RESEARCH AND PRACTICE

The Alcohol Policy Environment and Policy Subgroups as Predictors of Binge Drinking Measures Among US Adults Ziming Xuan, ScD, SM, MA, Jason Blanchette, MPH, Toben F. Nelson, ScD, Timothy Heeren, PhD, Nadia Oussayef, JD, MPH, and Timothy S. Naimi, MD, MPH

Excessive alcohol consumption is a leading cause of mortality, morbidity, social, and economic burden in the United States.1---6 Of the adverse consequences linked to excessive alcohol consumption, binge drinking accounts for approximately half of alcohol-attributable deaths, two thirds of years of potential life lost, and three fourths of economic costs.5---7 Patterns of excessive alcohol consumption, including binge drinking, vary substantially across US states.8---10 Alcohol control policies, which consist of the laws, regulations, and practices designed to reduce excessive alcohol use and related harms, also differ considerably among US states.11,12 A number of discrete alcohol policies have been shown to reduce excessive alcohol consumption and related problems at the population level.1,13---23 However, the alcohol policy “environment” in US states is complex, and any discrete alcohol policy coexists with many other policies. Because the presence of a policy may be correlated with other policies, and because the effect of a policy may depend on its interaction with other policies, the relationship between a policy and an outcome may be inadequately characterized by failing to account for other policies that are in effect. Therefore, assessing the effect of multiple policies is crucial not only for understanding the cumulative effects of alcohol policies, but also for understanding the independent effects of policy subgroups or individual policies. Policy scales have been used to examine policy environments in several public health concerns, including alcohol use,24 tobacco use25---27 and obesity.28,29 We recently developed the Alcohol Policy Scale (APS) to represent the alcohol policy environment in all 50 states and Washington, DC, from 1999 to 2011.11 Previous analyses validated the APS scores by demonstrating that states with higher APS scores (representing mixes of more effective or better implemented policies) were

Objectives. We examined the relationships of the state-level alcohol policy environment and policy subgroups with individual-level binge drinking measures. Methods. We used generalized estimating equations regression models to relate the alcohol policy environment based on data from 29 policies in US states from 2004 to 2009 to 3 binge drinking measures in adults from the 2005 to 2010 Behavioral Risk Factor Surveillance System surveys. Results. A 10 percentage point higher alcohol policy environment score, which reflected increased policy effectiveness and implementation, was associated with an 8% lower adjusted odds of binge drinking and binge drinking 5 or more times, and a 10% lower adjusted odds of consuming 10 or more drinks. Policies that targeted the general population rather than the underage population, alcohol consumption rather than impaired driving, and raising the price or reducing the availability of alcohol had the strongest independent associations with reduced binge drinking. Alcohol taxes and outlet density accounted for approximately half of the effect magnitude observed for all policies. Conclusions. A small number of policies that raised alcohol prices and reduced its availability appeared to affect binge drinking. (Am J Public Health. 2015;105: 816–822. doi:10.2105/AJPH.2014.302112) associated with a lower state-level prevalence of binge drinking in both repeated crosssectional generalized estimating equation (GEE) models and longitudinal models.11 However, analyses that rely on state-level outcomes can be susceptible to ecological fallacy.30 Furthermore, the effect of the policy environment among demographic groups and the independent effects of subgroups of alcohol policies have not been assessed previously. Our aims were (1) to determine the associations between the state alcohol policy environment and 3 individual-level binge drinking measures, overall and stratified by demographic factors; and (2) to assess the independent effects of policy subgroups and selected individual policies on these 3 binge drinking measures.

METHODS All of the data sources we used had uniform ascertainment methods across all 50 states and the District of Columbia.11 The primary source

816 | Research and Practice | Peer Reviewed | Xuan et al.

for 13 of the 29 policies was the Alcohol Policy Information System.12 We used 18 additional data sources to collect and code data about policies and provisions that were not included in Alcohol Policy Information System.11 Multiple data sources were available for some policies, which we cross-checked to ascertain consistency. Remaining discrepancies were resolved by a public health lawyer who used the legal research database WestlawNext.31 For 6 policies with missing data from 2004 to 2008, we used WestlawNext, in consultation with the public health lawyer, to conduct historical reviews and complete policy data for those missing years.

Alcohol Policy Scale Development Ten experts in alcohol policy assisted with the selection of alcohol policies for inclusion in the APS scale. Forty-seven alcohol control policies were nominated by panelists. The scale ultimately included 29 policies, which all had been consistently collected and had reliable cross-state data. Because our study examined

