, we test the sensitivity of findings to a host of secondary control variables that account for other drug policies affecting the supply or demand for substances. These include prescription drug monitoring program operations and mandates, pain clinic laws, medical and recreational cannabis legalization, law enforcement seizures of fentanyl, heroin, and oxycodone, beer taxes, cigarette taxes, the rate of SUD treatment providers, the Affordable Care Act Medicaid expansions, and reporting PSUPs and targeted SUD treatment funding PSUPs (Meinhofer et al., 2018, 2021a, 2021b).

3.5. Methods

Our identification strategy employs a difference‐in‐differences framework that exploits variation in the staggered implementation of PSUPs across states and over time using the effective dates in Table A1. Equation (1) is the baseline TWFE model that estimates the static effect of each PSUP on outcome Y s,t . Punitive s,t is an indicator equal to one if a punitive PSUP is effective in state s and time t, and zero otherwise. Priority s,t is an indicator equal to one if a priority treatment PSUP is effective in state s and time t, and zero otherwise. In regression models using HCUP, where the unit of analysis is a state‐year, policies are coded as in place in year t if the policy's effective date falls between January and June, and in year t+1 if the policy effective date falls between July and December. In regression models using NVSS Natality Files, where the unit of analysis is a state‐year‐quarter of conception, policies are coded as in place in the year‐quarter of implementation. θ s  are state fixed‐effects and δ t are time fixed‐effects. Threats to identification might remain if other determinants of Y s,t change differentially over time in treated states. To minimize potential threats, a vector of control variables X s,t  that accounts for demographic characteristics and health insurance policies is included in all specifications (see Section 3.4). Regressions are weighted with the number of births in a state‐year and standard errors are clustered at the state level. We take the log of outcomes to account for variable skewness. Equation (1) offers a formal representation of our regression model:

Ys,t=β0+β1Punitives,t+β2Prioritys,t+β3Xs,t+θs+δt+εs,t (1)

We also generate static estimates based on the more flexible dynamic event study specification in Equation (2) following Meinhofer et al. (2021b):

Ys,t=β0+λPreD1s,j∈Pre+∑−1−4λ1jD1s,j+∑14λ1jD1s,j+λPostD1s,j∈Post+λ2P2s,t+β3Xs,t+θs+δt+εs,t (2)

Equation (2) controls for four leads and lags of the policy P1 (either Punitive s,t or Priority s,t ) and bins distant relative periods. Lags and leads can help evaluate if there are dynamic treatment effects and whether the parallel trends assumption needed for the validity of difference‐in‐differences models appears reasonable. Lags and leads are captured in the dummy variables D1 s, j  (j = t − k + 1), where k is P1's effective date in state s, j is the period relative to P1's effective date, and P2 is an indicator for the other policy. D1 s, j ∈ Pre captures all relative periods such that j < −4 into a single indicator and D1 s, j ∈ Post captures all relative periods such that j > 4 into a single indicator. The reference group is j = 0, the period prior to the effective date. Using Equation (2), we generate overall treatment effect parameters during the first 4 years of exposure to the policy by averaging estimates from the first four lags since policy implementation. 2 Each year receives equal weight. Focusing on the first 4 years can help mitigate complications from the changing composition of treated states, which is more extreme for distant relative periods and may compromise the interpretation of dynamic effects as well as standard errors.

We also exclude always treated units from the sample and re‐estimate Equations (1) and (2) following Sun and Abraham (2021). In particular, when estimating β 1 in Equation (1), we run a regression that drops all states for which Punitive s,t  = 1 for all years in the sample period of 2008–2018. Likewise, when estimating β 2 in Equation (1), we run a different regression that instead drops all states for which Priority s,t  = 1 for all years in our sample period. We repeat this approach when estimating Equation (2). Finally, we graphically depict leads and lags estimated in Equation (2) to visually inspect any anticipation or differential pre‐trends and/or dynamics in treatment effects. This study was approved by Weill Cornell Medicine's Institutional Review Board.

4. RESULTS

4.1. Summary statistics

We present summary statistics for the HCUP data in Table 1 and for the NVSS data in Table 2. The rate of NDWS is 5.10 per 1000 newborn hospitalizations. The rate of NDWS is higher in births that are financed by Medicaid and self‐paid, in medium metropolitan, small metropolitan, and rural areas, and in communities in the bottom two income quartiles. The rate of NDWS is lower in births financed by private payers, in large center and fringe metropolitan areas, and in higher income areas. The proportion of births with low gestational age, low birth weight, very low birth weight, and with prenatal care is 0.081, 0.064, 0.011 and 0.984, respectively.

