(I'm giving you fair warning: this is a long piece, but I've divided it up for you. Each part will do fine on its own, but of course I'd like you to read both. You'll be a better person for it. --PalMD)
Part I: Uncommon Sense
Everything was OK until Christmas; before that she felt fine. Then she began to feel tired. She was having trouble sleeping so, she thought, maybe that was it. She tried some sleeping pills from the drugstore, but she still didn't feel right. Just doing a load of laundry wore her out. She and her husband normally got together with the same set of friends every New Year's Eve, but this year she was too tired. She hated New Year's Day, because every year she made another set of resolutions, usually about her weight. When she got on the scale this year, she was up twenty pounds. Twenty pounds! Impossible! Her pants fit more or less the same, although her shoes were tight. In fact, she'd been wearing unlaced shoes around the house and couldn't really fit into most of her socks. She went to bed disappointed.
That night she woke out of a sound sleep and sat bolt upright---she couldn't breathe. She ran to the window, threw it open, and gulped in cold air, slowly feeling better. Her husband called an ambulance.
Heart failure is a condition in which the heart isn't pumping well enough to meet the body's needs. Fluid can back up into the lungs, making it hard to breathe, the legs can become swollen; and depending on the severity, heart failure can lead to sudden death. One of the mainstays for the treatment of heart failure is a class of medications called diruetics which cause patients to urinate more, decreasing swelling and easing breathing. But diuretics don't really affect the heart itself, they just drain off fluid, alleviating some symptoms. It seems logical that if a failing pump is responsible for many of the symptoms of heart failure (and it is), then medications that improve the pumping action would be a good thing, and those that decrease it bad. It's common sense. It's intuitive.
Medical students and residents often dread discussions about statistics (a characteristic which I'm sure is not unique to these groups). And who can blame them, really? Statistical analysis is inherently non-intuitive. Its purpose is to separate us from our own natural inclinations to identify patterns (our "intuition") in order to systematically study relationships between variables of interest.
A number of years ago, doctors noted that a certain class of drugs ("inotropes") improved the pumping action of the heart. Patients with heart failure were given these drugs, and their heart function improved. Common sense, right? But when the topic was studied systematically, researchers found that these drugs actually increased mortality. Oops (in this case, "oops" means people dying).
Another group of drugs called beta-blockers can reduce the pumping action of the heart, and for years were assiduously avoided in heart failure---until study after study showed that beta-blockers actually decrease mortality in chronic heart failure.
Common sense can give us a starting point, but until the big questions are examined systematically, we are in danger of intuiting our patients to death. Beta blockers are now a mainstay of heart failure treatment, and inotropes a rarely-used footnote, a treatment reserved for a specific set of circumstances. But we didn't figure that out through common sense alone.
Part II: Does smoking pot cause cancer?
Since common sense is such a poor tool, we have to find other tools to hunt down the truth, and there are many ways to ask and answer important medical questions. Recently, I explored one way to look at risk reduction, in this case, vegetables and cancer, and this got me thinking---what if I wrote something in a bit more detail about how we evaluate the risk or benefit of a behavior? Would any sane person read it? Would I be able to make it make sense? What if I pick a sufficiently controversial topic? Let's run with it.
Smoking things is bad for the lungs. Cigarette smoking causes chronic obstructive lung disease and lung cancer. We can't predict exactly who will be affected, but we do know that the more you smoke, the higher the risk. Given that smoking cannabis is also popular, and that eventually it is likely to become legal, it is reasonable to ask if smoking cannabis causes cancer as well. Common sense would suggest it.
Tobacco contributes to the development of several types of cancer. When it is smoked it causes lung cancer. When it is chewed, it causes oral cancers. Both types of ingestion can contribute to head and neck cancers. Since tobacco can cause so many different kinds of cancer, it seems plausible that there may be something about tobacco itself, rather than just the smoke, that leads to lung cancer. Since pot is most often smoked without tobacco, maybe it's not as high-risk.
Still, lung cancer is pretty dreadful, and if we anticipate cannabis use increasing, it would be good to know if it also leads to cancer. There are a number of ways we can ask this question. There are also a number of difficulties. Lung cancer is relatively rare, so large numbers of people need to be studies. It also takes a long time to develop, so we need a lot of "person-years" or data.
The strongest way to look for an association would be a randomized controlled trial (RCT). In an RCT, a group of people are randomly selected and divided into groups. In this case, one would be given pot to smoke, and another group given something to smoke that is known not to cause cancer as a control. The two groups could be followed over time, and then at a certain point, cancer cases could be compared between the two groups.
But we aren't going to do this, even though this is the most robust way to study the question. For one thing, we don't have access to a good control substance, something that people could smoke, is enough like pot to fool them, but is known not to cause cancer. We also can't ethically subject subjects to something we think is reasonably likely to cause grave harm. An RCT is pretty much out.
