Secondary Logo

Journal Logo

Research Reports

Variations by Education Status in Relationships Between Alcohol/Pregnancy Policies and Birth Outcomes and Prenatal Care Utilization: A Legal Epidemiology Study

Roberts, Sarah C. M. DrPH; Mericle, Amy A. PhD; Subbaraman, Meenakshi S. PhD, MS; Thomas, Sue PhD; Kerr, William PhD; Berglas, Nancy F. DrPH

Author Information
Journal of Public Health Management and Practice: March/April 2020 - Volume 26 - Issue - p S71-S83
doi: 10.1097/PHH.0000000000001069

Abstract

Recent research categorizes state-level policies targeting alcohol use during pregnancy (alcohol/pregnancy policies) as either supportive or punitive.1–3 Conceptually, supportive policies offer information, services, and treatment that might prevent or treat use during pregnancy or protect women from prosecution. Punitive policies use threats of punishment or punish use during pregnancy by threatening and carrying out removal of children and involuntary treatment. State-level alcohol pregnancy policy environments have become more punitive over time.2,3 In 2013, more than 40 states had at least 1 alcohol/pregnancy policy.2 Categorizing alcohol/pregnancy policies into supportive versus punitive matters from ethical and legal perspectives may be less useful from an effectiveness standpoint. One recent study found that Mandatory Warning Signs policies may be associated with improvements in some birth outcomes, such as reductions in very low birth weights.4 In contrast, our previous research studying the effects of more than 40 years of alcohol/pregnancy policies found that, at best, most policies do not improve birth outcomes or prenatal care use.5 At worst, some policies (Mandatory Warning Signs and Child Abuse/Neglect) lead to increased adverse birth outcomes and decreased prenatal care use.5 Further, these relationships vary by race.6 Our research also found that most policies are not associated with alcohol use during pregnancy,7 and, to the extent there are relationships, they vary by race/ethnicity.8 When combined with birth outcomes findings, these findings may be better interpreted as indicators of women's willingness to self-report alcohol use in different policy environments rather than changes in alcohol use. Such results reflect concerns raised in arguments about potential detrimental effects of these policies.9,10 Findings from qualitative research and smaller quantitative studies support plausibility of these relationships, as they have found that fear of being reported to Child Protective Services (CPS) and losing children is a barrier to prenatal care use for women who use drugs11,12; alcohol warning label policies and warning signs may raise awareness that does not translate into behavior13; and fear of having already irreversibly harmed a baby is a reason for not making changes in alcohol/drug use after discovering pregnancy and is a barrier to prenatal care.11,14

Before discounting current alcohol/pregnancy policy approaches, though, additional research is needed to assess whether such policies benefit subsets of people. If effects are in different directions in socioeconomic status subgroups, overall results could mask differences and lead to erroneous conclusions. Effects could plausibly vary by socioeconomic status subgroups. Recent work emphasizes that (1) population-level interventions that improve health overall do not necessarily reduce and may increase disparities15; (2) approaches to population health that rely on individual agency and ability to apply knowledge and information to one's own behaviors, which may be more likely among higher socioeconomic status women, may increase disparities16–18; and (3) being involved with CPS is more common among women of lower socioeconomic status,19 and thus could make prenatal care avoidance more common.

Here, we examine whether effects of alcohol/pregnancy policies on low birth weight (LBW), preterm birth (PTB), prenatal care use (PNC), and alcohol use during pregnancy differ by socioeconomic status, measured as women's education status. Based on recent work described earlier, we expect benefits (ie, decreased adverse birth outcomes, increased PNC, and decreased alcohol during pregnancy) to be greater among women with more education. We expect harms (ie, increased adverse birth outcomes, decreased PNC and increased alcohol during pregnancy) to be greater among women with less education.

Methods

Data sources

This study uses 1972-2015 Vital Statistics data for birth outcomes (LBW and PTB) and PNC; 1985-2016 Behavioral Risk Factor Surveillance System (BRFSS) data for alcohol use during pregnancy; National Institute on Alcohol Abuse and Alcoholism's Alcohol Policy Information System (APIS)20 and original legal research for alcohol/pregnancy policies; and secondary sources for state-level controls.

Vital Statistics data were obtained from Natality Birth Data (http://www.nber.org/data/vital-statistics-natality-data.html) and requests to the Centers for Disease Control and Prevention for restricted use data from 2005 to 2015. These data include individual-level data for US births. Between 1972 and 1984, these data include a 50% to 100% sample of all births. Beginning in 1985, they include 100% of births. LBW and PTB are available throughout the study period, and PNC has been included in all states since 1980 and 43 states from 1972 to 1979. All singleton births during the study period among residents of 50 US states and District of Columbia were included. The dataset includes 155 446 714 births.

The Behavioral Risk Factor Surveillance System is an annual telephone survey that tracks health behaviors and health status of US adults. Pregnancy status has been assessed annually since 1985. The Behavioral Risk Factor Surveillance System has included questions about alcohol use since 1984, although data about alcohol use were not collected during even years in the 1990s. In 1993, participation rates were more than 70%; in the 2000s, participation rates were closer to 50%. Our analytic sample consists of female BRFSS respondents of reproductive age (18-44 years) who indicated they were currently pregnant and provided data on drinking (n = 57 194 between 1985 and 2016).

State-level alcohol/pregnancy policy statutory, regulatory, and effective date data were obtained from APIS and original legal research. We have described processes for obtaining and coding these data elsewhere.2 Briefly, processes involved (1) identifying and gathering relevant statutes and regulations; (2) identifying effective dates for each; (3) coding policies, including ensuring interrater reliability; and (4) checking with states and secondary sources to ensure accuracy of data gathering and coding.