4.2. Punitive prenatal substance use policies and neonatal outcomes

Table 3 reports estimates of the effect of punitive PSUPs on the log of perinatal outcomes, including neonatal drug withdrawal syndrome diagnoses per 1000 hospitalizations and the proportion of births with low gestational age, low birth weight, very low birth weight, and prenatal care. For each outcome, we provide estimates with always treated states dropped from the sample and with always treated states included. Coefficient estimates from Equation (1) are presented in Panel A and from Equation (2) in Panel B. Panel B estimates capture treatment effects in the first 4 years since policy implementation. We evaluate the parallel trends assumption with event study plots in Figure 1.

TABLE 3.

Effect of punitive prenatal substance use policies on the log of outcomes

(1) (2) (3) (4) (5)
Neonatal drug withdrawal Low gestational age Low birth Weight Very low birth Weight Prenatal care
Panel A: Static two‐way fixed effects specification
Drop always treated statessss 0.111* −0.007 0.002 0.017 −0.003
(0.065) (0.009) (0.007) (0.014) (0.002)
N 311 1452 1452 1452 1320
States 29 33 33 33 30
Keep all states 0.176** −0.011 0.002 0.012 −0.004*
(0.079) (0.009) (0.007) (0.012) (0.002)
N 490 2244 2244 2244 2112
States 46 51 51 51 48
Panel B: Event study specification
Drop always treated states 0.096** −0.004 −0.002 0.004 −0.004***
(0.044) (0.011) (0.008) (0.016) (0.001)
N 311 1452 1452 1452 1320
States 29 33 33 33 30
Keep all states 0.104** −0.007 −0.000 −0.000 −0.004***
(0.042) (0.011) (0.008) (0.014) (0.001)
N 490 2244 2244 2244 2112
States 46 51 51 51 48
Open in a new tab

Note:Effect of punitive prenatal substance use policies (PSUPs) on the natural log of neonatal drug withdrawal syndrome, low gestational age, low birth weight, very low birth weight, and prenatal care. Outcomes in Column (1) were drawn from the 2008–2018 HCUP FastStats and the unit of analysis is a state‐year. Outcomes in Columns (2)‐(5) were drawn from the 2008–2019 NVSS Natality Files and the unit of analysis is a state‐year‐quarter of conception for conception years 2008–2018. Models were estimated with least squares and weighted by the number of births in a state‐year. All models include control variables, state and year fixed‐effects, and control for priority SUD treatment PSUPs. State clustered standard errors are in parentheses.

***p < 0.01, **p < 0.05, *p < 0.1.

FIGURE 1.

FIGURE 1

Open in a new tab

Effect of punitive prenatal substance use policies (PSUPs) on the log of outcomes. The rate of neonatal drug withdrawal syndrome was drawn from the 2008–2018 Healthcare Cost and Utilization Project (HCUP) FastStats. The proportion of births with low gestational age, low birth weight, and any prenatal care were drawn from the 2008–2019 National Vital Statistics System (NVSS) Natality Files and reflect conception years 2008–2018. Coefficient estimates and 95% confidence intervals are based on event study models that include PSUP lags and leads (Equation 2). Models are weighted by the number of births in a state‐year, include State and year fixed‐effects, and account for control variables as well as for the priority treatment PSUP. Always treated punitive states are dropped

We find that punitive PSUPs increase the rate of NDWS by 11%–18% when estimating coefficients with the static TWFE specification, and by 10% when estimating coefficients with the event study specification. We find no statistically significant impact on low gestational age, low birth weight, and very low birth weight regardless of the specification. The coefficient estimates and standard errors are generally small, indicating large effects can be ruled out. We do find small but statistically significant reductions in prenatal care, with coefficient estimates ranging from −0.003 to −0.004, indicating that punitive PSUPs reduce prenatal care by approximately 0.03%–0.04%. Event study plots in Figure 1 are consistent with these findings and show no evidence of divergent pre‐period trends across treated and untreated.

4.3. Priority substance use disorder treatment prenatal substance use policies and neonatal outcomes

Estimates of the effect of priority SUD treatment PSUPs on neonatal outcomes are presented in Figure 2 and Table 4. Regression coefficient estimates in Table 2 show that priority SUD treatment does not impact NDWS, but does impact other neonatal outcomes. The proportion of births with low gestational age reduced by 1% in the event study specification and 2% in the TWFE specification when the sample includes all states; however, the coefficient estimate is not statistically significant when always treated states are dropped. Priority SUD treatment also reduces the proportion of births that are low birth weight by about 2% and the results are similar across specifications and samples. In the event study model in Panel B, coefficient estimates for Column (4) of −0.032 and −0.034 indicate that priority SUD treatment PSUPs reduce the proportion of births classified as very low birth weight by 3%. Effects on very low birth weight are not statistically significant in the TWFE model. Priority SUD treatment PSUPs are associated with a 0.4%–0.5% increase in prenatal care utilization across specifications. Event study plots in Figure 2 are consistent with these findings and show no evidence of divergent pre‐period trends across treated and untreated states.

FIGURE 2.

