A Scary Government Report Implies That Moderate Drinking Is Manifestly Reckless
The focus on the health risks of alcohol consumption gives short shrift to the reasons people like to drink.
A new report from a federal advisory panel charged with examining the evidence concerning the health outcomes associated with alcohol consumption paints a decidedly more alarming picture than a recent review from the National Academies of Sciences, Engineering, and Medicine (NASEM). Since both reports are supposed to guide this year's revision of the federal government's dietary advice, the question of which to believe is of keen interest to businesses that oppose a reduction in recommended drinking limits.
A coalition of alcohol-related trade groups yesterday slammed the scarier report, which was prepared by the Interagency Coordinating Committee on the Prevention of Underage Drinking (ICCPUD). The coalition, which includes groups representing distillers, vintners, brewers, restaurateurs, farmers, bartenders, and liquor retailers, argues that the ICCPUD can't be trusted because it is "tainted by bias" against alcohol. Those business interests understandably prefer the NASEM report. Critics of that report, meanwhile, complain that it was skewed by a pro-alcohol bias.
Questions of motivation aside, there are sound scientific reasons to treat both reports with caution, given the widely recognized weaknesses of the observational studies on which they rely. And regardless of which report you find more persuasive, the question of whether and how much to drink ultimately comes down to a value judgment that weighs individual risks against individual benefits. Neither public health bureaucrats nor physicians who follow their guidance are in a position to make that call, although their advice may factor into it.
The NASEM report concludes with "moderate certainty" that alcohol consumption within the limits recommended by the most recent version of the Dietary Guidelines for Americans—two drinks a day for men and one for women—is associated with a 10 percent increase in breast cancer risk. But it says "no conclusion could be drawn" regarding other cancers.
NASEM's panel of experts also found enough evidence to conclude with "moderate certainty" that drinkers who consume "moderate amounts of alcohol" face a lower risk of death from cardiovascular disease than teetotalers do. And the panel concluded, with the same level of confidence, that "moderate alcohol consumption is associated with lower all-cause mortality."
The ICCPUD report, by contrast, warns that "the risk of dying from alcohol use begins at low levels of average use." It concurs with a recent advisory from Surgeon General Vivek Murthy, who said alcohol consumption has been convincingly linked to "at least seven different types of cancer" and warned that even moderate drinking raises the risk of those diseases. But the ICCPUD report goes further, linking alcohol to a litany of terrifying outcomes, ranging from lethal overdoses and injuries to potentially deadly diseases such as liver cirrhosis and esophageal cancer.
The ICCPUD nevertheless concedes that moderate drinking is associated with a reduced risk of diabetes among women and a reduced risk of ischemic stroke for men and women. "For conditions such as ischemic heart disease, ischemic stroke, certain cancers (e.g., kidney and thyroid), and diabetes mellitus," it says, "there is mixed research about the potential protective effect on disease occurrence and mortality for people who consume relatively low amounts of alcohol and who do not engage in binge drinking."
In addition to its emphasis on all the bad things that might happen to you if you drink, the ICCPUD report is notably different from the NASEM report in the way it treats the inherent uncertainties of epidemiological research. Those uncertainties stem from problems such as the inaccuracy of self-reported alcohol consumption, the difficulty of controlling for all the variables that affect disease risks, and the fact that moderate drinkers differ from abstainers in other ways that may affect their health, including socioeconomic status, physical activity, tobacco use, and dietary habits.
The authors of the NASEM report are upfront about those limitations and careful to present their findings as associations that may or may not reflect causal relationships. The authors of the ICCPUD report are less cautious.
"Research causally links alcohol consumption, including non-excessive and excessive use, with more than 200 health conditions," the ICCPUD report declares. "The primary focus of these analyses was on morbidity and mortality from conditions that are considered causally related to alcohol." The authors purport to be considering "the relationship between alcohol consumption and the occurrence of disease and injury for diseases and injuries causally related to alcohol."
That characterization appears throughout the 81-page report. But on page 43, a careful reader will discover that conditions "considered causally related to alcohol" may not, in fact, be causally related to alcohol: "As more research is performed, conditions where there was previously thought to be a causal link with alcohol consumption may be found to be non-causal."
That section of the report also notes that "the relationships between alcohol and health in this report are informed primarily by observational studies." It adds that "the limitations of this literature are considerable and need to be acknowledged." But the upshot, according to the authors, is that "the literature may under-estimate alcohol-related risk."
How so? "People who drink are often compared to those who do not drink, and ideally to those who have never or only very rarely consumed alcohol," the report says. "Many of these non-drinkers are in fact former drinkers who quit drinking due to poor health (due to alcohol use or other reasons); as such they are misclassified and 'contaminate' the non-drinking reference group."
The NASEM report avoids that pitfall by limiting its analysis to studies where the comparison group consisted of lifetime abstainers. It nevertheless found that moderate drinking was associated with lower cardiovascular risk and lower all-cause mortality.
