NIH’s National Institute on Alcohol Abuse and Alcoholism (NIAAA) has funded a randomized trial comparing moderate alcohol consumption and abstention among adults age 50+ who are at above-average risk of cardiovascular disease. Although NIAA is the funder of record, the majority of the money is actually from the Foundation for the NIH (FNIH), which received $67.7 million for the study from five of the largest global beverage companies: Anheuser-Busch InBev, Heineken, Diageo, Pernod Ricard, and Carlsberg. The New York Times’s Roni Caryn Rabin recently reported on communications between NIAA staff, the study’s principal investigator, and these funders that make it harder to have full confidence in the study’s eventual findings.

The potential health benefits of moderate alcohol consumption have been widely reported, although there is also evidence that it increases the risk of breast cancer (and, of course, heavy drinking and drunk driving are well-known public health problems). However, most of the research comes from observational studies. Those who drink moderately are probably different enough from those who don’t that it’s hard to know how much of the difference in their outcomes is attributable to alcohol itself vs. the characteristics that shape their alcohol habits. A randomized trial is the best way to identify differences in health outcomes that can be attributed to the alcohol consumption itself.

The MACH15 study

The Moderate Alcohol and Cardiovascular Health Trial (MACH15) aims to enroll 7,800 participants from sites around the world and follow them for approximately six years. Those randomized to the active comparator arm will be instructed to consume one serving of alcohol – beer, wine, or spirits – each day, with partial reimbursement, while those in the control arm are told to avoid alcohol.

Participants must have consumed at least one drink in the past five years and must not be heavy or problem drinkers. Participants must also be at high risk for the occurrence of a new cardiovascular disease event (based on AHA/ACC risk score or medical records of clinical or subclinical CVD), but exclusion criteria include a long list of other diseases – so, for instance, participants can’t have had colon or liver cancer, and neither they nor their mothers or sisters can have been diagnosed with breast cancer that required surgery or chemotherapy. Outcome measures include cardiovascular disease (myocardial infarction, ischemic stroke, hospitalization for angina, or coronary/carotid revascularization), diabetes (progression of normoglycemic status to pre-diabetes, or pre-diabetes to diabetes), and death from any cause.

The study’s principal investigator is Kenneth J. Mukamal of Beth Israel Deaconess Medical Center, and study locations include the Johns Hopkins ProHealth Clinical Research Center, Wake Forest Sticht Center on Aging, Julius Clinical in the Netherlands, and the Center for Bioethics and Research in Nigeria.

Solicitation of industry funding

Roni Caryn Rabin reported in July 2017 that this $100 million clinical trial was underway, supported by $67.7 million from Anheuser-Busch InBev, Heineken, Diageo, Pernod Ricard, and Carlsberg. She quotes various experts who are concerned about the industry funding or study design, including NYU nutrition and food studies professor Marion Nestle, who warned, “Research shows that industry-sponsored research almost invariably favors the interests of the industry sponsor, even when investigators believe they are immune from such influence.” NIAA director George F. Koob told Rabin the trial would be immune from industry influence. PI Mukamal told Rabin he wasn’t aware that alcohol companies were providing financial support, stating, “We have had literally no contact with anyone in the alcohol industry in the planning of this.”

Rabin found evidence to the contrary. Her March 2018 story on the trial relies on emails and travel vouchers obtained through FOIA requests, as well as interviews with some of those involved. In 2013 and 2014, she reports, NIAA staff arranged meetings between scientists and industry representatives:

The presentations gave the alcohol industry an opportunity to preview the trial design and vet the investigators. Indeed, the scientist leading the meetings was eventually chosen to head the huge clinical trial.

They also made the industry privy to pertinent details, including a list of clinical sites and investigators who were “already on board,” the size and length of the trial, approximate number of participants, and the fact that they could choose any beverage. By design, no form of alcohol — wine, liquor or beer — would be called out as better than another in the trial.

… But [NIAA director George Koob’s] predecessor, Dr. Ken Warren, who helped organize and participated in some of the meetings as acting director of the alcohol abuse institute, acknowledged in an interview that the scientists’ presentations were meant to both “demonstrate to the industry that the study was feasible” and “to determine if they had interest in taking part” as funders.

… In an interview, Dr. Lorraine Gunzerath, a retired senior adviser to Dr. Warren, took credit for coming up with the idea of reaching out to the alcohol industry for funding.

… But, referring to the alcohol industry, Dr. Gunzerath said, “If we had a clinical trial, and it was a positive result — which we thought it might be, you sort of think you know where it’s going — they’d be happy.”

All the N.I.H. had to do was “make a business case to the industry that it would be to their benefit, even if they couldn’t actually control the trial’s outcome,” Dr. Gunzerath said.

Gunzerath also told Rabin that the study couldn’t have happened without industry funding, and that if the funders hadn’t liked research team suggested at the meetings, they would have declined to support the study.

