Pregnancy-specific alcohol policies and admissions to substance use disorder treatment for pregnant people in the USA

Abstract

Aims

We examined relationships between pregnancy-specific alcohol policies and admissions to substance use disorder treatment for pregnant people in the USA.

Methods

We merged state-level policy and treatment admissions data for 1992–2019. We aggregated data by state-year to examine effects of nine pregnancy-specific alcohol policies on the number of admissions of pregnant women where alcohol was reported as the primary, secondary, or tertiary substance related to the treatment episode (N = 1331). We fit Poisson models that included all policy variables, state-level controls, fixed effects for state and year, state-specific time trends, and an offset variable of the number of pregnancies in the state-year to account for differences in population size and fertility.

Results

When alcohol was reported as the primary substance, civil commitment [incidence rate ratio (IRR) 1.45, 95% CI: 1.10–1.89] and reporting requirements for assessment and treatment purposes [IRR 1.36, 95% CI: 1.04–1.77] were associated with greater treatment admissions. Findings for alcohol as primary, secondary, or tertiary substance were similar for civil commitment [IRR 1.31, 95% CI: 1.08–1.59] and reporting requirements for assessment and treatment purposes [IRR 1.21, 95% CI: 1.00–1.47], although mandatory warning signs [IRR 0.84, 95% CI: 0.72–0.98] and priority treatment for pregnant women [IRR 0.88, 95% CI: 0.78–0.99] were associated with fewer treatment admissions. Priority treatment findings were not robust in sensitivity analyses. No other policies were associated with treatment admissions.

Conclusions

Pregnancy-specific alcohol policies related to greater treatment admissions tend to mandate treatment rather than make voluntary treatment more accessible, raising questions of ethics and effectiveness.

Short Summary: Civil commitment policies and reporting requirements for assessment and treatment purposes were consistently associated with greater treatment admissions, while mandatory warning signs and priority treatment policies were, in some cases, related to fewer treatment admissions.

Introduction

For years, the scholarly and public health focus related to alcohol use during pregnancy has concentrated on epidemiologic studies that examine trends in alcohol consumption among pregnant people and on identifying harms and thresholds for harms related to drinking during pregnancy (Cdc. 1997; Cdc. 2009; O'leary and Bower 2012; Flak et al. 2014; Denny et al. 2019; Kesmodel et al. 2019). Until recently, research on interventions that could reduce harms related to drinking during pregnancy has rarely considered the policy level. During the 2010s, a few studies in the USA examined the relationship between state-level pregnancy-specific alcohol policies and alcohol use during pregnancy as well as adverse birth outcomes and prenatal care utilization. Overall, this research found little evidence that these policies are effective at reducing alcohol consumption during pregnancy (Roberts et al. 2019) and that some policies—particularly mandatory warning signs, child abuse and neglect, and priority treatment policies—may contribute to increases in adverse birth outcomes and decreased prenatal care utilization (Subbaraman et al. 2018). The general pattern of limited effectiveness of pregnancy-specific alcohol policies is consistent with research on effects of pregnancy-specific drug policies (Faherty et al. 2019; Austin et al. 2022; Tabatabaeepour et al. 2022), which is not surprising as most pregnancy-specific policies focus on both alcohol and drugs (Thomas et al. 2018).

One pathway through which pregnancy-specific alcohol policies could affect health behaviors and outcomes is through the extent to which they facilitate or hinder use of substance use disorder treatment for pregnant people. One way to answer this question is by examining the effects of policies that specifically focus on facilitating treatment for pregnant people, i.e. policies that give priority for entering treatment to those who are pregnant. The two studies that have examined priority treatment policies specifically related to alcohol use have found associations between these policies and higher odds of any drinking during pregnancy, low birthweight, and preterm birth, but lower odds of inadequate prenatal care utilization (Subbaraman et al. 2018; Roberts et al. 2019). The research examining effects of priority treatment policies that focus specifically on drugs or on substance use generally has, with one exception (Meinhofer et al. 2022), not found relationships between these policies and outcomes (Kozhimannil et al. 2019; Faherty et al. 2022; Maclean et al. 2022; Tabatabaeepour et al. 2022). One study did find that when priority treatment policies were implemented along with other pregnancy-specific policies, the proportion of treatment admissions was greater among pregnant women compared with admissions among all women of reproductive age (Kozhimannil et al. 2019). It is important to note that the reason that priority treatment policies do not appear to have intended effects may not be inherent to the policies themselves, but due to factors related to their implementation, such as lack of availability of treatment services, limited state funding to implement the policies, and co-existing policies that may hinder patient and provider willingness and ability to obtain treatment (White et al. 2022).

