economic, empirical research article

RESEARCH AND PRACTICE

Effects of Alcohol Tax Increases on Alcohol-RelatedDisease Mortality in Alaska: lime-Series AnalysesFrom 1976 to 2004I Alexander C. Wagenaar, PhD, Mildred M. Maldonado-Molina, PhD, and Bradley H. Wagenaar, BA

Morbidity and mortality associated with con-

sumption of alcoholic beverages constitute asubstantial public health burden in the UnitedStates. An estimated 85 000 deaths per yearare associated with drinking, including carcrashes, other unintentional injuries, homicide,suicide, and a range of diseases, particularly

those affecting the liver, pancreas, and heart.1-3

All states have taxes specific to alcoholic bever-ages; some of these taxes are imposed primarily

to generate revenue,4 whereas others are osten-sibly intended to promote public health andwelfare by limiting alcohol consumption.5'r

More than 100 studies have examined therelationship between alcoholic beverage prices(or alcohol tax rates as a surrogate for prices)and various indices of sales or consumption ofalcohol (see Babor et al.7 and Chaloupka et al.8

for recent reviews). With a few exceptions (e.g.,Salomaag), studies consistently find price or taxlevels to be inversely related to sales or con-sumption of alcoholic beverages, with the mag-nitude of the effect in terms of elasticities rangingfrom -0.2 to -2.0, depending on population,methods, time period, and specific beverage.

Substantially fewer studies have examinedthe effects of alcohol prices or taxes on mea-sures of alcohol-related morbidity or mortality,and of those that have, most have focused oninjury rather than disease. Of 18 studies thatexamined price or tax effects on traffic crashes,15 found that higher alcohol prices or taxeswere associated with fewer crashes,"'-24 but 3found no effect25 27 Sloan et al.2 5 found thatalcohol prices were not related to falls, fires, orother unintentional injury rates, but OhstWldt and

Morrisey 28 found that higher beer taxes were

related to lower rates of nonfatal industrialinjuries. In addition, 8 studies (6 of which were

conducted by 1 research team) found higheralcohol prices or taxes associated with lowermorbidity or mortality from intentional injuries,including assault, homicide, suicide, child abuse,and spouse abuse.2 9 3' However, I of these

Objective. We evaluated the effects of tax increases on alcoholic beverages in

1983 and 2002 on alcohol-related disease mortality in Alaska.

Methods. We used a quasi-experimental design with quarterly measures of

mortality from 1976 though 2004, and we included other states for comparison.

Our statistical approach combined an autoregressive integrated moving averagemodel with structural parameters in interrupted time-series models.

Results. We observed statistically significant reductions in the numbers and

rates of deaths caused by alcohol-related disease beginning immediately after

the 1983 and 2002 alcohol tax increases in Alaska. In terms of effect size, the

reductions were -29% (Cohen's d=-0.57) and -11% (Cohen's d=-0.52) for the 2

tax increases. Statistical tests of temporary-effect models versus long-term-

effect models showed little dissipation of the effect over time.

Conclusions. Increases in alcohol excise tax rates were associated withimmediate and sustained reductions in alcohol-related disease mortality in

Alaska. Reductions in mortality occurred after 2 tax increases almost 20 years

apart. Taxing alcoholic beverages is an effective public health strategy for

reducing the burden of alcohol-related disease. (Am J Public Health. 2009;99:1464-1470. doi:10.2105/AJPH.2007.131326)

studies found that higher beer taxes were not31specifically related to levels of robbery or rape.

Finally, a number of studies have found higher

alcohol prices or taxes to be associated withlower rates of alcohol dependence37' 38 and liverdcrrhosis. 23'39

-42 In contrast, Schweitzer et al.43

did not find a significant relationship betweenalcohol prices and rates of alcohol dependence.

To elicit more information about the effectof alcoholic beverage taxes on disease mortality,we examined patterns of alcohol-related diseasemortality in the state of Alaska over a 29-year

period to determine whether 2 major increasesin alcohol tax rates, 1 in 1983 and the otherin 2002, affected alcohol-related mortality in

the state.

METHODS

Research DesignThe research design of the study can be

represented as shown in Figure 1.We utilized a time-series quasi-experimental

design, which is similar to an experiment in thatit is intended to identify the effect of an

intervention while using comparison data se-ties to rule out or control for possible alterna-

tive explanations for the effect. 45 A good ex-perimental design has a number of benefits: itcan allow researchers to eliminate many con-founding factors as a threat to a causal interpre-tation of an observed relationship without theneed to identify, measure, and statistically controlfor all possible confounds; it can avert debates

about which of the numerous specific controlvariables to include; and it can prevent re-searchers from introducing biases into the studyby including only those control variables forwhich operational measures are available.

