 Hi, welcome back to history and philosophy of science and medicine. I'm matt brown today We're talking about values and disease screening and the science And decision-making behind disease disease screening in medicine. I Want to remind you about something we talked about last week having to do with inductive risk. So if you remember According to the way we talked about this last week You have if you have a hypothesis like Chemical X is safe for human consumption And that's a hypothesis that could be true or false It's also based on the evidence that you get from the the data you collect You could choose to accept or reject the hypothesis and therefore possible situations either you Accept the hypothesis and the hypothesis is true, which is great That's what you want or you reject the hypothesis and the hypothesis is false also great either way you've got the right answer but if you Accept the hypothesis, but the hypothesis is actually false you have a false positive result, right? That's a form of error Similarly, if you reject the hypothesis, but the hypothesis is actually true. That's a false negative error, right? presumably the chemical is safe or it's not to some criteria of safety and You accept or reject it or not and so use these four possibilities and two of them are forms of error Now if you have a false positive error The consequences of that could be risk to human life and health because if you accept the hypothesis that the That the chemical is safe and so you don't regulate it or ban it But it in fact is not safe then you're gonna have those risks those those consequences On the other hand if you reject the hypothesis if you if you say it's not safe And then you go ahead and regulate it or ban it then you're going to forgo the benefits of chemical x and they're going to be economic Losses to the producers of it and and so on and so forth, right? so these are the kinds of considerations that One has to weigh in order to make judgments about the evidence, right? So in designing Studies in determining what the criteria For accepting or rejecting the hypothesis is going to be you have to you have to weigh These possible consequences and the risks associated with them and set your standards accordingly That's what we talked about last time Now suppose we switch the case a little bit Our hypothesis now is not concerning some chemical safety question It's concerning of the situation of a particular patient, right? The the hypothesis in question is that patient that a patient has Condition x that's going to cause disease for example, they may have Some kind of growth of cells in their breast that is going to cause breast cancer disease We might have a test of this hypothesis Through a screening technology like screening mammography, right? And this screening test has the same structure of possible choices Either the either the condition is going to cause disease or not, right? at some at some for some criteria of disease and We can choose we can design our We can design our Screening test in various ways, but there's always going to be some probability of False positive or false negative air so false positive your tests come back comes back and says You have the disease or you have the condition that will lead to disease but you actually don't or False negative error. It says you're fine the test says you're fine, but in fact you have The disease or the conditions that are precursors to the disease, okay? Now The assumption is going to be that this test is is going to in many cases where there's a treatment available Is going to lead to certain kinds of treatment so Whenever we accept the hypothesis that the patient has the disease or has the condition that will cause disease We are presuming also that we're going to treat Treat that accordingly. There's going to be some kind of treatment decision made on the basis of that information okay, and Here again, we have the same kind of inductive risk issues, right? If you have a false positive error, then there are risks of over treatment of Stigma if the disease carries some form of social stigma of the mental distress of being told you have a disease when you in fact You don't have the disease all of these things are Consequences of a false positive error and the things we risk Based on the the the probability of a false positive error Similarly for a false negative error if we tell you you're fine, but you're not then we're risking increased morbidity and mortality, right? So these are the kinds of errors that need to be balanced in the design of of screening medical screening Techniques and technologies right so you might in the case of Cancer screening for example, you might start with an early detection test that Tries to minimize false negative errors, but has lots of false positives As long as you can follow that up in relatively short time with a second test That is that is less That may be more expensive or more difficult, but has a much lower false positive rate, right? Those are the kinds of choices there both economic Aspects to these choices as well as these sort of social ethical consequences to to human well-being at stake We should think also in the con in the context of our discussion today Not only about the the inductive risks involved in a particular screening test Which is which is definitely relevant to patient treatment and Medical decision-making we should also think about the inductive risk involved in the evidence about those screening tests So both of the articles that you read for today by Plutinska and by Koranian Fernandez Pinto look at breast cancer screening and Evidence that engaging in those screenings is effective, right? So right hypothesis in this case might be that screening for condition X say breast cancer is Effective for patients in group Y say women age 40 to 50, right? if we say if our if our sort of if the hypothesis if we accept the hypothesis, right and therefore we have a policy recommendation or a medical recommendation that women 40 to 50 and in that group Should receive the screening, right? We risk all the same sort of Consequences of a false positive error over treatment stigma mental distress, etc If we do not give the if we do not recommend the screening Then there's a risk of false negative error that that has the consequences of increased morbidity and mortality and and you know, we can compare the individual level here that we were talking about before with the with the mass level so the the The judgment about effectiveness in the