Influence and confidence

Rabin reports that Mukamal “emphasized in an interview that he was committed to reporting the results accurately based on the data.” This is an important point. NIH has enhanced transparency by requiring trials that receive agency funding to register their protocols prospectively and report summary results at ClinicalTrials.gov. Publication in a peer-reviewed journal can bring the findings to the attention of clinicians and researchers, but any member of the public can access the basic information at ClinicalTrials.gov (MACH15’s page is here).

My concern with this study isn’t primarily that investigators will meddle with the data or fail to publish findings unfavorable to the funders. What I’m most worried about is the possibility that knowledge of the need for industry sponsors could have influenced how the study was designed. For instance, past trials that randomized participants to moderate alcohol consumption or abstention have supplied participants with one type of alcohol (Everclear in the Postmenopausal Women’s Alcohol Study and separate groups receiving red and white wine in the CArdiovaSCulAr Diabetes & Ethanol trial). By contrast, MACH15 allows participants to drink beer, wine, or spirits. Such a wide range of options might make sense for a trial that’s much larger and longer-running than those studies. Mukamal and colleagues make the case for multiple alcohol options in a 2016 article in Alcoholism: Clinical & Experimental Research, noting that trials have found feasibility and adherence to be best when studies use beverages most culturally acceptable to the study population and request less-extreme dietary changes.

It’s entirely possible that investigators unaware of the likely role of industry sponsors would also have concluded that for best recruitment, adherence, and retention, participants in the alcohol arm should be allowed to drink beer, wine, or spirits. But I can’t help wondering whether if faced with compelling cases for a trial allowing beer or wine consumption only as well as for a trial that allowed all three kinds of alcohol, those involved in the decision might have decided to go with the option that would be more attractive to Diageo (maker of Johnny Walker, Smirnoff, Captain Morgan, and Tanqueray) and Pernod Ricard (whose brands include Beefeater, Chivas Regal, Absolut, Havana Club, and Jameson).

Rabin’s articles include comments from experts expressing concerns about other aspects of the study, such as the inclusion and exclusion criteria. With all of these elements, it’s possible that investigators unaware of potential funders would have made the same decisions after considering all the tradeoffs involved – and the project did undergo NIH’s review process, which includes evaluation by a study section of experts. But many of us will wonder about how knowledge of the need for industry funding might have subtly affected one or more of the many decisions that go into this kind of trial. As a result, we’ll have less confidence in the results once they’re published.

Policies and investigations

Three days after the Times published Rabin’s article about the 2013 and 2014 meetings NIAA arranged between scientists and industry representatives, a follow-up article (also by Rabin) reported that NIH director Francis Collins announced an agency investigation into whether the officials involved violated agency policy in their solicitation of industry funding, and a pending request for a standing advisory committee to review the study’s design and methodology.

In her article about the 2013 and 2014 meetings, Rabin links to section 1135 of the NIH Policy Manual, Gifts Administration. That policy states (emphasis added):

Currently, the only authorized financial intermediary for a third party donation of funds is the FNIH which operates under explicit statutory authority to solicit, accept, invest, and manage third party donations to support the NIH in its mission. … In the use of gift acceptance authority, employees and managers must determine whether acceptance of a gift would compromise or appear to compromise the integrity of the NIH or any of its employees.

The policy also refers to NIH Manual Chapter 1167, Public-Private Partnerships (PPPs) for details about how transfers of funds from FNIH are governed. That chapter states that each Institute or Center (IC) “is responsible for evaluating potential partnerships in the specific context of the IC mission” and provides some “guiding principles in the form of questions that may be useful in assessing whether or not to enter into a particular PPP.” Among those questions is this one:

Has there been an assessment of other business practices or organizational activities of the potential partner that may represent or suggest a direct or indirect conflict with NIH’s mission and/or NIH or HHS policies?

As these manual chapters make clear, it’s not just a question of whether a gift requires a compromise or represents a conflict – institutes also need to consider whether a gift might create the appearance of compromise or suggest a conflict. These things matter in how the results are received.

I’ll be interested to learn the findings of the NIH investigation into this matter. It may be that NIAA staff did not follow the appropriate procedures, and/or that the Obama administration officials in charge at the time failed to exercise appropriate oversight. Stronger policies or better training might be recommended to reduce the risks in future studies.

It’s impossible for us to know whether this study could have eventually happened without industry contributions; recently, Congress has become more generous with NIH funding, and spending-bill language sometimes encourages research on specific topics. We also can’t know whether a solicitation to beverage companies that didn’t include specifics about the study design or investigators, or suggestions about what the findings would likely be, would have resulted in any contributions. Where we’re left now is with an ambitious trial involving a great deal of effort from thousands of participants and many researchers. The findings will be important but will always have an asterisk of potential inappropriate influence attached.