Given that Alaska is unusual compared withother states in terms of weather, population,economy, health, and many other factors, noother single state is optimal for comparisonwith Alaska. Therefore, we elected to use theaggregate of all other states as the comparisongroup. This comparison ensures that effectsobserved in Alaska are not caused by anychanges in substantive factors (e.g., economy,society, policy) or measurement factors (e.g.,diagnosis coding categories) that could

1464 l Research and Practice I Peer Reviewed I Wagenaar et al. American Journal of Public Health I August 2009, Vol 99, No. 8


Experimental 01 0, 0 X 0 ] 0,0(Alaska) -, 0" 1 . .... ... "

Comparison 0,, 0,: 0,1 ... ... ......... ... ... .. .0(other states)

Note. Ot represents an observation at a given time t, with each t being one quarter of a calendar year. The first observation took place at time tj, the first quarter of 1976, and the seriesended at t116, the last quarter of 2004. X, represents an increase in alcohol tax in August 1983, when the excise tax on beer increased from $0.25 to $0.35 per gallon, the tax on wine increasedfrom $0.25 to $0.35 per gallon, and the tax on distilled spirits increased from $4.00 to $5.50 per gallon. 44

X2 represents a subsequent increase in alcohol tax in October 2002, when thetax on beer increased to $1.07 per gallon. the tax on wine increased to $2.50 per gallon, and the tax on spirits increased to $12.80 per gallon.44 m represents the number of quarterly observationsbefore the first tax increase (m =30). n represents the number of quarterly observations after the first tax increase and before the second tax increase (n =77). These 2 sets of observationsare summed with 9 follow-up quarterly observations after the second tax increase, producing a total of 116 observations.

FIGURE 1-Representation of study design.

affect observed alcohol-related mortality

over time.

We further strengthened the study's designby examining 2 separate increases in tax ratesoccurring almost 20 years apart. This element

of the study allowed us to test the replicabilityof observed effects, and it ensured that ob-

served effects were not caused by uniquecircumstances not generalizable overtime, thusruling out the "contemporaneous history"threat to internal validity first articulated byCampbell and Stanley.41' Moreover, a simulta-neous evaluation of the effects of the 2 taxincreases allowed us to perform a long-ternfollow-up evaluation of the earlier tax increase.

Data Collection

Data on alcohol tax rates in Alaska werecollected from a number of sources. The Al-cohol Policy Information System providessummaries of US state and federal tax rates formultiple classes of alcoholic beverages, detailson changes in rates, full legal citations, and textfrom the relevant codified statutes, startingwith calendar year 2003 .47-'9 Data for the

period prior to 2003 were collected by experi-enced research attorneys who used standardlegal research methods to search the records ofcodified statutes in Westlaw databases (http://www.westlaw.com, fee-based membership) when

available, and who searched law library hard-copy materials for earlier years.

Data on alcohol-related mortality outcomeswere based on death-ceitificate data recorded

by the National Vital Statistics System of theNational Center for Health Statistics. whichincludes all deaths occurTing Mithmn the United

States. We obtained the complete annual data

set containing 1 record on each deceased per-

son for each year from 1976 through 2004

from the National Bureau of Economic Re-

search.5" From these data we created quarterly

coimts of deaths stratified by underlying cause of

death, first for Alaska and then for the otherstates as a group.

We cumulated counts by cause of death into

3 outcome variables. (1) Alcohol-caused mor-

tality, which represents all deaths caused bydiseases for which the alcohol-attributable

fraction is 1.0 (e.g., alcoholic liver disease,

alcohol-induced chronic pancreatitis, alcoholpsychoses, alcohol abuse, alcohol dependence

syndrome, alcoholic polyneuropathy, alcoholic

cardiomyopathy, alcoholic gastritis, and acute

alcohol poisoning). (2) Alcohol-related mortal-ity, which represents all deaths caused bydiseases with alcohol-attributable fractions

0.35 and higher but less than 1.0 (i.e., othercirrhosis; cholelithiasis; acute and chronic

pancreatitis; malignant neoplasms of themouth, pharynx, esophagus, liver, and breast;

epilepsy; and cardiovascular diseases including

hypertension, ischemia, arrhythrnia, cerebro-

vascular disease, and ischemic and hemor-rhagic stroke). (3) The third variable is the sum

of the first 2 variables. Alcohol-attributable

fractions for diseases, categorized by Interna-tional Classification of Diseases (ICD)5" code, arebased on RehIm et al.,52 English et al,53 and

Shultz et al.,54 which in turn were based oncomprehensive reviews and meta-analyses of the

literature (for the ICD codes used, see the tableavailable as a supplement to the online version of

the article at http://Nw,ww.ajph.oi-g). Finally, in

addition to analyses of mortality frequency

counts, we also used annual estimates of popu-

lation from the US Census Bureau to calculate

death rates per 100 000 population aged 15years and older. ' 5-5

Statistical Analyses

Given the large number of repeated obser-

vations, we used an approach that combined aBox-Jenkins autoregressive integrated moving

average (ARIMA) model with structural pa-rameters, 59,6

(1) (1 -B 4 ) yt =o+o(111, +o,)212t +fiZ,

+i/jiX+(f - OB4)u1,

where Y,=I to Yj - are the 3 outcome measures

by quarter from t= (first quarter 1976) through

t=116 (last quarter 2004); a is a constant; ol is

the estimated effect of implementation of the1983 alcohol tax increase; Ilu is a step functionequal to 0 before the 1983 tax change took effectand 1 after the change; no2 is the estimated effectof implementation of the 2002 alcohol taxincrease; 12t is a step function equal to 0 before

the 2002 tax change took effect and 1 after thechange; [3 is the estimated effect of Z, the

frequency (or rate) of alcohol-related disease inthe comparison states; 4, is a vector of estinates

that controls for outliers X,; E) is the first-order

seasonal moving average parameter; ut is a

random (white noise) error component; and B is

the backshift operator such that B4 (yd equals

YI-4. We used SAS version 9.1 Proc ARIMA

(SAS Institute Inc, Cary, NC) to estimate allmodels, and we used a strict significance crite-nion of Pat less than .001 to evaluate all models

August 2009, Vol 99, No. 8 1 American Journal of Public Health Wagenaar et aLi. Peer Reviewed I Research and Practice I 1465


for effects of outliers. A maximum of 1 outlierwas detected and controlled in any model.