population right is a judgment about sort of the What's going to happen to the whole group of people that does or does not get this treatment, you know for individuals, right? There's always going to be some cases of breast cancer that are caught early and receive treatment And you know, if they get this if they get this screening their lives are going to be extended There are going to be other cases where people slip by as a false negative They're not going to get detected and they're not going to get treatment early on Similarly, they're going to be false positives. They're going to be those who suffer from over treatment Or or the mental distress that goes with over diagnosis But the question is sort of, you know stacking it all up in the in the mass, right for the whole population Is the risks are the risks associated worth it, right? Is it worthwhile for? for a person in this group to get the the screening Or are there odds of harm greater than their odds of being helped? Now there are a number of sources of uncertainty about the effectiveness of disease screening that are discussed in the readings Choice of end point, right? This this is a question of what it is. We're measuring for effectiveness Are we effect? Are we measuring survival time from detection? Say if we introduce an early detection technology Does that increase survival time? But that can be a biased end point, right? Because Others who are not detected at all Until later, you know There may it may be that you don't you don't actually extend life. You just extend the time That you know that they are going to have the disease Also, the choice of disease specific or all-cause mortality is a crucial Case of uncertainty There are some screening technologies where you can show that it makes it less likely that they're going to die from the disease That that's detected, right? But that they're also not they're not likely to live any longer So there's no no reduction in all-cause mortality That's a little bit puzzling, right? They they don't die from the thing you detected and treated them for but they still die just at the same age On average, you know, these are again. These are population level Results not individual level results. That's puzzling but but partly it's because It's it's harder to detect reductions to all-cause mortality because there are so many more factors involved There are biases and confounds of various kinds in trial design and in the results of those trials These are these are covered in great detail in the readings. There there are various ways in which Either the researchers The statistical techniques or even the patients can be biased by the trial design There's a variety of choices in your meta-analysis and your systematic reviews that means that Even when you're aggregating a large number of trials on the same question you can come to diverging Judgments for or against the effectiveness of a screening for a particular group There are also a lot of unknowns about the disease etiology and mortality so so You know what it what causes the disease? How does the disease progress and how many people in the general population now all actually die from the disease? Are things that we may not know a lot about in many cases. We we don't know a lot about What the causes of breast cancer are what the progression of breast cancer is at a biological level? and You know for many diseases our estimates of of Who dies from those from those diseases is a little bit uncertain right based on How we decide to measure it and estimate it There's also competing and uncertain evidence concerning over diagnosis rates so what constitutes an over diagnosis we have a clear definition You know someone is over diagnosed if that they are diagnosed as having a disease but the disease is not actually going to lead to lead to real symptoms detectable symptoms Or experienceable symptoms in the life course of the patient right so for example some cancers Right can be detected quite early, but they grow so slowly that the patient dies of natural causes Long before the disease the the cancerous growth would cause any disease There's also conflict over how we measure the harms of over diagnosis, right? so not only what is the rate which People are over diagnosed, but how bad is it and we don't agree about how to measure that One of the things to keep in mind is that research funding itself is biased towards detection and treatment not prevention, right? There's a there's a kind of crass economic reason for this There is no profit in prevention, but there is profit in providing treatment, right? So it's it's natural that The the medical Companies that provide the detection Technology would be interested in this and would provide funding for it Also, you know in some cases we don't you know the the the prevention is Hard to figure out because we don't know what the causes are Or there may not be preventable causes in some cases It's important. I think also to point out that early detection is not the same as prevention, right? So some some defend the importance of screening Technologies as a as a general argument in favor of preventative medicine Which is you know, no doubt important But but detecting something early And treating it early is not the same thing as preventing the disease from happening All right, it's just early intervention. It's not prevention Those aren't the same. They should be they should be you know categorized separately so You might ask based on these readings This should we not do screening magma mammography or should we not do it until you reach a certain age range and Here I think the authors and me personally don't have a specific medical recommendation to make Only to point out that though the way that this is being decided is is really fraught with these values And so there may be a need for a certain kind of Patient autonomy here or also stakeholder Consultation and consideration of a variety of perspectives before we make recommendations, right? What's clear is that the recommendations that are being made now that are in conflict with one another Are being made on the basis of somewhat different values and motivations and biases So that's what we're talking about today I look forward to hearing what you think about about this and the other issues raised by the reading so please Way in on discord I leave a comment on the video or on the discussion board And I look forward to seeing many of you in class Otherwise, I look forward to talking to you next week Bye