We obtained state-level control data from the US Census, US Centers for Disease Control and Prevention, APIS, published research,21–23 and original legal research.

Alcohol/pregnancy policy data were merged with individual-level Vital Statistics data based on month and year the woman became pregnant. Alcohol/pregnancy policy data were merged with individual-level BRFSS data based on year the survey was completed.

Measures

Outcomes

Birth outcomes include low birth weight (LBW, dichotomous outcome of born less vs at or more than 2500 g24) and preterm birth (PTB, dichotomous outcome of born before vs at or after 37 weeks' gestation25). We also examined prenatal care use (PNC, any prenatal care, late entry [ie, after the first trimester], and inadequate care [accounts for timing of entry and number of visits, based on Kotelchuck index26]). Alcohol use during pregnancy outcomes refers to use the month before the survey and includes (1) any alcohol (dichotomous, one or more drinks); (2) any binge drinking (dichotomous, 5 or more [4 or more beginning in 2006] drinks on an occasion); and (3) heavy drinking (frequency, quantity, and binge drinking frequency, using indexing27,28 and modeled as a dichotomous outcome of 16+ in the past month, roughly 4 or more drinks per week, a level at which there is well-documented harm29). Our modeling approach (fixed effects for year) accounts for changes in items/question wording over time.

Alcohol/pregnancy policy predictors

State-level alcohol/pregnancy policies are main independent variables. Mandatory Warning Signs, Prohibitions on Criminal Prosecution, Reporting for Data/Treatment Purposes, Priority Treatment for Pregnant Women, and Priority Treatment for Pregnant Women and Women with Children are considered supportive. Child Abuse/Neglect, Civil Commitment, and CPS Reporting Requirements are considered punitive. Policy variables are dichotomous.

Moderators

Education (less than high school, high school graduate or equivalent, more than high school, and other/missing) was chosen as the measure of socioeconomic status because it was included consistently and measured mostly consistently and because more education is a reasonable proxy for individual aptitude to apply knowledge and information to one's own behaviors.

Individual-level controls

Individual-level controls for Vital Statistics analyses include maternal age, race, marital status, education, nativity, and parity. For BRFSS, they include age, race, marital status, education, income, tobacco use, and physical activity.

State-level controls

State-level controls include state- and year-specific unemployment rate, poverty rate, per capita cigarette sales, retail control policies for wine and spirits, and per capita alcohol consumption. In Vital Statistics analyses, per capita alcohol consumption is used as a proxy for regional drinking culture and other alcohol policies that influence alcohol consumption in general.

Analysis

Multivariable logistic regression models included all policy indicators, fixed effects for state and year, adjusted for individual and state-level control variables, and accounted for clustering of standard errors according to state of residence. Analyses included state and year fixed effects to account for changes in data gathering over time and other events in those states and years. Birth certificate version indicator variables were included in Vital Statistics analyses. State-specific cubic time trends were in Vital Statistics analyses to address concerns with endogeneity. It was infeasible to include state-specific time trends in BRFSS analyses.

To assess for differential effects by education, we examined interactions between education and each policy in separate models; that is, although all models included all policy indicators, only 1 interaction term was examined at a time. Overall, interactions were considered statistically significant if a Wald test was P < .05. Group-specific interaction terms were considered statistically significant if both the Wald test and the education-specific policy interaction term were P < .05. Regression model output and Wald test results are shown in Appendix A (see Supplemental Digital Content, available at http://links.lww.com/JPHMP/A604). The postestimation margins command was used to obtain predicted values. Statistical significance of differences for subgroups was assessed by whether the main effect of the policy was statistically significant (ie, for high school graduates, the reference group) and whether predicted marginal differences between when laws were in effect versus not in effect for a subgroup were statistically significant at P < .05. All analyses were performed in Stata v15.1. Two models in Vital Statistics analyses would not converge; 1 converged when we switched the reference group; this is noted in the table. Descriptions of sensitivity analyses examining whether expanding policies to include those that also cover drugs as well as whether race coding in Vital Statistics data affected findings and their results are in shown in Appendix B (see Supplemental Digital Content, available at http://links.lww.com/JPHMP/A605).

Results

Supportive policies

Tables Supp1a and Supp1b (see Supplemental Digital Content, available at http://links.lww.com/JPHMP/A604) present logistic regression model output, including Wald test results and coefficients for interaction terms for supportive policy findings. Tables 1 through 3 present predicted probabilities for all policy findings and show education-/group-specific effects.

