In some sense, of course, I know the answer to this: a combination of tort law and government mandated warnings, but I don’t think this really adequately explains why those warnings are there and especially not how they are worded.
Why are blanket statements equivalent to “do not drink” used so routinely on prescriptions? For prescriptions in British Columbia, the warnings against driving or operating heavy machinery for at least some medications read, “Do not drive or operate heavy machinery while using this medication until you know how it affects you.” And yet the anti-drinking warnings still read simply, “Do not consume alcohol while using this medication.”
The thing is, consuming alcohol while using such medications is entirely common. It is, in fact, the expected behaviour. No doctor who prescribes such medicines to 10 or more adult patients would give serious consideration to the idea that none of their patients have ever consumed alcohol while affected by such medications, and the behaviour of both 18-25 year olds and 25-40 year olds makes it likely that even prescribing to just one patient is enough to make it more-likely-than-not that a doctor has prescribed to someone who used alcohol while on a medication that warns against such use if the medicine was taken for long enough and/or by persons in particular demographics.
Kaitlin Bell Barnett has written about this issue in traditional outlets as well as on her own blog. It seems that to the extent that there are risks they are either small or less serious or simply not substantiated by any research. The exception is when medications are not used as prescribed: overdoses of certain medications combined with heavily intoxicating levels of alcohol can cause very serious effects, even death. Is this a reason to form a blanket ban, however? Especially since there appears to be very little reason to believe that these effects are likely with every type of medication that bears Do Not Drink warnings. Benzodiazepines, for instance, can have effects that at high doses combine with sufficiently high amounts of alcohol in the blood to produce death. But SSRIs typically carry an identical warning, and it does not appear likely that alcohol and SSRIs interact in any substantially similar way.
Barnett*1 does articulate a study that identifies increased risk of stroke involving bleeding into the brain while on SSRIs. But while Barnett documents the authors of the study stating this is a reason not to drink while using SSRIs (since binge drinking of alcohol is also shown to correlate with increased risk of this type of stroke), it appears that the authors were not warning the public not to take aspirin in conjunction with alcohol despite aspirin’s well known blood-thinning effects and associated risks, including risks of bleeding into the brain.
While Barnett worries that, “the dangers of mixing meds and alcohol rarely enter the public discussion” I worry instead that the dangers are hyped quite often. Barnett herself seems to vacillate between two seemingly contradictory positions. The first worries that studies are insufficient to determine the dangers and thus we are further endangered by our lack of knowledge. The second worries that society is being advised of dangers whose existences have never been established by methodologically sound studies.
I’d be happy to support concerns of the first type, save that such concerns really don’t need my amplification. The authors of the SSRI study were happy to recommend that those who wish to drink talk with their doctors about switching to another form of antidepressant. Yet, there is also some evidence of risks with other antidepressants’ combination with alcohol. That leads both to the question of whether or not funding by companies who make competing medications might have affected researchers public warnings (which, remember, went beyond what the study was actually able to scientifically conclude) and also to the reasonable observation that drug companies may be happy to fund research into the risks of competing drugs if popular coverage drives customers towards their own products.
Worries of the second type appear to need more of my support. In particular, the effects of alcohol appear to be relatively small without a great deal of consumption. This consumption might be sufficient to have measurable health effects if small in any given week but continued over many years. Or it might be sufficient if an overwhelming amount of alcohol is consumed in a single binge. In between, it seems that frequent binge drinking can increase health effects to measurable amounts even if no one binge drinking event damaged health in a directly measurable way.
This leads researchers to take data from binge drinking studies (since binge drinking is more likely to have a measurable effect over a study-relevant period of time) and act as if the conclusions of such studies apply equally to all drinking.
My problem with this is three-fold.
First, I support evidence based policies. Where we have no evidence, we should be engaging in further investigation, not producing policy based on unjustified fear that an effect will occur.
Second, this is not a case where we have no evidence: when studies are performed that find no measurable effect at a certain level of alcohol consumption, then we must be able to state that there is no measurable effect and therefore no measurable risk. Studies repeatedly find risks for binge drinking that are not found for more moderate drinking. While I support further study any time researchers have sound basis for hypothesizing a significant health risk to the public, we have also known since Paracelsus that the dose makes the poison. We cannot avoid in all contexts or amounts that has ever been shown to be dangerous in one context or amount. While nitrogen gas at 70% concentration in our atmosphere is fine, an atmosphere of 99.9% nitrogen would be lethal. We do not conclude that humans must avoid nitrogen gas (or that humans should avoid it while taking SSRIs).
Third and finally, whatever health effects exist for combining certain medications with certain levels of alcohol, the members of the general population are routinely willing to have a single glass of wine while using hydrocodone or a single beer with Prozac. They are willing to do that despite medical warnings against exactly this behaviour. Why? Barnett speaks of social pressures to drink and the stigma of mental illness as the reason why those taking psychiatric meds in particular might feel uncomfortable declining an offered glass of beer or wine in a social setting. But other health conditions are also stigmatized. An ongoing bad back might be equally awkward to discuss when declining. I believe that an under appreciated factor here is that drinking on its own has negative effects – headaches, nausea, dizziness and more – at certain doses and in many cases the public cannot perceive negative health effects when consuming typical quantities of alcohol while taking typically prescribed levels of certain medications.