Another way to consider how pregnancy-specific alcohol policies may relate to treatment is by examining the relationship between policies and treatment admissions for pregnant people, particularly those who are using alcohol. We are unaware of research that has examined the relationship among those who use alcohol, although there has been some research examining the relationship among pregnant people who use other drugs. This research has found that more punitive policies, such as those defining substance use during pregnancy as child abuse or neglect, were associated with lower quality treatment for pregnant women with opioid use disorder. Namely, the treatment did not include receiving medications for opioid use disorder (Angelotta et al. 2016), and a lower proportion of treatment admissions were found among pregnant women compared to all women of reproductive age when only punitive policies were in effect (Kozhimannil et al. 2019; Atkins and Durrance 2020). Research also found a higher proportion of admissions among pregnant women when policies that define substance use during pregnancy as child abuse or neglect or criminalize pregnant people’s substance use co-exist with other kinds of policies, such as priority treatment and mandated reporting to government authorities (Kozhimannil et al. 2019).

In this study, we examine the relationship between state-level pregnancy-specific alcohol policies and the likelihood of substance use treatment admissions related to alcohol for pregnant people in the USA. We use data from the Treatment Episode Data Set: Admissions (TEDS-A), a national data system of substance use treatment admissions collected annually since 1992. Prior studies have used TEDS-A data to examine the effects of pregnancy-specific drug policies on treatment admissions for pregnant women (Kozhimannil et al. 2019; Atkins and Durrance 2020), although methodological challenges limit the interpretation of findings. Such studies have not distinguished pregnancy-specific policies focusing on alcohol from those focusing on drugs and have not focused on treatment admissions that were primarily related to alcohol use. They have looked at subsets of policies or used policy data developed for political tracking rather than research purposes (Angelotta et al. 2016; Kozhimannil et al. 2019; Atkins and Durrance 2020), and, in some cases, did not control for other policies that may co-exist with the main policy of interest (Angelotta et al. 2016). Some studies have grouped all policies together (Kozhimannil et al. 2019), which may make sense conceptually but can also mask effects of specific policy subtypes. In addition, previous studies have focused on the proportion of women of reproductive age who were pregnant at the time of admission (Kozhimannil et al. 2019; Atkins and Durrance 2020). Conceptually, this is problematic because it assumes that similar numbers of people become pregnant and continue pregnancies across states.

Here, we bring together policy data collected and coded for research purposes along with substance use disorder treatment admissions data collected across the 50 states and the District of Columbia from 1992 to 2019 to examine the effect of nine different pregnancy-specific alcohol policies on treatment admissions. We focus specifically on policies that address alcohol use during pregnancy and concentrate analyses on pregnant people for whom alcohol was a named substance related to the treatment admission. We consider admissions of pregnant people relative to the number of pregnancies that resulted in a birth in the state and year, rather than among all women of reproductive age.

Methods

Data sources

This study combines data from the TEDS-A, state-level policy data from the Alcohol Policy Information System, and other secondary sources. All data are publicly available, deidentified, and collapsed so that the unit of analysis is state-year. The Institutional Review Board of the University of California, San Francisco considered this research exempt.

