Effects of Alcohol Retail Privatization on Excessive Alcohol Consumption and Related Harms

A Community Guide Systematic Review

Robert A. Hahn, PhD, MPH, Jennifer Cook Middleton, PhD, Randy Elder, PhD, Robert Brewer, MD, MSPH, Jonathan Fielding, MD, MPH, MBA, Timothy S. Naimi, MD, MPH, Traci L. Toomey, PhD, Sajal Chattopadhyay, PhD, Briana Lawrence, MPH, Carla Alexia Campbell, MHSc, and the Community Preventive Services Task Force
Context: Excessive alcohol consumption is the third-leading cause of preventable death in the U.S. This systematic review is one in a series exploring effectiveness of interventions to reduce alcoholrelated harms. Evidence acquisition: The focus of this review was on studies evaluating the effects of the privatization of alcohol retail sales on excessive alcohol consumption and related harms. Using Community Guide methods for conducting systematic reviews, a systematic search was conducted in multiple databases up to December 2010. Reference lists of acquired articles and review papers were also scanned for additional studies. Evidence synthesis: A total of 17 studies assessed the impact of privatizing retail alcohol sales on
the per capita alcohol consumption, a well-established proxy for excessive alcohol consumption; 9 of these studies also examined the effects of privatization on the per capita consumption of alcoholic beverages that were not privatized. One cohort study in Finland assessed the impact of privatizing the sales of medium-strength beer (MSB) on self-reported alcohol consumption. One study in Sweden assessed the impact of re-monopolizing the sale of MSB on alcohol-related harms. Across the 17 studies, there was a 44.4% median increase in the per capita sales of privatized beverages in locations that privatized retail alcohol sales (interquartile interval: 4.5% to 122.5%). During the same time period, sales of nonprivatized alcoholic beverages decreased by a median of 2.2% (interquartile interval: 6.6% to 0.1%). Privatizing the sale of MSB in Finland was associated with a mean increase in alcohol consumption of 1.7 liters of pure alcohol per person per year. Re-monopolization of the sale of MSB in Sweden was associated with a general reduction in alcohol-related harms.

Conclusions: According to Community Guide rules of evidence, there is strong evidence that privatization of retail alcohol sales leads to increases in excessive alcohol consumption.
(Am J Prev Med 2012;42(4):418 427) Published by Elsevier Inc. on behalf of American Journal of Preventive Medicine

Context
From the Community Guide Branch, Epidemiology and Analysis Program Offce (Hahn, Middleton, Elder, Chattopadhyay, Lawrence, Campbell), National Center for Chronic Disease Prevention and Health Promotion (Brewer), CDC, Atlanta, Georgia; Los Angeles County Department of Health Services (Fielding); University of Minnesota School of Public Health (Toomey), Minneapolis, Minnesota; and the Schools of Medicine and Public Health (Naimi), Boston University, Boston, Massachusetts Author affliations are shown at the time the research was conducted. Names and affliations of the Task Force members can be found at www.thecommmunityguide.org/about/task-force-members.html. Address correspondence to: Robert A. Hahn, PhD, MPH, Community Guide Branch, Epidemiology and Analysis Program Offce, CDC, 1600 Clifton Road, Mailstop E-69, Atlanta GA 30333. E-mail: [email protected]. 0749-3797/$36.00 doi: 10.1016/j.amepre.2012.01.002

xcessive alcohol consumption, including both binge drinking and underage drinking, is responsible for approximately 79,000 deaths per year in the U.S., making it the third-leading cause of preventable death in the nation.1 In 2009, approximately 23% of adult drinkers (aged 18 years) in the U.S. reported binge drinking (consuming fve or more drinks per occasion for men and four or more drinks per occasion for women) in the past 30 days, as did 25.2% of high school students.2,3 Among full-time college students in 2008, 48.6% of men and 34.4% of women reported binge drinking.4 In 2006,

418 Am J Prev Med 2012;42(4):418 427

Published by Elsevier Inc. on behalf of American Journal of Preventive Medicine

Hahn et al / Am J Prev Med 2012;42(4):418 427

419

the estimated economic cost of excessive drinking in the U.S. was $223.5 billion.5 The reduction of excessive alcohol consumption is thus a matter of major public health and economic interest. Following the end of Prohibition in the U.S. in 1933, some states continued prohibition at the state level. Control states allowed alcohol to be sold, but only through government-run retail stores; license states allowed retail sales by commercial interests.6 In all states, wholesale of alcoholic beverages is under state control. Currently, no states prohibit alcohol sales, and the number of states that have retained control over retail sales has been reduced through privatization. However, in all states with government control over certain beverage types, government control is restricted to off-premises sales outlets (i.e., outlets where alcohol is sold for consumption elsewhere); government control does not affect the on-premises sale of alcohol in any state. In the U.S. and Canada, privatization most often affects wine and spirits (e.g., vodka and whiskey). In the U.S., all states and counties that permit the sale of alcohol allow privatized retail sales of beer, and most allow privatized retail sale of all alcoholic beverages. In contrast, in the Scandinavian countries where most other studies of privatization have been conducted, privatization and remonopolization generally affect beer sales. The National Alcohol Beverage Control Association (NABCA; www. nabca.org) classifes state retail alcohol sales control policies as shown in Table 1. The predominant trend in the U.S. and elsewhere is toward relaxing government control over the sale of alcoholic beverages, including by privatizing alcohol sales. The formation of the European Union has also led to a loosening of national control and increased privatization in some member nations.7,8 However, because privatization could plausibly lead to increases in excessive alcohol consumption and related harms, a public health intervention of possible interest to some jurisdictions and decision makers may be the reversal of privatization (remonopolization) or the maintenance of government control where it exists currently. This review addresses three research questions related to the effect of privatizing retail sales of alcoholic beverages. (1) Does retail privatization of a specifc type of alcoholic beverage increase its excessive consumption and associated harms (e.g., alcohol-impaired driving, assaults, and cirrhosis of the liver)? (2) Does privatization of sales of one type of alcoholic beverage also reduce excessive consumption of alcoholic beverages for which sales are not privatized (e.g., does the retail privatization of wine sales lead consumers to reduce their consumption of liquor, if liquor sales are still subject to government control)? (3) Does the re-establishment of state control over the retail sales of an
April 2012