American Journal of Public Health | April 2015, Vol 105, No. 4

RESEARCH AND PRACTICE

state-level alcohol policies, we did not include federal policies, policies that did not vary across states (e.g., public intoxication laws), and policies without reliable data across states (e.g., mandatory substance abuse assessment for driving under the influence offenders). We developed standardized, idealized descriptions of each policy. Panelists then independently rated the efficacy of each policy based on a 5-point Likert scale (1 = low efficacy, 5 = high efficacy) for each of 4 distinct domains: (1) reducing binge drinking among adults, (2) reducing impaired driving among adults, (3) reducing drinking among underage youths, and (4) reducing drinking and driving among underage youths. The mean of the panelists’ ratings for each policy in a given domain was used as its efficacy rating (ER). For the purpose of this study, we used ERs for reducing binge drinking among adults.32 In consultation with the panelists who had expertise in particular policies, we also developed a legislative implementation rating (IR) for each policy based on provisions or characteristics of a particular policy.11 Factors that influenced the IR were primarily related to the policy’s statutory design, including provisions that enabled the policy to be effective, broadly applicable, and enforceable. For example, ratings of keg registration laws were based on the size of the keg to which the law applied, the amount of deposit required, and whether there were penalties for label destruction. In some cases, the IR was determined by its magnitude (e.g., state alcohol tax rates, alcohol outlet density, number of alcohol control enforcement personnel). The IR score for each policy by state and year ranged from 0.0 (no policy) to 1.0 (full implementation) for all policies. IR scores varied by state and year, whereas the scoring criteria applied to each policy were uniform across state and year. Additional details about the policy panelists and development of the Alcohol Policy Scale ERs and IRs are available in previous articles.11,32

Aggregating Policy Data to Generate Alcohol Policy Scale Scores We followed a previously used aggregation technique to calculate a composite index by first converting the ER to the inverse of its efficacy rank relative to other policies.11 We then multiplied each policy’s converted ER

(i.e., based on the inverse of its rank) by the same policy’s IR. We then summed across the 29 policies to obtain an overall score for each state-year. Mathematically, the formula to calculate the raw APS scores is: ð1Þ APS score jh ¼

n¼29 X

 ERk ; IRkjh ;

k¼1

where j = state; h = year; k = policy; ER = converted ER based on inverse of rank; and IR = legislative IR. Each raw APS score was then divided by the maximum possible score and multiplied by 100 to rescale it within a theoretical range from 0% to 100%. We also explored the independent associations of several policy subgroups with the 3 binge drinking measures. Methods for calculating modified APS scores to represent policy subgroups were the same as described previously, except that APS scores were based on a restricted number of policies rather than all 29 policies. Policies in the subgroups that were used for comparison with other policy subgroups were mutually exclusive; the policies included in each policy subgroup are listed as data available as a supplement to the online version of this article at http://www.ajph.org. General population policies were defined as policies that were not specific to targeting individuals under the legal drinking age (£ 20 years). Underage-specific policies were defined as policies to reduce or prevent access to alcohol specifically among individuals under the legal drinking age (£ 20 years). Consumption-oriented policies consisted of policies that regulated alcohol production, sales, consumption, or furnishing practices. Consumption policies were compared with impaired driving policies, which consisted of policies to prevent an already intoxicated person from driving a motor vehicle. We also compared the subgroup of policies with higher ERs with those with lower ERs. Finally, we compared policy subgroups based on their mechanism of action. Pricing policies alter the prices of alcoholic beverages. Physical availability policies alter the frequency, duration, location, or social context in which the sale and purchase of alcohol is permitted. We placed policies not falling into pricing or physical availability subgroups into an “other” group for comparison.

April 2015, Vol 105, No. 4 | American Journal of Public Health

Binge Drinking Measures We obtained individual-level data on binge drinking, binge drinking frequency, and maximum drinks consumed during 1 episode within the past month from the Behavioral Risk Factor Surveillance System (BRFSS) survey.33,34 The BRFSS is a state-based, random-digit dial telephone survey of people aged 18 years and older that is conducted monthly in all states, the District of Columbia, and 3 US territories. Data were weighted to be representative of state populations. Further details about the BRFSS and its methods are available at http://www. cdc.gov/brfss. Binge drinking measures were assessed in the 50 states and the District of Columbia from 2005 to 2010. Of the 2 397 742 adult respondents for the 6 study years (2005---2010), the median age of the weighted national sample was 45 years, and 48.6% were male. These years of data were assessed because the Centers for Disease Control and Prevention began to collect the maximum number of drinks data (the basis for 1 of the outcome measures) in 2005. Binge drinking was defined as 1 or more occasions of consuming 5 or more (men) or 4 or more (women) drinks during the past 30 days. To explore more frequent or intense measures of binge drinking, we created 2 additional dichotomous outcome variables: 1 variable was when the threshold for the number of drinks remained the same (‡ 5 for men, ‡ 4 for women), but the threshold for frequency of occasions increased to 5 or more episodes in the past 30 days; the other variable was when the threshold for the frequency remained the same (‡ 1), but the threshold for number of drinks increased to 10 or more based on the maximum number of drinks consumed by a respondent during a drinking occasion during the past 30 days.