TEDS-A is a dataset of annual admissions of individuals ages 12 and older to substance use disorder treatment facilities that receive public funds, reported since 1992 by states to the federal Substance Abuse and Mental Health Services Administration. Facilities that receive any public funding (whether from their state or from federal grants) are required to report TEDS-A data annually, irrespective of whether the facilities are publicly or privately administered. Data are based on admissions; thus, an individual may be included more than once if they are admitted to treatment multiple times during a calendar year. The dataset includes more than 50 million admissions over nearly three decades. Among the TEDS-A variables are demographic and socioeconomic characteristics; pregnancy status at the time of admission; and the primary, secondary, and tertiary substances used at the time of admission. The TEDS-A codebook defines the variable “gender” as the admitted individual’s biological sex and codes all individuals who were pregnant at admission as female; gender identity is not collected. For this analysis, we collapsed TEDS-A datasets across 50 states and the District of Columbia for 1992–2019. Some states did not report TEDS-A data in some years (n = 40 state-years), and some states did not report admissions of any pregnant women in some years (n = 97 state-years). The final dataset includes 1331 of a possible 1428 state-years.

Policy data on pregnancy-specific alcohol policies were obtained from the Alcohol Policy Information System. Data for state control variables were obtained from the Centers for Disease Control and Prevention, National Institute on Alcohol Abuse and Alcoholism, and US Bureau of Labor Statistics.

Measures

Outcomes

State-year outcome measures for the analyses are (i) the number of treatment admissions of pregnant women where alcohol was reported as the primary substance related to the treatment episode and (ii) the number of admissions of pregnant women where alcohol was reported as any substance (i.e. primary, secondary, or tertiary) related to the treatment episode. To account for differences in population size by state and year, we considered admissions of pregnant women relative to pregnancies that resulted in a birth in the state-year. We collected birth data from the National Center for Health Statistics by state-year and corrected multiple births (i.e. twins, triplets, etc.) to be included as a single pregnancy resulting in birth.

Policy variables

State-year pregnancy-specific alcohol policies are the independent variables of interest. We examined nine policies: child abuse and neglect, civil commitment, reporting requirements for child protective services (CPS) purposes, reporting requirements for data collection purposes, reporting requirements for assessment and treatment purposes, mandatory warning signs, prohibitions on criminal prosecution, priority treatment for pregnant women only, and priority treatment for pregnant women and women with children. Descriptions of these policies are given in Table 1, and policy trends are presented in Fig. 1. For sensitivity analyses, we examined the same pregnancy-specific policies related to either alcohol, drugs, or both alcohol and drugs. Each policy was measured dichotomously, coded as 0 if it was not in effect in a state for a given year and 1 if it was in effect at any time in that state-year.

Table 1.

State-level policies related to alcohol use during pregnancy.

Policy	Policy description
Child abuse and neglect	Address the legal significance of a woman’s conduct prior to birth of a child and of damage caused in utero and, in some cases, define alcohol use during pregnancy as child abuse or neglect
Civil commitment	Mandatory involuntary commitment of a pregnant woman to treatment or mandatory involuntary placement of a pregnant woman in protective custody of the state for the protection of a fetus from prenatal exposure to alcohol
Reporting requirements for CPS purposes	Mandated or discretionary reporting of suspicion of or evidence of alcohol use or “abuse” by women during pregnancy for child welfare purposes (i.e. child abuse/neglect investigation) to either CPS or to a health authority
Reporting requirements for data collection	Mandated or discretionary reporting of suspicion of or evidence of alcohol use or “abuse” by women during pregnancy for data gathering purposes to either CPS or to a health authority
Reporting requirements for assessment and treatment purposes	Mandated or discretionary reporting of suspicion of or evidence of alcohol use or “abuse” by women during pregnancy to a state agency for purposes of assessment or treatment of the child, the child’s parents, or immediate caregivers (NIAAA 2022); they are not specific to a referral to treatment and may involve treatment as a condition of not being reported to or investigated by child welfare agencies
Mandatory warning signs	Require that notices be posted in settings, such as licensed premises, where alcohol is sold. The warning language required across jurisdictions varies in detail, but in each case, warns of the risks associated with drinking during pregnancy
Prohibitions against criminal prosecution	Prohibit use of the results of medical tests, such as prenatal screenings or toxicology tests, as evidence in the criminal prosecutions of women who may have caused harm to a fetus or a child
Priority treatment	Mandate priority access to substance use disorder treatment for (i) pregnant women who “abuse alcohol” and (ii) pregnant women who “abuse alcohol” and women with children

CPS, child protective services.

