Estimating the impact of availability restrictions and taxation increases on alcohol consumption, 100% alcohol-attributable and all-cause mortality in the Baltic countries and Poland 2001-2020

INTRODUCTION

The European Union has one of the highest levels of alcohol consumption in the world, and the Baltic countries and Poland contribute to this. In 2019, all four countries were among the top 20 World Health Organization (WHO) Member States with respect to adult (defined as 15 years and older) alcohol per capita consumption (APC, Latvia: 13.1 litres (l) pure alcohol per capita; Lithuania: 11.8 l; Poland: 11.6 l; Estonia: 11.3 l) [1]. Accordingly, the proportions of alcohol-attributable disease and mortality burden have been high [2, 3], and alcohol control policies have been an important means by which to reduce this burden. Prior research has shown that major alcohol control policies in these countries were associated with marked immediate impacts on the level of consumption [4], all-cause mortality [5], and 100% alcohol-attributable mortality (see [6] for cause-specific mortality; for an overview, see [7]). In these analyses, the focus was on estimating the effect of major alcohol control policies, and thus were based on estimated average effects of taxa­tion and availability restrictions. The aim of the current study was to arrive at the best estimate of the effect for each policy category – i.e., taxation increases and availability restrictions – separately.

MATERIAL AND METHODS

SELECTION OF INTERVENTIONS

We selected all taxation increases which decreased the affordability of alcohol beverages, and all availability restrictions which decreased off-premise trading hours by at least 20%, in Estonia, Latvia, Lithuania, and Poland during the period 2001-2020 (see Appendix Table 1 for a listing of policies). Reductions in affordability were selected as a criterion, as it is the best marker for the underlying mechanism of action resulting from the taxation increases [8]. Affordability was established by considering changes in prices for alcoholic beverages, and disposable income [9]. Availability changes were based on time limitations placed on off-premise sales, defined as a reduction of at least 20% of opening hours per week, thus excluding minimal changes in availability. All three Baltic countries introduced a prohibition on alcohol sales during night time, which fulfilled these criteria [9]. The 20% threshold was established by structured expert interviews [9].

STATISTICAL METHODOLOGY

As full details of the statistical analysis strategy have been reported in earlier publications [4-6], we only give a summary of the methodology used here. In short, for all analyses, we used regression-based methodology, taking into consideration secular trends, and using trends in the other countries we studied that had no interventions as the control [7, 10].
More specifically, to assess the policy effects on yearly APC, we analyzed WHO data between the years 2000 and 2020. Outcome data were differenced and we analyzed the effects of dummy-coded variables (0.1) for taxation policies, availability policies, their interaction, and countries in a single model (see Table 1 for summary of effects; for the full model, see Appendix Table A2).
For both all-cause mortality and 100% alcohol-attributable, we used logarithmized monthly age-standardized mortality rates for the population 15 years and older as the outcome (based on 5-year-intervals and European standards [11]), and estimated the effects for one year after the respective intervention was implemented using a generalized additive mixed model [12]. All four countries were included in the analyses and were denoted by a dummy variable with Poland serving as the reference category.
For all outcomes, we also tested interactions to ensure that intervention effects did not differ by country. For the analyses of all-cause mortality, we computed separate models for males and females. Residuals of all models were examined using diagnostic graphs and R-squared was used to assist with selecting the most appropriate model. All analyses were performed by R version 4.2.3 [13]. Statistical significance was set at α = 0.05.

RESULTS

EFFECTS ON LEVEL OF CONSUMPTION

Table 1 provides an overview of the average effects of availability restrictions and taxation increases on APC. While taxation increases reducing affordability were on average associated with a decrease of almost one litre in APC from the previous year (point estimate: –0.89 l; 95% CI: –1.35 l, –0.43 l), availability restrictions were not significantly associated with APC within one year post the intervention (point estimate: –0.33 l; 95% CI: –1.06 l, 0.41 l). None of the interactions between the policy effect and country reached significance, indicating the taxation increases reducing affordability reduced level of alcohol consumption in the population similarly in all countries and independent of the specific year of enactment.

EFFECTS ON ALL-CAUSE AND 100% ALCOHOL-ATTRIBUTABLE MORTALITY RATES

Key results for the average effects of policy intervention on age-standardized all-cause mortality rates are summarized in Table 2. For males, the increase in taxation was associated with a significant reduction in the all-cause mortality rate, which was reduced by 1.30% (95% CI: –2.41%, –0.17%). Given the average number of deaths in the four countries, a 1.3% reduction in the mortality rate would correspond to 93 deaths postponed in Estonia, 166 in Latvia, 239 in Lithuania, and 2,606 in Poland, using the 2019 population data as the reference points. In addition, availability restrictions were associated with a reduction in 100% alcohol-attributable mortality by 11.42% (95% CI: –16.39%, –6.11%). The other associations were in the hypothesized direction, but were not statistically significant.
For females, neither taxation increases nor availability restrictions had statistically significant associations with all-cause mortality, even though both associations were in the hypothesized direction. Availability restrictions were association with an 8.17% reduction of 100% alcohol-attributable mortality (95% CI: –15.13%, –0.70%). The increase in taxation correlated with a significant decrease of the alcohol-attributable mortality rate, which was a reduction of 9.06% (95% CI: –14.02%, –3.83%).
For both sexes, taxation increases had a stronger impact on all-cause mortality rates compared to availability reductions. Interaction effects for countries having both taxation and availability policies were not statistically significant (Appendix Tables A6-A9), which suggested that the two policy categories did not change all-cause mortality rates differently across the four countries under investigation.
The situation was different for 100% alcohol-attributable mortality. Taxation differed in its association with this kind of mortality between countries for males, although there was no significant effect overall. For availability restrictions, there were sizable effects of an almost, or over, 10% average reduction in 100% mortality rates for both sexes.