Table 1. The National Alcohol Beverage Control Association (NABCA; www.nabca.org/) classication of state retail alcohol sales control policies
State
Alabama Idaho Iowa Maine Maryland (Montgomery County only) Michigan Mississippi Montana New Hampshire North Carolina Ohio Oregon Pennsylvania Utah Vermont Virginia Washington West Virginia Wyoming

Wine
Private Private Private Private Government Private Private Private Private Private Private Private Government Government and agents Private Private Private and government Private Private

Spirits
Government Government and agents Private Agents Government Private Private Agents Government Government Agents Agents Government Government and agents Agents Government Government and agents Private Private

alcoholic beverage (re-monopolization) reduce the excessive consumption of that beverage and the harms related to excessive consumption?

Findings, Recommendations, and Directives from Other Reviews and Advisory Groups
The 2010 WHO-sponsored review Alcohol: No Ordinary Commodity,8 a consensus narrative review of a broad array of alcohol interventions, assessed government control of the retail sale of alcohol, and rated it 3 (the highest rating) for effectiveness in reducing excessive alcohol consumption and related harms and for the extent of research supporting the fnding and 2 on crosscultural testing, suggesting generalizability across settings. The cost of implementing government control was rated as low. The reviewers concluded that the evidence is quite strong that off-premise government control systems limit alcohol consumption and alcohol-related problems,

420

Hahn et al / Am J Prev Med 2012;42(4):418 427

and that elimination of government off-premise monopolies can increase total alcohol consumption.8 The present analysis adds more recent evidence on effects of privatization and applies a more formal protocol to the evaluation and synthesis of available scientifc evidence on this topic. The intervention reviewed here may be helpful in addressing several national health objectives related to substance abuse prevention, as specifed in Healthy People 2020.9

Evidence Acquisition
The Guide to Community Preventive Services (Community Guide) systematic review process was used to assess whether privatization leads to increases in excessive alcohol consumption and related harms. More details on the Community Guide review process are presented elsewhere.10,11 In brief, this process involves forming a systematic review development team; developing a conceptual approach to organizing, grouping, and selecting interventions; prioritizing these interventions; searching for and retrieving the existing research evidence on the effects of the interventions; assessing the quality of each study; abstracting information from each study that meets qualifying criteria; drawing conclusions about the body of evidence on intervention effectiveness; and translating the evidence on effectiveness into recommendations. To help ensure objectivity, the systematic review development team consists of systematic review methodologists and subject matter experts from a range of agencies, organizations, and academic institutions. The review team works under the oversight of the nonfederal, independent Community Preventive Services Task Force (Task Force), which directs the work of the Community Guide. The systematic review development team collects and summarizes evidence on (1) the effectiveness of interventions in altering the healthrelated outcomes of interest; and (2) benefts and harms of the intervention on other health and nonhealth outcomes. When an intervention is shown to be effective, information is also included about (3) the applicability (or generalizability) of the evidence to diverse population segments and settings; (4) the economic impact of the intervention; and (5) barriers to implementation. The systematic review development team then presents the results of this review process to the Task Force, which considers all of the evidence presented and determines whether it is suffcient to warrant a recommendation for practice or policy.10 The rules of evidence under which the Task Force makes its determination require consideration of several aspects of the body of evidence, including the number of studies of different levels of design suitability and execution, as well as judgments regarding the consistency of the fndings, the public health importance of the overall effect size, and the balance of the outcome of interest with other consequences of the intervention.

Figure 1. Analytic framework

Note: Privatization of retail sales of alcoholic beverages: Oval indicates intervention; rectangles with rounded corners indicate mediators or intermediate outcomes; rectangles indicate health outcomes.

overall, prices tend to increase with privatization.13,14 However, the wider range of alcohol products typically available in privatized systems may result in more low-priced products that would appeal to high-volume or high-risk drinkers, even if the average price for all alcohol products were similar in state-owned and privatized systems. Fifth, because it introduces competition among alcohol outlets, privatization may lead to increased alcohol advertising in various venues, including TV, radio, billboards, and at the point of purchase.13 Sixth, because of increased numbers of outlets and less direct governmental control, privatization may lead to decreased enforcement of and compliance with sales regulations (e.g., minimum-drinking-age laws).15 Many of these consequences of privatization may increase the demand for the privatized beverage, substitution to or from other beverages, and access to alcoholic beverages. In turn, these consequences may affect excessive consumption and related harms. In contrast, re-monopolization is expected to have effects opposite to those of privatization.