Individual- and State-Level Covariates Individual-level measures included gender, age, race/ethnicity (non-Hispanic White, nonHispanic Black, Hispanic, and non-Hispanic others, including Asian, Native Hawaiian, or other Pacific Islander, American Indian, or Alaska Native), marital status (married, not married), level of education (< high school, high school or general equivalency diploma, some post-high school, college graduate), and

Xuan et al. | Peer Reviewed | Research and Practice | 817

RESEARCH AND PRACTICE

household income (< $15 000, $15 000--$24 999, $25 000---$34 999, $35 000---$49 999, ‡ $50 000) collected in the BRFSS. State-level covariates included state-level demographic characteristics (proportion of adults aged ‡ 21 years, gender, race/ethnicity), urbanization (proportion of urban population), median household income, religious composition (Catholic adherence rate), police officers per capita, and geographic region (Northeast, Midwest, South, West). Data for these covariates were obtained from several sources.35 We engaged in a multistep analysis process. We used GEE logistic regression to account for the clustering among respondents within states. We estimated a series of GEE models to assess the bivariate relationship between APS scores and adult individual-level binge drinking measures, with subsequent models controlling for individual-level covariates, year as a categorical variable, and state-level covariates. We computed odds ratios and robust standard errors with a significance level of a = 0.05 and 95% confidence intervals (CIs). We examined the odds of binge drinking for an absolute 10 percentage point (i.e., 10 percentage point of the total maximum possible APS) increase in the state APS scores. This 10 percentage point difference was similar to the interquartile range of APS scores among the 306 state-years of data (interquartile range = 11.2 percentage point of the total maximum possible APS score). We examined the stratified associations between the APS and the 3 binge drinking measures on the

basis of age, gender, and race/ethnicity in separate analyses. A 1-year lag between the policy exposure variable and binge drinking measures was included in all analyses (APS scores in year X were associated with binge drinking prevalence in year X + 1). We also examined the relationship between APS scores and binge drinking stratified by year.

RESULTS Of the 2 397 742 adult respondents for the 6 study years, after weighting, 14.0% reported binge drinking within the past 30 days, 3.4% reported 5 or more binge drinking episodes, and 3.7% reported consuming a maximum of 10 or more drinks during a drinking occasion in the past 30 days.

Alcohol Policy Scale Scores and Binge Drinking In bivariate analyses, a 10 percentage point increase in the APS score was associated with a 10% lower odds of binge drinking and binge drinking 5 or more times, and a 12% lower odds of consuming 10 or more drinks during an occasion (Table 1). In fully adjusted models, a 10 percentage point increase in the APS score was associated with a lower likelihood of binge drinking (adjusted odds ratio [AOR] = 0.92; 95% CI = 0.91, 0.93), frequent binge drinking (AOR = 0.92; 95% CI = 0.90, 0.94), and consuming 10 or more drinks (AOR = 0.90; 95% CI = 0.87, 0.92). We conducted a sensitivity

analysis by removing respondents between ages 18 and 20 years (under the legal drinking age), but we found no difference in the parameter estimates. The relationship between APS scores and binge drinking outcomes did not differ substantially by gender or age group (Table 2), whereas almost all of the CIs from the stratified analyses overlapped, except for the comparison between the 35 to 64 years and the 65 years and older age groups. However, there were significant differences by race/ethnicity, particularly for the association between the APS and binge drinking. Specifically, the adjusted odds of binge drinking were lower among nonHispanic White individuals (AOR = 0.87; 95% CI = 0.86, 0.88) and non-Hispanic others (AOR = 0.92; 95% CI = 0.87, 0.98; suggesting a greater effect of alcohol policies) than for non-Hispanic Black and Hispanic groups. Because the BRFSS began to use a gender-specific definition of binge drinking (‡ 5 drinks for men, ‡ 4 drinks for women) in 2006, we conducted sensitivity analyses by restricting the analyses from 2006 to 2010; the results were similar to those reported in Tables 1 and 2.

Policy Subgroups and Binge Drinking Measures The independent effects of modified APS scores representing policy subgroups are presented in Table 3. Subgroups that had the strongest independent relationship with binge drinking were policies targeting the general population versus those targeting youths, those

TABLE 1—Odds Ratios of Individual-Level US Adult Binge Drinking Measures Associated With an Absolute 10 Percentage Point Increase in the State-Level Alcohol Policy Scale (APS) Score: Behavioral Risk Factor Surveillance System, 2005–2010 Binge Drinking,a OR (95% CI)

‡ 5 Binge Drinking Episodes,a OR (95% CI)

Maximum No. of Drinksb ‡ 10, OR (95% CI)

Bivariate GEE model

0.90 (0.89, 0.90)

0.90 (0.89, 0.92)

0.89 (0.87, 0.91)

Adjusted GEE model (individual-levelc covariates)

0.89 (0.88, 0.90)

0.90 (0.88, 0.92)

0.90 (0.88, 0.91)

Adjusted GEE model (individual-levelc covariates and yeard)

0.89 (0.88, 0.90)

0.90 (0.88, 0.92)

0.90 (0.87, 0.92)

Adjusted GEE model (individual-levelc and state-levele covariates, and year)

0.92 (0.91, 0.93)

0.92 (0.90, 0.94)

0.90 (0.87, 0.92)

Models

Note. CI = confidence interval; GEE = generalized estimating equations; OR = odds ratio. A 1-year lag between the APS exposure variable and drinking measures was assessed (e.g., APS scores in year X were associated with binge drinking measures in year X + 1) using GEE models. a Binge drinking was defined in Behavioral Risk Factor Surveillance System Surveys (BRFSS) as ‡ 1 occasions of consuming ‡ 4 drinks for women or ‡ 5 drinks for men during the past 30 days. b Maximum number of drinks was defined in BRFSS as the maximum number of drinks a person consumed on 1 occasion during the past 30 days. c Individual-level covariates are gender, age, race/ethnicity, marital status, level of education, and household income. d Year modeled as a categorical variable. e State level covariates are proportion of adult (‡ 21 years) population, gender, race/ethnicity, urbanization, household income, religious composition (Catholic), police officers per capita, and geographic region.