TABLE 1 - Predicted Probability and Average Marginal Effects of Supportive and Punitive Policies on Birth Outcomes, by Educationa
Low Birth Weight Preterm Birth
No Policy Policy Average ME 95% CI No Policy Policy Average ME 95% CI
Supportive
MWS
Less than HS 0.076 0.074 −0.003 −0.005 to 0.000 0.111 0.110 −0.001 −0.005 to 0.002
HS graduate 0.060b 0.065 0.005 0.003 to 0.007 0.094 0.099 0.005 0.002 to 0.007
More than HS 0.048 0.053 0.005 0.003 to 0.007 0.082 0.086 0.004 0.002 to 0.006
Other/missing 0.070 0.075 0.005 0.002 to 0.008 0.106 0.109 0.003 0.000 to 0.006
RR_DTx
Less than HS 0.078 0.072 0.006 0.009 to −0.003 0.112 0.109 −0.003 −0.008 to 0.002
HS graduate 0.061 0.063 0.003 0.000 to 0.005 0.094 0.099 0.005 0.001 to 0.009
More than HS 0.048 0.051 0.003 0.001 to 0.005 0.081 0.087 0.006 0.002 to 0.009
Other/missing 0.070 0.074 0.004 0.001 to 0.007 0.105 0.111 0.006 0.001 to 0.011
PTxP
Less than HS 0.076 0.075 0.000 −0.004 to 0.004 0.110 0.114 0.004 −0.002 to 0.010
HS graduate 0.061 0.067 0.006 0.004 to 0.008 0.094 0.102 0.008 0.004 to 0.012
More than HS 0.048 0.055 0.007 0.004 to 0.009 0.082 0.089 0.007 0.004 to 0.011
Other/missing 0.070 0.078 0.008 0.004 to 0.012 0.106 0.115 0.009 0.003 to 0.015
PTxPWC
Less than HS 0.076 0.073 −0.002 −0.007 to 0.002 0.111 0.109 −0.002 −0.007 to 0.004
HS graduate 0.061 0.064 0.003 −0.001 to 0.006 0.095 0.100 0.004 0.000 to 0.009
More than HS 0.049 0.052 0.003 0.001 to 0.005 0.083 0.088 0.005 0.002 to 0.008
Other/missing 0.071 0.077 0.006 0.003 to 0.008 0.106 0.115 0.008 0.003 to 0.014
PCP
Less than HS 0.075 0.080 0.004 −0.001 to 0.010 0.110 0.116 0.006 0.000 to 0.012
HS graduate 0.061 0.068 0.007 0.002 to 0.011 0.095 0.106 0.010 0.005 to 0.016
More than HS 0.049 0.052 0.003 0.000 to 0.006 0.083 0.093 0.009 0.005 to 0.014
Other/missing 0.071 0.077 0.006 0.001 to 0.010 0.107 0.115 0.009 0.003 to 0.014
Punitive
CACN
Less than HS 0.075 0.075 0.000 −0.003 to 0.003 0.110 0.112 0.002 −0.002 to 0.006
HS graduate 0.061 0.066 0.005 0.003 to 0.007 0.094 0.102 0.008 0.005 to 0.012
More than HS 0.048 0.052 0.004 0.001 to 0.006 0.082 0.090 0.008 0.004 to 0.013
Other/missing 0.071 0.075 0.005 0.001 to 0.008 0.105 0.114 0.009 0.004 to 0.013
CC
Less than HS DNCc DNC DNC DNC 0.111 0.107 −0.004 −0.017 to 0.009
HS graduate DNC DNC DNC DNC 0.096 0.098 0.002 −0.008 to 0.013
More than HS DNC DNC DNC DNC 0.083 0.088 0.004 −0.006 to 0.015
Other/missing DNC DNC DNC DNC 0.107 0.107 0.000 −0.009 to 0.009
RR_CPS
Less than HS 0.077 0.071 0.005 0.009 to −0.001 0.113 0.104 0.009 0.013 to −0.004
HS graduate 0.061 0.062 0.001 −0.002 to 0.003 0.096 0.094 −0.002 −0.007 to 0.002
More than HS 0.049 0.050 0.002 −0.001 to 0.004 0.084 0.083 −0.001 −0.005 to 0.004
Other/missing 0.071 0.073 0.002 −0.001 to 0.004 0.108 0.105 −0.003 −0.008 to 0.002
Abbreviations: CACN, Child Abuse/Neglect; CC, Civil Commitment; CI, confidence interval; HS, high school; ME, marginal effect; MWS, Mandatory Warning Signs; PCP, Prohibitions on Criminal Prosecution; PTxP, Priority Treatment for Pregnant Women only; PTxPWC, Priority Treatment for Pregnant Women and Women with Children; RR_CPS, CPS Reporting Requirements; RR_DTx, Reporting Requirements for Data and Treatment purposes.
aModels display the predicted probability (predictive margins) of outcomes based on models testing the interaction of each policy and maternal education in separate logistic regression models that included fixed effects for state, year, and state-specific time trends and adjusted for individual- and state-level covariates, including all other pregnancy-specific alcohol policies.
bItalicized font indicates P < .05.
cDNC: One model predicting low birth weight did not converge.

Birth outcomes

Relationships between each supportive policy and LBW and PTB varied by education (see Table Supp1a, Supplemental Digital Content, available at http://links.lww.com/JPHMP/A604). The variation was mostly due to differences for women with less than high school education compared with high school graduates.

Only one supportive policy (Reporting Requirements for Data and Treatment Purposes) had a benefit; it was associated with lower LBW (0.6% lower) for women with less than high school education (Table 1). No other supportive policies had a benefit in terms of birth outcomes for any education subgroup.

All supportive policies had harms; 4 were associated with higher LBW and PTB for high school graduates, and all 5 were associated with higher LBW and PTB for women with more than high school education. Increases ranged from 0.3% to 1.0%.

With the exception of the one supportive policy that was associated with lower LBW for women with less than high school education, no supportive policies were associated with birth outcomes for women with less than high school education; and one supportive policy was not associated with LBW for women with high school education.

Prenatal care use

Relationships between each supportive policy and PNC outcomes varied by education, with 2 exceptions (see Table Supp1a, Supplemental Digital Content, available at http://links.lww.com/JPHMP/A604). Differences for women with less than high school education and high school graduates were found for all 5 policies for at least 1 PNC outcome. Differences between women with greater than high school education and high school graduates were found for 2 policies (Mandatory Warning Signs and Reporting Requirements for Data and Treatment Purposes) for at least 1 PNC outcome.

