MEASURING ALCOHOL CONSUMPTION—IS A REASONABLE CHANGE ALWAYS REASONABLE? RESPONSE TO KIEFER & SPANAGEL (2006)

We would like to thank Kiefer & Spanagel [1] for their thought-provoking letter related to improving alcohol measurement. Their point is provoking, because to our knowledge no cross-cultural research, including meta-analyses to measure relative-risks (e.g. estimates of mortality and morbidity such as those presented in the Report of the World Health Organization [2]), has applied adjustments for body weight. Even the Global Burden of Disease Study [3] did not use weight-adjusted cut-offs to define the three levels of alcohol intake, although they did use different cut-offs for men and women. It would be interesting to see whether the findings from this report would have been different if adjustments for body weight had been used. We fully agree with Kiefer & Spanagel [1] that adjustment for body weight or the use of a standard measure of ethanol intake as g/kg/day (gram pure ethanol per kg body weight per day) has the potential to avoid some alcohol measurement inequalities due to sex and ethnically or culturally based population differences. This may be particularly true for clinical research, as mentioned by Kiefer & Spanagel [1], where body weight can be measured objectively in the clinical setting. However, for cross-cultural survey research there are a number of considerations that could make the routine adjustment for body weight problematic. First, as noted previously regarding adjusting alcohol consumption measures for biological gender differences [4], adjusting for the drinker size of the person would be inappropriate if alcohol consumption differences that affect peak blood alcohol consumption (e.g. drinking pace and whether alcohol is consumed with meals) are correlated with the person’s size. For example, if smaller people tend to drink more slowly compared with larger people, their peak blood alcohol level might be the same or even lower than that of larger people who drank the same quantity but more quickly, even though the overall dose per body weight for the smaller person would be higher than for the larger person. In this scenario, adjusting for body weight would be inappropriate for estimating negative consequences related to the acute effects of alcohol. This is not to say that adjusting for body weight is not a good idea if all other things are equal. However, in cross-cultural comparisons all other things are rarely equal, and such an adjustment could produce erroneous results if body size and drinking pattern or style are confounded. On the other hand, as the world grows increasingly small and we engage increasingly in cross-national studies, it is good advice to take into consideration the impact of cultural differences in body size as well as other cultural differences that apply to how we measure and interpret alcohol consumption when making comparisons across various countries. It is possible that the extent of high-risk drinking is underestimated in some countries where smaller body weight and riskier drinking patterns coincide, as noted above. A second factor that affects interpretation of overall volume of alcohol consumption is drinking pattern—that is, the harmful effects of drinking 14 drinks per week are likely to depend on how the drinks were consumed (e.g. 14 on one day versus two drinks per day with a meal). Thus, drinking pattern is likely to be a much more important factor than adjusting for body weight, particularly for the acute effects of alcohol such as driving under the influence, and again drinking patterns are known to vary across cultures [3]. In addition, survey measures of alcohol consumption are far from perfect, and there are a number of other alcohol measurement biases that may affect differentially different countries for which the effects of adjusting for body weight would either be inappropriate or irrelevant. For example, survey instruments on alcohol consumption have been found to underestimate sales data by between 30 and 70%[5, 6]. Unfortunately, these factors are not constant in cross-country comparisons and thus influence comparisons. Given the magnitude of such a bias, body weight adjustments would be a minor factor for the increase in the validity of measurement instruments. Survey measures of body weight may also have systematic errors that would put into question routine adjustments for body weight. In particular, there is evidence in the literature that self-reports of body weight are often biased, and this bias does not necessarily apply equally to the population, with women and those who are overweight more likely to under-report weight and differential biases associated with different age groups [7, 8]. Finally, our study showed differences in estimates of mean consumption within a country of as much as 30%, depending on the type of alcohol measure used (e.g. generic versus beverage specific measurement, graduated frequency versus quantity–frequency). Furthermore, the type of measure producing the highest estimate was not consistent across countries. For this type of comparative research, the bias of not adjusting for body weight would be irrelevant, as we compared different instruments within a country and compared these differences between instruments across countries. For between-instrument comparisons within countries, body weight adjustment would be a constant factor applying to all instruments in the same way, and thus would not bias the between-instrument comparisons. To conclude, we think that Kiefer & Spanagel have raised an important issue. Specifically, however, there is a need to develop measures of alcohol consumption that are valid for cross-cultural comparisons before adjustment for drinker size can be applied.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.