FREE! Subscribe to News Fetch, THE daily wine industry briefing - Click Here

Sponsored by:

14 Reasons the new NIH/industry-funded moderate alcohol & health study could be a $100 Million CREDIBILITY flop

Click here for an editorial on the issues raised by this article: Alcohol science versus public credibility – Why MACH15 must correct its protocol in order to be worthwhile

This full premium article has been updated with comments from Dr. Kenneth Mukamal, the Principal Investigator of the study.

By Lewis Perdue, Publisher and Executive Editor


The new $100-million-dollar, government/industry-funded study on moderate alcohol consumption — first reported on by Wine Industry Insight almost three years ago and expanded upon this year by the NY Times — could set a record as one of the most flawed, least relevant, and least credible studies ever to be supported by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) — part of the National Institutes of Health (NIH).

The 92-page study protocol, highlighted by the editor is at this link. A shorter web version can be found at this link.

Key Problems

  1. Healthy participants are specifically excluded.
  2. Only people with diagnosed cardiovascular conditions and risks (including previous strokes and heart attacks) are allowed.
  3. No one below the age of 50 is allowed. That minimizes the relevance to most of the population.
  4. Industry funding and previous industry support of study investigators will deal a blow to credibility
  5. The drug treatments for CVD will vary from subject to subject. No way to separate effects of alcohol consumption/abstention from CVD pharmaceuticals and treatment.
  6. Women are mistakenly treated as men for alcohol dosage.
  7. Tobacco smokers are allowed.
  8. Cannabis use is not addressed despite the fact that 24% of the test demographic are users and the health effects of combined alcohol and marijuana use is unknown.
  9. Fails to consider the effects of lifelong moderate consumption.
  10. Vague or no requirements or controls regarding time or circumstances of consumption, diet, or exercise.
  11. Industry funding offers an appearance of conflicts of interest that assures the study will not be accepted as a valid, unbiased study by a large number of the general public, along with physicians, public policy makers and regulators.
  12. Potential compliance issues with smokers or reformed smokers.
  13. No provision for test subject bias or placebo effect.
  14. The study leaves a huge amount of science undone.
  15. No data sharing with outside researchers: thwarts transparency, no reproducibility, damages credibility. (Protocol on government website very recently corrected. See full text with screen capture and Principal Investigator Comments)

People aged 50 years and older with cardiovascular issue only — Healthy people excluded.

According to version 1.0 of the study’s 92-page pdf filed with the NIAAA, the six-year “Moderate Alcohol and Cardiovascular Health Trial” (MACH15), will follow approximately 7,800 non-healthy people who have consumed at least one alcoholic beverage in the previous five years. The study is restricted to participants aged 50 years or older who have been diagnosed with cardiovascular disease (CVD) or above-average risk.





The MACH15 protocol randomly assigns study subjects to one of two cohorts: an abstainer group which must refrain from consuming alcohol for the six-year period and a consumer group which must consume one (but no more than one) alcoholic beverage daily with approximately 15 grams of ethanol.

The study’s two objectives are to determine any difference in outcomes involving CVD and the development of diabetes between the groups. The primary outcome studied is the timing of a non-fatal CVD event or death from any cause.

Limited relevance: Sick people only

With no historical data on previous alcohol consumption levels or patterns the study fails to investigate how much (or little) should healthy people of all drinking ages consume to remain healthy? Buy focusing on an unhealthy, over age 50 cohort only, the results may offer relevant medical recommendations only to that select, sick age group.

The MACH15 document suggests that the results may be relevant for healthy adults and younger members of the population. However, no data is offered to make that claim credible.

In addition, there are no healthy test subjects as controls for the study.

Screen Shot 2017-11-28 at 8.51.51 AM

Sick people are good from a practical standpoint: They get sicker faster and die quicker (high background rate of events)

The following is the MACH15’s complete Rationale for selecting unhealthy, older people. Paragraph breaks have been added for readability along with emphasis and comments.

“Although epidemiologic evidence generally suggests that alcohol consumption tends to be associated with lower risk of coronary heart disease across a wide variety of populations, ethical, practical, and clinical considerations suggest that this hypothesis is most efficiently studied in high-risk individuals.

“From a practical standpoint, a high-risk population with a high background rate of events requires a smaller potential sample size, minimizing the number of participants needed to recruit and follow.

“This population also tends to have more extensive contact with health care systems, improving recruitment, and may have a particularly vested interest in cardiovascular prevention strategies, which may enhance adherence.”