Three supportive policies had a benefit for at least 1 education subgroup; they were associated with increased PNC for at least 1 outcome (Table 2). For women with less than high school education, Reporting Requirements for Data and Treatment Purposes was associated with decreased late PNC and decreased inadequate PNC, and Priority Treatment for Pregnant Women Only was associated with decreased late PNC. For women with more than high school education, Prohibitions on Criminal Prosecution was associated with decreased inadequate PNC. Decreases in inadequate PNC ranged from 1.1% to 3.4% and in late PNC from 2.5% to 2.7%.

TABLE 2 - Predicted Probability and Average Marginal Effects of Supportive and Punitive Policies on Prenatal Care Utilization, by Educationa
No Prenatal Care Late Prenatal Care Inadequate Prenatal Care
No Policy Policy Average ME 95% CI No Policy Policy Average ME 95% CI No Policy Policy Average ME 95% CI
Supportive
MWS
Less than HS 0.019b 0.022 0.003 0.000 to 0.006 0.288 0.281 −0.007 −0.022 to 0.008 0.202 0.203 0.001 −0.013 to 0.014
HS graduate 0.012 0.016 0.005 0.003 to 0.006 0.215 0.246 0.031 0.021 to 0.041 0.140 0.171 0.031 0.019 to 0.042
More than HS 0.008 0.011 0.004 0.002 to 0.006 0.166 0.203 0.037 0.028 to 0.045 0.104 0.137 0.033 0.024 to 0.042
Other/missing 0.018 0.030 0.013 0.006 to 0.019 0.217 0.261 0.044 0.025 to 0.064 0.152 0.201 0.049 0.019 to 0.079
RR_DTx
Less than HS 0.021 0.018 −0.002 −0.007 to 0.003 0.295 0.268 0.027 0.048 to −0.007 0.215 0.180 0.034 0.052 to −0.016
HS graduate 0.012 0.013 0.001 −0.002 to 0.004 0.221 0.229 0.008 −0.007 to 0.024 0.149 0.147 −0.002 −0.016 to 0.012
More than HS 0.008 0.009 0.001 −0.001 to 0.003 0.170 0.186 0.016 0.004 to 0.029 0.111 0.116 0.006 −0.006 to 0.017
Other/missing 0.018 0.026 0.008 0.001 to 0.015 0.224 0.236 0.012 −0.010 to 0.033 0.163 0.169 0.006 −0.023 to 0.035
PTxP
Less than HS 0.020 0.018 −0.002 −0.006 to 0.002 0.290 0.265 0.025 0.045 to −0.004 0.205 0.186 −0.019 −0.039 to 0.001
HS graduate 0.013 0.013 0.000 −0.002 to 0.003 0.222 0.233 0.011 −0.011 to 0.033 0.147 0.153 0.006 −0.015 to 0.027
More than HS 0.008 0.008 0.000 −0.002 to 0.001 0.175 0.188 0.013 −0.003 to 0.030 0.113 0.118 0.006 −0.009 to 0.020
Other/missing 0.019 0.024 0.005 −0.003 to 0.013 0.225 0.245 0.020 −0.010 to 0.049 0.161 0.182 0.021 −0.016 to 0.058
PTxPWC
Less than HS 0.020 0.019 −0.001 −0.005 to 0.004 0.287 0.272 −0.014 −0.029 to 0.000 0.203 0.191 −0.012 −0.030 to 0.005
HS graduate 0.013 0.014 0.001 −0.003 to 0.005 0.224 0.225 0.002 −0.013 to 0.017 0.148 0.151 0.003 −0.017 to 0.023
More than HS 0.009 0.008 0.000 −0.003 to 0.002 0.177 0.177 0.000 −0.021 to 0.021 0.114 0.116 0.002 −0.021 to 0.025
Other/missing 0.020 0.020 0.000 −0.004 to 0.004 0.229 0.209 0.020 0.032 to −0.008 0.163 0.162 −0.001 −0.029 to 0.027
PCP
Less than HS 0.020 0.020 0.001 −0.004 to 0.005 0.286 0.269 −0.017 −0.038 to 0.004 0.203 0.187 −0.015 −0.033 to 0.003
HS graduate 0.013 0.014 0.001 −0.001 to 0.004 0.224 0.220 −0.005 −0.021 to 0.012 0.148 0.140 −0.008 −0.019 to 0.003
More than HS 0.008 0.009 0.000 −0.001 to 0.002 0.177 0.168 −0.009 −0.023 to 0.005 0.114 0.104 0.011 0.020 to 0.001
Other/missing 0.020 0.022 0.003 −0.003 to 0.008 0.228 0.208 −0.020 −0.045 to 0.004 0.163 0.161 −0.002 −0.037 to 0.032
Punitive
CACN
Less than HS 0.020 0.020 0.000 −0.003 to 0.003 0.289 0.270 0.019 0.036 to −0.002 0.202 0.201 −0.001 −0.021 to 0.019
HS graduate 0.013 0.014 0.001 −0.001 to 0.003 0.225 0.221 −0.004 −0.014 to 0.006 0.146 0.155 0.009 −0.001 to 0.019
More than HS 0.008 0.010 0.002 0.000 to 0.004 0.175 0.182 0.006 −0.003 to 0.016 0.111 0.125 0.014 0.006 to 0.022
Other/missing 0.019 0.023 0.004 −0.001 to 0.009 0.229 0.220 −0.008 −0.020 to 0.004 0.161 0.171 0.010 −0.009 to 0.029
CC
Less than HS 0.020 0.021 0.001 −0.008 to 0.009 0.285 0.295 0.009 −0.016 to 0.035 0.202 0.181 −0.022 −0.083 to 0.040
HS graduate 0.013 0.015 0.003 −0.006 to 0.011 0.223 0.239 0.016 −0.012 to 0.043 0.148 0.133 −0.015 −0.054 to 0.024
More than HS 0.008 0.011 0.002 −0.005 to 0.010 0.176 0.209 0.033 −0.004 to 0.069 0.114 0.110 −0.004 −0.039 to 0.031
Other/missing 0.020 0.027 0.008 −0.008 to 0.023 0.227 0.250 0.022 −0.007 to 0.052 0.163 0.145 −0.018 −0.055 to 0.018
RR_CPS
Less than HS 0.020 0.020 0.000 −0.006 to 0.005 0.288 0.273 −0.016 −0.040 to 0.008 0.202 0.200 −0.002 −0.032 to 0.028
HS graduate 0.012 0.015 0.002 −0.002 to 0.006 0.221 0.235 0.013 −0.008 to 0.034 0.145 0.163 0.018 −0.001 to 0.037
More than HS 0.008 0.009 0.001 −0.002 to 0.003 0.173 0.192 0.019 0.002 to 0.036 0.110 0.130 0.019 0.003 to 0.035
Other/missing 0.019 0.029 0.010 0.002 to 0.017 0.224 0.245 0.021 −0.006 to 0.048 0.159 0.186 0.027 −0.001 to 0.056
Abbreviations: CACN, Child Abuse/Neglect; CC, Civil Commitment; CI, confidence interval; HS, high school; ME, marginal effect; MWS, Mandatory Warning Signs; PCP, Prohibitions on Criminal Prosecution; PTxP, Priority Treatment for Pregnant Women only; PTxPWC, Priority Treatment for Pregnant Women and Women With Children; RR_CPS, CPS Reporting Requirements; RR_DTx, Reporting Requirements for Data and Treatment purposes.
aModels display the predicted probability (predictive margins) of outcomes based on models testing the interaction of each policy and maternal education in separate logistic regression models that included fixed effects for state, year, and state-specific time trends and adjusted for individual- and state-level covariates, including all other pregnancy-specific alcohol policies.
bItalicized font indicates P < .05.