818 | Research and Practice | Peer Reviewed | Xuan et al.

American Journal of Public Health | April 2015, Vol 105, No. 4

RESEARCH AND PRACTICE

TABLE 2—Odds Ratio of Individual-Level US Adult Binge Drinking Outcomes Associated With an Absolute 10 Percentage Point Increase in the State-Level Alcohol Policy Scale (APS) Score, Stratified by Demographic Characteristics: Behavioral Risk Factor Surveillance System, 2005–2010 Binge Drinking,a AOR (95% CI)

‡ 5 Binge Drinking Episodes,a AOR (95% CI)

Maximum Number of Drinksb ‡ 10, AOR (95% CI)

0.92 (0.91, 0.94)

0.92 (0.91, 0.96)

0.90 (0.88, 0.93)

0.92 (0.90, 0.93)

0.91 (0.87, 0.95)

0.86 (0.81, 0.91)

18–20

0.89 (0.83, 0.96)

0.96 (0.85, 1.08)

0.95 (0.85, 1.06)

21–34

0.91 (0.89, 0.93)

0.92 (0.88, 0.96)

0.89 (0.86, 0.93)

35–64

0.93 (0.91, 0.94)

0.92 (0.89, 0.94)

0.89 (0.86, 0.92)

‡ 65

0.86 (0.83, 0.90)

0.85 (0.79, 0.91)

0.77 (0.68, 0.88)

Models Stratified by Demographic Characteristics Gender Male Female Age, y

Race/ethnicity Non-Hispanic White

0.89 (0.88, 0.91)

0.90 (0.88, 0.92)

0.87 (0.84, 0.89)

Non-Hispanic Black

1.02 (0.96, 1.07)

1.02 (0.91, 1.14)

1.07 (0.94, 1.23)

Hispanic

1.00 (0.94, 1.06)

1.04 (0.92, 1.16)

0.97 (0.87, 1.07)

Non-Hispanic others

0.92 (0.87, 0.98)

0.86 (0.77, 0.97)

0.85 (0.76, 0.96)

Note. AOR = adjusted odds ratio; CI = confidence interval. A 1-year lag between the APS exposure variable and drinking measures was assessed (e.g., APS scores in year X were associated with binge drinking measure in year X + 1) using generalized estimating equations models. Covariates adjusted in the model include year as a categorical variable, individuallevel covariates (gender, age, race/ethnicity, marital status, level of education, and household income), and state-level covariates (proportion of adult [‡21 years] population, gender, race/ethnicity, urbanization, household income, religious composition [Catholic], police officers per capita, and geographic region). a Binge drinking was defined in Behavioral Risk Factor Surveillance System Surveys (BRFSS) as 1 or more occasions of consuming 4 or more drinks for women or 5 or more drinks for men during the past 30 days. b Maximum number of drinks was defined in BRFSS as the maximum number of drinks a person consumed on 1 occasion during the past 30 days.

targeting alcohol consumption versus those targeting impaired driving, those with higher efficacy ratings versus those with lower ratings, and those policies that either raised the price of alcohol or reduced its physical availability versus other policies. Pricing and physical availability policies were moderately correlated within states (r = 0.36; P < .001; data not shown). After adjusting for other policies, a 10 percentage point increase in a combined measure that included all pricing and physical availability policies (n = 16 policies) was associated with reduced odds of binge drinking (AOR = 0.93; 95% CI = 0.92, 0.94). Among the 3 pricing policies, taxes (AOR = 0.97; 95% CI = 0.96, 0.97) had the most protective effect for binge drinking after controlling for the remaining 28 policies, whereas the effects for wholesale and retail pricing policies were relatively minimal in adjusted models. Among physical availability policies, outlet density (AOR = 0.96; 95%

CI = 0.95, 0.97) had a strong independent association with binge drinking after controlling for all remaining policies. After adjusting for the other 27 policies, a 10 percentage point increase in a combined measure based on tax and outlet density was associated with a 4% reduced odds of binge drinking (AOR = 0.96; 95% CI = 0.95, 0.96).