Inclusion and Exclusion Criteria

To qualify as a source of primary evidence for this review, a study had to meet several criteria. It had to evaluate retail privatization or renewed government control; be conducted in a country with a highincome economy16; be primary research (rather than a review of other research); be published in English; and have a comparison group and/or compare conditions before and after privatization. Optimally, alcohol consumption in locations that experienced changes in government control (e.g., privatization) would be compared with alcohol consumption in similar locations that did not experience this change. A study also had to report outcomes related to excessive alcohol consumption or related harms. Some specifc harms of interest included alcohol-related medical conditions (e.g., liver cirrhosis), alcohol-impaired driving and alcohol-related crashes, unintentional injuries, intentional injuries, and crime. Government reports were included, but unpublished papers were not, as they may be diffcult for others to access and are not peer-reviewed. When studies assessed multiple outcomes of interest, those with the strongest known association with excessive alcohol consumption were given more weight. Outcome measures in the studies that had the strongest association with excessive alcohol consumption included binge drinking, heavy drinking, liver cirrhowww.ajpmonline.org

Conceptual Approach and Analytic Framework

Six intermediate consequences of privatization are hypothesized to affect alcohol consumption (Figure 1). First, privatization generally leads to increases in the density of off-premises alcohol outlets.12 Second, privatization may increase the availability of alcohol by increasing the days and hours during which it is sold.13,14 Third, privatization may increase the availability of specifc types and brands of alcoholic beverages. Fourth, privatization may affect the retail price of alcohol because of various factors that may either increase or decrease prices;

Hahn et al / Am J Prev Med 2012;42(4):418 427 sis mortality, alcohol-related hospital admissions, and alcoholrelated motor vehicle crashes (or strong proxies for such crashes, e.g., single-vehicle nighttime crashes for alcohol-related motor vehicle crashes).17 Less-direct measures included per capita ethanol consumption, a well-recognized proxy for estimating the number of excessive drinkers in a population8,18,19; unintentional injuries; suicide; and crime (e.g., homicide and aggravated assault). Cross-sectional and panel studies in which the sequence of events was not taken into account (e.g., counting year as an ordinal variable) were considered secondary evidence in this review. Secondary evidence was regarded as useful for generating hypotheses and for strengthening or weakening conclusions based on primary evidence, but was insuffcient alone for assessing intervention effectiveness. Many of the studies qualifying for this review used per capita alcohol sales as a proxy for excessive drinking. In assessing the quality of study execution, a penalty (described below) was assigned to studies that assessed changes in population-level consumption rather than changes in excessive drinking by individuals. This was a conservative approach because excessive drinking and per capita alcohol consumption are strongly related both theoretically and empirically.18 20 This empirical relationship is conceptualized in the single distribution theory, which asserts that excessive drinkers, including binge and heavy drinkers, account for a consistent proportion of the drinking population in a given setting,19 such that the prevalence of excessive alcohol consumption is directly related to per capita alcohol sales. Further, the theory proposes that the relationship between per capita sales and heavy or excessive drinking is quadratic, meaning that, If population A has twice the average consumption of population B, then A has about four times (i.e., 22) the prevalence of heavy drinking. Cook and Skog19 report evidence from multiple countries supporting this proposition. Thus, changes in per capita alcohol consumption would be expected to have a greater effect on excessive alcohol consumption, including binge drinking, than on non-excessive alcohol consumption. This evidence supports the Task Forces use of per capita alcohol consumption based on sales as an outcome measure for assessing the impact of privatization on excessive alcohol consumption.

421

The quality of studies that were candidates as primary evidence for this review was evaluated both in terms of design and execution. Studies with greatest design suitability were those in which data on exposed and comparison populations were collected prospectively; studies with moderate design suitability were those in which data on exposed and comparison populations were collected retrospectively or in which there were multiple pre- or post-intervention measurements, but no concurrent comparison population; and studies with least-suitable designs were cross-sectional studies or those in which there was no comparison population or only a single pre- and post-measurement in the intervention population. On the basis of the number of threats to validitysuch as poor measurement of exposure or outcome, lack of control of potential confounders, or high attritionstudies were characterized as having good (at most one threat to validity), fair (two to four threats), or limited (fve or more threats) quality of execution. Studies with good or fair quality of execution and any level of design suitability (greatest, moderate, or least) qualifed for the body of evidence. Effect estimates were calculated as relative percentage change using the following formulas:

For studies with beforeafter measurements and concurrent comparison groups: Effect estimate Ipost Cpost Ipre Cpre 1 100%

where: Ipost last reported outcome rate or count in the intervention group after the intervention; Ipre last reported outcome rate in the intervention group before the intervention; Cpost last reported outcome rate in the comparison group after the intervention; Cpre last reported outcome rate in the comparison group before the intervention.

Search for Evidence

The following databases were searched for this review: Econlit, PsycINFO, Sociology Abstracts, MEDLINE, Embase, and EtOH. All years were searched up to October 2007. (Details of the search strategy are available at www.thecommunityguide. org/alcohol/supportingmaterials/SSalcoholuse.html.) This search was updated in the ISI Web of Knowledge through December 2010 by the Alcohol Epidemiology Program at the University of Minnesota. Reference lists of articles reviewed as well as lists in review articles were searched, and subject matter experts were consulted.

For studies with beforeafter measurements but no concurrent comparison: Effect estimate

Ipost Ipre Ipre 100% .

Assessing the Quality and Summarizing the Body of Evidence on Effectiveness

Each study that met the criteria for candidate studies was read by two reviewers, who used standardized criteria (available at www. thecommunityguide.org/about/methods.html) to assess suitability of study design and threats to validity. Uncertainties and disagreements between reviewers were reconciled by consensus among the team members. The teams classifcation of the designs of studies reviewed corresponds with the research questions of the review and the standards of the Community Guide review process11 and may differ from the classifcation reported in the original studies. April 2012

Several events of privatization (e.g., the privatization of wine in Iowa in 1985) were assessed by more than one team of researchers, thus resulting in multiple studies of the same event. Effect estimates are reported for each research group, noting which were associated with a single event. Median effect sizes are calculated using the means of privatization events with differing fndings from different researchers.