FIGURE 2

Open in a new tab

Effect of priority substance use disorder (SUD) treatment prenatal substance use policies (PSUPs) on the log of outcomes. The rate of neonatal drug withdrawal syndrome was drawn from the 2008–2018 Healthcare Cost and Utilization Project (HCUP) FastStats. The proportion of births with low gestational age, low birth weight, and any prenatal care were drawn from the 2008–2019 National Vital Statistics System (NVSS) Natality Files and reflect conception years 2008–2018. Coefficient estimates and 95% confidence intervals are based on event study models that include PSUP lags and leads (Equation 2). Models are weighted by the number of births in a state‐year, include state and year fixed‐effects, and account for control variables as well as for the punitive PSUP. Always treated priority treatment states are dropped

TABLE 4.

Effect of priority substance use disorder treatment prenatal substance use policies on the log of outcomes

(1) (2) (3) (4) (5)
Neonatal drug withdrawal Low gestational age Low birth Weight Very low birth Weight Prenatal care
Panel A: Static two‐way fixed effects specification
Drop always treated states −0.050 −0.011 −0.022*** −0.010 0.004**
(0.105) (0.007) (0.006) (0.011) (0.002)
N 363 1672 1672 1672 1584
States 34 38 38 38 36
Keep all states −0.060 −0.022*** −0.022*** −0.012 0.004**
(0.117) (0.005) (0.005) (0.009) (0.002)
N 490 2244 2244 2244 2112
States 46 51 51 51 48
Panel B: Event study specification
Drop always treated states −0.063 −0.008 −0.016*** −0.032** 0.005
(0.049) (0.006) (0.004) (0.015) (0.003)
N 363 1672 1672 1672 1584
States 34 38 38 38 36
Keep all states −0.062 −0.013*** −0.016*** −0.034** 0.005*
(0.051) (0.005) (0.002) (0.012) (0.003)
N 490 2244 2244 2244 2112
States 46 51 51 51 48
Open in a new tab

Note: Effect of priority substance use disorder treatment prenatal substance use policies (PSUPs) on the natural log of neonatal drug withdrawal syndrome, low gestational age, low birth weight, very low birth weight, and prenatal care. Outcomes in Column (1) were drawn from the 2008–2018 HCUP FastStats and the unit of analysis is a state‐year. Outcomes in Columns (2)‐(5) were drawn from the 2008–2019 NVSS Natality Files and the unit of analysis is a state‐year‐quarter of conception for conception years 2008–2018. Models were estimated with least squares and weighted by the number of births in a state‐year. All models include control variables, state and year fixed‐effects, and control for punitive PSUPs. State clustered standard errors are in parentheses.

***p < 0.01, **p < 0.05, *p < 0.1.

4.4. Effects of prenatal substance use policies on NDWS by subpopulation

Effects of PSUPs may vary across subpopulations due to differences in substance use, prenatal care, and access to SUD treatment, among other factors. In Figure 3, we present estimates of punitive PSUPs on NDWS separately for births financed by Medicaid and private insurance, by geographic area, and by community income quartile. For most subgroups, estimates are imprecize and we are not able to rule out economically meaningful effects. However, punitive PSUPs do increase NDWS among births financed by Medicaid (10% increase) and in lower‐income communities (10% increase). Figure 4 presents estimates of priority treatment PSUPs on NDWS by subpopulation. Priority treatment PSUPs have no effect on NDWS for any subpopulation studied, except for a reduction in NDWS in communities in the second‐lowest income quartile (14% decrease).

FIGURE 3.

FIGURE 3

Open in a new tab

Effect of punitive prenatal substance use policies (PSUPs) on the log of neonatal drug withdrawal syndrome, by subpopulation. The rate of neonatal drug withdrawal syndrome was drawn from the 2008–2018 Healthcare Cost and Utilization Project (HCUP) FastStats. Coefficient estimates and 95% confidence intervals are a linear combination estimates from the four intervention years from event study models (Equation 2). Models are weighted by the number of births in a state‐year, include state and year fixed‐effects, and account for control variables as well as for the priority substance use disorder (SUD) treatment PSUP. Always treated punitive states are dropped

FIGURE 4.

FIGURE 4

Open in a new tab

Effect of priority substance use disorder (SUD) treatment prenatal substance use policies (PSUPs) on the log of neonatal drug withdrawal syndrome, by subpopulation. The rate of neonatal drug withdrawal syndrome was drawn from the 2008–2018 Healthcare Cost and Utilization Project (HCUP) FastStats. Coefficient estimates and 95% confidence intervals are a linear combination estimates from the four intervention years from event study models (Equation 2). Models are weighted by the number of births in a state‐year, include state and year fixed‐effects, and account for control variables as well as for the punitive PSUP. Always treated priority SUD treatment states are dropped

4.5. Robustness checks

We assess the robustness of Equation (1) estimates in Tables A2–A10. In Tables A2–A5, we estimate regressions for each policy separately with and without controls in columns (1) through (4). Column (5) contains the TWFE coefficient estimates with all states from Tables 3 and 4 for comparison purposes. Estimates for punitive and priority SUD treatment PSUPs in Tables A2–A5 are similar to main estimates in Tables 3 and 4.