[Given previous studies and media attention to the potential benefits of moderate consumption, participants assigned to the teetotaler cohort may likewise have a vested interest to cheat and have an occasional drink]

“Ethically, a high-risk population minimizes the number of participants placed at risk in a randomized trial, and with declines in problem drinking behavior with older age, tends to reduce potential harms from alcohol while maximizing potential benefits.

“From a clinical and public health perspective, this population is also the most likely to benefit if the primary or secondary hypotheses of MACH15 are confirmed, for these high-risk individuals stand to benefit the most from any given decrease in relative risk and would therefore have the most favorable risk-benefit ratio for alcohol consumption.

“Concordantly, results in a global, diverse, high-risk population are apt to be generalizable to a wide variety of lower-risk populations in terms of relative risk reduction, even if any absolute risk reduction is necessarily smaller.”

[“Apt to be” is not a scientific term unless backed up by credible data. A footnote would be useful here to define the likelihood and magnitude of “apt.”]

Dr. Mukamal commented:

“It is, of course, routine to study difficult, complex exposures in high-risk populations, both for practicality and because the issue of whether alcohol prevents (or causes) coronary heart disease is most salient for those individuals. That is how virtually every dietary factor ever proven to influence risk of heart disease has been tested; for example, the widely-touted Mediterranean diet has been unequivocally proven to lower risk of heart disease, but only in high-risk primary and secondary prevention like those enrolled here.”

Study & industry funding first reported by Wine Industry Insight.

As first reported by Wine Industry Insight, in 2014, the NIAAA actually solicited funds and approval of a beer and spirits-funded clinical study on the health effects of moderate alcohol consumption that is now estimated to cost $100 million.

Documents provided to Wine Industry Insight in 2014 originally estimated the funding at $36 million to $54 million.

The current funding will be provided through the NIH Foundation which is a separate 501(c)(3) corporation set up to accept non-governmental outside funding. The Foundation has solicited funds from beer, spirits, wine, insurance and drug companies.

The analysis of funding sources is unclear due to the lack of transparency of the NIH Foundation which is a private non-profit which raises money for the NIH. It has no legal requirement to list the names of donors.

According to the most recent publicly available IRS Form 990’s filed by the NIH Foundation (2015 and 2014 tax years), it has assets of more than $74 million and in the past two years has received $74,5 million and $34.6 million, respectively.

Images below from ProPublica Non-Profit Explorer which has full form 990s back to 2001.

Screen Shot 2017-11-28 at 3.13.36 PM

All donors are anonymous.

Screen Shot 2017-11-28 at 3.25.32 PM

The projects supported are described in nearly impossible-to-read type designed to deter interested parties

Screen Shot 2017-11-28 at 3.08.37 PM

Industry funding delivers a credibility blow from the beginning.

Wine Industry Insight’s 2014 article touched on potential credibility issues associated with the alcohol industry’s previous financial support of this study’s investigators.

Roni Rabin at the New York Times found multiple instances where MACH15’s scientists had previously received funding from the alcoholic beverage industry

From Rabin’s July 4, 2017 article:

“Five companies that are among the world’s largest alcoholic beverage manufacturers — Anheuser-Busch InBev, Heineken, Diageo, Pernod Ricard and Carlsberg — have so far pledged $67.7 million to a foundation that raises money for the National Institutes of Health, said Margaret Murray, the director of the Global Alcohol Research Program at the National Institute on Alcohol Abuse and Alcoholism, which will oversee the study.

“The decision to let the alcohol industry pay the bulk of the cost has raised concern among researchers who track influence-peddling in science.”

More specific information from Rabin about the study’s investigators at the New York Times: “Is Alcohol Good for You? An Industry-Backed Study Seeks Answers”

Americans have a dim view of industry-financed studies

While there are no formal conflicts of interests by the investigators running this trial — not as defined by the NIH or other research organizations — the funding by an industry with a stake in the results offers the appearance of a conflict which affects the general public’s view of potential bias.

Industry funding is perceived as a conflict by the general public who are far more skeptical of a study’s outcomes and conclusions. That means that its final results — regardless of its conclusions — will not be seen as a valid, unbiased study by a large number of the general public, along with physicians, public policy makers a regulators.

Screen Shot 2017-11-28 at 10.23.29 AM

The chart, above, is from Smithsonian Magazine (People Don’t Trust Scientific Research When Companies Are Involved) which created the graphic based on data from a study funded by Michigan State University: (Perceived conflict of interest in health science partnerships).

