The last time I saw my doctor, she asked me how many drinks I have per week. “I don’t know, maybe 10,” I said. She wrote my answer on a clipboard. “No wait, I guess it’s more like 15. Or 20?”
“Well, which is it?”
I had no idea, and that seemed like a problem. My fuzzy recollection corresponded to a range that includes both low-risk (or even healthy) levels of consumption—one or two drinks per day for men—and more dangerous behavior. Did my habit put me in the normal range, or had I unwittingly become a heavy drinker?
In the months that followed I kept a tally of my tipples, I think more out of curiosity than dread and discovered that my intake changes through the year: boozy winter, sober spring. But overall the numbers gave me some relief. It turns out that I’m not a lush at all, but a somewhat temperate fellow who has an average of 11 drinks per week.
That sense of normalcy didn’t last very long. A couple of weeks ago, the Washington Post put out a shocking chart that shows how much alcohol U.S. adults consume, and it was plain to see where I fit into the distribution. My weekly total, still shy of a dozen drinks, puts me in the very drunkest quarter of Americans. Almost 30 percent of the country never touches alcohol, the chart declares, and another 30 percent do so only on special occasions—at most once every couple of weeks. And then there are the rest of us, the inebriate winos like me and all my friends and almost everyone we know.
The more I thought about these data, the stranger they seemed. My 11-drink-per-week habit corresponds almost exactly to the national average, 556 drinks per year. But according to this chart, that number would be grossly misleading, since it combines the huge number of people who never quaff at all with the much smaller group who drink like it’s their job. (Those in the top decile of consumption swig an incomprehensible average of 74 drinks per week.) When I compare my habit to the median instead of the mean, I find that I’m dosing myself with 10 times more alcohol than the typical American.
At first I thought the figures must be wrong. But the author of the chart, Christopher Ingraham, had pulled them from a 2007 book by the Duke University economist Philip J. Cook, who drew from fieldwork conducted by the Census Bureau for the National Epidemiologic Survey on Alcohol and Related Conditions.* Cook’s NESARC data are a little old—they come from 2000 and 2001—but by all accounts they’re unlikely to have changed. In fact, the pattern they represent, with a tiny minority responsible for almost all the nation’s drinking, has remained more or less a constant over many decades, and it seems to hold for other countries, too. It’s like a golden rule for alcohol research: No matter where or when in history you look, consumption fits into the same, skewed distribution.
Like all good things related to alcohol, the “single-distribution theory” got its start in France. After World War II, an Algerian-born ex–artillery officer named Sully Ledermann started trying to figure out why his French countrymen lived such shorter lives than men in other European countries. By 1946 he’d figured out that the higher mortality rates were concentrated among men of middle age: A 40-year-old Frenchman was three times more likely to die than an age-matched Dutchman and twice as likely to die as an Englishman. Ledermann blamed these early deaths on booze.
His evidence was straightforward: Ledermann noticed that, over time, death rates in France corresponded to the country’s total consumption of alcohol and that people died most often in the regions where they drank the most—Brittany and Normandy. According to an excellent article by the late Norwegian sociologist Ole-Jorgen Skog, Ledermann’s idea went against the prevailing view among his colleagues in English-speaking countries. In England and America, most researchers believed that alcoholics—the people most likely to die from too much drinking—were the victims of a special predisposition or disease and that their condition had little to do with the standards for “normal” drinking where they came from. If that were true, then the rate of deaths from alcoholism should be fairly constant from one place to another. It wouldn’t matter if the average person were drinking one glass of wine per day or two or three or five; the drunks would still be drunks.
Ledermann’s data suggested otherwise. To him, it looked like the rate of alcohol-related mortality rose and fell atop a social tide, instead of floating somewhere on its own. In fact, there seemed to be a predictable relationship between the number of alcoholics in a given place or culture and the amount of booze consumed overall. He argued that a single type of curve—the log-normal distribution—could describe the nature of this relationship, and then he made an even bolder claim: That standard curve would keep its shape even as overall consumption habits changed. If Frenchmen started drinking more, that trend would turn some abstainers into social drinkers, and some social drinkers into sots.
His theory helped explain why at the end of the 1930s, when so many Frenchmen were dying young, the country’s alcohol consumption peaked at an average of almost five drinks per day, or 33 per week. Since then, French intake has declined by more than half—as of 2009, it’s down to two drinks per day—but it’s still much higher than in most other European countries. (Only the Czech, Estonians and Irish take more liquor.) A recent study blamed French drinking habits for 49,000 early deaths per year and a full 13 percent of the nation’s male mortality.
For our part, Americans drink a little less than the French on average, and researchers here have attributed 9.8 percent of all deaths among working-age adults to excessive drinking. The Centers for Disease Control and Prevention tracks deaths from 54 different alcohol-related causes, including falls, car accidents, acute toxicity, liver disease, stroke, and multiple types of cancer.
As it happens, Ledermann died in middle age himself, in 1967. In the decades since, his single-distribution theory has been tweaked but never fully contradicted. (Ole-Jorgen Skog gave the theory its most thorough working over in the 1980s.) The alcohol-consumption curve does seem to take a roughly standard form, but its shape isn’t as precise as Ledermann claimed. Similarly, his argument that any change in mean consumption will change the alcoholism rate in predictable ways doesn’t always hold. But when it comes to assessing public policy, it’s not a bad rule of thumb.
Ledermann’s work may be as interesting, though, for what it says about why we drink the way we do. He argued that the standard, skewed pattern of alcohol use arises from the way that drinking spreads: as a social contagion. If I was surprised to learn that my “modest” drinking habit places me well outside the American mainstream, that’s just because I see my friends’ behavior as the norm. When you’re sitting on the slope of that consumption curve, it can be hard to keep things in perspective.
*Correction, Oct. 8, 2014: This article originally misidentified economist Philip J. Cook's academic affiliation. He was at Duke University, not Stanford.