In Tables A6 and A7, we show the robustness of results to including additional secondary control variables discussed in Section 3.4. In Table A6, we repeatedly estimate Equation (1) but including the additional control listed in the first column. Note that each coefficient estimate is estimated with a different regression. The coefficient estimates are robust to the addition of other drug policy variables in the model. In Table A7, we estimate Equation (1) with the main controls and all the secondary controls listed in Table A6. Punitive PSUPs increase NDWS, although the magnitude falls slightly to be in a similar range as the event study coefficient estimates. As before, punitive PSUPs have no effects on low gestational age, low or very low birth weight, or prenatal care receipt. In line with previous findings, priority SUD treatment PSUPs reduce the proportion of births with low gestational age and low birth weight, and increase prenatal care. In this model, priority SUD treatment PSUPs reduce very low birth weight, which is not a previously statistically significant result for Equation (1) coefficient estimates.

In Table A8, we show the robustness of results to alternative PSUP definitions as described in Section 3.3. Using variation from stronger PSUPs does not affect the results. Some states' laws included sunset provisions or were repealed, which is a source of variation that we did not use in the main specification. 3 In Table A9, we account for the repeals by setting policy variables to zero in the year that a law was no longer in place. Results are similar when accounting for repeals, except the increase in prenatal care from priority treatment PSUPs disappears.

When treatment effects are constant across states and time, Equation (1) can estimate the average treatment effect on the treated (ATT) under the standard parallel trends assumption (Goodman‐Bacon, 2021). However, Equation (1) can generate biased estimates when treatment effects are heterogenous across states or time, even when the parallel trends assumption holds. This bias emerges as TWFE estimates are a weighted sum of ATTs in each state and time, which compares the evolution of the outcome between consecutive time periods across pairs of states. In staggered implementation designs such as ours, the “comparison group” in some of those comparisons is treated in both periods and its treatment effect in the second period gets differenced out, generating negative weights (Goodman‐Bacon, 2021). The negative weights can lead to biased estimates, including sign reversals, when effects are heterogeneous. While the event study specification in Equation (2) is robust to treatment effect heterogeneity over time, it is not robust to treatment effect heterogeneity across states. In Table A10, we estimate difference‐in‐differences coefficients using the methods described in Callaway and Sant’Anna (2021), which address bias from heterogeneity across states or time. Callaway and Sant’Anna propose a “Group Average Treatment Effect on the Treated” (GATT). To estimate the GATT, treated units that take treatment at the same date are segregated into groups, always treated units are excluded from the sample, and “clean” comparisons (i.e., comparing groups that take treatment with groups that are untreated) are constructed, these clean comparisons are then weighted by group size to construct the overall GATT. We use the “not treated yet” group as the comparison group and apply doubly robust regressions proposed by Sant’Anna and Zhao (2020). Standard errors are estimated using a bootstrap procedure (999 repetitions) and account for within‐state clustering. Following Callaway and Sant’Anna (2021), we omit covariates. While estimates are generally similar when using this approach, we do find a statistically significant decrease of 15% in NDWS following priority treatment PSUP adoption.

On the whole, Tables A2–A10 show that the main findings that punitive PSUPs harm infant health through increases in NDWS and priority treatment PSUPs improve infant health by reducing low gestational age and low birth weight births are robust. Prenatal care results may be somewhat sensitive to the choice of model (i.e., event study or TWFE) and whether repeals are accounted for.

5. DISCUSSION

This study provides comprehensive evidence of the impact of punitive and priority treatment PSUPs on birth outcomes and is the first to examine the mediating role of prenatal care as well as heterogeneity by socioeconomic characteristics. Our findings suggest that punitive PSUPs are not effective in reducing adverse birth outcomes and instead may increase neonatal drug withdrawal syndrome and reduce prenatal care utilization. Back‐of‐the envelope calculations imply that on average, punitive PSUP implementation would result in approximately 49 additional infants born with NDWS and 270 fewer women seeking prenatal care in the state each year. Priority treatment PSUPs do not appear to impact neonatal drug withdrawal syndrome, but may lead to small reductions in low gestational age, low birth weight, and very low birth weight, and increases in prenatal care utilization. Back‐of‐the envelope calculations imply that on average, priority treatment PSUPs would result in 80 fewer low gestational age births, 88 fewer low birth weight births, and 326 more women engaging in prenatal care in an implementing state each year. As states move toward more punitive policies, our findings are concerning.