DISCUSSION Our study is the first to examine the relationship between the state-level alcohol policy environment (defined as the cumulative effect of multiple concurrent policies) and individual-level binge drinking behaviors in the United States. We found that a 10 percentage point higher score on the strength of the state policy environment corresponded to an approximate 8% lower adjusted odds of binge drinking and binge drinking 5 or more times, and a 10% lower adjusted odds of consuming 10 or more

April 2015, Vol 105, No. 4 | American Journal of Public Health

drinks within 1 episode during the past 30 days. The evidence of a strong inverse relationship between the strength of the policy environment and individual-level binge drinking constituted another step in the validation of the APS and is indicative of an environment--behavior linkage of substantial public health significance. The strong and consistent inverse relationships between higher APS scores and highfrequency and high-intensity binge drinking (in addition to binge drinking) suggested that alcohol control policies might have a similar impact on more frequent and more intense binge drinking. This lends additional evidence35 contradicting the assertion that alcohol policies, including taxes, might have relatively smaller effects on highly risky alcohol consumption patterns.36,37 Significant associations between the policy environment and binge drinking were observed across age, gender, and racial/ethnic groups, although the relationship was the least consistent on the basis of race/ethnicity. The lack of a statistically significant association between the APS and binge drinking among Blacks and Hispanics warrants future study. However, it was possible that the lower prevalence of binge drinking among Blacks and Hispanics compared with Whites10 might have some bearing on this finding. It was possible that individual-level covariates played a lesser role in altering the strength of the association than did state-level covariates. These results further emphasized the importance of state-level factors—policy related and otherwise—to alcohol consumption patterns among individuals. Understanding the independent effects of policy subgroups or individual policies within the context of the larger policy environment can assist in policy planning and priority setting. We explored independent effects of policy subgroups with the expectation that independent inverse associations with adult binge drinking would be stronger for (1) generalpopulation policies compared with underage-specific policies, (2) consumptionoriented policies compared with drivingoriented policies, and (3) policies with higher comparative ERs compared with those with lower ERs. The results were consistent with our expectations and, therefore,

Xuan et al. | Peer Reviewed | Research and Practice | 819

RESEARCH AND PRACTICE

TABLE 3—Odds Ratios of Individual-Level US Adult Binge Drinking Outcomes Associated With an Absolute 10 Percentage Point Increase in the State-Level Alcohol Policy Scale (APS) Score, Defined by Policy Subgroups: Behavioral Risk Factor Surveillance System, 2005–2010 Policy Groups All policiesd (n = 29)

Binge Drinking,a AORb (95% CI)

‡ 5 Binge Drinking Episodes,a AORb (95% CI)

Maximum Number of Drinksc ‡ 10, AORb (95% CI)

0.92 (0.91, 0.93)

0.92 (0.90, 0.94)

0.90 (0.87, 0.92)

Age-targeted policiese General-population policies (n = 19)

0.92 (0.91, 0.93)

0.93 (0.91, 0.95)

0.91 (0.89, 0.94)

Underage-specific policies (n = 10)

1.01 (0.99, 1.02)

0.98 (0.96, 1.01)

0.97 (0.95, 0.997)

Consumption versus driving orientationf Consumption-oriented policies (n = 21)

0.93 (0.92, 0.94)

0.93 (0.91, 0.95)

0.91 (0.89, 0.93)

0.98 (0.97, 0.99)

0.97 (0.95, 0.99)

0.97 (0.95, 0.99)

High rating (n = 14)

0.94 (0.93, 0.95)

0.94 (0.92, 0.96)

0.92 (0.90, 0.94)

Low rating (n = 15)

0.98 (0.97, 0.99)

0.96 (0.94, 0.98)

0.95 (0.93, 0.98)

Driving-oriented policies (n = 8) Policy efficacy ratingg

Mechanism of actionh Pricing policies (n = 3)

0.96 (0.96, 0.97)

0.97 (0.96, 0.99)

0.97 (0.95, 0.98)

Physical availability policies (n = 13) Other policies (n = 13)

0.96 (0.95, 0.97) 0.99 (0.98, 1.00)

0.96 (0.94, 0.99) 0.97 (0.96, 0.98)

0.96 (0.93, 0.98) 0.96 (0.95, 0.97)

Note. AOR = adjusted odds ratio; CI = confidence interval. a Binge drinking was defined in Behavioral Risk Factor Surveillance System Surveys (BRFSS) as ‡ 1 occasions of consuming ‡ 4 drinks for women or ‡ 5 drinks for men during the past 30 days. b A 1-year lag between the APS exposure variable and drinking measures was assessed (e.g., APS scores in year X were associated with binge drinking measure in year X + 1) using generalized estimating equations models. All models included year as a categorical variable, individual-level covariates (gender, age, race/ethnicity, marital status, level of education, and household income), and state-level covariates (proportion of adult [‡ 21 years] population, gender, race/ethnicity, urbanization, household income, religious composition [Catholic], police officers per capita, and geographic region). c Maximum number of drinks was defined in BRFSS as the maximum number of drinks a person consumed on 1 occasion during the past 30 days. d The estimates for “All policies” are the same estimates from the fully adjusted models in Table 1. e The adjusted model included both subgroups of policies. General-population policies consisted of policies that are intended to target the general population. Underage-specific policies consisted of policies that are intended to target the underage population. See data available as a supplement to the online version of this article at http://www.ajph.org for list of policies. f The adjusted model included both subgroups of policies. Consumption-oriented policies consisted of policies that do not regulate driving, or can otherwise be viewed as policies that regulate alcohol production, sales, consumption, or furnishing. The driving policies group consisted of policies that are intended to act by preventing or removing an already intoxicated person from driving, or can otherwise be viewed as policies that regulate driving. See data available as a supplement to the online version of this article at http://www.ajph.org for list of policies. g The adjusted model included both subgroups of policies. Policy groups were determined by efficacy ratings assigned by a panel of 10 alcohol policy experts. See data available as a supplement to the online version of this article at http://www.ajph.org for list of policies. h The adjusted model included all 3 subgroups of policies. Pricing policies consisted of those that theoretically affect the consumer price of alcohol. Physical availability policies are those that theoretically establish restrictions on the availability of alcohol. Other policies group consisted of policies that are intended to act by preventing or removing an already intoxicated person from driving, or can otherwise be viewed as policies that regulate driving. See data available as a supplement to the online version of this article at http://www.ajph.org for the list of policies.