DISCUSSION

Both taxation increases reducing affordability, and availability reductions affecting at least 20% of the off-premise purchasing hours for alcoholic beverages were associated with reductions in APC and all-cause mortality, but it was only for taxation increases that the associations proved to be significant. In contrast, for 100% alcohol-attributable mortality, only availability restrictions were significantly associated for both males and females.
Before we discuss implications, we would like to discuss causality as the first potential limitation. Clearly, these results are based on non-experimental and ecolo­gical data, and thus in principle alternative interpretations cannot be excluded [10]. However, the causal language used at times in this article still appears to be justified, as it seems extremely unlikely that in four countries over a 20-year period, alcohol policy changes happened at the same time as any potential alternative events which could equally well explain the decreases in APC in the same years. Additionally, these alternative events would have to result in immediate decreases in all-cause or 100% alcohol-attributable mortality in the month following the intervention and lasting for at least 12 months. In total, we recorded 13 taxation increases, of which 12 were included in our models, as well as four implementations of availability restrictions (see Table A1). Thus, the alternative explanation would have to be 16 events which happened exactly in the same months, the taxation increases and the availability restrictions were enacted, and this is clearly extremely unlikely.
Moreover, the unique design of using the respective trends in the countries that lacked interventions at a particular point in time as a control allows for strong statistical control [10, 14]. Poland seems to serve as a useful control in particular, as only two taxation increases and no availability reductions were put in place in the time period observed against a backdrop of a dismantling of alcohol policy [15, 16]. This led to a situation where despite reductions in all-cause mortality, alcohol-attributable mortality increased in this country.
Other limitations of our study include the limited number of time points for the analyses with APC serving as the outcome. APC data are collected on a yearly basis, and this means that by using differenced data for four countries over 20 years, we only have a total of 80 time points, which limits the number of variables which can be included in the analysis. Finally, the number of covariates for which comparable monthly data exist in all four countries is limited as well.
Despite the potential limitations, our results are consistent: increasing alcohol taxation to reduce affordability was related to decreases in the level of alcohol consumption by 0.89 l, and consequently to decreases in male all-cause mortality of 1.3%. In addition, availability restrictions impacted on 100% alcohol-attributable mortality in both sexes.
Overall, these results for the Baltic countries and Poland underscore the important role of alcohol taxation as a key alcohol policy for public health, corroborating the choice of the WHO European Region to establish increases in taxation as their signature project [17]. While the importance of implementing alcohol taxation increases to keep up with affordability cannot be overstated, it also must be mentioned that acceptance of taxation interventions requires a supportive environment. Key here is to increase the general public’s awareness that alcohol is “no ordinary commodity” in modern societies, but rather a substance with toxic and health-threatening properties to the drinkers themselves and to their families [18].
As for availability restrictions, while the impact was not as strong as taxation on APC and on all-cause mortality, it was the more important policy with respect to 100% alcohol-attributable mortality. This is in line with reviews of the literature which have clearly indicated its role in reducing alcohol-attributable harm [18, 19], especially injuries. In addition, a recent analysis on the last implementation of availability restrictions in Lithuania demonstrated that such restrictions were associated with reductions in cardiovascular mortality [20]. In summary, availability restrictions seem to be more linked to patterns of drinking – most notably, heavy episodic drinking – which have been shown to be linked to the specific outcomes above: injury [21]; for cardiovascular mortality [22]. In addition, heavy episodic drinking is clearly linked to the most important underlying causes of 100% alcohol-attributable mortality: alcohol poisoning, which is by definition an acute effect, which depends on ingestion of large quantities of alcohol in one event; and alcohol-related liver cirrhosis [23].

CONCLUSIONS

Our analyses confirm that taxation increases and availability restrictions are key alcohol control policies, as already identified by the WHO when they were established as “best buys” to reduce alcohol consumption and attributable harm [7, 24]. Both alcohol control policies were shown to be important in reducing consumption and alcohol-attributable mortality. However, as they produce partly different outcomes, they should be implemented according to the epidemiological profile of the country and the desired effects being sought.

ACKNOWLEDGEMENTS

This research was funded by the U.S. National Institute on Alcohol Abuse and Alcoholism, Grant/Award Number: 1R01AA028224. The authors would like to thank Ms. Astrid Otto for referencing and copy-editing.

DISCLOSURE

The authors report no conflict of interest.

References