Our findings for punitive PSUPs are in line with previous qualitative studies suggesting that by inducing fear of stigma or punishment, punitive policies may unintendedly deter pregnant women from seeking SUD treatment and prenatal care (Angelotta et al., 2016; Christian, 2004; Figdor & Kaeser, 1998; Jarlenski et al., 2018; Jessup et al., 2003; Miranda et al., 2015; Roberts et al., 2010, 2011; Stone, 2015; Thomas et al., 2018). Our findings are also in line with quantitative studies suggesting punitive PSUPs induce worse outcomes, including reductions in SUD treatment utilization among pregnant women (Atkins & Durrance, 2020; Kozhimannil et al., 2019) and increases in foster care entries among infants (Atkins & Durrance, 2021; Sanmartin et al., 2019, 2020). A study using variation from eight states and five changing punitive PSUPs find a 24%–30% increase in NDWS in the first years of implementation (Faherty et al., 2019). Our estimates using variation from 46 states and 11 changing punitive PSUPs implied similar, albeit smaller, effects. Similarly, while not statistically distinguishable from zero, Atkins and Durrance (2020) report a 10%–14% increase in NDWS following adoption of a punitive PSUP. Our findings are comparable to coefficient estimates reported in previous related work. Women of lower socioeconomic status can face greater exposure of their conditions as they seek publicly funded services, which may explain our findings that Medicaid beneficiaries and newborns in lower income communities are most affected by punitive PSUPs.

To our knowledge, our study documents the first estimates of the impact of priority treatment PSUPs on birth outcomes. There are at least four possible explanations for our null effects on NDWS but decreases in low gestational age and low birth weight following the implementation of priority SUD treatment PSUPs. First, priority treatment PSUPs may induce variation in NDWS primarily through changes in prenatal exposure to opioids and corresponding OUD treatment, while variation in low gestational age and low birth weight may result from changes in prenatal exposure to any substances and corresponding SUD treatment. As such, differences in the price elasticity of demand from opioids versus other substances, or in the noxiousness of opioids versus the noxiousness of other substances imply that the effect of PSUPs on prenatal substance use and subsequent birth outcomes may depend on the type of substance being used. Second, OUD treatment for pregnant women with OUD should entail behavioral therapy and methadone or buprenorphine, which are opioid medications for treating OUD. These medications are considered the gold standard for OUD treatment in pregnancy, but can be associated with NDWS risk despite improving other pregnancy outcomes (Goodman et al., 2019). As such, NDWS need not change if priority treatment PSUPs reduce in illegal opioid use and increase methadone or buprenorphine treatment, despite improvements in other pregnancy outcomes. Third, another explanation is that opioid treatment programs, which provide OUD treatment for persons with opioid use disorder, already require that pregnant women receive priority access to the opioid treatment program. As such, priority treatment PSUPs may have greater impacts on pregnant women with other substance use disorders aside from opioids. Lastly, given statistically significant decreases in NDWS following priority treatment PSUP adoption when using the estimator proposed by Callaway and Sant’Anna (2021), it is possible that bias from treatment effect heterogeneity may explain statistically insignificant effects when using the baseline TWFE estimator.

Our findings have important policy implications. A recent amendment to the CAPTA made by the Comprehensive Addiction and Recovery Act in 2016 has been interpreted by some states as requiring all substance‐exposed newborns to be reported to CPS for abuse or neglect, resulting in more stringent PSUPs. 4 Leading medical organizations including the American Congress of Obstetricians and Gynecologists and the American Medical Association oppose punitive PSUPs. They argue these policies may deter pregnant women from seeking care, increasing risks of fetal harm. Our findings support these recommendations. Another key implication is the need to focus on enhancing behavioral healthcare services for pregnant women (Meinhofer et al., 2020).

This study has several limitations. First, the design and enforcement of PSUPs may differ across states, and these variations might impact outcomes. Second, our sample period spans 11 years and is based on recent data, which captures variation from late‐adopted PSUPs. As such, our findings may not be representative of studies using variation from early‐adopted PSUPs. This difference in policy variation is noteworthy since the first wave of PSUPs primarily occurred in the 19980s and 1990s. Third, characterizing complex and heterogenous legal statutes with a binary indicator likely misses some nuanced differences in PSUP stringency and scope that policymakers should consider (i.e., impact of weaker PSUPs), and we lack power to test for many differential effects that would be of interest. Fourth, some of our estimates are only marginally statistically significant. Finally, although we control for a large set of potential confounders, unobserved factors could be biasing our estimates.

In conclusion, we find that punitive PSUPs are associated with increases in adverse perinatal health outcomes while priority treatment PSUPs are associated with reductions in adverse perinatal health outcomes. More research is needed to understand the impact of PSUPs using different outcomes and subpopulations.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Supporting information

Supporting Information S1

Click here for additional data file. (75.1KB, docx)

ACKNOWLEDGEMENTS

We thank Yaralin Negron and Nadia Tabatabaeepour for their research assistance. This work was supported by the Robert Wood Johnson Foundation RWJF77962, the National Institute on Drug Abuse K01DA051777, and the University of North Carolina Wilmington Charles L. Cahill Grant.