Figure 1.

Number of U.S. states with pregnancy-specific alcohol policies, 1992–2019.

State-level control variables

Control variables include state- and year-specific poverty, unemployment, per capita cigarette consumption, and total alcohol consumption per capita ages 14 and older.

Main analyses

First, we fit nine separate Poisson regression models to examine the unadjusted association between each policy variable and each outcome, including fixed effects for state and year. Then, we fit Poisson regression models that included all nine policy indicators simultaneously, state-level control variables, fixed effects for state and year, and state-specific quadratic time trends. We fit all models with an offset variable (specifically, the natural log of number of pregnancies in the state-year) to account for differences in population size and fertility by state and year. By including an offset variable in a Poisson model, the exponentiated regression coefficient of the policy variable reports how much the expected rate changes multiplicatively for a unit increase in the variable. Standard errors were clustered by state in all models. Analyses were conducted using Stata v16.

Sensitivity analyses

We conducted several sensitivity analyses to inform the interpretation of results. First, we considered possible effects of states’ pregnancy-specific drug use policies. With the exception of mandatory warning signs, most policies that address alcohol use during pregnancy also address drug use (Thomas et al. 2018). Due to this overlap, we determined it was infeasible to include the pregnancy-specific alcohol policy variables and equivalent drug policy variables as separate variables in each model. Rather, we created new variables to examine the associations between alcohol and/or drug use policies and compared these results to the main analysis models that included alcohol policies only.

In addition, we conducted sensitivity analyses to account for differences in how measures were created. As an alternative outcome, we used the number of women of reproductive age (ages 15–44) in the state-year as the offset variable using Current Population Survey data (Flood et al. 2021). As alternative policy measures, we coded policies as in effect if they were in effect for at least 6 months during the year, rather than at any point during the year.

Results

Sample description

The final sample includes 582 436 women who were pregnant at the time of admission, over 1331 state-years. See Table 2. Most were ages 21–24 (24%), 25–29 (30%), or 30–34 (20%). The majority reported their race as white (61%); about one-quarter (24%) were Black. Thirteen percent reported Hispanic ethnicity. Forty-three percent had not completed high school, 38% had a high school degree or equivalent, and 17% had completed at least some college. Half were not in the labor force (50%), 37% were unemployed, and 13% were employed either part time or full time. Eighteen percent reported alcohol as the primary substance related to the treatment admission, and 39% reported alcohol as the primary, secondary, or tertiary substance related to the treatment admission.

Table 2.

Characteristics of pregnant women in TEDS-A sample, 1992–2019 (N = 582 436).

	N	%
Age
<18 years	21 688	4
18–24 years	194 611	33
25–29 years	171 831	30
30–34 years	115 374	20
35+ years	78 569	13
Missing	363	0
Race
White	355 864	61
Black or African American	138 227	24
Alaska Native/American Indian	21 025	4
Asian/Pacific Islander	5639	1
Other single race	47 309	8
Two or more races	8210	1
Missing	6162	1
Hispanic ethnicity
Hispanic	76 976	13
Not Hispanic	488 378	84
Missing	17 082	3
Education
Less than Grade 9	42 347	7
Grades 9–11	208 898	36
High school diploma or equivalent	219 351	38
Some college or vocational school	86 656	15
4 years of college or more	13 091	2
Missing	12 093	2
Employment status
Full time	39 052	7
Part time	35 691	6
Unemployed	210 204	37
Not in labor force	283 335	50
Missing	14 154	2
Substance use at admission
Alcohol as primary substance reported	103 236	18
Alcohol as any substance reported	225 123	39

Figure 1 displays the enactment of the nine pregnancy-specific alcohol policies over the study period. In 1992, eight of the nine policies had been enacted in at least one state; civil commitment policies were first enacted in 1998. As of 2019, the most common policies were reporting requirements for CPS purposes (27 states), reporting policies for data collection (27), reporting requirements for assessment and treatment (25), mandatory warning signs (25), and child abuse and neglect (24). Fewer states had policies for priority treatment for pregnant women (12), prohibitions on criminal prosecution (7), civil commitment (5), and priority treatment for pregnant women and women with children (4). These policies changed little over time.