For each Alaska outcome measure (alcohol-

caused mortality, alcohol-related mortality, and

the sum of both), 4 specific models were esti-

mated: (1) frequency, (2) rate per 100000

population aged 15 years and older, (3) rate per

100000 population including the comparison

states covariate, and (4) natural logs of rate per

100000 population including the comparison

states covariate. Frequencies were examined

for initial evidence of effects. Rates per popu-

lation were examined to assess whether ob-

served differences were caused by population

changes. Then the comparison states covariate

was included to assess whether observed ef-

fects in Alaska may have been caused by other

factors changing over time across states. The

natural logs were modeled to ensure estimates

were not affected by heteroscedasticity

(changing variance in death rates over time).

Finally, to statistically test whether the effects

were permanent or temporary, we estimated an

alternative specification of the time-series

model that used a first-order transfer function

on the differenced tax-change variable:

(2) (co/i-6B)(1-B)I.

where (o is the estimated shift effect of It,

representing the 1983 alcohol tax increase; 6 is

the estimated rate of decay of the initial effect;

and B is the backshift operator, such that B(I)

equals I,-,. If the estimated value for 6 in such a

model is very close to unity, the effect is long-

term and does not dissipate. If the estimate of 6

is less than 1, the immediate effect decays over

time. See chapter 3 of McCleary and Hay 61 for a

detailed description of transfer-function model-

ing that is accessible to nonstatisticians.

All of the models fit the data well and

explained substantial proportions of the vari-

ance in deaths over time. Models of frequencies

had R2 of 0.68 to 0.72 (after adjustment for

number of degrees of freedom used in the

model). Given that rates or ratios naturally

have substantially higher measurement error

do than the numerator frequencies alone, ad-

justed R2 for those models was somewhat

lower, ranging from 0.13 to 0.33. Finally, pa-

rameter estimates in terms of change in fre-

quency or population rate of death were

transformed into 2 standardized metrics of

effect: percentage change in the outcome based

TABLE 1-Effects of Alcohol-Tax Increases on Alcohol-Related Disease Mortality: Alaska,

1983 and 2002

Estimate PercentageR2 (SE) t P Change Cohen's d

All alcohol-related and alcohol-caused mortality (AAF_>0O35)

Effects of 1983 alcohol-tax increase

Frequency 0.77 -5.65 (1.73) -3.27

Rate per 100000 population 0.33 -1.37 (0.50) -2.77

Rate per 100000 population with other states 0.33 -1.19 (0.52) -2.28

covariate

Log rate per 100000 population with other states 0.26 -0.19 (0.08) -2.23

covariate


Frequency 0.77 -5.15 (2.11) -2.44

Rate per 100000 population 0.33 -1.23 (0.57) -2.17

Rate per 100000 population with other states 0.33 -1.36 (0.57) -2.39

covariate

Log rate per 100000 population with other states 0.26 -0.19 (0.09) -2.08

covariate

Subset analysis: alcohol-caused mortality (AAF= 1.0)


Frequency 0.68

Rate per 100000 population 0.30

Rate per 100000 population with other states 0.31

covariate

Log rate per 100000 population with other states 0.22

covariate


Frequency 0.68

Rate per 100000 population 0.30

Rate per 100000 population with other states 0.31

covariate

Log rate per 100000 population with other 0,22

states covariate

-2.93 (1.48)-0.77 (0.38)

-0.74 (0.38)

.00

.01

.02

.03

.02

.03

.02

.04

-1.98 .05-2.05 .04

-1.96 .05

-0.13 (0.10) -1.33 .19

-2.19 (1.75)

-0.59 (0.44)

-0.55 (0.44)

-1.25 .21

-1.35 .18

-1.25 .21

-0.11 (0.11) -1.00 .32

-28.6

-22.5

-19.6

-0.57

-0.88

-0.77

-17.0 -0.79

-11.3-13.0

-14.5

-0.52

-0,79

-0.88

-17.1 -0.80

-24.4

-20.8

-19,9

-12.1

-7.0

-9.1

-8.5

-0.38

-0.59

-0.56

-0.40

-0.29

-0.45

-0.42

-10.8 -0.36

Subset analysis: alcohol-related mortality (AAF > 0.35 < 1.0)


Frequency 0.31

Rate per 100000 population -0.15

Rate per 100000 population with other states -0.13

covariate

Log rate per 100 000 population with other states-0.04

covariate


Frequency 0.31

Rate per 100000 population -0.15

Rate per 100000 population with other states -0.13

covariate

Log rate per 100000 population with other states -0.04

covariate

-2.45 (1.17)

-0.44 (0.31)

-0.42 (0.36)

-2.13 .04

-1.41 .16

-1.19 .24

-0.27 (0,13) -2.21 .03

-1.81 (1.40)

0.01 (0.36)

-0.22 (0.37)

-1.29 .200.04 .97

-0.58 .57

-0.16 (0.13) -1.20 .23

-31.6-18.2

-17.7

-24.0

-12.9

0.4

-7,4

-0.654053

-0.52

-0.75

-0.48

0.02

-0.26

-14.9 -0.44

Note. AAF=alcohol-attributable fraction. Results derived from autoregressive integrated moving average models combinedwith structural parameters. /2 adjusted for degrees offreedom used in model; calculated negative e

2 in second subset analysis

results from increased variance heterogeneity.