Two supportive policies had harms for at least 1 education subgroup. Mandatory Warning Signs was associated with increased no PNC for all education subgroups and increased late PNC and inadequate PNC for high school graduates and women with greater than high school education (ranging from 0.3% to 3.7%). Reporting Requirements for Data and Treatment Purposes was associated with increased late PNC for women with greater than high school education (1.6%). No other supportive policies were associated with PNC for any education subgroup.

Alcohol consumption during pregnancy

Relationships between supportive policies and alcohol consumption did not vary by education, with one exception (see Table Supp1b, Supplemental Digital Content, available at http://links.lww.com/JPHMP/A604).

Four supportive policies had a benefit for at least 1 education subgroup (Table 3). Mandatory Warning Signs was associated with decreased binge drinking for women with less than high school education and women with more than high school education. Two policies (Reporting Requirements for Data and Treatment Purposes and Priority Treatment for Pregnant Women and Women with Children) were associated with decreased heavy drinking and 1 policy (Prohibitions on Criminal Prosecution) was associated with decreased any drinking for high school graduates. Decreases ranged from 0.9% in binge and heavy drinking to 3.6% in any drinking. No other supportive policies were associated with self-reported drinking during pregnancy for any education subgroup.

TABLE 3 - Predicted Probability and Average Marginal Effects of Supportive and Punitive Policies on Alcohol Use, by Educationa
Any Drinking Binge Drinking Heavy Drinking
No Policy Policy Average ME 95% CI No Policy Policy Average ME 95% CI No Policy Policy Average ME 95% CI
Supportive
MWS
Less than HS 0.078 0.085 0.007 −0.018 to 0.032 0.021b 0.011 0.009 0.19 to −0.000 0.020 0.013 −0.007 −0.020 to 0.005
HS graduate 0.103 0.111 0.008 −0.019 to 0.035 0.028 0.015 −0.012 −0.026 to 0.002 0.023 0.018 −0.004 −0.021 to 0.012
More than HS 0.154 0.141 −0.012 −0.034 to 0.010 0.035 0.022 0.013 0.023 to −0.003 0.032 0.024 −0.008 −0.019 to 0.003
RR_DTx
Less than HS 0.076 0.085 0.008 −0.014 to 0.031 0.015 0.016 0.001 −0.007 to 0.009 0.018 0.016 −0.002 −0.010 to 0.005
HS graduate 0.104 0.108 0.004 −0.020 to 0.028 0.023 0.019 −0.003 −0.014 to 0.007 0.024 0.016 0.009 0.018 to −0.000
More than HS 0.153 0.144 −0.008 −0.030 to 0.013 0.027 0.029 0.003 −0.008 to 0.013 0.028 0.028 0.000 −0.010 to 0.009
PTxP
Less than HS 0.082 0.077 −0.005 −0.033 to 0.023 0.016 0.014 −0.002 −0.012 to 0.007 0.016 0.018 0.002 −0.010 to 0.013
HS graduate 0.107 0.101 −0.006 −0.031 to 0.018 0.022 0.019 −0.003 −0.012 to 0.005 0.020 0.023 0.002 −0.007 to 0.012
More than HS 0.150 0.142 −0.008 −0.035 to 0.019 0.030 0.023 −0.007 −0.021 to 0.008 0.029 0.025 −0.003 −0.018 to 0.009
PTxPWC
Less than HS 0.080 0.097 0.017 −0.010 to 0.044 0.016 0.014 −0.002 −0.010 to 0.007 0.018 0.009 −0.008 −0.018 to 0.001
HS graduate 0.104 0.130 0.027 0.008 to 0.045 0.022 0.014 −0.007 −0.017 to 0.002 0.022 0.010 0.012 0.021 to −0.004
More than HS 0.147 0.161 0.014 −0.003 to 0.031 0.029 0.023 −0.006 −0.017 to 0.006 0.029 0.022 −0.007 −0.018 to 0.003
PCP
Less than HS 0.082 0.055 −0.028 −0.059 to 0.003 0.015 0.015 0.000 −0.112 to 0.011 0.017 0.011 −0.006 −0.017 to 0.004
HS graduate 0.108 0.072 0.036 0.056 to −0.016 0.021 0.023 0.002 −0.012 to 0.016 0.021 0.022 0.002 −0.009 to 0.013
More than HS 0.150 0.127 −0.022 −0.051 to 0.007 0.028 0.029 0.000 −0.015 to 0.016 0.028 0.023 −0.005 −0.015 to 0.004
Punitive
CACN
Less than HS 0.084 0.073 −0.010 −0.037 to 0.017 0.017 0.012 −0.005 −0.012 to 0.002 0.020 0.010 0.010 0.018 to −0.002
HS graduate 0.104 0.113 0.009 −0.013 to 0.032 0.024 0.015 0.009 0.018 to −0.001 0.023 0.013 0.010 0.016 to −0.004