Effect of pregnancy-specific alcohol policies

When alcohol was reported as the primary substance related to the admission, child abuse and neglect policies and civil commitment policies were each associated with greater rates of treatment in the separate unadjusted policy models. In adjusted models including all policies, civil commitment policies were associated with greater rates of treatment admissions [incidence rate ratio (IRR) 1.45, 95% CI: 1.10–1.89, P = .008]. Reporting requirements for assessment and treatment purposes were also associated with greater rates of treatment admissions [IRR 1.36, 95% CI: 1.04–1.77, P = .022]. See Table 3.

Table 3.

Association between state policies targeting alcohol use during pregnancy and treatment admissions (1992–2019).

	Alcohol reported as primary reason for admission				Alcohol reported as any reason for admission
	Single policy models		All policy model		Single policy models		All policy model
	IRR	95% CI	IRR	95% CI	IRR	95% CI	IRR	95% CI
Child abuse and neglect	1.31	1.03, 1.67	1.04	0.81, 1.32	1.40	1.09, 1.81	1.05	0.87, 1.28
Civil commitment	1.80	1.59, 2.04	1.45	1.10, 1.89	2.01	1.73, 2.34	1.31	1.08, 1.59
Reporting requirements for CPS purposes	1.05	0.82, 1.35	0.88	0.69, 1.12	0.95	0.73. 1.25	0.97	0.81, 1.16
Reporting requirements for data collection	0.86	0.69, 1.08	0.89	0.75, 1.07	0.90	0.72, 1.13	0.89	0.74, 1.05
Reporting requirements for assessment and treatment purposes	1.34	0.96, 1.86	1.36	1.04, 1.77	1.28	0.92, 1.78	1.21	1.00, 1.47
Mandatory warning signs	1.15	0.92, 1.45	0.83	0.67, 1.01	1.15	0.90, 1.47	0.84	0.72, 0.98
Prohibitions on criminal prosecution	0.90	0.70, 1.17	0.84	0.52, 1.34	0.97	0.70, 1.35	0.85	0.59, 1.21
Priority treatment for pregnant women only	0.95	0.75, 1.19	0.91	0.73, 1.15	1.07	0.89. 1.28	0.88	0.78, 0.99
Priority treatment for pregnant women and women with children	1.25	0.83, 1.86	0.85	0.53, 1.38	1.30	0.79, 2.14	0.78	0.49, 1.25

Abbreviations: IRR, incidence rate ratio; CI, confidence interval; CPS, child protective services. Single policy models include one policy indicator and fixed effects for state and year. All policy models include nine policy indicators simultaneously, state-level control variables, fixed effects for state and year, and state-specific quadratic time trends. All models include the number of pregnancies in the state-year as an offset variables, and standard errors are clustered at the state level. Bold indicates P < .05.

When alcohol was reported as a primary, secondary, or tertiary substance related to the admission, child abuse and neglect policies and civil commitment policies were each associated with greater rates of treatment in the separate unadjusted policy models. In adjusted models including all policies, civil commitment policies [IRR 1.31, 95% CI: 1.08–1.59, P = .007] were again associated with higher rates of treatment admissions, as were reporting requirements for assessment and treatment purposes [IRR 1.21, 95% CI: 1.00–1.47, P = .047]. Mandatory warning signs [IRR 0.84, 95% CI: 0.72–0.98, P = .029] and priority treatment for pregnant women [IRR 0.88, 95% CI: 0.78–0.99, P = .044] were associated with lower rates of treatment admissions.

No other state alcohol policies were significantly associated with treatment admissions.

Sensitivity analyses: effect of pregnancy-specific alcohol and/or drug policies

Findings when examining pregnancy-specific policies that covered alcohol and/or drugs were mostly similar in the direction and magnitude of effects, although statistical significance sometimes was P <.10. The exception was for priority treatment for pregnant women, where the relationship between the policy and treatment admissions was no longer statistically significant. See Table 4.

Table 4.

Association between state policies targeting alcohol and/or drug use during pregnancy and treatment admissions (1992–2019).