Evidence Synthesis
Intervention Effectiveness
The effects of 12 distinct privatization events were assessed in 17 studies and reported in 13 publications.14,2132 In addition, there was one study of remonopolization, described separately below.33 The privatization events assessed were in seven U.S. states (Alabama, Idaho, Iowa [two events], Maine, Montana,

422

Hahn et al / Am J Prev Med 2012;42(4):418 427

New Hampshire, West Virginia); two Canadian provinces (Quebec [two events] and Alberta); and Finland. Several publications described a single privatization event, and several publications each assessed more than one privatization event. All studies used alcohol sales data as an index of population-level alcohol consumption except one25 that assessed changes in individual-level consumption (in Finland). Fitzgerald and Mulford14,22 also asFigure 2. Percentage change by location of privatization event in consumption of sessed changes in selfprivatized alcoholic beverages attributable to privatization reported consumption in Note: Bars on the point estimate represent 95% CIs when reported or calculated. addition to sales data (in Iowa). However, as they quence of relative ineffciencies of scale (e.g., multiple note, their measures of alcohol consumption were probsmaller outlets and increased overhead expenses).35 In lematic (e.g., separate cross-sectional studies collected contrast to other intermediate outcomes, higher priunder different sampling procedures with different ces would be expected to result in decreased consumpinterview procedures), and thus only the assessments tion. However, Fitzgerald and Mulford21 assessed of changes in alcohol sales from their study are inwhether the increase in the price of spirits following cluded in this review. privatization of retail sales in Iowa had affected consumIn the U.S., privatized beverages were limited to ers purchasing behavior, and found that only 37.4% of wine and spirits, as beer was already privatized. In those surveyed who purchased liquor in the past month Canada, sale of beer was privatized in addition to sales recognized that prices had increased, and 2% of Iowa of wine and spirits, and in Scandinavia, privatization consumers reported increasing their purchase of liquor (and re-monopolization) focused on beer sales. The from adjacent states. privatization events assessed in these studies occurred between 1950 and 2000. Three studies used autoregresEffects of privatization on consumption of privatized sive integrated moving average (ARIMA) time sebeverages. Overall, the median increase in per capita ries.2729 Fourteen studies (presented in eight publicasales of privatized beverages was 44.4%, with an interquartile tions)14,23,26 30,32 were of greatest design suitability; interval of 4.5% to 122.5% (Figure 2). Some differences, three studies (presented in two publications)24,25 were of however, were observed across studies in the impact of moderate design suitability. All studies were of fair qualprivatization on retail sales of privatized beverages and in the ity of execution. relationship between privatization and other public health The qualifying studies provided information on sevoutcomes, described below. eral of the intermediate consequences of privatization Studies assessing the effects of the privatization of wine discussed above (Figure 1). These consequences include (1985) and spirits (1987) sales in Iowa had inconsistent increased numbers of alcoholic beverage outlets, infndings. Wagenaar and Holder31 reported that wine concreased hours and days of sale, advertising, greater brand sumption increased 93.0% (95% CI69.3%, 120.2%) selection, and acceptance of alternate forms of payment from baseline to 44 months after privatization of wine (e.g., credit card).13,14,23,26 30,32,34 All of these intermedisales in Iowa, with no decrease in spirits or beer conate outcomes would be expected to result in increased sumption. Following the subsequent privatization of consumption. spirits sales in Iowa 2 years later, these researchers23 reMost studies reviewed reported generally higher prices ported a 9.5% (95% CI3.5%, 15.9%) increase in spirits for alcoholic beverages in the privatized than in the state consumption, along with a 12.1% (95% CI 20.6%, control setting. The higher prices may be the conse2.7%) decrease in wine consumption and no change in
www.ajpmonline.org

Hahn et al / Am J Prev Med 2012;42(4):418 427

423

beer consumption. They also found no evidence that privatization affected alcohol purchasing across state lines (effect estimate 0.1%, 95% CI 3.1%, 6.2%). In contrast, Mulford and Fitzgerald27 found that wine privatization was associated with a nonsignifcant long-term increase of only 0.5% (95% CI 6.8%, 8.3%) in wine sales, and that spirits privatization was associated with a nonsignifcant long-term increase of 0.7% (95% CI 1.9%, 3.4%) in spirits sales. For both beverages, the nonsignifcant long-term effects were preceded by 3-year spikes in sales. The differences between the conclusions of these two research teams about the effects of the privatization of wine and spirits in Iowa may be the consequence of different modeling strategies, different time periods covered, and different forms of alcoholic beverages includedparticularly the inclusion of wine coolers in measures of wine sales by Wagenaar and Holder.31,32 Finally, Makela25 assessed the impact on alcohol consumption of a law in Finland that allowed the sale of mediumstrength beer (MSB) in grocery stores. This was the only study included in the review that assessed changes in selfreported alcohol consumption by individuals over time. Survey participants were specifcally asked about their levels of consumption before implementation of the new law and then again in the year following its implementation. The researchers stratifed their fndings based on drinking patterns of respondents before and after privatization. Consumption of all alcoholic beverages (not just the privatized beverage) increased by a mean of 1.7 L of pure alcohol per year per person interviewed (approximately 137 ounces of 80 proof liquor). Makela reports that 86% of the increase in overall alcohol consumption was attributable to increases in the privatized beverage (MSB). The greatest increase in alcohol consumption after privatization was observed among those who reported drinking between 17 and 68 ounces of pure alcohol per year at frst interview. However, there was also an increase in consumption in the population that reported no alcohol consumption within the past 30 days when frst interviewed. Effects of privatization on alcohol-related harms. Two studies assessed the association between retail privatization and motor vehicle crashes. One study estimated that incremental privatization over a 20-year period was associated with a nonsignifcant 11.3% (95% CI 33.9%, 19.0%) decrease in traffc fatalities in Alberta, Canada.29 This study estimated the degree of privatization over a long period preceding fnal privatization in 1994, had only 1 year of follow-up, and used a proxy outcome measure. A second study assessed changes in alcohol-related harms associated with the Iowa privatization of wine in 1985 and spirits in 1987.14,26 The researchers compared the period before 1985 with the period after 1989, when
April 2012