This 2009 book chapter from the Institute of Medicine (US) Committee on Conflict of Interest in Medical Research, Education, and Practice, offers a deeper look: Conflicts of Interest in Biomedical Research.

Drug and treatments regimes for CVD will vary and confound results

Selecting only unhealthy trial subjects means that most or all will come to the study with existing treatment and pharmaceutical regimes. Those will vary from patient to patient depending upon the specific condition, state of illness and treating physician. Such variables range from the effects of highly specialized pharmaceuticals which have their own set of interactions to more common drugs such as statins, aspirin, and the reported deleterious CVD effects of some other NSAIDs.

Separating data regarding effects of moderate alcohol consumption from the effectiveness of a medical treatment regime will be impossible.

Dr. Mukamal commented:

“Large randomized trials differ substantially from small, highly-controlled feeding studies. Randomization itself controls for all measured and unmeasured confounding factors, so long as the trial is large enough. It provides an unbiased, real-world answer irrespective of other cardiovascular risk factors, including diet, exercise, smoking, social status, etc.”

Fails to consider the effects of lifelong moderate consumption

Cardiovascular disease (CVD), for most people, progresses slowly over a period of decades. If moderate consumption has a preventative effect for CVD, it is possible that a consumption may need to take place over as many years as CVD develops and not over the limited lifespan of this study.

Several dozen observational studies have concluded that people who drink alcohol in moderation have a lower overall death rate than abstainers or heavy drinkers. However, none of those has been acknowledged as sufficiently valid by governmental bodies and anti-alcohol advocates.

Dr. Mukamal commented:

“Several observational studies have already assessed the impact of changes in alcohol intake and their time course on subsequent risk of CVD and suggest effects well within the timeframe of MACH15; those have been summarized in at least one review as well.

“While a longer study would offer a greater opportunity to determine the effects of alcohol intake on the full progression of atherosclerosis and subsequent coronary heart disease, it would also require randomized assignment for a far longer period of time. As a result, any trial must balance practical demands on participants with the expected time course of effect.”

Fails to recognize that men and women are not physiologically identical

The fact that government regulators recommendation a maximum of one drink a day for women versus two for men is a macro observation that women are different. However, more than a simple size difference, women tend to have a higher percentage of weight in adipose tissue. Alcohol, on the other hand, is most soluble in water which means that on a pure molar level, a given amount of alcohol will result in a higher blood alcohol concentration in women versus men of a similar weight.

In addition, women’s alcohol consumption is closely linked to breast cancer via a number of estrogen cellular pathways that are still being investigated. For one pathway, see, “Tamoxifen represses alcohol-induced transcription of RNA polymerase III-dependent genes in breast cancer cells.”

However, a 2016 study, “Alcohol intake and breast cancer in the European prospective investigation into cancer and nutrition” found that

“Taking 0 to 5 g/day as reference, alcohol intake of >5 to 15 g/day was related to a 5.9% increase in breast cancer risk (95% CI: 1–11%). Significant increasing trends were observed between alcohol intake and ER+/PR+, ER−/PR−, HER2− and ER−/PR−HER2− tumors.”

Given the results of that study, the 15-grams of alcohol which women would be receiving in the study would raise their cancer risk.

Dr. Mukamal commented:

“NIAAA and other bodies recommend the same upper limits of intake for women as for men 65 years of age and older, who will comprise an important share of the trial population.”

Tobacco smokers welcome

Accepts tobacco smokers. Entirely plausible that smoking in the “drink a day” cohort could grossly overshadow any effects that alcohol consumption might have … thus biasing results toward the negative.

Because alcohol and smoking are inseparable for some people, smokers who drink and are assigned to the teetotaler cohort may not comply with the protocol.

No protocol adjustment to limit smokers to the national smoking population average of 15.1% percent of the population (CDC: Burden of Tobacco Use in the U.S.). Indeed, given the potency of tobacco contaminates, the protocol should consider limiting the percentage of smokers to an appropriate level adjusted for age, sex, race, and education status using current CDC data.

Potential compliance issues with smokers or reformed smokers

While the study design indicates it it aware of “synergistic effects” between tobacco and alcohol, it does not address the co-morbidity of the two substances, specific neurotransmitter connections, or significant behavior connections between the two substances.

An extensive multi-author, multi-institution study (Including the NIH/NIAAA and VA) published in the Proceeding of the National Academy of Sciences (PNAS) indicates (among other findings) that smoking cigarettes seems to make quitting alcohol harder.