	Alcohol reported as primary reason for admission				Alcohol reported as any reason for admission
	Single policy models		All policy model		Single policy models		All policy model
	IRR	95% CI	IRR	95% CI	IRR	95% CI	IRR	95% CI
Child abuse and neglect	1.24	0.98, 1.57	1.13	0.93, 1.38	1.26	0.99, 1.59	1.13	0.96, 1.33
Civil commitment	1.80	1.59, 2.04	1.36	1.00, 1.85	2.01	1.73, 2.34	1.26	1.02, 1.56
Reporting requirements for CPS purposes	1.02	0.80, 1.29	0.86	0.67, 1.10	0.93	0.72, 1.21	0.95	0.79, 1.15
Reporting requirements for data collection	0.86	0.70. 1.07	0.91	0.78, 1.06	0.90	0.73, 1.11	0.90	0.78, 1.03
Reporting requirements for assessment and treatment purposes	1.32	0.92, 1.88	1.29	0.98, 1.71	1.20	0.83, 1.73	1.18	0.97, 1.43
Mandatory warning signs	1.15	0.92, 1.45	0.81	0.66, 1.00	1.15	0.90, 1.47	0.83	0.71, 0.97
Prohibitions on criminal prosecution	0.89	0.69, 1.14	0.90	0.58, 1.41	0.98	0.73, 1.31	0.91	0.64, 1.30
Priority treatment for pregnant women only	0.92	0.75, 1.11	0.96	0.79, 1.17	1.06	0.89, 1.26	0.91	0.80, 1.02
Priority treatment for pregnant women and women with children	1.21	0.84, 1.75	0.85	0.54, 1.34	1.21	0.75, 1.97	0.79	0.50, 1.25

Abbreviations: IRR, incidence rate ratio; CI, confidence interval; CPS, child protective services. Single policy models include one policy indicator and fixed effects for state and year. All policy models include nine policy indicators simultaneously, state-level control variable, fixed effects for state and year, and state-specific quadratic time trends. All models include the number of pregnancies in the state-year as an offset variables, and standard errors are clustered at the state level. Bold indicates P < .05.

Sensitivity analyses: alternative measures

Sensitivity analyses that used alternative measures showed similar patterns of results as the main analyses. Findings for civil commitment were robust to model specification. Findings for reporting requirements for assessment and treatment and mandatory warning signs did not vary in magnitude or direction of effects, although statistical significance varied in some cases. Relationships that were statistically significant at a P <.05 level were as follows. When an offset variable of women of reproductive age was used, civil commitment policies [IRR = 1.45, 95% CI: 1.08–1.96, P = .013], reporting requirements for assessment and treatment [IRR = 1.38, 95% CI: 1.04–1.83, P = .024], and mandatory warning signs were again associated with treatment admissions [IRR = 0.80, 95% CI: 0.65–1.00, P = .047]. Similarly, when policy measures were coded if they were in effect for at least 6 months during the year, both civil commitment policies and mandatory warning signs were associated with treatment admissions [IRR = 1.34, 95% CI: 1.12–1.60, P = .001 and IRR = 0.80, 95% CI: 0.66–0.97, P = .024, respectively]. Findings for priority treatment for pregnant women were not robust to model specification.

Discussion

Given that few pregnant people who need substance use disorder treatment receive treatment (Martin et al. 2020), it is important from a public health perspective to identify policies that may facilitate utilization. In this study, we found greater rates of alcohol-related treatment admissions associated with civil commitment policies and reporting requirements for assessment and treatment purposes, and lower rates of treatment admissions associated with mandatory warning signs and priority treatment for pregnant women, although the priority treatment findings were not robust across different modeling assumptions. Notably, the policies that were most strongly and consistently associated with greater treatment rates (i.e. civil commitment and reporting requirements for assessment and treatment) either directly mandate treatment or use the threat of child welfare involvement to push people to enter treatment.

