Wednesday, 26 January 2011

the concept of causality

there's been a death, in the opposite house,
as lately as today,
i know it, by the numb look
such houses have always.

this is the first paragraph of emily dickinson's famous poem relating to death and dealing with a death in the family. i write such as an introduction because it is the root of what has brought me to the following conversation - a mourning period in the family - and also as a tribute to a great person whom i had not known much of until the twilight of his years.

in any case, this post does not revolve around such incident, but instead is about various other things; from the scientific approach to causality to evidence and, interestingly enough, death.

let me set the premise, and how i will argue for and against the point - something that a lot of my medic friends will find familiar (and even slightly laughable) - 'smoking causes cancer'. now, this was brought up by my cousin, jo jackson, and my continual exposure to colleagues and friends based in the medical and biological sciences fields has given me a very biased take on the issue from the start. indeed, we have discussed this issue to the proverbial death so much, that nowadays i write with assumed knowledge that it is true. however, ask your local cigarette tender and he will beg to differ, and it is this take on the subject that prompts such an analytical approach today. again, let me state, though, that i may be biased academically, i am not at all socially - i believe it is one's right to smoke should one want as long as it does not infringe upon the health and convenience of any other (and this may lead upon an entirely different and tangential discussion, so we shall leave that for another day).

now, i structure my argument based on the spectrum of arguments that may lead us to believe that smoking does indeed cause cancer - this ranging from the very molecular basis to a near-purely statistical claim of causality. this is not a comprehensive list, but the very ends of a spectrum should, if anything, lead a reader to fill in the gaps himself, should he be inclined.

for those familiar with cigarettes, nicotine is the first drug that comes to mind - and here we first ask: what is nicotine? i'm sure you can click that wikipedia link and warrant your own discussion about what it is, but for our intent and purpose here today, the points i'm trying to get across are:

1. nicotine is an addictive substance.
2. nicotine in itself does not cause cancer.
3. nicotine's primary action is as a psychoactive drug, and mimics the action of acetylcholine (ach) at neural synapses. this point is more important to the more molecularly-inclined people, so for those who think this is over their heads, just understand that it works like a drug.
4. nicotine is found naturally in some plants, and the significance in drug addiction and adverse effects lies in its dosage (much like many other things).

so, nicotine in itself isn't a horrible thing (which is a common misconception in the 'cigarettes cause cancer' debate). however, cigarettes aren't made of pure nicotine. there's a plethora of chemical substances in cigarettes which you can google up yourself (yay for using google as a verb!), and you can narrow down their actions to a few major mechanisms:

1. polyaromatic carcinogens. aromatic compounds are organic molecules which have the benzene ring structure, though most of them are not directly benzene-derivatives. the mechanism of action here is complex and well beyond the scope of our discussion today, but there's more than compelling evidence that aromatic compounds can (and do) cause cancer. they accumulate in the cell nucleus and have complex interactions with dna (amongst other things), leading to programmed and unintended cell death. the whole crux of the matter is, a lot of the chemicals in cigarettes, some of which are yet uncategorised) accumulate in smokers' cells and cause them to mutate. a small proportion of these cells go on to mutate in a very particular way that leads to disregulation of the cell cycle, leading to mutagenesis and cell death.

2. reactive oxygen species (ros). interestingly enough, this point was not brought up in our conversation by yours truly, but by jo jackson himself! and, yes, it is a very prominent factor in the causation of cell cancers. for the chemically-inclined, these are radicals which have an unpaired electron, which is very unstable and tends to react with other chemicals. this is particularly devastating in cells undergoing division, and in cell mitochondria, where ros are carefully regulated to create energy for our everyday use. again, the pathways are complex and numerous, well beyond the grasp of my description today, but in a nutshell: excess ros, which are created in a smoking environment (externally, in the lungs and on the cellular level), directly causes carcinogenesis. bad news for everyone.