Another way to go about asking the question is a cohort study. In a cohort study, we could gather a group of pot smokers, and identify a similar group of non-pot-smokers and follow them for a specified period of time (or until a particular outcome happens). We can then look at lung cancer rates in in each group too see if there is a significantly larger number of cases in the pot-smoking group. This isn't a terrible way to look at the question, but probably would require a lot of patients over a long period of time.
An easier choice would be a case-control study. In a case-control study, we don't have to wait for enough people to develop lung cancer---we start off by picking a group of lung cancer patients, and then pick out a group of patient similar to them but without lung cancer. We can then look at all sorts of variables and compare the groups. This is how the first large study of lung cancer and smoking was done. But there are a lot of limitations to this sort of study, not the least of which is having to obtain accurate historical information. There will inevitably be some element of recall bias, in which subjects recall their histories in ways that systematically differ from what actually happened. It's easy to believe that someone just diagnosed with lung cancer might be hyper-attuned to possible past risky behaviors. Still for rare diseases that take a long time to develop, this is often the only design that is practical. And while this sort of study can identify correlations, case-control studies are not great for identifying causation. But they can lead to further studies with different designs, studies that would not seem worthwhile without initial favorable evidence.
And this was the approach taken by researchers in New Zealand. They identified a group of patients with lung cancer, and compared them with a similar group of people taken from a voter registry. The then divided them into "pot smokers" and "pot non-smokers" (which included some light pot-smokers), and further divided the pot group by how much they recalled smoking. The results were fairly striking. When comparing non-pot-smokers to the heaviest smokers (more than 10.5 "joint-years", with a joint-year being one joint per day for a year) the odds of developing lung cancer were 19 times higher in the heavy smoking group (these relationships held even when controlled for tobacco smoking). Depending on what estimates you use, you would only have to have a few people become heavy pot smokers to cause an extra case of lung cancer.
These data are similar to those found for cigarette smoking: the more you smoke, the higher your risk. This case-control study cannot establish cause beyond doubt, but the evidence is pretty strong. It is biologically plausible that smoked cannabis causes lung cancer, and this case-control studies and others support the association. Perhaps future cohort studies will give us more information, but at this point, people smoking significant amounts of pot who would otherwise have normal life-expectancies might want to reconsider.
Part 3: Synthesis
My grandfather was, by all accounts, an exceptional man. I never had the chance to know him; he died of lung cancer when I was very young. I remember my dad sitting me down on the bed and telling me that Papa had died. I also remember going to a strange, spooky building to return his oxygen tanks when he no longer needed them. But that's pretty much it. The rest of my memories belong to others. Many years later, not long ago in fact, I was up north with my family for a week. It's been a family tradition for generations, and that particular year, the family was large. My aunt (Papa's daughter) was there, despite her advancing lung cancer. I said goodbye to her on the front porch the evening before leaving, and said, "I'm really glad you came." She said, "Me too," and we both knew exactly what we meant. That was the last time I saw her before her death.
Heart failure can be a miserable disease, but the last couple of decades have seen a revolution in its treatment. Some of our biggest gains against heart disease have been due to identifying its cause and helping people avoid risks. Lung cancer, though less common, is a dreadful disease, one where we've made some gains, but not enough. Eighty-ninety percent of lung cancer deaths are due to smoking. This makes these deaths preventable.
Everyone has to die of something, but modern medicine has given us tools to help prevent suffering and death. Morally, I've got nothing against marijuana. But if it really is as potent a risk factor as suggested by this study, people might want to think twice before sparking up.
Packer M, Carver JR, Rodeheffer RJ, Ivanhoe RJ, DiBianco R, Zeldis SM, Hendrix GH, Bommer WJ, Elkayam U, & Kukin ML (1991). Effect of oral milrinone on mortality in severe chronic heart failure. The PROMISE Study Research Group. The New England Journal of Medicine, 325 (21), 1468-75 PMID: 1944425
Fonarow, G., Abraham, W., Albert, N., Stough, W., Gheorghiade, M., Greenberg, B., O'Connor, C., Sun, J., Yancy, C., & Young, J. (2008). Influence of Beta-Blocker Continuation or Withdrawal on Outcomes in Patients Hospitalized With Heart FailureFindings From the OPTIMIZE-HF Program Journal of the American College of Cardiology, 52 (3), 190-199 DOI: 10.1016/j.jacc.2008.03.048
DOLL R, & HILL AB (1950). Smoking and carcinoma of the lung; preliminary report. British medical journal, 2 (4682), 739-48 PMID: 14772469
Aldington, S., Harwood, M., Cox, B., Weatherall, M., Beckert, L., Hansell, A., Pritchard, A., Robinson, G., Beasley, R., & , . (2008). Cannabis use and risk of lung cancer: a case-control study European Respiratory Journal, 31 (2), 280-286 DOI: 10.1183/09031936.00065707