1st BATCH OF FILES AVAILABLE NOW!

Over a year after sweeping cuts at the Dept of Health & Human Services effectively emptied out many alcohol-focused corners of the US government public health infrastructure, a certain cadre of cast-aside researchers returned with a vengeance. This week, a journal that has long favored the work of anti-industry researcher-advocates published a previously "sidelined" study designed to push US drinking guidance closer to a "no safe level" stance. And so the infamous Alcohol Intake & Health (AIH) study coordinated by the Interagency Coordinating Committee on the Prevention of Underage Drinking (ICCPUD) returned, albeit in a drastically altered form. And though the six core authors of the work made quite a splash with the re-release, their actual analysis continues to belie their insistence that they found "no protective effect at low levels" of consumption. Doubling Down on "No Protective Effect" The language authors use to discuss their results breaks from their data, just as it did when they released the draft version of the study in January 2025. This time, the researchers doubled down on their insistence that "there was no protective net effect of alcohol observed at any level of alcohol consumption." In fact, the study headlines "No Protective Effect at Low Levels" in the version published by the Journal of Studies on Alcohol & Drugs (JSAD) this week. Co-author Dr Katherine Keyes repeats the assertion in a press release issued by her institution, Columbia University's Mailman School of Public Health. The research "overturns a common misconception that alcohol can protect health," the release continues, insisting that "potential benefits are outweighed by the risks even at 7 drinks per week." Ref 1But once again, it glosses over the finding that below 6-7 drinks per week benefits do outweigh the risks. The headline assertions of the study are based on the risk curves illustrated in a pair of charts that plot the estimated alcohol-attributable deaths per 1,000 people at various numbers of average drinks consumed per week (assuming drinks of 13.6 g of ethanol). Authors generated separate risk curves by sex, identifying a "point estimate," or an average, for each. Figure 1 in the study (reprinted below) plots these averages as well as the ranges within which authors are 95% confident the actual results fall.Authors emphasize that the average curves cross the 1 death per 1,000 threshold at 6.5 drinks per week for males and 7 drinks per week for females. They also write that "at 14 drinks per week for males, the risk of an alcohol-attributable death was approximately 4%," which is a lot in their view. Female drinkers consuming half that much, 7 drinks per week, have just "a 0.01% chance" of dying from an alcohol-attributable cause, the authors wrote. While the findings are the same, none of those acknowledgments appear so clearly in the draft published last year. (Figure 1, taken directly from the study as published by JSAD.)Low Confidence and a Hidden J Furthermore, neither the draft nor the new version of the study pay much attention to what happens at even lower levels of consumption. When results come up negative (fewer than 0 deaths per 1,000 people), that implies a "protective" effect, or a net benefit to health. Researchers acknowledge that they identified "small protective associations at up to 3 drinks per week for females and males," but that they were "not statistically significant."That's in part because of the wide range of "confidence intervals," indicated on the charts by shading on either side of the average. In other words: the real results could land anywhere in the shaded regions. And the upper bounds of those possible results suggest 5-10 deaths per 1,000 even at 1 drink per week. Yet authors never acknowledge the opposite uncertainty. The lower bounds of the ranges do not cross 0 until between 11 and 12 drinks per week for female drinkers and 13 to 14 drinks per week for male drinkers. That implies the possibility of net health benefits at up to 1.5-2 drinks per day.Take one more look at the far left side of the risk curves above to notice the gap in the curve between 0 and 1 drinks per week. Presumably, the number of people who consume 0 drinks per week and die from an alcohol-attributable cause is 0. But the authors don't connect those dots. If they did, their curve would be J-shaped, or the kind of curve that implies net health benefits at low levels of drinking, just like hundreds of studies demonstrated over many decades and which these authors and others work to "debunk."Modeling Makes for "Weak-to-Moderate" Quality This is all assuming that the data analysis completed by the authors can be relied upon. And that is an open question. Recall, they use quite a lot of "statistical modeling" to arrive at their results, not all of which is explicitly explained. They also employed "expert panels" to determine which of 7,200 prior articles should be used to develop risk curves for each individual health outcome. That list was first whittled down to 56 studies and panelists ultimately decided to use just 16, authors acknowledge. And though they explain their process for selecting panelists, the process used by each one to select their favored studies remains a black box. (Note too that the 6 core authors are joined by 19 others in this version of the study, presumably adding the previously anonymous panelists as co-authors, though that isn't explicitly stated.)These unknowns and the extensive use of modeling earned the study a "methodological quality" rating of "weak-to-moderate" from the International Alliance for Responsible Drinking (IARD). The org's research team summarized that "the authors' conclusions and discussion overstate the implications of the study's results, the strengths of its methodology, and do not sufficiently reflect the uncertainty of the risk estimates."Higher Risk for Men Driven by Questionable Injury Assumptions One area where the authors' modeling is particularly controversial and heavily impacts overall results is their inclusion of injuries from motor vehicle crashes, interpersonal violence, poisonings, suicides and more in their analysis. They use their own calculations to shift from prevalence of those harms and average consumption into risk curves comparable to those determined by more traditional epidemiological studies. Charts combine and stratify risk curves by individual causes of death, demonstrating just how much estimates for injuries impact the assumed risks, especially for male drinkers. Authors somehow model that the risk of those injuries increase in an essentially linear fashion beginning at just 1 drink per week. And those assumed risks pile up quickly.But these are the kinds of acute alcohol-related harms that most drinkers understand well. And some critics argue that these outcomes have no place in an