The ICCPUD report also notes that "non-drinkers tend to have other risk factors for ill health that are unrelated to alcohol," which are "difficult to account for statistically." That is a valid point, and the NASEM report acknowledges it. But what's strange about the ICCPUD's discussion of research limitations is that the report's authors never consider the possibility that methodological problems might result in findings that exaggerate the risk posed by moderate drinking. In the authors' telling, the only danger is that the research "may under-estimate alcohol-related risk."
Consider underreporting of alcohol consumption. The ICCPUD report acknowledges that problem. "Self-reported alcohol use" is "subject to recall and social desirability bias, which can result in underreporting of alcohol consumption," the authors say. "To address potential underestimation of alcohol consumption in cohort studies when modeling the lifetime risk of alcohol-attributable deaths, we assumed that 10% of alcohol consumed by cohort participants was not captured in these studies."
There is reason to wonder whether that adjustment was adequate. A 2014 Canadian study, for example, compared responses in a survey about alcohol use to actual consumption reflected in liquor, wine, and beer sales. The researchers found that "spirits consumption was underestimated by 65.94% compared with sales data, wine by 38.35% and beer by 49.02%."
In light of such underreporting, the data that inform official advice about alcohol may be systematically biased toward finding health risks at relatively low levels of consumption. Furthermore, that advice may itself increase the likelihood of underreporting: When men know that the government says they should not consume more than two drinks a day, for example, they are apt to report that they keep to that limit, whether or not that is true. None of this seems to interest the authors of the ICCPUD report.
Like Murthy's advisory, the ICCPUD report never explicitly mentions the NASEM review, which Congress commissioned to help inform the revision of the Dietary Guidelines. But in what may be a veiled criticism of the NASEM panel, the ICCPUD report says "the use of all-cause mortality studies to examine population estimates about alcohol and health is problematic" because such studies "include deaths from conditions that have no causal relationship with alcohol, thereby increasing the risk of confounding and reducing the specificity of the findings."
The authors say it is therefore "necessary to estimate mortality risk based on cause-specific risks (i.e., specific risks from conditions causally related to alcohol), as was done in the present study, to inform public health prevention efforts because only these direct risks are preventable." That formulation again conflates correlation with causation, although animal studies and theories about how alcohol consumption might affect disease risk can help draw that distinction. In any case, this argument does not discredit the NASEM report, which considers not only all-cause mortality but also health outcomes in specific areas, including weight change, cancer, cardiovascular disease, neurocognition, and maternal alcohol consumption during lactation.
Even drinkers who take the ICCPUD report at face value may not be deterred by the dangers it describes. "In the United States, males and females have a 1 in 1000 risk of dying from alcohol use if they consume more than 7 drinks per week," the report says. "This risk increases to 1 in 100 if they consume more than 9 drinks per week." Although a tenfold increase in risk is nothing to sneeze at, the absolute risk remains pretty low even if you overlook the questionable assumptions underlying that estimate.
Is a 1 percent risk of alcohol-related death "acceptable"? The report's authors clearly do not think so. "Studies on alcohol and health are often based on the risk of harms and what level of risk is 'acceptable,'" they note. They concede that "acceptable risk levels may vary from person to person and be context-dependent," hinging on questions such as whether "there a beneficial tradeoff for exposure to a risk." But "for environmental hazard regulations," they note, "an involuntary risk of 1 in 1,000,000 lifetime deaths is the standard definition of an acceptable threshold."
That "standard definition" can easily lead to regulations that impose exorbitant costs for each death that is theoretically prevented. But leaving that point aside, there is a crucial, policy-relevant difference between an "involuntary risk" from, say, air pollution and a voluntary risk such as the potential health hazards of drinking. In a free society, the former is the government's business, while the latter is not.
The report cites research suggesting that "the public is willing to accept risks from voluntary behaviors that are 1,000 times greater than the risks from involuntary behaviors." In Australia and the United Kingdom, the authors note, the government has "used an alcohol-attributable lifetime mortality risk of 1 per 100 people" in "determining the threshold for acceptable risk." But "this level of risk," they complain, "is substantial and seems incompatible with public health objectives." They ruefully admit that "society" may be "more willing to accept a higher risk of death associated with alcohol consumption compared to other voluntary activities."
If you are looking for evidence of anti-alcohol bias, this discussion, together with the report's general tone and its treatment of methodological issues, will help you make your case. But the most striking thing about these comments is the unexamined collectivist assumption that decisions about drinking depend on what "the public," "society," or the government deems "acceptable." In reality, those decisions are made by individuals, who may consider official advice but do not necessarily view it as decisive.
Crazy as it may seem to the ICCUPD, people may indeed perceive "a beneficial tradeoff" from alcohol consumption, which could extend beyond measurable health effects to include relaxation, conviviality, or even the pleasure of tasting a good wine or whiskey. Americans may therefore choose to continue drinking regardless of what the Dietary Guidelines say. I bet they will.