further supported the construct validity of the APS. Among policies grouped by their mechanism of action, policies that raised the price of alcohol and those that limited the availability of alcohol had significantly stronger inverse associations with binge drinking compared with other policies. Particularly noteworthy was that the pricing policies group consisted of only 3 policies, whereas the other 2 groups consisted of 13 policies each. Between them, pricing and physical availability policies accounted for almost 90% of the effect magnitude associated with all 29 policies, and among those 16 policies, taxes and outlet density accounted for approximately half of the effect magnitude associated with all 29 policies. These findings not only were consistent with other research

that demonstrated that pricing and physical availability policies were highly effective,1,21,35 but also demonstrated that most of the impact of the alcohol policy environment was accounted for by a relatively small number of policies.

Study Limitations Our study is subject to caveats and limitations. For our study and other policy studies, caution about causation versus association is warranted. However, although less alcohol consumption or binge drinking might be associated with increased public opinion in favor of alcohol policies (suggesting possible reverse causation), the primary evidence base for the Centers for Disease Control and Prevention comprehensive alcohol policy reviews were conducted by the Guide to Community

820 | Research and Practice | Peer Reviewed | Xuan et al.

Preventive Services. This guide covered many of the policies used in our APS scales (including taxes and alcohol outlet density) and was based on longitudinal studies in which the effect of a policy was assessed after its enactment. This then controlled for the prevalent attitudes that led to their earlier passage, and demonstrated a temporal relationship between policy implementation and subsequent changes in drinking behavior or drinking-related outcomes.16,17,21---23,38,39 Using lagged analyses in repeated cross-sectional GEE models that accounted for years as fixed effects also helped establish temporal relationships in our study. We studied state-level policy environments, and therefore, excluded a number of potentially effective policies that were not

American Journal of Public Health | April 2015, Vol 105, No. 4

RESEARCH AND PRACTICE

promulgated at the state level (e.g., federal policies, city-level alcohol taxes, and mass media alcohol advertising policies). Furthermore, some state-level policies nominated as efficacious were not feasible to include in our study because there were no reliable data available across all 50 states for those policies.32 The ERs and legislative IRs were determined by a group of 10 alcohol policy experts and the research team, based on an incomplete and limited evidence base. A different set of experts and a different research team might have established different ratings and a different scale. Finally, our implementation ratings for some state-level policies (e.g., retail pricing policies, regulation of hours of sale) might not reflect the degree to which local policies might reinforce or counteract the effects of corresponding statelevel policies. Enforcement of existing policies might modify the effects of those policies, but because there were no reliable, publicly available cross-state data on enforcement, we were unable to take into account the level of policy enforcement. This limitation was addressed by including policy provisions that made particular policies enforceable, by including the number of Alcoholic Beverage Control officials with enforcement capability as an alcohol policy in our scales, and by controlling for the number of police officers per capita as a state-level control variable. Although the alcohol policy environment was inversely associated with binge drinking among the underage adult population in BRFSS (i.e., those aged 18---20 years), underage-specific policies were not significantly associated with adult binge drinking, independent of general population policies. This was likely because of the limited age range of underage respondents in BRFSS. The impact of youth-specific policies should be further ascertained in other underage populations with a larger age range and younger participants. Finally, BRFSS estimates were subject to survey noncoverage and nonresponse biases, but were reliable for comparisons across states,40,41 which was the focus of our analyses. Limitations related to the imprecision of the exposure or outcome variables might have biased the results toward the null hypothesis, particularly because the methods used here for policy ascertainment,

policy scoring systems, and determining binge drinking prevalence were uniform across states.

Note. The content and views expressed in this article are those of the authors and do not necessarily represent those of the National Institute on Alcohol Abuse and Alcoholism or the National Institutes of Health.

