This is Section D now, the case control study. This is from Ernest Wynder who we talked about earlier. And this was a littler earlier than the cohort study because it was easier to do. So basically you'll find cases of lung cancer and you compare them to cases with no lung cancer. And so if you're in a hospital, for example, or in a clinic, there are cases of lung cancer right there. All you have to do is find some number of controls, and in this case they found roughly equivalent number of controls. They had 583 cases of lung cancer. And we call those exposed, and 576 controls who are similar to those cases as we can make them, and you have a nonsmoker or maybe a light smoker who we think of as not exposed. So now we can't actually calculate the incidence rate, because we don't have the entire population. So we're stuck with this relative odds rate which I discussed. So, for example, this relative odds is going to be 583 divided by 576. That's the odds for a smoker having lung cancer. The odds for a nonsmoker having lung cancer is 22 divided by 204. And when we compare those relative odds, we discover that the odds for having cancer if you're a smoker are 9.4 times as large as the odds for a nonsmoker. And that 9.4, there are many studies in the literature of smoking and lung cancer, and they mostly all come up with relative odds of about 9 or 10 or 15. And all of this suggests that lung cancer is related to a history of smoking. Here is a case control study of depression. This may be the most famous case control study in the literature on depressive disorder. And it was done by George Brown, a gifted sociologist in London. And he, with Tirril Harris, wrote a book called the Social Origins of Depression. And this is one of the books in 1978, it's an old book, that convinced people, finally, that, yes, depression did have social origins. And this is a case control study which has been replicated probably 30 or 40 or more times all with very very similar results. So basically in East London, George Brown had cases of depressive disorder and these were women. He was a clever guy. He knew that depression occurred more commonly in women. So he said, look, let's, why mess around with the men, you have to pay a lot and they don't have as much depression, so there's a lot more power per, dollar per dollar, the women are worth a lot more in a case control study of depressive disorder. So he focused on women. And you can see that of the cases, 52 cases of depressive disorder onset, 68% had a stressful life event just prior to the onset. And in the controls, who did not have depressive disorder, 20% had one or more severe stressful life events, so 68% versus 20%. And you can see I've calculated the relative odds. There is 8 or so. And if you talk to a clinician, a psychiatrist, or a psychologist treating people with depressive disorder, they can tell you that, not every single time, but almost all of the time it is only a short period of therapy or discussion before they learn of a stressful life event in the relatively immediate past history of the individual. So this is the evidence that stress and particular loss events, so widowhood, bereavement, people leaving the home, breaking up a romantic relationship are related to onset of depressive disorder. So these are strengths and weaknesses of the case control study. The strengths are it requires really no hypothesis, and the cases can come from clinics. And there are only a need for a small number of controls. So a classic example of the strength of the case control study is the cases of vaginal cancer that started appearing in Boston in the 1970s, and they basically had no idea what was going on. This was a disease that they hadn't seen before. And if, well, let's do a case control study. And they had no idea what was causing it, so it was like a kitchen sink. And a big questionnaire asked the women with vaginal cancer, you know, about their experiences, what they had been doing, what they had been eating, any medications, everything, basically. And they had 8 total cases and 32 controls. It's miniscule, if you can imagine, 8 cases, 32 controls. But the finding was so strong that there were four zeros after the decimal point for the significance of the finding. That is, it would only happen one in a hundred thousand times or so by chance that the women who had ingested a hormone treatment for morning sickness were the ones who had the vaginal cancer. So after this case control study with a total of 40 cases in controls, they basically eliminated that hormone treatment for morning sickness and had a huge public health benefit. So that's an example of the strengths. There are weaknesses, and these weaknesses are perhaps most important for the mental disorders, and that is, for example, biased recall. So say you're asking a woman whose son has depressive disorder, and you're asking her about the experience of giving birth to that son. Well, she is trying to explain something that's important to her, why does her son have depressive disorder. So it could be that she will think harder about the birth experience of the son, and sometimes we call that effort after meaning. She's trying to construct a narrative which makes sense. And she may work harder and locate more birth complications in that history than a woman who has nothing to explain with the son who does not have depressive disorder. So that's biased recall. And that's a case control study especially for mental disorders is subject to that. We also have the problem of temporality. We don't know, sometimes we're asking, but we don't know which came first. So if we're talking about stress, well, you have a divorce, maybe that leads to depression. But was there depression, or part of depression, or symptoms of depression prior to the divorce which contributed to the divorce? We have trouble figuring out what exactly causes what. And then finally, we sometimes have controls who are not exactly the same as the cases. So we have people in a clinic with depressive disorder, and then we solicit volunteers. Perhaps those volunteers are from a different neighborhood, a different social class. They are more or less depressed for various different reasons. And so we have to worry about whether there is something called confounding, which I'm going to talk to you about shortly. So that's out discussion of three basic research designs for depressive disorder from the classic designs in epidemiology. That is the cohort design, the case control design, and the ecological design. And now we're going to shift to trying to understand something we're going to call the web of causation, that is, how do we make sense of the data that we get from these various research studies.