Average volume of alcohol consumption, patterns of drinking and mortality among young Europeans in 1999

Sir—We try here to estimate alcohol-related mortality for young people in Europe for the year 1999. Young people were defined as between 15 and 29 years of age. Alcohol has causal relations to more than 60 disease consequences (English et al. 1995; Single et al. 1999; Gutjahr & Gmel 2001). These relations concern chronic and acute outcomes, which are in part beneficial (Ashley et al. 2000) and in part detrimental, and which depend on average volume of alcohol consumed as well as on other aspects of drinking usually summarized as patterns of drinking (Rehm et al. 1996). To give one example: in a mortality follow-up of a representative US survey, it was shown that patterns of drinking mediated the effect of average volume to a considerable degree. Within the same category of light-to-moderate average drinking (defined as 0–2 drinks a day), the risk for all-cause mortality almost doubled for males if they had occasional heavy drinking occasions (Rehm et al. 2001). Patterns of drinking are related in particular to accidents and injuries (Robson et al. 1998; Single et al. 1999; Greenfield et al. 2001). Consequently, patterns of drinking were included into the current analyses. Patterns can be seen as a weighting factor, which modifies (i.e. statistically interacts) the influence of average volume on mortality and other disease outcomes (Rehm et al. 2001). Unfortunately, whereas there are series on meta-analyses on the influence of average volume on different diseases and causes of mortality (English et al. 1995; Single et al. 1999; Gutjahr & Gmel 2001), nothing comparable exists for patterns. The reason for this gap is that the discussion on the influence of patterns of drinking on mortality is relatively new (Rehm et al. 1996), and thus most medical epidemiological studies published had not yet included measures for patterns (Rehm 1998; Feunekes et al. 1999). To derive a pattern weight for this study, despite these problems, an aggregate-level analysis was performed on data from European countries (Rehm & Gmel 2000; Gmel et al. 2001). The following bullet points give an overview on the data sources to arrive at the estimates (for details see Rehm & Gmel 2001a). Mortality data for Europe for 1999 by sex and cause of death, taken from the WHO EIP databank. Estimates for the relative risk of cause-specific mortality associated with average volume of consumption taken from the Swiss cost study (Gutjahr & Gmel 2001), which itself was based on English et al. (1995) and Single et al. (1999). An estimate on average volume of consumption including unrecorded consumption for each country taken from the WHO (1999) Global Database on Alcohol. A pattern weight for each country, calculated based on the relation between a descriptive pattern value for each country and mortality from hierarchical analysis (Rehm et al. 2001; Gmel et al. 2001). The pattern values were obtained from a survey on relevant drinking characteristics to key informants by WHO in 2000 and from representative survey data (Rehm et al. 2001). The pattern weight proved significant only for males. All analyses were carried out by region. Regions were determined by WHO (2000) as part of the World Health Report 2000, based on clustering adult and child mortality rates around the world. For the purpose of this report, the region Euro B was subdivided further into two subregions (B1 and B2) with B2 including the formerly Socialist economies further east: EURO A (very low child, very low adult mortality): Andorra, Austria, Belgium, Croatia, Czech Republic, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Israel, Italy, Luxembourg, Malta, Monaco, Netherlands, Norway, Portugal, San Marino, Slovenia, Spain, Sweden, Switzerland, United Kingdom. EURO B1 (low child, low adult mortality): Albania, Bosnia and Herzegovina, Bulgaria, Georgia, Poland, Romania, Slovakia, the Former Yugoslav Republic of Macedonia, Turkey, Yugoslavia. EURO B2 (low child, low adult mortality): Armenia, Azerbaijan, Kyrgyzstan, Tajikistan, Turkmenistan, Uzbekistan. EURO C (low child, high adult mortality): Belarus, Estonia, Hungary, Kazakhstan, Latvia, Lithuania, Republic of Moldova, Russian Federation, Ukraine. In the age group considered (15–29 years of age), there was a clear effect for sex (see Table 1). Independent of region, males showed a higher mortality rate than females, in total more than twice as high. Furthermore, there were dramatic differences in standardized mortality rates between regions. For females, the region Euro A had about half the mortality rate of the other regions, which showed about the same level of mortality. For males, the differences are even more marked. Euro C had the highest mortality rate, about three times the rate of Euro A. Table 2 gives an overview on the alcohol indicators by region. Alcohol consumption indicators vary considerably between regions. Euro A with all EU countries has high average consumption, mainly wine and beer. Patterns of drinking are relatively favourable in most of these countries, and unrecorded