1466 1 Research and Practice ý Peer Reviewed I Wagenaar et al. American Journal of Public Health I August 2009, Vol 99, No. 8


on the average of the 4 quarters immediatelyprior to a given tax rate change, and Cohen'sd,6 2 the effect size in standard deviation units,calculated with the raw standard deviation of theoutcome over the entire data series. Percentagechange is a commonly used, easily understoodmetric of interest to prevention practitioners andpolicymakers, and the Cohen's d effect size per-mits comparison of effect sizes across a widerange of intervention and outcome domains.

RESULTS

Results show statistically significant reduc-tions in the numbers of deaths caused byalcohol-related disease beginning immediatelyafter the 1983 and 2002 alcohol tax increasesin Alaska (Table 1). Estimated reductions inmortality are of clear substantive importance:the 1983 tax increase was followed by a -29%change in number of deaths (23 deaths avertedper year), and the 2002 tax increase was fol-lowed by a -11% change in deaths (an addi-tional 21 deaths averted per year). In terms ofCohen's d effect size, the reductions were-0.57 and -0.52 for the 2 tax increases, re-spectively, in chronological order. These effectsare large enough to be clearly discernible from

60

50

40

30

20

10

0

Ez

1983 tax increase

1976 1978 1980 1982

a plot of the data (Figure 2), especially forthe 1983 policy change, for which a 19-yearfollow-up period is available.

We analyzed the death rate to control forchanges in size of the population over time, butthis analysis did not appreciably change theestimated effects (Figure 3). Percentage reduc-tions in alcohol-related mortality declinedslightly to -23%/o for the 1983 tax increase andincreased slightly to -13%/o for the 2002 in-crease. Estimated Cohen's d effect sizes for thetax increases were larger for the rate per100 000 population measure than for theanalyses of raw numbers of deaths (-0.88 and-0.79). These results confirm that observedeffects of the tax policy changes are not attrib-utable to overall changes in population.

We obtained a third set of effect estimates byadding the comparison states to the model, todetermine whether observed alcohol-tax ef-fects in Alaska reflected mortality reductionscaused by any of a number of other possiblefactors operating across states. Again, there wasno appreciable change in the estimates. The1983 tax increase was associated with a -20%change in mortality (d=-0.77), and the 2002increase was associated with a -150/o change inmortality (d=-0.88). Thus, the comparison

2002 tax increase

1984 .19 86 1988 1990 19 92 1994 1996 1998 2000 2002 ' 2004

states did not experience the mortality declinesthat Alaska experienced, and the effects ob-served for Alaska cannot be attributed tobroader trends or other factors experiencedacross the other states. Repeating this modelon the natural logs of the death rate did notchange the findings. Finally, the estimates andstandard errors in Table 1 show that the mag-nitudes of the effects of the 2 tax changes werenot significantly different from each other.

The second and third panels of Table 1 pro-vide these same estimates of effect separatelyfor 2 subsets of the overall outcome measure:(1) alcohol-caused mortality, for which thealcohol-attributable fraction is 1.0, and (2)alcohol-related mortality, for which the alcohol-attributable fraction ranges from 0.35 to 0.99.In the second subset it is not known withcertainty whether any individual death wascaused by alcohol, but as the sample increases insize we are increasingly confident in the pro-portion of deaths that were caused by alcohol.We found no significant differences in themagnitudes of the estimated effects of the 2 taxchanges across these 2 outcome subsets (Table1). The sole obvious consequence of disaggre-gating the outcome is increased variability of theestimates (increased standard errors), an ex-pected result of the smaller and more variablemortality counts in the subset analyses.

An examination of the actual death countsby quarter (Figure 2) could be interpreted tomean that the obvious sudden effect of the1983 tax increase dissipated over time, be-cause after 5 years the average numbers ofdeaths per quarter returned to the levels theywere immediately before the policy change.Simple comparisons of the linear slopes of thepre-1983 and post-1983 periods show thatthey are similar (pre-1983 slope=0.21 vspost-1983 slope=0.32), but the slightly largerpost-1983 slope might be interpreted as adissipation of the effect seen in the suddenintercept difference in 1983. However, addi-tional analyses that directly tested alternativehypotheses assuming temporary versus per-manent effects demonstrated the effect didnot dissipate over time. When we used thetemporary-impact model specification, theestimate of (o remained essentially the same(-5.06, SE=1.83), but the estimate of thedecay parameter 8 was 0.98 (SE=0.03), verynear unity, demonstrating that the effect did

August 2009, Vol 99, No. 8 American Journal of Public Health

S•Actual --- Linear trend -- Model predicted]

Year

FIGURE 2-Frequency of quarterly alcohol-related disease mortality: Alaska, 1976-2004.

Wagenaar et at. I Peer Reviewed I Research and Practice 1 1467


TABLE 2-Alaska's Alcohol-Tax Rates per Gallon in Constant 2006 Dollars

1982 (Preincrease) 1983 (Postincrease)

Beer

Wine

Spirits

$0.46

$0.46

$7.28

$0.63

$0.63

$9.83

not decay over time.6' Moreover, the residualvariance of the temporary-effect model waslarger than that of the permanent-effect model,indicating poorer fit of the temporary-effectmodel.

When we repeated this procedure on thepopulation-rate outcome variable, we againfound no dissipation of effect over time. Similarstatistical tests of permanent versus temporaryeffects for the 2002 tax change are not yetpossible, because only 9 postchange datapoints are currently available.