3. heavy metals and radioactive elements. now, i'm quite the fan of very loud music, but this is, unfortunately, not what we're talking about here. also, when i say radioactive elements, i'm not talking about some material that, upon ingestion or inhalation will give you spider-senses or an adverse reaction to kryptonite. nay, we are all at the mercy of modern science, and i guess this would be a good place to introduce the concept of radioactivity. radioactive substances are unstable isotopes of most day-to-day elements (and sometimes compounds), which will decay to their stable isotopes, and in the process, emit various forms of energy. cigarettes contain the radioactive isotopes of Po, Cu, Pb, and various other 2-letter words that you can probably come up with yourself and be right. in any case, they decay and emit gamma rays, which is kind of like having a continuous x-ray machine going off in your lungs. needless to say, that's cancer just waiting to happen (as i'm sure your doctor has told you the risks of being x-rayed continually, and in high doses).

4. direct gene regulation. ok, so i admit that this is a relatively new area for me, and i won't go on about it too much. a lot of chemicals (again, those found in cigarettes) have been shown to directly interact with genetic material and regulate their expression. what this means for smokers is that there is a complex modulation of your cells and what they do upon inhaling cigarette smoke. this could mean one (or more) of a thousand things, from reduced cellular respiration to increased cell division to creation of rainbow unicorns. only God knows what the hell is going on here but as any master chef can tell you, putting unlabelled ingredients into the wok is a recipe for disaster. yes, i've degraded to talking in analogies, so it's time we move on.

on the other end of the spectrum, which we are dealing with now, is the statistical correlation between smoking and cancer. it is important here, to distinguish between correlation and confounding. a confounding factor is one that effects both your exposing factor and outcome, while not necessarily having a modulating effect on the relationship between exposure and outcome. what the what what? wait, it's not that confusing, let us use an example:

let's say i did a study about drinking coffee and developing cancer. let's say i just took the raw numbers of what proportion of people drink coffee, and what weightage of these people go on to develop cancer. assuming i haven't corrected for any other factors, i find that drinking coffee predisposes a person to developing cancer by 10-fold... now this would be horrible news for starbucks. but, as with any illustrative example, this is not the case. i may have forgotten to take into consideration that people who drink coffee also have a tendency to smoke cigarettes (obviously, just for example). and that, it is not the coffee that causes cancer, but the cigarettes. so, in this case, cigarette-smoking is a confounder and has messed up my data set :(

okay, on to the statistical argument. the point here is that simply having a statistical relationship between the two (cigarette smoking and developing cancer) is insufficient. and, this is exactly the fall-back argument that big cigarette conglomerates will use to get out of a sticky situation (tar puns aside). that and the fact that smoking is a choice activity (keep in mind we're talking about a behaviour-modifying and addictive substance here, which i find to be counter-intuitive, or at the very least controversial).

anyway, how much statistical correlation will we need to prove an indisputable relationship between exposure and outcome? apparently there is no magic number, or even a sufficient one. as long as there are people out there fighting the cause that cigarette smoking does not cause cancer, there will always be a shadow of a doubt, rendering statistical correlation insufficient (though it may still be significant). let us not talk about p-values or how social acceptance modifies this, but instead, let me just put forward that, after reading countless papers on the issue, i think that it lies beyond any reasonable doubt that the statistical evidence is just there. however, much like any tenet and belief, if one believes something with such passion, there is no convincing otherwise.

anyway, i think i have ranted more than my fair share for today. i guess this isn't really what i set out to write - i was hoping for a more impartial argument on the topic but i fear that i have portrayed myself as an anti-smoking person (which i am, but that is besides the point). am i trying to preach to people out there the fallacies and dangers of smoking? no, i am not that holier than thou. am i trying to warn people from an addictive substance from which they may have no return? no, i believe people are more determined than that, and can bend their wills as they wish. am i trying to lay out the facts as they verily are? yes, but apparently i have not done a well enough job. am i trying to pique your interest such that you do your own unbiased research and come up with your own conclusions to be used for yourself (as opposed to trying to justify a pre-conception)? yeah, let's go with that one.

No comments: