Friday, March 18, 2011

Latest Age of Onset Study Is Less than Meets the Eye

Just in time for St. Patrick's Day, a well-known drinking holiday, a new study comes out about the controversial relationship between age of onset of drinking and later alcohol dependence.

What the study apparently found was that 1) those who began drinking at all before the age of 11 (!) were statistically more likely to develop alcohol dependence than those who began later, and 2) those who began drinking regularly before age 21 were statistically more likely to develop alcohol dependence (but not alcohol misuse) than those who began later, with interestingly no statistically significant difference between those who began in early adolescence (11-14), middle adolescence (15-17), and those who began between 18-20 years of age. 

However, there is less here than meets the eye.  First of all, for regular drinking (defined as 2 or more drinks per week, an arbitrary distinction and not a real lot) they lumped together those who drank lightly or moderately with those who drank heavily, and lumped together those who started at exactly 21 with those who started much later.  Well DUH--those who had any interest at all in alcohol would likely have begun drinking at least occasionally before 21, and those who would actually wait that long tend to be nondrinkers for the most part. People uninterested in alcohol are unlikely to become alcoholics, plain and simple.  This is especially true since those individuals who began drinking regularly after 21 but did not initiate any drinking before 21 were excluded from the study.  Also since the study only followed participants to age 33, those who began drinking after 21 may not have had enough time for problem drinking to show up in the study.  As for those who began drinking regularly between 18-20, it is likely that most of them had their first drink well before 18 since most of that study's participants began drinking very early, but this distinction was not made among the reportedly regular drinkers.  And those who had their first drink at 18-20 were found to be 65% less likely to become alcoholics than those who began before the age of 14, making it rather curious why the researchers did not adjust for this in the analyses of regular drinking.

Other flaws existed as well.  The study did not adequately control for general deviant tendencies, and deviant people are more likely to break the law (such as the MLDA), more likely to begin drinking much earlier (or later) than their peers, and more likely to become alcoholics.  The only control variables were race/ethnicity, gender, poverty, tobacco, and illicit drug use, despite the fact that many other variables (e.g. genetics, childhood trauma, peer group, IQ, etc.) may very well have had an effect.  Also, the sample size was fairly small and was likely not representative of the general population, since all participants were initially recruited from schools in high-crime areas of Seattle.  Reporting bias, especially on the question of regular drinking, may also be an issue. Another issue was the apparently non-monotonic relationship between age of onset of regular drinking and alcoholism (starting at age 18-20 had a slightly higher odds ratio than at age 14 and under, though statisically insignificant), which is rather suspicious and runs counter to the idea that the relationship is primarily due to differences in brain development.  Finally, the adjusted odds ratios were 1.6-1.8, and all honest epidemiologists know that odds ratios below 2.0 should be taken with at least a grain of salt, if not a whole pound.

To really control for as many confounders as possible, twin studies are the best choice since their genetics and environment are about as similar as one can get.  A recent twin study (Agrawal et al. 2009) with a larger sample size in Australia found that, while drinking before age 15 was associated with a significantly greater incidence of alcohol dependence, the effect of age of onset diminishes after that point and practically disappears from age 18 onwards. Those who began drinking at 18 were not significantly more likely to become alcoholics than if they had started at 19, 20, 21, or even 23+. Funny how the effect of age on later alcoholism levels off precisely at the country's legal drinking age, 18.  And in a country that has roughly the same rate of alcoholism that we do despite their lower drinking age.

To prove causation as per the Bradford-Hills criteria, one crucial criterion is "biological plausibility."  While that is not exactly the same thing as "truth", without it one cannot have a great deal of confidence that a relationship is truly causal.  Animal studies are often used for this purpose, but rodents are hardly ideal to test a hypothesis like this since their adolescence occurs far too quickly to distinguish between specific ages, not to mention the obvious fact that rats are not people.  Thus, such studies have been rather inconsistent.  Nonhuman primates, such as monkeys, are better suited to such a task.  And the only known monkey study of whether earlier-onset drinkers are more likely to get hooked than later-onset ones shows no connection with age of onset per se, even though the monkeys were dosed with "binge" amounts.  This suggests that genetic, social, and cultural factors are most likely far more important than age of onset for humans, if the latter is even causal at all, and that the difference between an age of onset at 18 vs. 21 is practically negligible.

Also, one ought to take a look at a previous post about our neighbor to the north.  Again, they also have a roughly equivalent alcoholism rate as the USA, despite Canada's lower drinking age as well as their demographic and geographic features more conducive to heavy drinking.  Besides, it's not like the 21 drinking age really stops 18-20 year olds from drinking--even the latest study showed that nearly 9 out of 10 people will do so before age 21, many of them having initated at or before age 15.  Thus, it is unrealistic to expect everyone to not drink a drop until 21 or later, and far better to teach moderation and responsible drinking.  Yes, there were a few people who waited until 21 or later before drinking, but that was also true in the Australian study, in a country whose drinking age has been 18 for decades (and over a century in some parts).  It's time for Americans to stick their heads out of their current anatomically-impossible positions and join the rest of the civilized world.