Meinhofer, A. , Witman, A. , Maclean, J. C. , & Bao, Y. (2022). Prenatal substance use policies and newborn health. Health Economics, 31(7), 1452–1467. 10.1002/hec.4518

ENDNOTES

2

Overall treatment effect parameters generated with the post estimation Stata command lincom.

3

The punitive PSUP in Tennessee had a sunset clause in 2016 and Oklahoma repealed in 2018. The priority SUD treatment PSUP in Louisiana was repealed in 2015, in Maryland was repealed in 2015, and in Texas was repealed in 2015.

DATA AVAILABILITY STATEMENT

The HCUP FastStats Neonatal Abstinence Syndrome data that support some of the findings of this study are made openly available by the AHRQ at https://www.hcup‐us.ahrq.gov/faststats/NASServlet. The NVSS Natality data that support some of the findings of this study are available from the Centers for Disease Control and Prevention. Restrictions apply to the availability of these data, which were used under license for this study.

REFERENCES

  1. Administration for Children and Families . (2017). Guidance on amendments made to the child abuse prevention and treatment act (CAPTA) by public law 114‐198, the comprehensive addiction and Recovery act of 2016. https://www.acf.hhs.gov/cb/policy‐guidance/pi‐17‐02 [Google Scholar]
  2. Administration for Children & Families . (2021). State statutes search, definitions of child abuse and neglect. Child Welfare Information Gateway. https://www.childwelfare.gov/topics/systemwide/laws‐policies/state/?CWIGFunctionsaction=statestatutes:main&CWIGFunctionspk=1 [Google Scholar]
  3. Angelotta, C. , Weiss, C. J. , Angelotta, J. W. , & Friedman, R. A. (2016). A moral or medical problem? The relationship between legal penalties and treatment practices for opioid use disorders in pregnant women. Women's Health Issues, 26(6), 595–601. [DOI] [PubMed] [Google Scholar]
  4. Atkins, D. N. , & Durrance, C. P. (2020). State policies that treat prenatal substance use as child abuse or neglect fail to achieve their intended goals: Study examines the effect of state policies that treat prenatal substance use as child abuse or neglect on the incidence of neonatal abstinence syndrome and other factors. Health Affairs, 39(5), 756–763. [DOI] [PubMed] [Google Scholar]
  5. Atkins, D. N. , & Durrance, C. P. (2021). The impact of state‐level prenatal substance use policies on infant foster care entry in the United States. Children and Youth Services Review, 106194. [Google Scholar]
  6. Behnke, M. , Smith, V. C. , Behnke, M. , Smith, V. C. , Levy, S. , Ammerman, S. D. , Gonzalez, P. K. , Ryan, S. A. , Smith, V. C. , Wunsch, M. M. J. , Papile, L.‐A. , Baley, J. E. , Carlo, W. A. , Cummings, J. J. , Kumar, P. , Polin, R. A. , Tan, R. C. , & Watterberg, K. L. , & Committee on Substance Abuse . (2013). Prenatal substance abuse: Short‐and long‐term effects on the exposed fetus. Pediatrics, 131(3), e1009–e1024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Callaway, B. , & Sant’Anna, P. H. (2021). Difference‐in‐differences with multiple time periods. Journal of Econometrics, 225(2), 200–230. [Google Scholar]
  8. Center for Behavioral Health Statistics and Quality . (2018). 2018 national survey on substance use and health: Detailed tables. Substance Abuse and Mental Health Services Administration. [Google Scholar]
  9. Christian, S. M. (2004). Substance‐exposed newborns: New Federal Law raises some old issues. National Conference of State Legislatures, Children's Policy Initiative.
  10. Desai, R. J. , Huybrechts, K. F. , Hernandez‐Diaz, S. , Mogun, H. , Patorno, E. , Kaltenbach, K. , Kerzner, L. S. , & Bateman, B. T. (2015). Exposure to prescription opioid analgesics in utero and risk of neonatal abstinence syndrome: Population based cohort study. BMJ, 350. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Faherty, L. J. , Heins, S. , Kranz, A. M. , Patrick, S. W. , & Stein, B. D. (2022). Association between punitive policies and neonatal abstinence syndrome among Medicaid‐insured infants in complex policy environments. Addiction, 117(1), 162–171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Faherty, L. J. , Kranz, A. M. , Russell‐Fritch, J. , Patrick, S. W. , Cantor, J. , & Stein, B. D. (2019). Association of punitive and reporting state policies related to substance use in pregnancy with rates of neonatal abstinence syndrome. JAMA Network Open, 2(11), e1914078e1914078. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Figdor, E. , & Kaeser, L. (1998). Concerns mount over punitive approaches to substance abuse among pregnant women. The Guttmacher Report on Public Policy, 1(5), 3–5. [Google Scholar]
