In vino veritas — Part Two

In vino veritas — Part One here.

This past week, women around the world were frightened by hundreds of news stories, all telling them that even a single alcoholic drink a day was dangerous and could increase their risks for cancer. If we believed the media, the largest study ever conducted had found no amount of alcohol consumption to be safe.

This is a case where the evidence could not support the anti-alcohol message, even after reworking the data and ignoring the fact that nondrinkers were associated with higher risks for all cancers than 95% of drinkers. The public was left with the impression the study found something completely different than it actually had.

The source of this news was a study published in the Journal of the National Cancer Institute, led by epidemiologist Naomi E. Allen at the University of Oxford. This report was a secondary computer analysis of data from the Million Women Study, sponsored by Cancer Research UK. The Million Women Study had recruited about 1.33 million healthy British women, aged 50 to 64 years (average 55), having breast cancer screenings through the UK National Health Service Breast Screening Programme from May 1996 to 2001. This observational study had originally been designed to study hormone replacement therapy and mortality from cancers, heart disease and stroke. At recruitment and three years later, the women had filled out questionnaires asking social-demographic information, including how many alcoholic drinks they drank on average each week.

For this report, the authors used the database to look for correlations between alcohol consumption and cancer incidences. After excluding the women who already had cancer and those with missing information, the authors had 1.28 million questionnaires for their analysis. The study participants were flagged on the National Health Service central registry and deaths and incidences for 21 cancers were monitored for an average of 7.2 years. [Invasive breast cancers were not differentiated from other breast cancers coded on the medical records.]

Women were categorized into five groups, said the authors, according to the total number of alcoholic drinks they reported consuming per week on their recruitment questionnaire. Overall, the women had low to moderate alcohol consumption, the authors found: 24% reported being nondrinkers, 29% said they drank 2 or less drinks each week, 23% reported drinking 3–6 drinks per week, 19% reported drinking 7–14 drinks per week, and only 5% reported drinking 15 or more drinks every week. The average alcohol intake was 4.4 drinks per week among all the women and 7.1 drinks per week among the drinkers. The prevalence of women actually drinking heavily was so low, the authors said, they were unable to assess the risks of heavy, sustained drinking on cancer.

There were 68,775 incidences of all cancers among the women during those years. None of the risks for cancers associated with any of the alcohol consumption levels were tenable. This was actually a negative study. It found that among more than 1.2 million women there was no credible link between alcohol consumption and all cancers.

But the news and authors are claiming otherwise. According to the widely syndicated article from the Washington Post:

A new study involving nearly 1.3 million middle-age British women - the largest ever to examine whether alcohol increases a woman's risk of cancer - found that just one glass of chardonnay, a single beer or any other type of alcoholic drink per day poses a danger. “That's the take-home message,” said Naomi Allen.

“If you are regularly drinking even one drink per day, that's increasing your risk for cancer….There doesn't seem to be a threshold at which alcohol consumption is safe.”

Except, that is not what the study found. Besides finding no tenable correlations, this was an epidemiological study showing associations, and correlation cannot imply causation (as in “increasing your risk”). Another glaring inaccuracy to claims of a link between cancer and alcohol was that there was no dose response between the number of drinks the women consumed and their risk for all cancers. Women drinking no alcohol at all had higher incidences for all cancers than 95% of the drinking women.

The actual incidents of all cancers was 5.7% among the nondrinkers. The cancer incidents were lower among the women drinking up to 15 drinks a week: 5.2% among those consuming ≤2 drinks/week; 5.2% of those drinking 3-6 drinks/week; and 5.3% among those drinking 7-14 drinks a week. [Table 1.]

In other words, women drinking as many as two drinks a day were associated with lower actual incidences of all cancers compared with the nondrinkers.. This is the exact opposite of what has been widely reported. Although the differences in actual cancer risks were tiny, nondrinking was associated with a 0.4% higher incidence of all cancers compared to women drinking two drinks a day.