Interpretation of these findings requires attention to the broader evidence related to effects of these policies, as well as the ethical issues at play. Certainly, evidence indicating that state policies can increase rates of utilization of substance use disorder treatment among pregnant people is compelling from a public health perspective. Yet, previous research has also found that some pregnancy-specific reporting requirements are associated with increased adverse birth outcomes and that civil commitment policies are associated with increased late prenatal care (Subbaraman et al. 2018). Regarding policies that require reporting for assessment and treatment purposes, there is also a question as to whether the treatment admission related to reporting is voluntary or part of the ‘voluntary’ services that pregnant people are told to engage in to avoid child welfare involvement. By definition, these policies provide for “a report or notification to a state department for purposes of assessment or treatment of the child, the child’s parents, or immediate caregivers” (National Institute on Alcohol Abuse and Alcoholism 2022); they are not specific to a referral to treatment and may involve treatment as a condition of not being reported to or investigated by child welfare agencies.

There are significant ethical concerns pertaining to autonomy and justice related to using threats of jail, such as in the case of civil commitment policies, or removal of children, as in the case of the reporting requirements for assessment and treatment, as a means to compel pregnant people to obtain treatment (American Public Health Association 2019). Health professional association statements focus on expanding access to substance use treatment as the goal, rather than mandating use of treatment (e.g. Livingston et al. 2022). Research suggests that choice—for example, in terms of where treatment is received—is a key component of trauma-informed alcohol treatment approaches for pregnant people (Morton Ninomiya et al. 2023) and increases acceptability of treatment (Joshi et al. 2021), while fear of losing child custody remains a barrier to treatment (Frazer et al. 2019). Given that child welfare agencies investigate more Black and Indigenous children than white children, including infants reported related to pregnant people’s substance use, these fears and thus this barrier to treatment may be particularly acute for Black and Indigenous pregnant people (Edwards et al. 2021; 2023).

Questions as to whether it is ethical or effective to mandate substance use disorder treatment are not specific to pregnancy and are part of ongoing conversations among health professionals and among elected officials (Greenwald 2021; Frederique and Madras 2022). Our findings do not provide information about treatment outcomes, and thus do not provide evidence as to whether compelled treatment is effective at reducing harms related to pregnant people’s alcohol use. However, as other research indicates that while civil commitment policies are associated with decreased infant morbidities associated with alcohol use during pregnancy, both policies associated with greater treatment admissions in this study—civil commitment and reporting requirements for assessment and treatment purposes—are also associated with increased infant maltreatment (Roberts et al. 2023), suggesting that any increases in treatment during pregnancy associated with these policies may not result in longer-term improved health outcomes for infants.

Understanding the relationship between mandatory warning signs and fewer treatment admissions is less clear. Possible explanations include that mandatory warning signs contribute to a reduction in alcohol use disorder during pregnancy and thus less need for treatment; alternatively, mandatory warning signs could increase stigma and also scare people and lead them to avoid treatment. Previous research suggests that while mandatory warning signs may be associated with decreased binge drinking during pregnancy, they also appear to lead to a decrease in prenatal care utilization and an increase in adverse birth outcomes (Subbaraman et al. 2018; Roberts et al. 2019). Although this previous research did not examine the relationship between mandatory warning signs and need for treatment during pregnancy, it suggests that each explanation is plausible. Future research should seek to disentangle which mechanism may be operating, as this is relevant to whether mandatory warning signs policies reduce or increase harms related to alcohol use during pregnancy.

Importantly, priority treatment policies for pregnant people were not robustly related to treatment admissions. It is possible that this finding is due to the smaller number of priority treatment policies and the fact that most priority treatment policy changes occurred at the beginning of the study period (i.e. in the early 1990s) with little change in the years after. That said, this finding is consistent with other research indicating a lack of relationship between priority treatment policies and treatment admissions (Kozhimannil et al. 2019; Faherty et al. 2022; Maclean et al. 2022; Tabatabaeepour et al. 2022), as well as known challenges with implementing these policies (White et al. 2022). It is also worth noting that to the extent that priority treatment policies were associated with admissions, the direction of the findings suggested that priority treatment for pregnant women was associated with fewer rather than more treatment admissions. Unless these challenges are addressed, priority treatment should not be considered a policy likely to reduce public health harms related to pregnant people’s alcohol consumption.