consumption is relatively low. Euro B1 and C also both have relatively high average consumption, but relatively detrimental patterns of drinking and relatively high unrecorded consumption. In comparison with Euro B1, Euro C has even more detrimental patterns, higher unrecorded consumption with a domination of spirits consumption. Relative risk estimates for chronic causes of death were taken from the meta-analysis of Gutjahr & Gmel (2001) and Ashley et al. (2000) and combined with prevalence information to yield attributable fractions. For accidents and injuries, attributable fractions (AFs) of the published literature were converted into odds ratios, then multiplied by the pattern weight and the average volume estimates for each country for the period after 1995, and then reconverted to obtain sex- and cause-specific AFs per country (Rehm et al. 2001). The AFs per region were calculated as population-weighted averages of the country AFs, and mortality attributable to alcohol by region was estimated as the sum of cause-specific mortality (i.e. adding up the cause-specific attributable deaths, which were calculated by multiplying the region- and sex-specific fraction with the respective number of dead people for each cause). A description of the procedure without pattern weights can be found in the CRA guidelines (http://www.ctru.auckland.ac.nz/CRA/main.html) or in Single et al. (1996, 1999). Table 3 gives an overview of estimated deaths by region and sex. Almost 58 000 deaths among young people (15–29 years old) in Europe in 1999 can be attributed to alcohol. There are clear differences in the proportion by sex and region: males have higher proportions than females in mortality attributed to alcohol (25.3% versus 9.3%), and the western countries have lower rates of alcohol mortality compared to the formerly Socialist countries with the exception of the mainly Muslim Asian countries (see Table 3). Detailed analyses indicated that 19% of all alcohol-related deaths could be attributed to unrecorded consumption, again with higher proportions (28%) attributable in Euro C. Concerning cause of death, accidents and injuries accounted for the overwhelming majority of the deaths with the following rank order: transport accidents, poisonings, self-inflicted injuries and homicide. It should be noted that war-related deaths were given an AF of 0 in this estimation. Cirrhosis of the liver was the main alcohol-related cause of death within chronic diseases, but the absolute numbers were relatively small compared to accident numbers (total of 1442 deaths estimated attributable to alcohol-related liver cirrhosis or 2.5% or all deaths). Alcohol is a major contributor to deaths in young people in Europe. It could be shown that for males, patterns of drinking played an important role in causing death, in addition to average volume of drinking. It can be assumed that patterns also play a role in females, but the methodology used to derive pattern weights did not yield significant results. As females consume only a small portion of the overall per capita consumption (Simpura & Karlsson 2001), it can be speculated that the lack of finding a relationship for females may be due to the fact that this indicator is neither reliable nor valid for this group. It should also be noted that the pattern weights were derived by an aggregate level (time-series) analysis, which may be prone to several limitations and biases (Rehm & Gmel 2001b). A replication of estimating these pattern weights with individual level data would be valuable, although probably impossible in the near future due to the absence of individual level studies for most of the countries. Overall, however, the aggregate level results of the influence of patterns are in line with recent individual-level results of drinking patterns and mortality (Rehm et al. 2001). The role of unrecorded consumption is considerable. It is estimated that slightly less than one-fifth of the alcohol-attributable deaths can be attributed to this factor. Thus, unrecorded consumption should be included in future estimates on alcohol-related burden of disease. There are tremendous differences in mortality, as well as sex differences, between different regions of Europe. Alcohol seems to be the reason for a substantial portion of these differences, both for sex and for the difference between established market economies in the western part and the formerly Socialist republics in the East, with the exception of course of the mainly Muslim Asian countries. Thus, alcohol may be the biggest behavioural risk factor for this age group. All of these deaths are, in principal, avoidable. A large proportion of them could be avoided if existing laws and regulations on drinking and driving and on work-place safety were enforced (Edwards et al. 1994). Other proven alcohol policy measures such as taxation (Edwards et al. 1994) may help in further reducing the toll of alcohol not only in formerly Socialist economies. Given the severity of the problem, as well as the possibility of evidence-based policy tools to reduce them, why can the current situation not be turned around?