While the PNAS study focused on alcohol dependent smokers, the protocol should recognize and address the possibility that smokers — or recently reformed smokers — who are currently drinkers may have trouble complying if assigned to the teetotaler cohort.

No consideration of cannabis

In addition to tobacco smoking, the study protocol has no provision for dealing with cannabis use.

According to Marist Poll, April 17, 2017: 14% of Americans 18 years of age or older use marijuana regularly. Even more relevant to this study is the fact that nearly a quarter of the study’s target study subjects — Boomers aged 51-69 years — are regular cannabis users.

Screen Shot 2017-11-20 at 9.01.23 AM

Right-click table to view a larger image.

No provision for bias or placebo effect

Will the 19% assigned to the drinking who believe moderate consumption is healthy have better outcomes due to a placebo effect? Will that 19% do worse if assigned to the teetotaler cohort? Will the 26% who believe it is unhealthy do worse if assigned to the drinking group?

Will the outcomes of the study be more accurate if it corrects for possible pro or anti bias and accepts only those who believe it makes no difference?

Significantly more Americans believe moderate alcohol consumption is bad for one’s health (26%) than believe it is good (19%). A slight majority of Americans, 51%, believe drinking in moderation has no effect on health.

Screen Shot 2017-11-28 at 11.23.14 AM

Source: Gallup

Moderate consumer subject randomization & compliance issues

People who have been enjoying one or two drinks a day over an adult lifetime — up and to inclusion in the study, are less likely to comply with the single drink level and even more so if assigned to the teetotaler cohort. They are also less likely to volunteer as test subjects. If life-time moderate consumers opt-out, then that removes them as data points representing that behavior.

One baseline to be established is to determine the percentage of unhealthy study subjects who have been life-long moderate consumers and compare that percentage to a healthy population cohort with the same age profile.

This modified cohort of healthy moderate consumers could serve as an outcomes control.

Failure to assess subject selection criteria

How are people who have had only one or two drinks in the past five years different from those who are already moderate consumers? If years have elapsed since alcohol consumption, was the abstinence prompted by health reasons?

In addition, the United States has a diverse population. Demographic factors such as race, education, and income influence diet as well as eating and drinking habits.

Failure to share data with other researchers damages credibility

Screen capture, above, made Nov. 30, 2017.

No data sharing prevents proper examination of the results and prevents investigations to replicate the results.

If a study can’t be reproduced, it’s not really credible science. And it cannot be reproduced if data is not shared.

Dr. Mukamal commented:

“As with any NIH-funded trial, all data will be deposited in a repository for sharing at the conclusion of a trial. That is not true for industry-led trials and highlights one of many ways that the two differ. We have already been working with an NIH repository to ensure our data standards will synchronize seamlessly with theirs. Any individuals with appropriate ethics board approval will be able to analyze our data themselves.”

Further clarification on web site information from Dr. Mukamal:

“When the information was first submitted, we were still finalizing which repository the data would be stored in; NIH has several.

“That field will be updated, along with other elements of the template, at least once more before participants are recruited, as we plan to add some other information that is not yet posted there.

“Nonetheless, ALL NIH clinical trials must have a data sharing plan, and thus we do as well. In our case, NIAAA has an ongoing relationship with the NIMH data archive (, and thus we have worked closely with the archive to ensure that they are prepared to (eventually) receive our data. It will be fully accessible there.”

A Pragmatic Approach for Reproducible Research With Sensitive Data: “Reproducible research is important for assessing the integrity of findings and disseminating methods, but it requires making original study data sets publicly available.”

Incentivizing Data Sharing and Collaboration in Medical Research: “Sharing research data is a widely embraced ideal of academic medicine. Providing broad access to research data has the potential to stimulate progress in medical science and improve public health by revealing previously overlooked critical information and reducing redundant work. Making data available to peers also encourages researchers to improve the quality of their data collection, helps uncover errors, and increases the reproducibility and confidence of findings.”

Data sharing in clinical trials: An experience with two large cancer screening trials: “Broad sharing of clinical trial data is important for ensuring reproducibility, transparency, and maximal use of the data by the research community.”

Another example of why replication is important in science: “An endless stream of new discoveries makes science thrilling. But, as any seasoned researcher knows, such novelties are worthless unless they can be replicated. ”

Metascience: Reproducibility blues

Social Interaction Effects

Alcohol is frequently consumed in social settings with friends, family, and colleagues. Instead of being consumed like medicine, alcohol serves to facilitate conversation and acts as a common source of conversation and group cohesion.