Conclusions Characterizing the overall alcohol policy environment might be useful for future studies that focus on the independent effects of other individual policies or other policy subgroups. Future studies of alcohol policies in the United States could explore the correlations and interactions of individual policies and policy subgroups within the context of the overall policy environment across US states. The study strongly corroborates other evidence that alcohol policies reduce binge drinking, including frequent and high-intensity binge drinking, at the population level. j

About the Authors Ziming Xuan is with the Department of Community Health Sciences, Boston University School of Public Health, Boston, MA. Timothy Heeren is with the Department of Biostatistics, Boston University School of Public Health. Nadia Oussayef is with the Department of Health Law, Bioethics & Human Rights, Boston University School of Public Health. Timothy S. Naimi and Jason Blanchette are with the Section of General Internal Medicine, Boston Medical Center, Boston, MA. Toben F. Nelson is with the Division of Epidemiology and Community Health, University of Minnesota School of Public Health, Minneapolis. Correspondence should be sent to Ziming Xuan, ScD, SM, MA, Department of Community Health Sciences, Boston University School of Public Health, 801 Massachusetts Avenue, Boston, MA 02118 (e-mail: zxuan@bu. edu). Reprints can be ordered at http://www.ajph.org by clicking the “Reprints” link. This article was accepted May 29, 2014.

Contributors All of the authors contributed to the design and development of the study. Z. Xuan originated the study, oversaw the statistical analysis, and led the overall writing and revisions of the article. J. Blanchette participated in scale development, statistical analysis, and drafting of the article. T. F. Nelson and T. Heeren provided advice on interpretation of the findings and contributed to development of the article. N. Oussayef provided legal expertise in policy scale development. T. S. Naimi was the Principal Investigator of the project and oversaw the study and co-wrote the article.

Acknowledgments This research was supported by National Institutes of Health grant AA018377. We wish to acknowledge the enormous contributions of the following individuals who served as alcohol policy consultants for this project: Thomas Babor, PhD, Robert Brewer, MD, MSPH, Frank Chaloupka, PhD, Paul Gruenewald, PhD, Harold Holder, PhD, Michael Klitzner, PhD, James Mosher, JD, Rebecca Ramirez, MPH, Robert Reynolds, MA, and Traci Toomey, PhD.

April 2015, Vol 105, No. 4 | American Journal of Public Health

Human Participant Protection This study involved secondary analyses of publicly available, de-identified data. The institutional review board at Boston University School of Public Health reviewed and approved the study.

References 1. Babor T, Caetano R, Casswell S, et al, eds. Alcohol: No Ordinary Commodity: Research and Public Policy. 2nd ed. New York, NY: Oxford University Press; 2010. 2. Mokdad AH, Marks JS, Stroup DF, Gerberding JL. Actual causes of death in the United States, 2000. JAMA. 2004;291(10):1238---1245. 3. Rehm J, Mathers C, Popova S, Thavorncharoensap M, Teerawattananon Y, Patra J. Global burden of disease and injury and economic cost attributable to alcohol use and alcohol-use disorders. Lancet. 2009;373(9682): 2223---2233. 4. World Health Organization. Global Status Report on Alcohol and Health 2011. Geneva, Switzerland: World Health Organization; 2011. 5. Bouchery EE, Harwood HJ, Sacks JJ, Simon CJ, Brewer RD. Economic costs of excessive alcohol consumption in the US, 2006. Am J Prev Med. 2011;41(5):516---524. 6. Sacks JJ, Roeber J, Bouchery EE, Gonzales K, Chaloupka FJ, Brewer RD. State costs of excessive alcohol consumption, 2006. Am J Prev Med. 2013;45(4)474---485. 7. Centers for Disease Control and Prevention. Alcohol-Related Disease Impact (ARDI). 2008. Available at: http://www.cdc.gov/alcohol/ardi.htm. Accessed September 28, 2013. 8. LaVallee RA, Yi H. Apparent Per Capita Alcohol Consumption: National, State, and Regional Trends, 19772009: National Institute on Alcohol Abuse and Alcoholism. Survillance Report #92. Bethesda, MD: National Institutes of Health; 2011. 9. Kerr WC. Categorizing US state drinking practices and consumption trends. Int J Environ Res Public Health. 2010;7(1):269---283. 10. Kanny D, Liu Y, Brewer RD, Lu H. Binge drinking— United States, 2011. MMWR Surveill Summ. 2013; 62(suppl 3):77---80. 11. Naimi TS, Blanchette J, Nelson TF, et al. A new scale of the US alcohol policy environment and its relationship to binge drinking. Am J Prev Med. 2014;46(1)10---16. 12. National Institute of Alcohol Abuse and Alcoholism. Alcohol Policy Information System (APIS) 2013. Available at: http://www.alcoholpolicy.niaaa.nih.gov. Accessed January 14, 2014. 13. Klingemann HK, Holder HD, Gutzwiller F. Research on and prevention of alcohol-related trauma: phases, patterns, and issues. Addiction. 1993;88(7):861---866. 14. Babor TF. Linking science to policy. The role of international collaborative research. Alcohol Res Health. 2002;26(1):66---74. 15. Wagenaar AC, Tobler AL, Komro KA. Effects of alcohol tax and price policies on morbidity and mortality: a systematic review. Am J Public Health. 2010;100 (11):2270---2278.

Xuan et al. | Peer Reviewed | Research and Practice | 821

RESEARCH AND PRACTICE

16. Hahn RA, Kuzara JL, Elder R, et al. Effectiveness of policies restricting hours of alcohol sales in preventing excessive alcohol consumption and related harms. Am J Prev Med. 2010;39(6):590---604.