23,31

both wine and spirits sales were privatized. Despite increased per capita sales of both wine and spirits, there was a reported 1.6% decline in nighttime motor vehicle crashes and a 5.5% decline in liver cirrhosis. However, initial mortality data were for 1985the same year in which the privatization of wine occurred; thus these data included deaths both prior to and following privatization, weakening the analysis. Moreover, no comparison data were provided to adjust for national or regional trends in these outcomes over the time period evaluated. Effects of privatization on the consumption of nonprivatized alcoholic beverages. Many of the studies14,25,26,31,32,36 reviewed also assessed the effect of privatizing the sale of one type of alcoholic beverage on the sale of other nonprivatized beverages. In the seven settings assessed, the sales of nonprivatized alcoholic beverages decreased a median of 2.2%, with an interquartile interval ranging from a decrease of 6.6% in sales to a decrease of 0.1% (Figure 3). These decreases are not of suffcient magnitude to offset the overall increase in per capita sales of privatized beverages. Effects of re-monopolization on alcohol-related outcomes. One study in Sweden33 directly assessed effects of a 1977 re-monopolization of the sale of MSB (2.26% 3.50% alcohol by volume; beer in the U.S. is generally 4% 6% alcohol by volume). The study was of moderate design suitability and fair execution. Re-monopolization resulted in a substantial decline in the number of outlets for MSB, from 11,550 to 300. The effects of this policy change on hospitalization for several alcohol-related outcomes (alcoholism, alcohol intoxication, alcoholic psychosis, hospitalizations for acute alcohol intoxication, suicides, falls, motor vehicle crashes, and assaults) were assessed using time-series design, comparing the 4 years before and after re-monopolization. The results were stratifed by four age categories (10 19 years, 20 39 years, 40 59 years, and 60 years), and the study did not provide data to allow aggregation across age groups. The researchers identifed a number of positive changes in health outcomes following re-monopolization. Hospital admissions for the treatment of alcoholism, alcohol intoxication, and alcohol psychosis decreased across all age groups (p0.05), and there was a 20% decline (p0.05) in these outcomes among people aged 10 19 years. Hospitalizations for acute alcohol intoxication decreased across all ages from 3.5% to 14.7% (p0.05). Suicides decreased from 1.7% to 11.8% (p0.05). Falls decreased from 3.6% to 4.9% (p0.05). Motor vehicle crashes decreased 14% (p0.05) for three age categories (10 19 years, 40 59 years, and 60 years) and by 4.4% for those aged 20 39 years (p0.05). In contrast, assaults increased from 6.9% to 14.8% (p0.05)

424

Hahn et al / Am J Prev Med 2012;42(4):418 427

Place (year)

in three of four age categories, and decreased by 1.4% among those aged 20 39 years (p0.05). In summary, the remonopolization of MSB in Sweden was associated with reductions in most of the alcohol-related harms assessed across all age groups; however, many of these effects were not signifcant.

Undesirable direction
West Virginia (1981)

Iowa (1985) New Hampshire (1978) Montana (1979) Maine (1971) Alabama (1980)
Finland (1969) Finland (1969) West Virginia (1981) Iowa (1985)

Spirits

Light Beer

New Hampshire (1978)

Montana (1979) Maine (1971) Alabama (1980)
-60 -40 -20 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340

Beer

Cross-sectional studies, Median -2.2%; Interquartile Interval (-6.6, -0.1) Median panel studies, and summarized studies in transInterquartile lation. The team found interval 20 cross-sectional and Figure 3. Percentage change in consumption of nonprivatized alcoholic beverages panel studies29,35,3754 asattributable to privatization of other alcoholic beverages sessing the association of Note: Bars on the point estimate represent 95% CIs when reported or calculated. privatization with alcohol consumption and related harms; fve studies35,38,43,47,48 had multiple outcomes. The researchers found that when beer sales were (Panel studies were included in this section if they did not privatized, there were increases in alcohol consumpinclude time as an ordinal covariate or compare alcoholtion and alcohol-related harms such as arrests for related outcomes before and after events of privatizadrunkenness and alcohol-related illnesses. They also tion.) Seventeen studies29,35,37,38,40,42 48,50 54 found that found that re-monopolization generally resulted in deprivatization was associated with greater consumption creased population-level consumption of the affected alcoholic beverage, decreases in excessive alcohol con(nine were signifcant,37,40,42 44,46,48,52,53 seven were sumption, and decreases in alcohol-related harms. In not,29,35,38,45,47,50,54 and one51 did not report signifaddition, they found that when a particular strength of cance). Four studies found decreased consumption (three 40,43,44 38 37,41,50 beer became more or less accessible, consumers tended to were signifcant, one not ). Three studies purchase the beverage type that was more readily accesassessed the association of privatization with cirrhosis sible. This beverage substitution effect appeared to be mortality; all were positive and two37,41 were signifstronger among different strengths of beer than among cant. Finally, two studies assessed the association of 39,46 different types of alcoholic beverages. privatization with motor vehicle fatalities, and both found a positive, but nonsignifcant association; one study found a negative, signifcant association of Summary of Intervention Effectiveness privatization and drunk driving. Overall, this evidence is consistent with evidence from primary studies indiAcross the studies qualifying for this review, the privaticating positive associations between privatization and zation of off-premises retail sales of an alcoholic beverage increased population-level consumption and between was associated with a median 44.4% increase (interquarprivatization and alcohol-related harms. tile interval 4.5%, 122.5%) in the per capita sales of the Makela, Rossow, and Tryggvesson36 published (in beverage and with a 2.2% decline (interquartile interval English) a review of studies conducted in Finland, Swe6.6%, 0.1%) in the per capita sales of beverages for den, and Norway that were not published in English which sales were not privatized. One study (in Finland)25 translation and thus were not included in this review. found that the increases in consumption occurred among These studies examined the effect of either privatizing or drinkers at all consumption levels. A single study33 evalre-monopolization of the sales of medium or strong beer uating the effects of re-monopolization of alcohol sales between the mid-1960s and the early 1990s on various found that this change was associated with a subsequent kinds of alcohol-related harms (e.g., arrests for drunkendecrease in several alcohol-related harms (e.g., hospitalizations for acute alcohol intoxication). ness and alcohol-related illnesses).
www.ajpmonline.org