DISCUSSION

We found that increases in alcoholic bever-age tax rates were associated with significantand substantial reductions of alcohol-relateddisease mortality in Alaska. Reductions inmortality were observed for 2 tax increases 19years apart, indicating that the observed effectscannot be attributed to a single historical pe-riod or event, atypical or otherwise. The long-term follow-up after the first tax increaseallowed us to determine that the effect was nottemporary, but was maintained over time. Ourquasi-experimental research design included

other states as a comparison group, demon-strating that the effects observed for mortalityin Alaska were not caused by broader nationaltrends, factors affecting mortality in commonacross states, or a sudden change experiencedacross states in 1983 or 2002.

Our results are consistent with basic eco-nomic theory and econometric analyses of theprice elasticity of alcoholic beverages, whichhold that an increased tax on beverage alcoholis presumed to raise the price to consumers,who respond by purchasing and drinking lessalcohol. Lower alcohol consumption is thenthought to reduce risk of death caused by arange of alcohol-related diseases, resulting in adecline in mortality counts and rates. Our studyvalidated this account by demonstrating a clearlink between the first and last factors in this

2001 (Preincrease) 2002 (Postincrease)

$0.40

$0.40

$6.28

$1.20

$2.80

$14.35

presumed mechanism of effect (alcohol-tax in-creases and alcohol-related mortality). Com-plete and accurate quarterly measures of the 2central intervening factors-retail prices anddrinking behaviors-are not available for the29-year period studied. However, the theoryunderlying this mechanism of effect is so wellestablished that a lack of measures for eachintervening effect does not reduce the plausi-bility of the findings.

A colleague and a reviewer both sought toexplain the fact that the effect of the 1983 taxincrease appears to be larger than the effect ofthe 2002 tax increase, as shown in the per-centage-change column in Table 1. Severalreasonable hypotheses quickly come to mind.For example, general price inflation means that

any given alcohol tax increase in dollars pergallon (as alcohol taxes often are measured,rather than as a percentage of the sale price)would likely be a larger proportion of the totalretail price of a product in 1983 than in 2002.However, close examination of the estimatesand standard errors in Table 1 shows no sig-nificant differences between estimates of theeffects of the 2 changes. Moreover, we only hadaccess to 2.25 years of follow-up data for the2002 tax change. Determination of whether the2 tax changes had significantly different long-

term magnitudes of effect, and the possible causesof such differences (if any), requires waiting foradditional follow-up data to become available.

Some questions might be raised aboutthe timing of observed effects of alcohol-tax increases. For instance, scientists andpractitioners who focus on individual-leveldisease may find the claim that immediatepopulation-level effects on mortality can becaused by modest changes in the social envi-ronment to be biologically implausible and"counterintuitive," as a reviewer of this article

said. A few of the specific causes of deathincluded in our outcome measures are theresult of the acute toxic effects of ethanolingestion (e.g., poisoning), but most are chronic

conditions that result from decades of highexposures to ethanol (e.g., cirrhosis or cancer).However, mortality caused by long-term,chronic alcohol use responds immediately to achange in drinking levels, because at any giventime there is a reservoir of individuals in thepopulation who are about to die from a chronicalcohol-related disease.63 Even modest reduc_

tions in current drinking immediately retardsprogression of alcohol-related disease for thispopulation, resulting in an immediate reductionin the death rate, as found in the present study.

In this regard, alcohol-related disease is notunlike some other diseases, such as chronicobstructive pulmonary disease (COPD). Chronic,long-term exposure to air pollution is one causeof COPD.6 4 Development of COPD resultingfrom exposure to urban air pollution levels usu-ally takes decades. 64 Nevertheless, sudden re-

ductions in ambient air pollution result in im-mediate reductions in COPD mortality. Popeet al. reviewed 18 "striking studies ... that

observed changes in daily death counts asso-clated with short-term changes in particulate airpollution.' 65(r475) This pattern of immediatepopulation-level changes in mortality occurs forboth COPD and alcohol-related disease, and forthe same reason: at any given moment in time,there are many in the population whose cumu-lative exposure (to pollution or alcohol) is nearthe threshold that will cause death, and a re-duction in current exposure delays death. Atrajectory or cascade of delayed individual

deaths results in population mortality declinesthat begin immediately when the exposure re-ductions occur.

Another question that could be raised aboutthe time-ordered pattern of observed effects isrelated to the continual gradual reduction inreal (inflation adjusted) alcohol taxes over time.Table 2 shows the Alaska tax rates immediatelybefore and after each tax change in constant2006 dollars (adjusted by the consumer priceindex for urban consumers in Anchorage 66).For example, the beer tax increased from $0.46

per gallon in 1982 to $0.63 in 1983 at the timeof the legislated increase, but then graduallydeclined to $0.40 by 2001 because of inflation,before increasing to $1.20 as a result of the 2002law. Given such a pattern, one might haveexpected any effect of the 1983 tax increase onmortality to have dissipated after a decade and ahalf of inflation. However, we found little

1468 I Research and Practice I Peer Reviewed I Wagenaar et al. American Journal of Public Health I August 2009, Vol 99, No. 8


I Acal --- Linear rend - Model pred-ite]

1983 tax increaw

1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 ' 1998 2000 ' 2002 ' 2004

Year

FIGURE 3-Rate of quarterly alcohol-related disease mortality per 100000 population aged15 years and older: Alaska, 1976-2004

evidence of such dissipation of effect Perhaps thesudden decline in drinking caused by increasedprices established new normative and behavioralpatterns that then were maintained in the face ofa gradual elimination of the original price increasethat stimulated the reduction in consumption.There is a need for additional study of the effectsof sudden tax or price increases on drinking-related normative expectations and behaviorpatterns. The reduction of the real tax rate inAlaska also raises the plausible hypothesis that thesubstantial effects observed here would be sig-nificantly larger if such tax increases were main-tWined in real dollars via indexing to inflation.

