Alcohol consumption is a loaded topic. Most people know that it is probably not very good for your health, but how bad can it be? We know that some people die due to alcohol consumption, but that probably happens only to drunkards in old age. Moderation is the key. Moreover, there’s been a notion in popular culture for years, that a drink or two might be actually good for you! Everybody does it. Millions of people cannot be wrong, of course. These claims will be explored in the article below.
The paper that I will be studying today is Alcohol Use and Burden for 195 countries and territories, 1990-2016: a systematic analysis of the Global Burden of Disease Study 2016. This paper, published in the prestigious journal Lancet, is one of the most highly cited scientific papers in recent years, and disproves a lot of myths around alcohol consumption.
Most researchers concur that alcohol is a leading cause for much illness and disease. However, some evidence suggests that a low intake of alcohol has a protective effect on some conditions like ischemic heart disease (the most common form of heart disease) and diabetes. The method of data collection for this “evidence” was self-reporting by consumers of alcohol, which has been found to be unreliable. Some recent studies have instead focused on alcohol sales numbers, prevalence of drinking habits, etc.
This study improves on the prevalent data collection process in the following ways: 694 individual and population-level data sources have been consulted. Alcohol consumption levels have been adjusted to include consumption by tourists (which probably accounts for a family large percentage of alcohol consumption in touristy places). New methods have been devised to better approximate illicit alcohol sales (that are not recorded in stores). 23 health-related outcomes of alcohol consumption, like cirrhosis, tuberculosis, cancer etc have been identified and individually studied. Also, relative risk curves have been used to determine the ideal level of alcohol consumption for maximal health benefits. What is a relative risk curve, you ask? If not drinking alcohol can be thought to have a “risk” factor of for the advent of tuberculosis, then drinking 4 glasses of beer everyday can be thought to have a relative risk factor of . Hence, drinking 4 glasses of beer everyday might make you times more likely to contract tuberculosis.
How do you study the effects of alcohol? By comparing a group of people who drink, with a group of people that does not. This group of people which does not drink, with which other categories of people are compared, can be called the “reference category” in this context. Previous studies didn’t care about the composition of this reference category. Hence, this reference category for them included people who drank heavily in the past (and hence had less than ideal health conditions). On comparing this reference category with the category of drinkers, they didn’t find much difference, and hence concluded that drinking doesn’t change one’s health that much. This study removes this bias by carefully controlling the membership of this reference category.
Also, previous studies just used the assumption that the amount of drinking that minimizes harm is 0%, and then determined population statistics and error margins from this. This should not just be an assumption, but needs to be proven. Moreover, such an assumption reduces the accuracy of the statistical study. This paper does not make such an assumption, and hence better captures the effects of alcohol on health outcomes. Moreover, possibly the confidence intervals of the conclusions are larger.
The data used in this study has been collected from 195 locations across the world from 1990 to 2016, for both sexes, and for 5 year age groups between the ages of 5 and 95 years or older. Moreover, the risk of alcohol consumption for 23 diseases and health disorders has also been studied.
Alcohol consumption is measured in grams of pure ethanol consumed. Hence, a couple of light beers might be equal to a whiskey shot. For estimating consumption of alcohol, 121029 data points from 694 resources were pooled. For estimating disease risk, risk estimates from 592 studies were combined. A total of 28 million individuals were involved in these 592 studies.
The total amount of alcohol in a community is approximated from the total amount of alcohol sold in stores. The relative numbers of drinkers and abstainers are approximated through surveys. Also, the amount of alcohol consumed at an individual level, across age groups, is also approximated through surveys. This data is corrected for alcohol consumed by tourists. Also, in order to get a conservative estimate for the amount of illicit alcohol sold (outside of stores), the authors collate estimates from multiple studies on the sale of illicit alcohol, and construct a uniform probability distribution based on these multiple studies. Hence, each study has an equal probability of being correct. They then construct a new probability distribution, whose range is 0 to the average of the above probability distribution, and then take a sample of 1000 from this new probability distribution, which they then average. This average is obviously lower (probably) than the average of the first distribution. Hence, they get a conservative estimate of the amount of illicit alcohol sold.
In 2016, 32.5% of the world population were drinkers (25% of females and 39% of males). Hence, 2.4 billion people in the world drank alcohol. The mean amount of alcohol consumed was 0.73 standard drinks daily for females and 1.7 standard drinks for males. The actual consumption varied with the Socio-Demographic Index, or SDI, which can be thought of as similar to an index of economic and social prosperity.
On the y-axis of both these graphs, the disability-adjusted life-years, or DALYs have been plotted. These count the number of years spent handicapped or otherwise disabled, per 100,000 years, due to alcohol consumption. Note that these are only global averages. The data varies considerably with changing SDIs. Note that in both instances, alcohol consumption only increases the amount of disability in both genders across 23 diseases and disorders. On the other hand, the prevalence of Ischaemic heart disease in older age groups reduces slightly in both genders. Also, diabetes in females also becomes slightly better with alcohol consumption. Note that these are the only exceptions out of 23 diseases.
In 2016, 2.8 million deaths were attributed to alcohol consumption. This is 6.8% of all deaths amongst males, and 2.2% of all deaths amongst females. Also, alcohol was the cause of 6% of all DALYs amongst males (years spent incapacitated), and 1.6% of all DALYs amongst females. These percentages were even higher for groups of people between 15-49 years. The three leading causes of death that alcohol contributed to, for both females and males in this age group, were tuberculosis (1.4%), road injuries (1.2%) and self-harm (1.1%). For the age groups 50-80, the most prevalent disease in high SDI regions that was contributed to by alcohol consumption was cancer (27.1%). In poorer societies, tuberculosis was the most prevalent such disease. In age groups beyond 80, hemorrhage and hypertensive heart disease became much more prevalent.
If the risk of alcohol consumption with respect to these 23 diseases could be “averaged out” and then plotted, the graph would look something like this:
Past studies had shown that although risks associated with alcohol consumption increased with an increase in alcohol intake, consuming a small amount of alcohol daily was actually good for the body. Subsequent studies, conducted using mendelian analysis and multi-variate meta analysis, had then shown this to be a flawed conclusion. This paper also reaffirms the finding that no amount of alcohol is good for the body. For optimum health, the amount of alcohol we should be drinking daily is zero. However, the graph shown above suggests that drinking a maximum of one drink everyday is not all that bad, although still worse than not drinking at all.
Alcohol is a leading risk factor for disease world wide, and accounts for nearly 10% of deaths worldwide. The government should make the buying of alcohol more expensive, and should restrict its availability in stores. An extreme example of unchecked alcohol consumption is Russia, where in the 1980s, 75% of male deaths were attributed to excessive alcohol consumption. The authors concede that they don’t have data for multiple things, like the fraction of road accidents outside the US, and interpersonal harm caused due to alcohol. Also, they don’t have data on drinking below the age of 15. However, they conclude that having this data could only increase the estimate of health risk associated with alcohol consumption, and not reduce it.