Relative percent change in consumption

Hahn et al / Am J Prev Med 2012;42(4):418 427

425

Other Harms and Benets

Government control over retail alcohol sales generally results in lower alcohol outlet density. In addition to potential public health benefts, lower outlet density may improve quality of life by reducing property damage and public disturbance (e.g., public intoxication). However, the studies reviewed did not assess these effects. The review team did not postulate any serious harms associated with the maintenance of government control over retail alcohol sales.

Applicability
Consistent evidence of the association between privatization and increased per capita alcohol sales comes from studies done in multiple settings in the U.S., Canada, and Europe. Most of these studies evaluated the effects of privatizing the sales of wine and spirits. Only one Swedish study33 specifcally assessed the impact of re-monopolization (of MSB). The fndings from the current review apply most directly to the impact of privatizing the sale of wine and spirits in high-income nations such as the U.S.

Economic Efciency
The present systematic economic review identifed one study55 in Canada that used simulation modeling to estimate healthcare and law enforcement costs and costs of lost productivity due to disability and premature mortality in the event all Canadian provinces and territories were to privatize alcohol sales. The study was judged to be satisfactory by Community Guide economics criteria (www.thecommunityguide.org/about/EconAbstraction_ v5.pdf). Study authors concluded that these costs were substantially greater than the tax and mark-up revenue gained from increased sales associated with privatization; however, beneft data were not documented.

such events to occur in the U.S. or other high-income nations. Privatization has assumed different forms in different states and localities. Thus, it would be useful to determine how the effects of privatization observed in this review vary by the degree of government regulation and other specifc parameters of the privatization. Although, in general, government control establishes a greater degree of regulation over retail alcohol sales than systems in which sales have been privatized, Her et al.13 have noted that privatization might involve a change from a very restrictive alcohol management system to a loosely regulated private one; it would also potentially involve a change from a commercially orientated public system to a private sector operation that is heavily regulated. No peer-reviewed studies were found that evaluated economic effects of privatizing the sale of alcoholic beverages in the U.S. The anticipated effects of privatization include a large, but short-term, source of revenue to states; a potential increase in healthcare and criminal justice costs; and productivity losses from expected increases in excessive alcohol consumption owing to greater availability and/or lower prices. Studies assessing these economic impacts would be useful for informing future discussions of this issue. It would be useful to assess the effects of different specifc approaches to privatization on state revenues associated with sales and taxes on alcoholic beverages.

Conclusion
The evidence consistently showed that privatization of retail alcohol sales was associated with a substantial increase in per capita sales of the privatized beverages, a well-established proxy for excessive alcohol consumption. There was also evidence that re-monopolization is associated with a decrease in alcohol-related harms. Therefore, according to Community Guide rules of evidence, there was strong evidence that retail privatization of alcohol sales leads to increases in excessive alcohol consumption. In the U.S., many states have privatized the retail sales of alcoholic beverages. Currently, three states control the off-premises consumption retail sales of both wine and spirits, and an additional ten states maintain control over the retail sale of spirits alone. In addition, one county in the state of Maryland has county-level control over the retail sale of spirits and wine. The fndings of the present report are based solely on evidence related to the public health consequences of privatization, which may be one of several factors considered in making decisions on whether to privatize retail alcohol sales. The maintenance of government control of off-premises sale of alcoholic

Research Gaps
Although the studies reviewed have demonstrated an association between privatization and increases in the per capita consumption of the privatized beverages without substantial reductions in consumption of other alcoholic beverages, additional research is needed to clarify the relationship between privatization and various patterns of excessive alcohol consumption (e.g., binge drinking) as well as harms related to it. Most useful would be cohort studies in the U.S. similar to the one conducted by Makela et al.25 in Finland, assessing the effects of privatization on patterns of excessive alcohol consumption (e.g., binge drinking) and alcohol-related harms. It would also be useful to evaluate the impact of increased government control over alcohol sales (e.g., re-monopolization) on excessive alcohol consumption and related harms, were
April 2012