More than HS 0.147 0.152 0.004 −0.023 to 0.032 0.030 0.024 −0.007 −0.013 to 0.000 0.030 0.023 −0.008 −0.017 to 0.002
CC
Less than HS 0.082 0.060 −0.022 −0.049 to 0.006 0.015 0.026 0.011 −0.003 to 0.025 0.017 0.043 0.027 −0.000 to 0.054
HS graduate 0.106 0.089 −0.018 −0.040 to 0.004 0.021 0.029 0.008 −0.016 to 0.031 0.020 0.037 0.016 −0.013 to 0.046
More than HS 0.149 0.121 −0.028 −0.058 to 0.003 0.028 0.042 0.014 −0.012 to 0.039 0.028 0.044 0.016 −0.021 to 0.053
RR_CPS
Less than HS 0.081 0.081 0.000 −0.024 to 0.023 0.013 0.022 0.009 0.000 to 0.018 0.014 0.023 0.008 −0.006 to 0.023
HS graduate 0.108 0.102 −0.006 −0.024 to 0.012 0.019 0.025 0.006 −0.004 to 0.016 0.021 0.020 −0.001 −0.009 to 0.008
More than HS 0.150 0.144 −0.006 −0.028 to 0.016 0.028 0.030 0.002 −0.006 to 0.011 0.028 0.029 0.001 −0.009 to 0.011
Abbreviations: CACN, Child Abuse/Neglect; CC, Civil Commitment; CI, confidence interval; HS, high school; ME, marginal effect; MWS, Mandatory Warning Signs; PCP, Prohibitions on Criminal Prosecution; PTxP, Priority Treatment for Pregnant Women only; PTxPWC, Priority Treatment for Pregnant Women and Women With Children; RR_CPS, CPS Reporting Requirements; RR_DTx, Reporting Requirements for Data and Treatment purposes.
aModels display the predicted probability (predictive margins) of outcomes based on models testing the interaction of each policy and educational attainment in separate sample-weighted logistic regression in models that included fixed effects for state and year and adjusted for individual- and state-level covariates, including all other pregnancy-specific alcohol policies.
bItalicized font indicates P < .05.

Punitive policies

Tables Supp1c and Supp1d (see Supplemental Digital Content, available at http://links.lww.com/JPHMP/A604) present logistic regression model output, including Wald test results and coefficients for interaction terms for punitive policy findings. Tables 1 through 3 present predicted probabilities for all policy findings and show education-/group-specific results.

Birth outcomes

Relationships between each punitive policy and LBW and PTB varied by education (see Table Supp1c, Supplemental Digital Content, available at http://links.lww.com/JPHMP/A604). Variation was due primarily to differences between women with less than high school education compared with high school graduates, with one exception.

One punitive policy had a benefit for one education subgroup. For women with less than high school education, LBW and PTB were lower (0.5% and 0.9%) when CPS Reporting Requirements were in effect (Table 1).

One punitive policy had a health harm. For high school graduates and women with more than high school education, LBW and PTB were higher (ranging from 0.4% to 0.8%) when Child Abuse/Neglect was in effect. No other punitive policies were associated with LBW or PTB for any education subgroup.

Prenatal care use

Most relationships between each punitive policy and PNC varied by education (see Table Supp1c, Supplemental Digital Content, available at http://links.lww.com/JPHMP/A604). Variation was due both to differences between women with less than high school education compared with high school graduates and to differences between women with greater than high school education compared with high school graduates.

One punitive policy had a benefit for one education subgroup; Child Abuse/Neglect was associated with decreased late PNC (1.9%) for women with less than high school education (Table 2). Two punitive policies had harms for women with more than high school education; Child Abuse/Neglect was associated with increased inadequate PNC, and CPS Reporting Requirements was associated with increased late PNC and inadequate PNC (ranging from 1.4% to 1.9%). No other punitive policies were associated with PNC for any education subgroup.

