Studies of human health are usually snapshots of a large population over a small fraction of their lives, gathering data so that researchers can use statistical methods to make inferences and identify correlations between lifestyles or genetics and health outcomes. There are many pitfalls here, not least of which is the tendency to lump together several groups with very different risks into one group, because the researchers didn't have the resources or the necessary data to dig deeper. For example, there is the business of risk levels between groups at various levels of alcohol consumption quoted below (wherein alcohol consumption habits are probably themselves strongly associated with lifestyle packages, wealth, conscientiousness, and other harder-to-measure line items that influence how well people tend to make use of medical services and how well they tend to take care of their health).
In an age of rapid progress in biotechnology, lifestyle choices like whether your drink a little or less than a little will soon become irrelevant to the general trajectory of your future health. Your future life span near-entirely depends on how fast rejuvenation therapies such as those featured in the Strategies for Engineered Negligible Senescence (SENS) proposals can be built. I point out this research by way of an example of one of the many systematic ways in which scientific work can be incomplete, misleading, or flawed, and encourage you think of it every time you look at another epidemiological study, so as to wonder what the authors there might be missing.
Multiple studies have shown that the likelihood of dying for people who drink increases as they consume more alcohol. Those same studies have shown that a person's mortality risk also increases at the other end of the spectrum - among people who choose not to drink at all - though the risk is still much less than for heavy drinkers. Some researchers have hypothesized that the increased mortality among nondrinkers could be related to the fact that light alcohol consumption - drinking, on average, less than one drink a day - might actually protect people from disease and reduce their stress levels.
But [other researchers] decided to examine whether characteristics of different subgroups of nondrinkers could explain the increased mortality risk. "Among nondrinkers, people have all sorts of background reasons for why they don't drink. We wanted to tease that out because it's not really informative to just assume that nondrinkers are a unified group." [The researchers] lied on data collected in 1988 by the National Health Interview Survey about the drinking habits of more than 41,000 people from across the United States. The researchers also had access to information about which respondents died between taking the survey and 2006. During the survey, nondrinkers were asked to provide their reasons for not drinking.
The research team divided nondrinkers into three general categories: "abstainers," or people who have never had more than 12 drinks in their lives; "infrequent drinkers," or people who have fewer than 12 drinks a year; and "former drinkers." Each category was further divided using a statistical technique that grouped people together who gave similar clusters of reasons for not drinking. The team then calculated the mortality risk for each subgroup compared with the mortality risk for light drinkers, and they found that the risks varied markedly.
Abstainers who chose not to drink for a cluster of reasons that included religious or moral motivations, being brought up not to drink, responsibilities to their family, as well as not liking the taste, had similar mortality risks over the follow-up period to light drinkers. Former drinkers, however, had the highest mortality risk of all nondrinkers. Former drinkers whose cluster of reasons for not drinking now included being an alcoholic and problems with drinking, for example, had a 38 percent higher mortality risk than light drinkers over the follow-up period. "So this idea that nondrinkers always have higher mortality than light drinkers isn't true. You can find some groups of nondrinkers who have similar mortality risks to light drinkers."