426

Hahn et al / Am J Prev Med 2012;42(4):418 427

11. Zaza S, Wright-De Aguero LK, Briss PA, et al. Data collection instrument and procedure for systematic reviews in the Guide to Community Preventive Services. Am J Prev Med 2000;18(1S):44 74. 12. Campbell AC, Hahn RA, Elder RA, et al.; Task Force on Community Preventive Services. The effectiveness of limiting alcohol outlet density as a means of reducing excessive alcohol consumption and alcoholrelated harms. Am J Prev Med 2009;37(6):556 69. 13. Her M, Giesbrecht N, Room R, Rehm J. Privatizing alcohol sales and alcohol consumption: evidence and implications. Addiction 1999; 94(8):112539. 14. Fitzgerald JL, Mulford HA. Consequences of increasing alcohol availability: the Iowa experience revisited. Br J Addict 1992;87(2):26774. 15. Rossow I, Karlsson K, Raitasalo K. Old enough for a beer? Compliance with minimum legal age for alcohol purchases in monopoly and other off-premise outlets in Finland and Norway. Addiction 2008;103:1468 73. 16. World Bank. World Development Indicators 2006. devdata. worldbank.org/wdi2006/contents/cover.htm. 17. Gruenewald PJ, Millar AB, Treno AJ, Yang Z, Ponicki WR, Roeper P. The geography of availability and driving after drinking. Addiction 1996;91(7):967 83. 18. Nelson D, Naimi TS, Brewer RD, Roeber J. U.S. State alcohol sales compared to survey data, 19932006. Addiction 2010;105:1589 96. 19. Cook PJ, Skog OJ. Alcool, alcoolisme, alcoolisation comment. Alcohol Health Res World 1995;19(1):30 1. 20. Greenfeld TK, Rogers JD. Who drinks most of the alcohol in the U.S.? The policy implications. J Stud Alcohol 1999;60(1):78 89. 21. Fitzgerald JL, Mulford HA. Privatization, price and cross-border liquor purchases. J Stud Alcohol 1993;54(4):462 4. 22. Fitzgerald JL, Mulford HA. Alcohol availability, drinking contexts and drinking problems: the Iowa experience. J Stud Alcohol 1993;54(3):3205. 23. Holder HD, Wagenaar AC. Effects of the elimination of a state monopoly on distilled spirits retail sales: a time-series analysis of Iowa. Br J Addict 1990;85(12):161525. 24. Macdonald S. The impact of increased availability of wine in grocery stores on consumption: four case histories. Br J Addict 1986;81:3817. 25. Makela P. Whose drinking does the liberalization of alcohol policy increase? Change in alcohol consumption by the initial level in the Finnish panel survey in 1968 and 1969. Addiction 2002;97(6):701 6. 26. Mulford HA, Fitzgerald JL. Consequences of increasing off-premise wine outlets in Iowa. Br J Addict 1988;83(11):12719. 27. Mulford HA, Ledolter J, Fitzgerald JL. Alcohol availability and consumption: Iowa sales data revisited. J Stud Alcohol 1992;53(5):48794. 28. Smart RG. The impact on consumption of selling wine in grocery stores. Alcohol Alcohol 1986;21(3):233 6. 29. Trolldal B. An investigation of the effect of privatization of retail sales of alcohol on consumption and traffc accidents in Alberta, Canada. Addiction 2005;100(5):66271. 30. Trolldal B. The privatization of wine sales in Quebec in 1978 and 1983 to 1984. Alcohol Clin Exp Res 2005;29(3):410 6. 31. Wagenaar AC, Holder HD. A change from public to private sale of wine: results from natural experiments in Iowa and West Virginia. J Stud Alcohol 1991;52(2):16273. 32. Wagenaar AC, Holder HD. Changes in alcohol consumption resulting from the elimination of retail wine monopolies: results from fve U.S. states. J Stud Alcohol 1995;56(5):566 72. 33. Ramstedt M. The repeal of medium-strength beer in grocery stores in Sweden: the impact on alcohol-related hospitalizations in different age groups. In: Room R, ed. The effects of Nordic alcohol policies: what happens to drinking and harm when alcohol controls change? Finland: Nordic Council for Alcohol and Drug Research (NAD), 2002:69 78. 34. Mustonen H, Sund R. Changes in the characteristics of drinking occasions resulting from liberalization of alcohol availability: A reanalysis of the 1968 and 1969 Finnish panel survey data. In: Room R, ed. The effects of Nordic alcohol policies: what happens to drinking and harm when alcohol controls change? Nordic Council for Alcohol and Drug Research, 2002:49 55.

beverages is one of many effective strategies to prevent or reduce excessive consumption, which is one of the leading causes of preventable death and disability.
We would like to thank Steve Wing (SAMHSA) and Ralph Hingson, ScD, MPH, National Institute of Alcohol Abuse and Alcoholism, for participation in this review. We also thank our team of consultants for their comments: Edward Bernstein, MD, Boston University School of Public Health; Richard J. Bonnie, LLB, University of Virginia; Raul Caetano, MD, MPH, PhD, University of Texas at Dallas; Frank Chaloupka, PhD, University of Illinois at Chicago; Norm Giesbrecht, PhD, Centre for Addiction and Mental Health, Canada; Thomas Greenfeld, PhD, National Alcohol Research Center/Alcohol Research Group, California; Joel Grube, PhD, and Harold Holder, PhD, Pacifc Institute for Research and Evaluation (PIRE), California; Mel Kohn, MD, Oregon Department of Human Services; Jurgen Rehm, PhD, Centre for Addiction and Mental Health, Canada; Robin Room, PhD, Stockholm University; Henry Saffer, PhD, National Bureau of Economic Research; Leslie Snyder, PhD, University of Connecticut; Tim Stockwell, PhD, University of Victoria; Alex Wagenaar, PhD, University of Florida College of Medicine. No fnancial disclosures were reported by the authors of this paper.