Personal experience, then, is overwhelming vague and ubiquitous medical warnings, and that is something I find potentially very dangerous. I worry in particular that researcher’s conflation of the effects of binge drinking with more moderate drinking which is supposed to lead the public to treat all drinking as equally dangerous as binge drinking is succeeding in a spectacularly harmful way. The two are likely being treated as equally dangerous by many people, but unfortunately I see many people apparently setting binge drinking’s dangers to be equal to those of moderate drinking, rather than considering the default risk to be that associated with binge drinking.
In other words, when professionals are unwilling to draw a distinction between binge drinking and moderate drinking, the public may also be unwilling to draw such a distinction, but with most experiences with alcohol being moderate and (in the short term) apparently safe the extrapolation that may seem most sensible is treating all drinking as safe.
Researchers can scream all they like that, no, all drinking is dangerous because we can’t quantify the danger. The public, however, has enough personal experience to confuse anecdotes with data. We as a society simply do not appear to be acting as if researchers warning held real relevance for us. That worries me.
Negative results are important to publish in any area of science, but I think that here in particular we need more research released. In combination with anti-depressants, alcohol over time may diminish any beneficial effect the anti-depressant might provide. Yet as we know about depression, isolation kills. A doctor might be very distressed to hear a patient on SSRIs is home alone the vast majority of days and drinking even a single beer a week in that isolated context. On the other hand, a doctor might be pleased to hear that a patient on SSRIs who has not had alcohol in months broke away from isolation to have two and a half glasses of wine over four hours while having dinner with a few good friends. Publishing the negative results will give doctors firmer guidelines to help them decide when they should and when they shouldn’t worry about their patients’ consumption of alcohol. Perhaps that one beer a week isn’t enough to worry about, but three would be. We simply don’t know.
But most importantly, better information for patients is crucial to maintaining credibility with the public. If our doctors do not appear to be aware that the dose makes the poison (and the medicine), we may be less inclined to take their advice. And, frankly, if they are giving advice that contradicts repeated negative results those doctors deserve to be ignored. The failure to heed negative results when seeking a measurable effect from the interaction of one weekly drink with regular use of aspirin*2 is just as anti-science as the failure to heed negative results when seeking a measurable effect from the introduction of shark cartilage to cancer patients.
I think that, where warranted, warnings could be readily understandable, relatively brief, and much more credible if reworded to be consistent with the evidence. A hypothetical example on an imaginary medication might be:
Take care when consuming alcohol with this medication. Increased risks of serious (short term/long term) effects have been observed when this medication is combined with more than one drink per 80 pounds of body mass at any one time and also when combined with more than 12 drinks per month regardless of body mass. Speak to your doctor if you find yourself consuming alcohol more than one time per week while on this medication and especially if you consume even moderate amounts of alcohol while on more than one medication bearing a similar warning.
I want us all to get the medications we might need to be healthy. I don’t mean here to focus on psychiatric meds specifically, though it appears that there is better research on certain painkillers and benzodiazepines than on most antidepressants, anti-psychotics, and non-benzo anti-anxiety medications. Even that better research, however, tends to focus on the effects of large doses as the effects of small doses appear (from what we can tell where negative results are not generally published) to be beneath clinical relevance.
I also want us all to get good medical advice, and part of getting good medical advice is getting advice that hasn’t warn out its power through overuse and extrapolation of risk into areas where experiments and studies show no measurable risks. For any given medication, there may be risks, even risks that manifest only in conjunction with alcohol, that researchers have not yet studied. The fact that no risk is currently identified does not in itself mean that a drug is risk free. However, scientific uncertainty is leading researchers to warn about dangers that we have studied and that have not been found to exist to any observable degree.
I believe public discussion of health studies, and particularly studies linking medication/alcohol combinations to risks, has struck the wrong balance between expressing uncertainty about whether or not a drug is “safe”*3 and recognition of areas where studies have been performed and found no measurable effect. I think we’re better enacting evidence-based policies and enhancing the credibility of medical researchers when we avoid blanket statements of risk that do not conform to what we know from research, especially (but not only) when personal experience will tend to lead to widespread societal disbelief in the seriousness, accuracy, or both of such statements.
In the meantime, listen to your doctors, but feel free to question them. I’ve had two doctors in the past year that were particularly good at answering my questions on various topics with actual citations to research and (where necessary because of my lack of expertise) then interpreting them, rather than simply treating me as if their personal opinions should be sufficient without any research to back them up. I wouldn’t make these criticisms if I didn’t feel we can do better.
*1: or Bell Barnett, I’m not sure which is correct: I’ll just go with Barnett for now.
*2: If no measurable effects are found: I don’t actually know the state of this research, I mean it to be a hypothetical.
*3: quotation marks highly intentional: we really don’t know what people mean when they say “safe”. Water skiing as practiced today is generally safe