Removing social support elements from people’s lives has been shown to affect health and longevity: Social relationships and physiological determinants of longevity across the human life span. Among that study’s conclusions: “Physiological impacts of structural and functional dimensions of social relationships emerge uniquely in adolescence and midlife and persist into old age.”

Worthy of particular interest to this study is a 2017 study — “Perceived social support predicts lower cardiovascular reactivity to stress in older adults,” — which concluded that, “The results suggest that quality, but not quantity, of perceived social support predicts reduced blood pressure reactivity to stress in the laboratory.

In relationship to compliance with the protocol, this affects both drinkers and non-drinkers with the greatest potential impact on the teetotaler cohort. Temptation could result in teetotaler drinking or increased consumption among drinkers. Less likely, but possible, may be an inclination to avoid some previous social activities.

Failure to assess/correct effects of BAC and gender

Differences in subject gender, size, and percent body fat will cause BAC variations. If health effects are proportional to BAC, then dosage could make the difference.

No protocol for when or how the “dose” is administered

No protocol for how drink is consumed … with food? Drinking alcohol while eating slows the absorption of alcohol. Effect of Food and Food Composition on Alcohol Elimination Rates in Healthy Men and Women

No protocol for when during a day the drink is consumed: Mid-day? Evening? Possible sleep issues with proximity to bed time. Alcohol and Sleep I: Effects on Normal Sleep

Inadequate control and consideration of diet

No control on diet (“eat healthy”, with no specific definition or advice, is the only requirement other than an optional online survey), no attempt to standardize. Many studies have indicated a connection between cardiovascular and diabetic syndromes and processed and/or fast foods vs fresh, effects of olive oil and/or Mediterranean diet, micronutrients, epigenetics, and the environmental chemicals that leach from plastics.

Screen Shot 2017-11-29 at 5.55.22 AM

Folates — as just one example — can have dramatic effects on health.

While direct connections between folates and alcohol consumption have been equivocal, a 2017 study published in a Nature publication indicated that low folate levels can promote insulin resistance which is an early indicator of Type 2 Diabetes.(Association between Serum Folate and Insulin Resistance among U.S. Nondiabetic Adults).

In addition to its role in diabetes, a 2014 paper in Nature Endocrinology found insulin resistance a factor in in cardiovascular disease development.

Closely associated with folates is the amino acid methionine which is found primarily in meat, fish and poultry. A study of men aged 42-60 indicated that high dietary methionine intake increases the risk of acute coronary events.

Those studies just scratch the surface of the hundreds of other well-conducted studies in the scientific literature indicating important connections between diet and both cardiovascular disease and diabetes.

Failure to more closely control dietary factors could overwhelm the data in this study and confound the statistical significance of the data.

Inadequate control and consideration of exercise

Failure to address confounding science on relationship between EDC and T2 Diabetes.

A July, 2017 study quickly summarizes hundreds of previous studies which have solidly established the links between exercise and health:

“Physical inactivity increases all-cause mortality and has been identified as a modifiable risk factor for cardiovascular disease and type 2 diabetes.”

Failure to more closely control exercise factors could overwhelm the data in this study and confound the statistical significance of the data.

Optional online surveys as currently prescribed in the protocol are insufficiently accurate.

Study protocol wastes golden opportunity, leaves a huge amount of science on the table

Even if all the flaws and omissions above were corrected, the study wastes a golden opportunity to bring a fresh investigative look at the issue. A growing body of research indicates that many other confounding factors can cause or promote both CVD and diabetes. Those include changes in the gut microbiome and environmental chemicals to name just two.

Adding a microbiome test — such as the FDA-approved uBiome diagnostic panel — would add little to the overall panel and offer potential new indicators for the study. Unlike blood tests, the kit is also non-invasive.

In addition, because serum and urine are part of the protocol, three other tests could also shed light on the subject: double-stranded DNA breaks and specific epigenetic profiles.

Dr. Mukamal commented:

“As an NIH-funded trial, interested investigators (whether or not they are part of our current trial team) have the opportunity to collaborate and write ancillary studies to add extra measurements and collect more granular data. That is underway and, in an unexpected coincidence, includes both epigenetics AND microbiome assessments, along with perhaps a dozen others. That is, of course, routine in large trials in which the primary protocol funds the infrastructure upon which other studies can build.”