If facilitating admissions to treatment for pregnant people is part of a public health strategy to reduce harms related to alcohol consumption during pregnancy, findings here indicate that seven of the nine pregnancy-specific polices we examined are ineffective at accomplishing this goal. Those that are effective use the power of the state to coerce treatment, but also may contribute to increases in adverse birth outcomes or maltreatment (Subbaraman et al. 2018; Roberts et al. 2019) and/or may violate accepted public health ethics (American Public Health Association 2019).

There are a few options for moving this work forward. One might be to address the issues that affect successful implementation of priority treatment policies (e.g. lack of funding; co-existence with policies that threaten punishment) and then conduct further evaluation to see whether improved priority treatment policies then facilitate treatment utilization (White et al. 2022). Another approach might be to work with people who drink during pregnancy and their families and communities to identify policy solutions, including treatment-relevant policies, that are appropriate for them (Roberts et al. 2021). Finally, other research suggests that more general population alcohol policies, such as mandatory legal drinking age and restrictions on Sunday sales, may also be relevant for reducing harms related to alcohol use during pregnancy (Fertig and Watson 2009; Zhang and Caine 2011; Subbaraman et al. 2023).

Limitations

Despite its usefulness in examining treatment admissions across the USA over time, there are limitations inherent to using TEDS-A data. Reporting is mandated for facilities that receive state and federal block grant funding, yet data are missing for some states in some years (137 of 1428 state-years). This could bias results if data are missing as a policy change occurs and if the policy’s effects are concentrated in those years. Moreover, there are differences in data reporting across states (Substance Use and Mental Health Services Administration 2021). The use of all available state data across 28 years helps address this concern, as does the inclusion of state fixed effects and state-specific time trends in models to account for confounding due to unobserved differences. In addition, pregnant people may receive treatment services at locations that are not captured by TEDS-A, such as with private non-specialty providers; thus, these findings do not generalize to all treatment admissions. There may also be differential rates of alcohol consumption among pregnant people by state and year, and thus differential need for treatment. The inclusion of state-year fixed effects in our model should address this concern as well.

As noted previously, we rely on TEDS-A variables indicating biological sex and pregnancy status at admission to form our dataset. We have no means of knowing the gender identity of the pregnant individuals admitted to treatment centers and thus cannot discern differences based on gender identity in this study. Research on the impact of alcohol policies and interventions on pregnant people who do not identify as women is sorely lacking and an important topic for future research.

In addition, we chose to use the number of pregnancies that resulted in birth as the offset variable in our models to account for population differences by state-year. However, this does not account for the socioeconomic status of pregnant people in need of substance use disorder treatment at facilities receiving public funding, relative to the overall pregnant population. If there are differences in the proportion of pregnant people who would be eligible for publicly funded treatment across states, this could also affect findings. Again, state-year fixed effects should address this concern. Data for this analysis were aggregated at the state-year level, and therefore we could not adjust for individual-level confounders.

Conclusions

Few pregnancy-specific alcohol policies are associated with substance use disorder treatment admissions related to alcohol use among pregnant people. Policies shown to be related to increased admissions tend to mandate treatment rather than make voluntary treatment more accessible, raising critical questions of ethics as well as questions of the effectiveness of the treatment admissions that are occurring. Alternative strategies to reduce public health harms related to alcohol use during pregnancy are needed.

Author contributions

Nancy Berglas (Conceptualization-Supporting, Data curation-Lead, Formal analysis-Lead, Writing—original draft-Equal, Writing—review and editing-Equal), Meenakshi Subbaraman (Conceptualization-Supporting, Data curation-Supporting, Formal analysis-Supporting, Methodology-Equal, Writing—review and editing-Supporting), Sue Thomas (Conceptualization-Supporting, Data curation-Equal, Writing—review and editing-Supporting), and Sarah Roberts (Conceptualization-Lead, Data curation-Supporting, Formal analysis-Supporting, Funding acquisition-Lead, Project administration-Lead, Writing—original draft-Equal, Writing—review and editing-Equal)

Conflict of interest: None declared.

Funding

This work was supported by the US National Institute on Alcohol Abuse and Alcoholism at the National Institutes of Health [Grant 2R01AA023267]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.