  14. Goodman, D. , Whalen, B. , & Hodder, L. C. (2019). It’s time to support, rather than punish, pregnant women with substance use disorder. JAMA Network Open, 2(11), e1914135. [DOI] [PubMed] [Google Scholar]
  15. Goodman‐Bacon, A. (2021). Difference‐in‐differences with variation in treatment timing. Journal of Econometrics, 225(2), 254–277. [Google Scholar]
  16. Guttmacher Institute . (2019). State policies on substance use during pregnancy. Guttmacher Institute. [Google Scholar]
  17. HCUP FastStats . (2019). Neonatal abstinence syndrome (NAS) among newborn hospitalizations. https://www.hcup‐us.ahrq.gov/faststats/NASServlet?radio‐2=on&location1=US&characteristic1=01C11&location2=&characteristic2=01C11&expansionInfoState=hide&dataTablesState=hide&definitionsState=show&exportState=hide [Google Scholar]
  18. Huybrechts, K. F. , Bateman, B. T. , Desai, R. J. , Hernandez‐Diaz, S. , Rough, K. , Mogun, H. , Kerzner, L. S. , Davis, J. M. , Stover, M. , Bartels, D. , Cottral, J. , & Patorno, E. (2018). Risk of neonatal drug withdrawal after intrauterine co‐exposure to opioids and psychotropic medications: Cohort study. BMJ, 358, j3326. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Jarlenski, M. , Hogan, C. , Bogen, D. L. , Chang, J. C. , Bodnar, L. M. , & van Nostrand, E. (2018). Characterization of us state laws requiring health care provider reporting of perinatal substance use. Women's Health Issues, 27(3), 264–270. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jessup, M. A. , Humphreys, J. C. , Brindis, C. D. , & Lee, K. A. (2003). Extrinsic barriers to substance abuse treatment among pregnant drug dependent women. Journal of Drug Issues, 33(2), 285–304. [Google Scholar]
  21. Kozhimannil, K. B. , Dowd, W. N. , Ali, M. M. , Novak, P. , & Chen, J. (2019). Substance use disorder treatment admissions and state‐level prenatal substance use policies: Evidence from a national treatment database. Addictive Behaviors, 90, 272–277. [DOI] [PubMed] [Google Scholar]
  22. Kutner, M. H. , Nachtsheim, C. J. , & Neter, J. (2004). Applied linear regression models (4th ed.). McGraw‐Hill Irwin. [Google Scholar]
  23. Leech, A. A. , Cooper, W. O. , McNeer, E. , Scott, T. A. , & Patrick, S. W. (2020). Neonatal abstinence syndrome in the United States, 2004–16. Health Affairs, 39(5), 764–767. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Maclean, J. C. , Mallatt, J. , Ruhm, C. J. , & Simon, K. (2020). Economic studies on the opioid crisis: A review [National Bureau of Economic Research Working Paper Series]. National Bureau of Economic Research. [Google Scholar]
  25. Maeda, A. , Bateman, B. T. , Clancy, C. R. , Creanga, A. A. , & Leffert, L. R. (2014). Opioid abuse and dependence during pregnancy: Temporal trends and obstetrical outcomes. The Journal of the American Society of Anesthesiologists, 121(6), 1158–1165. [DOI] [PubMed] [Google Scholar]
  26. Meinhofer, A. , & Angleró‐Díaz, Y. (2019). Trends in foster care entry among children removed from their homes because of parental drug use, 2000 to 2018. JAMA Pediatrics, 173(9), 881–883. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Meinhofer, A. , Hinde, J. M. , & Ali, M. M. (2020). Substance use disorder treatment services for pregnant and postpartum women in residential and outpatient settings. Journal of Substance Abuse Treatment, 110, 9–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Meinhofer, A. , & Rubli, A. (2021a). Illegal drug market responses to state recreational cannabis laws. Addiction, 116(12), 3433–3443. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Meinhofer, A. , & Witman, A. E. (2018). The role of health insurance on treatment for opioid use disorders: Evidence from the Affordable Care Act Medicaid expansion. Journal of Health Economics, 60, 177–197. [DOI] [PubMed] [Google Scholar]
  30. Meinhofer, A. , Witman, A. E. , Hinde, J. M. , & Simon, K. (2021b). Marijuana liberalization policies and perinatal health. Journal of Health Economics, 80, 102537. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Miranda, L. , Dixon, V. , & Reyes, C. (2015). How states handle drug use during pregnancy. ProPublica. [Google Scholar]
  32. Patrick, S. W. , Faherty, L. J. , Dick, A. W. , Scott, T. A. , Dudley, J. , & Stein, B. D. (2019). Association among county‐level economic factors, clinician supply, metropolitan or rural location, and neonatal abstinence syndrome. JAMA, 321(4), 385–393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Patrick, S. W. , Schumacher, R. E. , Benneyworth, B. D. , Krans, E. E. , McAllister, J. M. , & Davis, M. M. (2012). Neonatal abstinence syndrome and associated health care expenditures: United States, 2000‐2009. JAMA, 307(18), 1934–1940. [DOI] [PubMed] [Google Scholar]