References
1. CDC. Alcohol-attributable deaths and years of potential life lostU.S., 2001. MMWR Morb Mortal Wkly Rep 2004;53(37):866 70. 2. Miller JW, Naimi TS, Brewer RD, Jones SE. Binge drinking and associated health risk behaviors among high school students. Pediatrics 2007;119(1):76 85. 3. National Center for Health Statistics. Health, U.S., 2010, with special feature on death and dying. www.cdc.gov/nchs/data/hus/hus10.pdf. 4. Johnston LD, OMalley PM, Bachman JG, Schulenberg JE. Monitoring the future: National survey results on drug use, 19752008. Volume II, College students and adults ages 19 50. Bethesda MD: National Institute on Drug Abuse, 11-1-2009. NIH Publication No. 09-7403. 5. Bouchery EE, Harwood HJ, Sacks JJ, Simon CJ, Brewer RD. Economic costs of excessive alcohol consumption in the U.S., 2006. Am J Prev Med 2011;41(5):516 24. 6. Aaron P, Musto D. Temperance and prohibility in America: an historical overview. In: Moore MH, Gerstein DR, editors. Alcohol and public policy: beyond the shadow of prohibition. Washington DC: National Academy Press, 1981:127 81. 7. Holder HD, Giesbrecht N, Horverak O, et al. Potential consequences from possible changes to Nordic retail alcohol monopolies resulting from European Union membership. Addiction 1995;90(12):160318. 8. Babor TF. Alcohol: no ordinary commodity. Research and public policy. 2nd ed. New York: Oxford University Press, 2010. 9. DHHS. Healthy People 2020. www.healthypeople.gov/2020/default. aspx. 10. Briss PA, Zaza S, Pappaioanou M, et al. Developing an evidence-based Guide to Community Preventive Servicemethods. Am J Prev Med 2000;18(1S):35 43.

www.ajpmonline.org

Hahn et al / Am J Prev Med 2012;42(4):418 427

35. Simon JL. The economic effects of state monopoly of packaged-liquor retailing. J Polit Econ 1966;74:188 94. 36. Makela P, Rossow I, Tryggvesson K. Who drinks more and less when policies change? The evidence from 50 years of Nordic studies. In: Room R, editor. The effects of Nordic alcohol policies: what happens to drinking and harm when alcohol controls change? Finland: Nordic Council for Alcohol and Drug Research, 2002:4255. 37. Andreano RL, Li JY. Death rates from liver cirrhosis in alcohol monopoly and free-market states. Q J Stud Alcohol 1974;35(Pt A):649 54. 38. Beard TR, Gant PA, Saba RP. Border-crossing sales, tax avoidance, and state tax policies: an application to alcohol. South Econ J 1997; 64(1):293306. 39. Colon I. The influence of state monopoly of alcohol distribution and the frequency of package stores on single motor vehicle fatalities. Am J Drug Alcohol Abuse 1982;9(3):32531. 40. Goel RK, Morey MJ. The interdependence of cigarette and liquor demand. South Econ J 1995;62(2):4519. 41. Heien D, Pompelli G. Stress, ethnic and distribution factors in a dichotomous response model of alcohol abuse. J Stud Alcohol 1987; 48(5):450 5. 42. Hoadley JF, Fuchs BC, Holder HD. The effect of alcohol beverage restrictions on consumption: a 25-year longitudinal analysis. Am J Drug Alcohol Abuse 1984;10(3):375 401. 43. Kenkel DS. Drinking, driving, and deterrence: The effectiveness and social costs of alternative policies. J Law Econ 1993;36(2):877913. 44. Kenkel DS. New estimates of the optimal tax on alcohol. Economic Inquiry 1996;34(2):296 319. 45. McCornac DC, Filante RW. The demand for distilled spirits: an empirical investigation. J Stud Alcohol 1984;45(2):176 8.

427

46. Miller T, Snowden C, Birckmayer J, Hendrie D. Retail alcohol monopolies, underage drinking, and youth impaired driving deaths. Accid Anal Prev 2006;38(6):11627. 47. Nelson JP. State monopolies and alcoholic beverage consumption. J Regulatory Econ 1990;2(1):8398. 48. Nelson JP. Advertising bans, monopoly, and alcohol demand: testing for substitution effects using state panel data. Rev Industrial Org 2003;22(1):125. 49. Ornstein S, Hanssens D. Alcohol control laws and the consumption of distilled spirits and beer. J Consum Res 1985;12(2):200 13. 50. Popham RE, Schmidt W, de Lint J. Government control measures to prevent hazardous drinking. In: Ewing JA, Rouse BA, editors. Drinking: alcohol in American society: issues and current research. Chicago: Nelson-Hall, 1978:239 66. 51. Spellman WE, Jorgenson MR. Liquor control and consumption. J Stud Alcohol 1983;44(1):194 7. 52. Stockwell T, Zhang J, Macdonald S, Pakula B, Gruenewald P, Holder H. Changes in per capita alcohol sales during the partial privatization of British Columbias retail alcohol monopoly 20032008: a multi-level local area analysis. Addiction 2009;104:182736. 53. Trolldal B. Alcohol price elasticities in control and license states in the U.S., 1982-99. Addiction 2005;100(8):1158 65. 54. Zardkoohi A, Sheer A. Public versus private liquor retailing: an investigation into the behavior of the state governments. South Econ J 1984;50(4):1058 76. 55. Popova S, Patra J, Sarnocinska-Hart A, Gnam WH, Giesbrecht N, Rehm J. Cost of privatization versus government alcohol retailing systems: Canadian example. Drug Alcohol Rev 2011;Epub ahead of print.

April 2012