The incidents of all cancers among the women drinking the most alcohol (15 or more alcoholic drinks a week), who made up only 5% of the cohort, was 5.8%, nearly the same as the nondrinkers. These women differed from the other groups of drinkers in other ways, such as being more likely to smoke, to exercise strenuously, have used oral contraceptives and be currently using hormone replacement therapy, have higher socioeconomic status, and lower BMIs. So considerable potential confounding factors are involved in trying to interpret this correlation.

In deriving relative risks [Table 2], Allen and colleagues then used as their reference level the women who drank two or less drinks a week — this group was given a relative risk of 1.00 (null). Entering into the realm of untenable splitting hairs, their computer modeling revealed that nondrinkers still had a 4% higher relative risk for all cancers than the women drinking two drinks a week, 2% higher than women drinking up to six drinks a week, and nearly identical risks to women drinking as many as 14 drinks a week.

This is clearly not the politically acceptable public message.

The authors then excluded all of the women who reported not drinking any alcohol and re-analyzed the data using only the drinkers, looking for correlations between alcohol intake and various cancers. They still found no tenable correlations with alcohol consumption and any of the cancers:

All of the relative risks hugged either side of null and none proved beyond random chance, statistical error or attributable to other factors. In fact, in contrast to suggestions made in other epidemiological studies mostly on male heavy drinkers, the authors stated, among nonsmokers in this study, taken together “there is little or no effect of moderate alcohol consumption on the risk of cancers of the upper aerodigestive tract.”

Nor did they find that the type of alcoholic drink — wine, spirits or beer — made any difference.

The most glaring omission in this study was the failure to report all-cause mortality, despite the fact the authors had that information. Other research has consistently shown lower rates of heart disease and all-cause mortality associated with alcohol consumption.

But, as the Washington Post noted, “officials have long worried about sending the ‘wrong message’” about drinking, even though “it’s true that studies have indicated that moderate drinking may cut the risk of heart disease and other ailments.” The newspaper went on to report:

“[Moderate drinking] is a level of consumption that generally has been found in scientific studies to be associated with a relatively low risk of harms,” said Robert Brewer of the federal Centers for Disease Control and Prevention.* “But low risk does not mean no risk.”

* The newspaper failed to disclose that Dr. Brewer is manager and director of the Alcohol Team at the CDC; serves as Principal Investigator for Alcohol-Related Disease Impact software, funded by the Robert Wood Johnson Foundation; and has authored numerous publications on alcohol and alcohol-related health effects.

It appears the anti-alcohol agenda overtook accountability to accurately report the study’s actual results. By carefully-worded reporting of the study, women have come away believing this study found that a single drink a day increases their risk for cancer. But read closely. For example, PBS reported the study as showing:

They found that drinking just one alcoholic drink per day increased the risk of breast, liver and rectal cancer. For women who also smoked, drinking increased the risk of mouth and throat cancer as well.

This and other news stories took their lead from the press release sent out by the journal, which said:

Low to moderate alcohol consumption among women is associated with a statistically significant increase in cancer risk and may account for nearly 13 percent of the cancers of the breast, liver, rectum, and upper aero-digestive tract combined, according to a report in the February 24 online issue of the Journal of the National Cancer Institute.

Did you catch the omissions? This study had examined 21 different cancers, yet the news gave us only part of the story. They didn’t tell us that none of the associations were tenable, nor that the study had also found that women drinking one to two drinks a day were associated with lower risks for other cancers — of esophageal adenocarcinoma, stomach, malignant melanoma, cervix, ovary, renal cell carcinoma, bladder, brain, thyroid, non-Hodgkin’s lymphoma, multiple myeloma and leukemia — with no overall cancer risk.

The bottom line is that scary claims that “there is no level of alcohol consumption that can be considered safe,” simply was not supported by the data. This study actually found no credible link between alcohol consumption and cancers at all. Or, if you want to split hairs and believe the small computed numbers, it found that the lowest risk for cancers was associated with women drinking up to 1-2 drinks a day.

© 2009 Sandy Szwarc