  34. Reddy, J. , & Schiff, D. (2022). A response to “Association between punitive policies and neonatal abstinence syndrome among Medicaid‐insured infants in complex policy environments”. Addiction, 117(3), 833–834. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Roberts, S. C. , Berglas, N. F. , Subbaraman, M. S. , Mericle, A. , Thomas, S. , & Kerr, W. C. (2019). Racial differences in the relationship between alcohol/pregnancy policies and birth outcomes and prenatal care utilization: A legal epidemiology study. Drug and Alcohol Dependence, 201, 244–252. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Roberts, S. C. , & Nuru‐Jeter, A. (2010). Women's perspectives on screening for alcohol and drug use in prenatal care. Women's Health Issues, 20(3), 193–200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Roberts, S. C. , & Pies, C. (2011). Complex calculations: How drug use during pregnancy becomes a barrier to prenatal care. Maternal and Child Health Journal, 15(3), 333–341. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Sanmartin, M. X. , Ali, M. M. , & Lynch, S. (2019). Foster care admissions and state‐level criminal justice‐focused prenatal substance use policies. Children and Youth Services Review, 102, 102–107. [Google Scholar]
  39. Sanmartin, M. X. , Ali, M. M. , Lynch, S. , & Aktas, A. (2020). Association between state‐level criminal justice–focused prenatal substance use policies in the US and substance use–related foster care admissions and family reunification. JAMA Pediatrics, 174(8), 782–788. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Sant’Anna, P. H. , & Zhao, J. (2020). Doubly robust difference‐in‐differences estimators. Journal of Econometrics, 219(1), 101–122. [Google Scholar]
  41. Shaw, M. R. , Grant, T. , Barbosa‐Leiker, C. , Fleming, S. E. , Henley, S. , & Graham, J. C. (2015). Intervention with substance‐abusing mothers: Are there rural–urban differences? American Journal on Addictions, 24(2), 144–152. [DOI] [PubMed] [Google Scholar]
  42. Sheather, S. (2009). A modern approach to regression with R. Springer. ISBN 978‐0‐387‐09607‐0. [Google Scholar]
  43. Stone, R. (2015). Pregnant women and substance use: Fear, stigma, and barriers to care. Health & Justice, 3(1), 2. [Google Scholar]
  44. Subbaraman, M. S. , Thomas, S. , Treffers, R. , Delucchi, K. , Kerr, W. C. , Martinez, P. , & Roberts, S. C. (2018). Associations between state‐level policies regarding alcohol use among pregnant women, adverse birth outcomes, and prenatal care utilization: Results from 1972 to 2013 vital statistics. Alcoholism: Clinical and Experimental Research, 42(8), 1511–1517. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Sun, L. , & Abraham, S. (2021). Estimating dynamic treatment effects in event studies with heterogeneous treatment effects. Journal of Econometrics, 225(2), 175–199. [Google Scholar]
  46. Thomas, S. , Treffers, R. , Berglas, N. F. , Drabble, L. , & Roberts, S. C. (2018). Drug use during pregnancy policies in the United States from 1970 to 2016. Contemporary Drug Problems, 45(4), 441–459. [Google Scholar]
  47. TN code § 39‐13‐107. 2014. https://law.justia.com/codes/tennessee/2014/title‐39/chapter‐13/part‐1/section‐first/ [Google Scholar]
  48. Villapiano, N. L. , Winkelman, T. N. , Kozhimannil, K. B. , Davis, M. M. , & Patrick, S. W. (2018). Rural and urban differences in neonatal abstinence syndrome and maternal opioid use, 2004 to 2013. JAMA Pediatrics, 171(2), 194–196. [DOI] [PubMed] [Google Scholar]
  49. Winkelman, T. N. , Villapiano, N. , Kozhimannil, K. B. , Davis, M. M. , & Patrick, S. W. (2018). Incidence and costs of neonatal abstinence syndrome among infants with Medicaid: 2004–2014. Pediatrics, 141(4). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

Supporting Information S1

Click here for additional data file. (75.1KB, docx)

Data Availability Statement

The HCUP FastStats Neonatal Abstinence Syndrome data that support some of the findings of this study are made openly available by the AHRQ at https://www.hcup‐us.ahrq.gov/faststats/NASServlet. The NVSS Natality data that support some of the findings of this study are available from the Centers for Disease Control and Prevention. Restrictions apply to the availability of these data, which were used under license for this study.

Articles from Health Economics are provided here courtesy of Wiley

RESOURCES