Alcohol consumption during pregnancy

Relationships between punitive policies and alcohol consumption did not vary by education (see Table Supp1d, Supplemental Digital Content, available at http://links.lww.com/JPHMP/A604).

One punitive policy had a benefit; Child Abuse/Neglect was associated with decreased binge and heavy drinking for high school graduates and decreased heavy drinking for women with less than high school education (ranging from 0.9% to 1.0%) (Table 3).

One punitive policy had a health harm; CPS Reporting Requirements was associated with increased binge drinking for women with less than high school education (0.9%). No other punitive policies were associated with self-reported drinking for any education subgroup.

Discussion

This study analyzed more than 40 years of data with more than 150 million births and found that effects of alcohol/pregnancy policies vary by education.

However, general hypotheses regarding directions of differential effects were unsupported. Adverse effects of alcohol/pregnancy policies appear concentrated among those with high school or greater education, while women with less than high school typically do not experience these harms and, in a few cases, appear to experience some benefits. That findings are inconsistent with hypotheses corresponds with findings examining effects of alcohol/pregnancy policies by race, where harms were concentrated among more advantaged (white) and health benefits among less advantaged (black) women.6

Although our study does not explore reasons for differences, there are plausible explanations. Exposure to information about policies might vary by education; women with more education might have more resources and, thus, may be in bars, restaurants, and other alcohol venues more often and may have greater exposure to Mandatory Warning Signs. Second, women of different education levels might behave differently when exposed to information about harms of alcohol use or the possibility of having their children removed by the state. Third, we found that both reporting policies—data/treatment and CPS—were associated with improved birth outcomes for women with less education. As more women of lower socioeconomic status receive direct public services, and assuming that services benefit health, women with less education may benefit more from reporting policies.

Among punitive policies, Child Abuse/Neglect and CPS Reporting Requirements generally had effects in different directions. Child Abuse/Neglect had adverse effects for women with more education, while CPS Reporting Requirements had beneficial effects for women with less education (although there was a different pattern for self-reported alcohol use). Regardless of observed beneficial impacts of CPS Reporting Requirements here, other research has documented adverse community-level effects of high levels of CPS involvement30 and that fear of being reported to CPS leads pregnant women to avoid both treatment31 and PNC.11 The opposite effects of Child Abuse/Neglect and CPS Reporting Requirements are worth examining. This could reflect Child Abuse/Neglect focus on legal aspects of child removal and CPS Reporting Requirements plausibly lead to provision of direct, helpful services or priority for other services instead of a focus on child removal. Results for Reporting Requirements for Data/Treatment that mirror those for CPS Reporting Requirements, though, suggest health benefits from reporting to and thus offering services through other agencies that do not have the implied coercion or threat of CPS may be possible.

Although patterns across birth outcomes and PNC are mostly consistent, patterns for alcohol outcomes are mostly not. One explanation is that alcohol outcomes are self-reported—and may reflect women's level of willingness to disclose use. Also, alcohol analyses rely on smaller samples and do not control for state-specific time trends.

These analyses have limitations. First, the large sample size has power to detect small effects. For birth outcomes, small effects matter, but it is unclear how important small effects on prenatal care are from a public health perspective. Second, with the exception of Mandatory Warning Signs, most policies address both alcohol and drugs.3 Because of this policy overlap, this study cannot distinguish whether the focus on alcohol or drug use during pregnancy matters for birth outcomes. As confirmed by sensitivity analyses (see Appendix A, Supplemental Digital Content, available at http://links.lww.com/JPHMP/A604], with the exception of Mandatory Warning Signs policies, findings can be interpreted as applying to alcohol/pregnancy and to drug/pregnancy policies. Third, our measure of socioeconomic status addresses 1 domain—education. Other measures, such as income or insurance type, might result in different patterns.

Conclusions

Effects of alcohol/pregnancy policies on birth outcomes vary by education, with women with more education experiencing health harms and women with less experiencing no effect, and, in a few cases, health benefits. New policy approaches that reduce harms related to alcohol use during pregnancy are needed.

Implications for Policy & Practice

  • State-level alcohol/pregnancy policies that aim to improve outcomes appear to have no effect or some beneficial effects for women with less education and adverse effects for women with more education.
  • New policy approaches to alcohol use during pregnancy that do not have adverse health effects are needed.
  • Public health professionals should take the lead on identifying and developing policy approaches that reduce harms related to alcohol use during pregnancy.

