A few weeks ago, the US Surgeon General issued a new advisory regarding alcohol and its link with cancer. He wanted to convey the mounting evidence that alcohol causes cancer. And it is true. Alcohol can indeed lead to certain types of cancer (refer to the diagram below). However, healthcare is highly nuanced, and this subject is no exception.
The purpose of any advisory is to scare straight the people at high risk, where the evidence is strong, and AVOID scaring senseless the people at low risk, where the evidence is weak and often wrong. It's a bit of a balancing act. Unfortunately, considering the majority of drinkers (60-70%) are light drinkers (up to 1 drink/day for females and 2 drinks/day for males), and heavy drinkers are already aware of the risks, I believe this advisory might be alarming the wrong group. In my own practice, I've received feedback about this advisory only from the light drinkers, not the heavy ones.
Limitations of Observational Studies
Studying alcohol and its effects on health is difficult. Like smoking, we're unable to perform randomized controlled trials, and therefore, we're left to make inferences from weaker kinds of evidence. With alcohol, we mainly use observational studies. These studies are conducted by finding a large group of people, asking them how much they drink, and coming back several years later to see what diseases the drinkers vs the non-drinkers picked up along the way.
Observational studies suffer from two main limitations: people are notoriously unreliable reporters of how much they drink, and drinkers are quite different from nondrinkers, a problem known as confounding. For example, drinkers often smoke more and exercise less, abstainers often stop because they were already sick ("sick-quitters"), and moderate drinkers tend to screen more for cancers (ascertainment bias). All these other factors, not the alcohol, may be driving differences in cancer rates.
In fact, we've already been down this road before. It was observational studies that erroneously led us to believe that light-moderate drinking was healthy for the heart. Consequently, unless the data is very strong, we should be skeptical of observational studies.
The best measure of the strength of an observational study is the RR or relative risk. For example, cigarette smoking is VERY strongly associated with lung cancer. The RR=20, meaning smokers are 20X more likely to get lung cancer than non-smokers. This number is so high it's hard to ignore. Unfortunately, the RR of most associations in healthcare are much lower. An RR=1.0 means no association. Most of the time the RR is less than 2.0, and frequently less than 1.25. Such associations are likely noise, not signal.
A strong association ONLY with heavy drinking in less common cancers
When it comes to alcohol and cancer, there are only a few cancers with strong associations. These are cancers in structures with direct contact with alcohol: the oropharynx with an RR=5, and the esophagus, also with an RR=5, and when combined with smoking the RR goes up to 30.
However, these numbers are only for heavy drinkers. (By convention, the reported RR is for the highest percentile of consumption). When we dig out the RR for light drinkers it is considerably smaller. For instance, the RR of oropharyngeal cancer in light drinkers is 1.17.
The other important consideration is that oropharyngeal and esophageal cancers aren't very common. Your lifetime risk of getting oropharyngeal cancer is 1 in 100. Increasing that risk to 1.17 in 100 with light drinking, barely moves the needle.
Breast cancer and colon cancer, on the other hand, are much more common. However, the RR for these cancers are so small for light drinkers (RR=1.05), it's almost not worth mentioning. Instead of a 1 in 10 risk of breast cancer, a light drinker has a 1.05 in 10 risk of breast cancer.
All-cause mortality
How do these small increased cancer risks in light drinkers impact their overall lifespan? It appears, not much. In fact, in observational studies of all-cause mortality , light drinking is associated with a slightly lower all-cause mortality. Let me repeat that. Despite an increase risk of cancer, light drinkers have a slightly lower risk of dying than abstainers.
There are several reasons for this discrepancy: cancer is often treatable, the majority of cancers are unrelated to alcohol, and there are many ways to die other than cancer. However, the main reason for this discrepancy brings us back to the limitations of the observational study:
It's just too dull an instrument to accurately discern small differences in small exposures.
Individualized benefit-harm analysis
Because there's no clear signal of increased cancer or mortality, it's reasonable for light drinkers to bring other factors into consideration when evaluating their drinking. Things that can only be determined case-by-case, and by personal experience. For some, alcohol improves: sugar control, social engagement, relaxation, and enjoyment of life. Yet in others, it may lead to: weight gain, poor sleep, violence or escalation to heavy drinking. Things you wouldn't know without self-experimentation: testing life with alcohol, and life without alcohol.
Lastly, if people don't drink, what will they do instead? Marijuana, mushrooms, ketamine, nicotine, sugar? Whatever it is, it will have its own concerns and benefit-harm analysis.
Take home message
Light drinking does not appear to be a problem; however, to safeguard against cancer, keep up with mammograms, colonoscopies, and oral checks from your dentist. AND DON'T SMOKE. But you knew that already.
This topic highlights the importance of assessing the strength of evidence. As the philosopher David Hume famously stated, "A wise person proportions their beliefs to the strength of the evidence". There's strong evidence that heavy drinking is bad for your health AND causes cancer. Take heed. Conversely, there's only weak and conflicting evidence that light drinking is bad. So if you're a light drinker, and you feel it brings more good to your life than bad... cheers.
ความคิดเห็น