34. Nelson DE, Holtzman D, Bolen J, Stanwyck CA, Mack KA. Reliability and validity of measures from the Behavioral Risk Factor Surveillance System (BRFSS). Soz Praventivmed. 2001;46(suppl 1):S3---S42.

17. Rammohan V, Hahn RA, Elder R, et al. Effects of dram shop liability and enhanced overservice law enforcement initiatives on excessive alcohol consumption and related harms: two community guide systematic reviews. Am J Prev Med. 2011;41(3):334---343.

35. Xuan Z, Nelson TF, Heeren T, et al. Tax policy, adult binge drinking, and youth alcohol consumption in the United States. Alcohol Clin Exp Res. 2013;37(10):1713---1719.

18. Shults RA, Elder RW, Sleet DA, et al. Reviews of evidence regarding interventions to reduce alcohol-impaired driving. Am J Prev Med. 2001;21(4 suppl 1):66-- 88. 19. Transportation Research Board of the National Academies. Traffic Safety and Alcohol Regulation: A Symposium. 2006. Available at: http://onlinepubs.trb. org/onlinepubs/circulars/ec123.pdf. Accessed January 15, 2013. 20. Bonne RJ, O’Connell ME, eds. Reducing Underage Drinking: A Collective Responsibility. Washington, DC: The National Academies Press; 2004. 21. Elder RW, Lawrence B, Ferguson A, et al. The effectiveness of tax policy interventions for reducing excessive consumption and related harms. Am J Prev Med. 2010;38:217---229. 22. Campbell CA, Hahn RA, Elder R, et al. The effectiveness of limiting alcohol outlet density as a means of reducing excessive alcohol consumption and alcohol-related harms. Am J Prev Med. 2009; 37(6):556---569. 23. Middleton JC, Hahn RA, Kuzara JL, et al. Effectiveness of policies maintaining or restricting days of alcohol sales on excessive alcohol consumption and related harms. Am J Prev Med. 2010;39(6):575---589.

36. Distilled Spirits Council of the United States. Increasing Alcohol Taxes Punishes the Entire Hospitality Industry. Available at: http://www.discus.org/policy/ taxes. Accessed February 17, 2014. 37. Beer Institute. Beer Tax Facts: The Economic and Societal Impacts of State and Federal Taxes on Beer. Available at: http://www.beerinstitute.org/assets/uploads/ Beer_Tax_Facts1.pdf. Accessed February 22, 2014. 38. Centers for Disease Control and Prevention. The Community Guide: Preventing Excessive Alcohol Consumption. Available at: http://www.thecommunityguide. org/alcohol/index.html. Accessed January 23, 2014. 39. Centers for Disease Control and Prevention. The Community Guide; Motor Vehicle-Related Injury Prevention: Reducing Alcohol-Impaired Driving. Available at: http://www.thecommunityguide.org/mvoi/aid/index. html. Accessed January 23, 2014. 40. Nelson DE, Naimi TS, Brewer RD, Roeber J. US state alcohol sales compared to survey data, 1993---2006. Addiction. 2010;105(9):1589---1596. 41. Paschall MJ, Ringwalt CL, Gitelman AM. The validity of state survey estimates of binge drinking. Am J Prev Med. 2010;39(2):179---183.

24. Brand DA, Saisana M, Rynn LA, et al. Comparative analysis of alcohol control policies in 30 countries. PLoS Med. 2007;4:e151. 25. Gilpin EA, Stillman FA, Hartman AM, Gibson JT, Pierce JP. Index for US state tobacco control initial outcomes. Am J Epidemiol. 2000;152(8):727---738. 26. Joossens L, Raw M. The Tobacco Control Scale: a new scale to measure country activity. Tob Control. 2006;15(3):247---253. 27. Schaap MM, Kunst AE, Leinsalu M, et al. Effect of nationwide tobacco control policies on smoking cessation in high and low educated groups in 18 European countries. Tob Control. 2008;17(4):248---255. 28. Taber DR, Chriqui JF, Perna FM, Powell LM, Chaloupka FJ. Weight status among adolescents in states that govern competitive food nutrition content. Pediatrics. 2012;130(3)437---444. 29. Nanney MS, Nelson T, Wall M, et al. State school nutrition and physical activity policy environments and youth obesity. Am J Prev Med. 2010;38(1):9---16. 30. Robinson WS. Ecological correlations and the behaviour of individuals. Am Sociol Rev. 1950;15(3):351-- 357. 31. Thomson Reuters. WestlawNext. Available at: http://next.westlaw.com. Accessed October 5, 2011. 32. Nelson TF, Xuan Z, Babor T, et al. Rating the efficacy and strength of evidence of alcohol control policies for the US: a comparative assessment. Am J Prev Med. 2013;45(1):19---28. 33. Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System. Available at: http://www.cdc.gov/brfss. Accessed January 30, 2013.

822 | Research and Practice | Peer Reviewed | Xuan et al.

American Journal of Public Health | April 2015, Vol 105, No. 4