References

1. Drabble LA, Thomas S, O'Connor L, Roberts SC. State responses to alcohol use and pregnancy: findings from the Alcohol Policy Information System (APIS). J Soc Work Pract Addict. 2014;14(2):191–206.
2. Roberts SCM, Thomas S, Treffers R, Drabble L. Forty years of state alcohol and pregnancy policies in the USA: best practices for public health or efforts to restrict women's reproductive rights? Alcohol Alcohol. 2017;52(6):715–721.
3. Thomas S, Treffers R, Berglas N, Drabble L, Roberts S. Drug use during pregnancy policies in the United States from 1970-2016. Contemp Drug Probl. 2018;45(4):441–459.
4. Cil G. Effects of posted point-of-sale warnings on alcohol consumption during pregnancy and on birth outcomes. J Health Econ. 2017;53:131–155.
5. Subbaraman MS, Thomas S, Treffers R, et al. 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. Alcohol Clin Exp Res. 2018;42(8):1511–1517.
6. Roberts S, Berglas NF, Subbaraman MS, Mericle AA, Thomas S, Kerr WC. Racial differences in the relationship between alcohol/pregnancy policies and birth outcomes and prenatal care utilization: a legal epidemiology study. Drug Alcohol Depend. 2019;201:244–252.
7. Roberts S, Mericle A, Subbaraman M, et al. State policies targeting alcohol use during pregnancy and alcohol use among pregnant women 1985-2016. Women Health Iss. 2019;29(3):213–221.
8. Roberts SCM, Mericle AA, Subbaraman MS, et al. Differential effects of pregnancy-specific alcohol policies on drinking among pregnant women by race/ethnicity. Health Equity. 2018;2(1):356–365.
9. Golden J. “A tempest in a cocktail glass”: mothers, alcohol, and television, 1977-1996. J Health Polit Policy Law. 2000;25(3):473–498.
10. Murphy S, Rosenbaum M. Pregnant Women on Drugs: Combating Stereotypes and Stigma. New Brunswick, NJ: Rutgers University Press; 1999.
11. Roberts SCM, Pies C. Complex calculations: how drug use during pregnancy becomes a barrier to prenatal care. Matern Child Health J. 2011;15(3):333–341.
12. Wu M, Lagasse LL, Wouldes TA, et al. Predictors of inadequate prenatal care in methamphetamine-using mothers in New Zealand and the United States. Matern Child Health J. 2013;17(3):566–575.
13. Abel EL. Prevention of alcohol abuse-related birth effects—I. Public education efforts. Alcohol Alcohol. 1998;33(4):411–416.
14. Kaskutas LA. Understanding drinking during pregnancy among urban American Indians and African Americans: health messages, risk beliefs, and how we measure consumption. Alcohol Clin Exp Res. 2000;24(8):1241–1250.
15. Keppel K, Bilheimer L, Gurley L. Improving population health and reducing health care disparities. Health Aff (Millwood). 2007;26(5):1281–1292.
16. Sumar N, McLaren L. Impact on social inequalities of population strategies of prevention for folate intake in women of childbearing age. Am J Public Health. 2011;101(7):1218–1224.
17. McLaren L, McIntyre L, Kirkpatrick S. Rose's population strategy of prevention need not increase social inequalities in health. Int J Epidemiol. 2010;39(2):372–377.
18. Frohlich KL, Potvin L. Transcending the known in public health practice: the inequality paradox: the population approach and vulnerable populations. Am J Public Health. 2008;98(2):216–221.
19. Sendlak AJ, Mettenburg J, Basena M, et al. Fourth National Incidence Study of Child Abuse and Neglect (NIS-4): Report to Congress. Washington, DC: US Department of Health and Human Services, Administration for Children and Families; 2010.
20. National Institute on Alcohol Abuse and Alcoholism. Alcohol Policy Information System. http://www.alcoholpolicy.niaaa.nih.gov/. Published 2016.
21. Martinez P, Kerr WC, Subbaraman MS, Roberts SCM. New estimates of the mean ethanol content of beer, wine, and spirits sold in the U.S. show a greater increase in per capita alcohol consumption than previous estimates. Alcohol Clin Exp Res. 2019;43(3):509–521.
22. Kerr WC, Greenfield TK, Tujague J. Estimates of the mean alcohol concentration of the spirits, wine, and beer sold in the United States and per capita consumption: 1950 to 2002. Alcohol Clin Exp Res. 2006;30(9):1583–1591.
23. Kerr WC, Barnett SBL. Alcohol retailing systems: private versus government control. In: Giesbrecht N, Bosma LM, eds. Preventing Alcohol-Related Problems: Evidence and Community-Based Initiatives. Washington, DC: APHA Press; 2017.
24. Centers for Disease Control and Prevention. Birthweight and gestation. http://www.cdc.gov/nchs/fastats/birthwt.htm. Published 2010.
25. Behrman RE, Stith Butler A, eds. Preterm Birth: Causes, Consequences, and Prevention. Washington, DC: Institute of Medicine, National Academies Press; 2007.
26. Kotelchuck M. An evaluation of the Kessner Adequacy of Prenatal Care Index and a proposed Adequacy of Prenatal Care Utilization Index. Am J Public Health. 1994;84(9):1414–1420.
27. Armor DJ, Polich JM. Measurement of alcohol consumption. In: Pattison EM, Kaufman E, eds. Encyclopedic Handbook of Alcoholism. New York, NY: Gardner Press; 1982:72–80.
28. Stahre M, Naimi T, Brewer R, Holt J. Measuring average alcohol consumption: the impact of including binge drinks in quantity-frequency calculations. Addiction. 2006;101(12):1711–1718.
29. O'Leary CM, Bower C. Guidelines for pregnancy: what's an acceptable risk, and how is the evidence (finally) shaping up? Drug Alcohol Rev. 2012;31(2):170–183.
30. Roberts DE. Shattered Bonds: The Color of Child Welfare. New York, NY: Basic Books; 2002.
31. Falletta L, Hamilton K, Fischbein R, Aultman J, Kinney B, Kenne D. Perceptions of child protective services among pregnant or recently pregnant, opioid-using women in substance abuse treatment. Child Abuse Negl. 2018;79:125–135.
Keywords:

alcohol; disparities; legal epidemiology; policy; pregnancy

Supplemental Digital Content

© 2020 Wolters Kluwer Health, Inc. All rights reserved.