analysis designed to influence dietary guidelines. Factor those out and the cardiovascular benefits of light-to-moderate drinking wipe out other chronic health risks until around 14 drinks per week for men and around 10-11 drinks per week for women, according to the study.Urges One Drink Per Day Guidance for All, Emphasis on Binge Drinking Though the data published this week is mostly the same as the draft report published in early 2025, this "final" version reveals just how rough that previous draft was. The study's introduction, explanation of its methods and discussion of its import were all fully rewritten, the results completely reorganized, with some findings left out entirely. Among the clearest changes is its embrace of a one drink per day recommendation for all adults, regardless of sex.Recall, under the auspices of ICCPUD and as part of the DGA process, the draft report explicitly avoided making specific recommendations. Divorced of that context, the new study gets straight to it: "The current results support changing the 2025-2030 US Dietary Guidelines on alcohol to recommend that current adult drinkers consume no more than 1 drink per day," the study reads. It's too late for that, of course. But that didn't stop authors from joining the scores of other health researchers and advocates aiming to undermine the new edition of the DGA and general health guidance under the Trump Admin.The study's policy recommendations don't stop there. "The reduction in alcohol-related harms requires population-level reductions in consumption, which are best achieved by implementing effective alcohol control policies, particularly in the areas of pricing and physical availability," the authors wrote. Though unstated here, such control policies have played no role in the "relatively rapidly declining" consumption trends observed in just "the past 2 to 3 years," as described by co-author Dr Keyes in a brief note published by the journal Addiction earlier this year. Ref 2After those more expected policy recommendations (like 1 drink per day and higher prices), the new study takes a more unusual turn. "Public health policy and programming should prioritize the risks associated with drinking patterns, particularly binge drinking, rather than focusing exclusively on average intake," authors wrote. They go on to recommend support for "routine screening" and mechanisms to better track different drinking patterns.Former ICCPUD Coordinator Attempts to Rewrite History, Blame Industry The updated AIH study isn't the only ICCPUD-related content published by JSAD this week. Accompanying the study is an editorial by former worker at the Substance Abuse and Mental Health Services Administration (SAMHSA), Robert Vincent, who served as ICCPUD's coordinator. In it, he purports to explain why the government never published the final AIH study and why he no longer works for the government. And in his view, the answer to both is simple: industry influence, or "commercial interests," as he calls it.Vincent paints the work he did at the ICCPUD and the AIH study as sound, implying that the industry stopped the work short because it threatened sales, making the usual, obligatory comparison to tobacco. Yet he also describes the loss of his job on April 1, 2025 as part of a "targeted reduction in force." And while it's true that many of the relatively small number of HHS employees who focused on alcohol, like him, lost their jobs that day, Vincent does not acknowledge that so did thousands of others. Far from "targeted," newly appointed leadership at HHS gutted the department in sweeping cuts in Spring 2025.But that's not Vincent's only questionable claim. "I never expected that doing my job&mdash;carefully, transparently, and in accordance with federal law&mdash;would end my federal service. Yet this is precisely what happened," his editorial begins. That might be how he feels like it unfolded. Others present different views of ICCPUD's work on his watch.First, it's not clear that embarking on an analysis of adult drinking was his or ICCPUD's job to do. Vincent echoes the after-the-fact justification that ICCPUD's appointment to do the AIH study "reflected evidence linking parental drinking to adolescent alcohol use and early initiation." That's not how the project began, though. It was a suggestion by ICCPUD's key outside contractor and former chair of anti-industry advocacy group the US Alcohol Policy Alliance, Alicia Sparks, who suggested ICCPUD oversee "dietary guidelines research" that "largely follows the Canadian model," according to meeting notes published as part of a House Oversight subcommittee report early this year.That report also described the study as "irretrievably flawed." Recall, the subcommittee published those findings after spending over a year trying to get information about ICCPUD's effort and the AIH authors operated in secret for many months both before and after the rough plan for the project became public. That undermines Vincent's claims about how he did his job. It's worth noting, too, that it wasn't just ICCPUD's operations that lacked transparency; the AIH continues to suffer from it (as noted above). And it wasn't just industry advocates who spotted its major scientific flaws. The Center for Science in the Public Interest (far from a friend of the industry) took issue with both the AIH's modeling techniques and found omissions of crucial details "especially problematic."There's plenty more to question in Vincent's editorial. But his closing "call to action" section repeats an especially dubious claim also embraced by certain authors of the AIH. Last year and early this year, some of the 6 authors attempted to describe the findings of the report as settled science. Here, Vincent claims that "the challenges confronting alcohol policy today are not rooted in scientific uncertainty."Researchers focused on the topic feel differently, however. In fact, while the NY Times and other outlets repeated Vincent's claims and the AIH study's findings this week, most ignored another large review of the evidence published late last month. And the abstract of that paper begins: "The relationship between alcohol and health is complex, and the evidence relating alcohol consumption to various cardiovascular diseases, cancers and other conditions is evolving."Stay tuned for a closer look at the findings of that paper and a couple of other recent publications that undermine the AIH study's assertions. Cheers,References1 George, S, et al, "Alcohol Intake and Health Study: No Protective Effect at Low Levels, With Mortality Increasing to 1 in 25 at 14 Drinks Per Week," Journal of Studies on Alcohol and Drugs, June 20262 Keyes, KM, "Are Americans sobering up? New trends in alcohol consumption in the United States," Addiction, 2026

ICCPUD Report Resurrected, Rhetoric Ratcheted Up Despite Same Data Disconnects

Publishing Info