Our study limited its focus to an evaluation ofthe effects of changing alcohol taxes on alcohol-related disease; we did not study mortalitycaused by the many categories of unintentionalor intentional injury that have substantial frac-tions attributable to alcohol (e.g., traffic crashes,homicide, suicide). A replication of this researchfocused on injury morbidity and mortality iswarranted. In addition, we limited our scope tomortality changes in Alaska; replication of thestudy for other states is also warranted.

In conclusion, Alaska's alcohol tax increasesresulted in large (ranging from 11% to 29%)reductions in deaths caused by alcohol-relateddisease. The effect sizes are large compared

with other efforts to prevent negative outcomesrelated to alcohol consumption. Three meta-analyses of efforts to prevent individual- andschool-level alcohol problems show effect sizesranging from d at -0.02 to d at -0.3 ,669meaning that alcohol taxes had an effect 2 to 4times larger than did other common preventionefforts. The size of the alcohol-tax effect is evenmore noteworthy given that state tax policyaffects the entire population of a state, ratherthan the relatively small numbers of individualsaffected by most other prevention programs. Inaddition, state alcohol taxation systems are al-ready in place, so there are virtually no additionalimplementation costs associated with the largebenefit to public health to be obtained by in-creases in alcohol taxes. m

About the AuthorsAt the time of the study, Alexander C. Wagenaar, MildredM. Maldonado-Molina, and Bradley H. Wagenaar werewith the Department of Epidemiology and Health PolicyResearch, College of Medicine, University of Florida,Gainesville.

Requests for reprints should be sent to Alexander CWagenaar, College of Medicine, University of Florida,1329 SW 16th St, Room 5130, Box 100177Z Gaines-ville, FL 32610-0177 (e-mail: [email protected]),

This article was accepted Apnl 19, 2008.Note. Findings and conclusions are solely the authors'

and do not necessarily represent the views of the Universityof Horida.

12

10

August 2009, Vol 99, No. 8 I American Journal of Public Health

75

0

6

ContributorsA. C_ Wagenaar designed the study and led the writing ofthe article. M-M. Maldonado-Molina conducted the sta-tistical analyses. B.H. Wagenaar designed the outcomemeasures by alcohol-attributable fraction and collectedmortality data. All authors participated in editing andrevising the article.

AcknowledgmentsThis study was funded by Robert Wood Johnson Foun-dation (grant 058005).

The authors appreciate the assistance of Linan Mawith data management and Amy L. Tobler with prepa-ration of the article and are thankful for the helpfulcomments made by the reviewers and Frank Cbaloupka.

Note. Findings and conclusions are solely the authors'and do not necessarily represent the views of the RobertWood Johnson Foundation.

Human Participant ProtectionThe University of Florida institutional review boardreviewed this study and deemed that its study protocol didnot require approval because we had no contact withparticipants, data were based on public mortality files, andanalyses involved aggregate mortality counts by state.

References1. Midanik LT, Chaloupka FJ, Saitz R, et al. Alcohol-attributable deaths and years of potential life lost-UnitedStates, 2001. M111VR Morb Mortal Wly Rep.2004;53:866-870.

2. Mokdad AB, Marks JS, Stroup DF, Gerberding JL.Actual causes of death in the United States, 2000.JAMA.2004;291:1238-1245.

3. National Institute on Alcohol Abuse and Alcohol-ism. I Oth Special Report to the US Congress on Alcohol andHealth. Washington, DC: US Dept of Health and HumanServices; 2000.

4. Giertz JF, Excise taxes. Tax Policy Center Web site.Available at: lhttp://www.taxpolicycenter.org/taxtopics/encyclopedia/Excise-Taxes.cfm Accessed July 7, 2008.

5. Treating alcohol problems and state alcohol taxrevenues. Ensuring Solutions to Alcohol Problems Website. Available at: http://www.ensuringsolutions.org/resources/resources show.htm?doc id=339039.Accessed July 7, 2008.

6. Tremper C, Mosher JF. Assessing State Readiness toAct on Alcohol Tax Research Findings. Felton, CA: PacificInstitute for Research and Evaluation; 2004.

7. Babor T, Caetano R, Casswell S, et al. Alcohol: NoOrdinary Commodity. New York, NY: Oxford UniversityPress; 2003.

8. Chaloupka FJ, Grossman M, Saffer H- The effects ofprice on alcohol consumption and alcohol related prob-lems. Alcohol Res Health. 2002;26:22-34.

9. Salomaa J. Neither medium-strength beer nor priceexplains the discrepancy in alcohol consumption betweenFinland and Sweden. Nordisk Alkoholtidskrift. 1993;10:330-346.

10. Adrian M, Ferguson BS, Her M. Can alcohol pricepolicies be used to reduce drunk driving? Evidence fromCanada. Subst Use Misuse. 2001;36:1923-1957.

11. Chaloupka FJ, Laixuthai A. Do youths substitutealcohol and marijuana? Some econometric evidence.Eastern Econj 1997;23:253-276.

Wagenaar et al. I Peer Reviewed ý Research and Practice 1 1469


12. Chaloupka FJ, Saffer H, Grossman M. Alcohol-control policies and motor-vehide fatalities. J Legal Stud1993;22:161-186.

13. Dee TS. State alcohol policies, teen drinking andtraffic fatalities. J Public Econ. 1999;72:289-315.

14. Evans WN, Neville D, Graham JD. General deter-rence of drunk driving: evaluation of recent Americanpolicies. Risk Anal. 1991, 11:279-289.

15. Mast BD, Benson BL, Rasmussen DW. Beer taxationand alcohol-related traffic fatalities. Southern Econj1999;66:214-249.

16. Mullahy j, Sindelar JL. Do drinkers know when tosay when? An empirical analysis of drunk driving. EconInq. 1994;32:383-394.

17. Rubin CJ. Economic conditions and alcohol prob-lems. j Health Econ. 1995;14:583-603.

18. Ruhm CJ. Alcohol policies and highway vehiclefatalities. J Health Econ. 1996;15:435-454.

19. Saffer H. Alcohol Consumption and Tax Differentialsbetween Beer, Wine and Spirits. Cambridge, MA: NationalBureau of Economic Research; 1989. Working paper 3200.

20. Saffer H. Alcohol advertising and motor vehiclefatalities. Review Econ Stat 1997;79:431-442.

21. Saffer H, Grossman M. Beer taxes, the legal drinkingage, and youth motor vehicle fatalities.] Legal Stud1987;16:351-374.

22. Saffer H, Grossman M. Drinking age laws andhighway mortality rates: Cause and effect Econ Inq.

1987;25:403-417.

23. Smart RG, Mann RE. Treatment, Alcoholics Anon-ymous and alcohol controls during the decrease in alco-hol problems in Alberta: 1975-1993. Alcohol Alcohol.1998;33:265-272.

24. Young DJ, Bielinska-Kwapisz A. Alcohol prices,consumption, and traffic fatalities. Southern Econj2006;72:690-703.

25. Sloan FA, Reilly BA, Schenzler C. Effects of prices,

civil and criminal sanctions, and law enforcement onalcohol-related mortality.j Stud Alcohol. 1994;55:454-465.

26. Whetten-Goldstein K, Sloan FA, Stout E, Liang L.

Civil liability, criminal law, and other policies andalcohol-related motor vehicle fatalities in the United States:1984-1995. Accid Anal Prev. 2000;32:723-733.

27. Young D1, Likens TW. Alcohol regulation and autofatalities. Int Rev Law Econ. 2000;20:107-126.

28. Ohsfeldt RL, Morrisey MA. Beer taxes, workerscompensation, and industrial injury. Rev Econ Stat.1997;1:155-160.

29. Birckmayer J, Hemenway D. Minimum-age drinkinglaws and youth suicide, 1970-1990. AmJ Public Health.1999;89:1365-1368.

30. Grossman M, Markowitz S. Alcohol Regulation andViolence on College Campuses. Cambridge, MA: NationalBureau of Economic Research; 1999. Working paper7129.

31. Markowitz S. The price of alcohol, wife abuse, andhusband abuse. Southern Econj 2000;67:279-303.

32. Markowitz S. Alcohol, drugs and violent crime. IntRev Law Econ. 2005;25:20-44.

33. Markowitz S, Chatterji P, Kaestner R. Estimating theimpact of alcohol policies on youth suicides.jMent HealthPolicy Econ. 2003;6:37-46.

34. Markowitz S, Grossman M. Alcohol regulation anddomestic violence towards children. Contemp Econ Pol.1998;16:309-320.

35. Markowitz S, Grossman M. The effects of beer taxesonphysical child abuse.jHealth Econ. 2000; 19:271-282.

36. Yamasaki A, Chinami M, Suzuki M, Kaneko Y, FujitaD, Shirakawa T. Tobacco and alcohol tax relationships withsuicide in Switzerland. Psychol Rep. 2005;97:213-216.

37. Henderson C, Liu XH, Roux AVD, Link BG, HasinD. The effects of US state income inequality and alcohol

policies on symptoms of depression and alcohol depen-dence. Soc Sci Med. 2004;58:565-575.

38. Skog 0, Melberg HO. Becker's rational addictiontheory: an empirical test with price elasticties for distilledspirits in Denmark, 1911-31. Addiction. 2006; 10l11444-1450.

39. Cook PJ, Tauchen G. The effect of liquor taxes onheavy drinking. Bell]Econ. 1982;13:379-390.

40. Heien D, Pompells G. Stress, ethnic, and distributionfactors in a dichotomous response model of alcoholabuse. J Stud Alcohol. 1987;48:450-455.

41. Rush B, Steinberg M, Brook R. Relationshipsamong alcohol availability, alcohol consumption and

alcohol-related damage in the province of Ontarioand the state of Michigan. Adv Alcohol Subst Abuse.1986;5:33-45.

42. Ponicki WR, Gruenewald PJ. The impact of alcoholtaxation on liver cirrhosis mortality. ] Stud Alcohol.2006;67:934-938.

43. Schweitzer SO, Intriligator MD, Salehi H. Alcohol-ism: an econometric model of its causes, its effects, and itscontrol. In: Grant M, Williams A, Plant M, eds. Economicsand Alcohol. London, England: Croom Helm;1983:107-127.

44. Alaska Stat ,j60.010. Alcoholic Beverage Tax Act of1983 and Alcoholic Beverage Tax Act of 2002.

45. Cook TD, Campbell DT. Quasi-Experimentation:Design and Analysis Issues for Field Settings. Boston, MA:Houghton Mifflin; 1979.

46. Campbell DT, Stanley JC. Experimental and Quasi-experimental Designs for Research. Chicago, IL: RandMcNally; 1963.

47. Alcohol beverage taxes: beer. Alcohol Policy Informa-tion System Web site. Available at: http://www.alcoholpolicy.niaaa.ni.gov/Tax-Beer. Accessed July 7, 2008.

48. Alcohol beverage taxes: wine. Alcohol Policy In-formation System Web site. Available at: http://www.alcoholpolicyniaaani,.gov/Tax-Wine. Accessed July 7,2008.

49. Alcohol beverage taxes: distilled spirits. AlcoholPolicy Information System Web site. Available at: http://www.alcoholpolicy.niaaa.nih.gov/Tax-Spirits. AccessedJuly 7, 2008.

50. National Bureau of Economic Research. Mortalitydata-vital statistics: NCHS's multiple cause of death data,1959-2005. Available at: http://www.nber.org/data!multieausehtml. Accessed July 7, 2008.

51. International Classification of Diseases and RelatedHealth Problems, 1 Oth Revision. Geneva, Switzerland:World Health Organization and German Institute ofMedical Documentation and Information; 2007.Available at: http://www.who.int/classifications/apps/icd/icdl 0online. Accessed September 4, 2008.

52. Rehm J, Patra J, Popova S. Alcohol-attributablemortality and potential years of life lost in Canada 2001:

implications for prevention and policy. Addiction.2006;101:373-384.

53. English DR, Holman CDJ, Milne E, et al. TheQuantification of Drug-Caused Morbidity and Mortality inAustralia 1995. Canberra, Australia: CommonwealthDept of Human Services and Health; 1995.

54. Shultz JM, Rice DP, Parker DL, Goodman RA, StrohG, Chalmers N. Quantifying the disease impact of alcoholwith ARDI software. Public Health Rep. 1991;106:443-450.

55. Intercensal estimates of the total resident populationof states: 1970 to 1980. Washington, DC: US CensusBureau; 1995. Available at: http://www.census.gov/popest/archives/1 980s/st708Ots.txt. Accessed July 7,2008.

56. Intercensal estimates of the total resident populationof states: 1980 to 1990. Washington, DC: US CensusBureau; 1996. Available at: http://www.census.gov/popest/rchives/1980s/st8090ts.txt AccessedJuly 7,2008.

57. State population estimates: annual time series, July1, 1990 to July 1, 1999. Washington, DC: US CensusBureau; 1999. Available at: http://www.census.gov-popest/archives/1 990s/ST-99-03.txt. Accessed July 7,2008.

58. Annual estimates of the population for the UnitedStates and states, and for Puerto Rico: April 1, 2000 to

July 1, 2004. Washington, DC: US Census Bureau; 2004.Available at: http://www.census.gov/popest/states/tables/NST-EST2004-0l.pdf. Accessed July 7, 2008.

59. Box GEP, Jenkins G. Time Series Analysis: Fore-

casting and Control. New York, NY: Holden-Day; 1976.

60. Wei WWS. Time Series Analysis: Univariate andMultivariate Methods. 2nd ed. Boston, MA: Pearson; 2006.

61. McCleary R, Hay RA. Applied Time Series Analysisfor the Social Sciences. Beverly Hills, CA: Sage; 1980.

62. Cohen J. Statistical Power Analysis for the BehavioralSciences. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Asso-dates; 1988.

63. Edwards G, Anderson P, Babor TF, et al. AlcoholPolicy and the Public Good. New York, NY: Oxford Uni-versity Press; 1994.

64. Viegi G, Maio S, Pistelli F, Baldacci S, Carrozzi L.Epidemiology of chronic obstructive pulmonary disease:health effects of air pollution. Respirology. 2006;11:523-532.

65. Pope CA Bates DV III, Raizenne ME. Health effectsof particulate air pollution: time for reassessment? Envi-ron Health Perspect. 1995;103:472-480.

66. Annual average consumer price index, all items-all urban consumers, 1960-present. Alaska Departmentof Labor & Workforce Development Web site. Avail-able at: http://almis.labor.state.ak.us/?PAGEID=67&SUBID=241. Accessed July 7, 2008.

67. Carey KB, Scott-Sheldon LAJ, Carey MP, DeMartiniKS. Individual-level interventions to reduce college stu-dent drinking: a meta-analytic review. Addict Behav.2007;32:2469-2494.

68. Tobler NS, Roona MR, Ochshorn P, Marshall DG,Streke AV, Stackpole KM. School-based adolescent drugprevention programs: 1998 meta-analysis. J Prim Prev.2000;20:275-336.

69. Tobler NS, Stratton HH. Effectiveness of school-based drug prevention programs: a meta-analysis of theresearch. JPnrm Pree. 1997;18:71-128

1470 1 Research and Practice I Peer Reviewed I Wagenaar et ai. American Journal of Public Health I August 2009, Vol 99, No. 8

COPYRIGHT INFORMATION

TITLE: Effects of Alcohol Tax Increases on Alcohol-RelatedDisease Mortality in Alaska: Time-Series Analyses From1976 to 2004

SOURCE: Am J Public Health 99 no8 Ag 2009

The magazine publisher is the copyright holder of this article and itis reproduced with permission. Further reproduction of this article inviolation of the copyright is prohibited. To contact the publisher:http://www.apha.org/

economic, empirical research article

Documents