The Question Is: What Is the Question? | Understanding Medical Research

[Music] one of my former professors used to say this all the time the question is what is the question and that is the question we're going to be answering today in this brief video but first let's talk about our goals we're going to discuss the exposure outcome paradigm of medical studies and show that the question you want to be asked in a study is not always the question that is asked we'll also introduce you to something called causal pathways that you can use to analyze what's really going on within a study but before we get started i want to remind you of a jaunty little character from the history of saturday night live named pat so um are you gonna change your name which one are you asking either no no no pat and i don't believe in that kind of sexist tradition dewey hun oh yeah but wouldn't it be so romantic to have the same last name oh gee i i don't know what do you think i think we should i don't know do you well it's up to you well it's kind of your decision no i really think you should decide oh okay i think this is a conversation we should be having at home you saw in that skit that people couldn't ask the question to pat that they wanted to ask they had to kind of tiptoe around the edges because they felt uncomfortable asking you know are you a man or a woman so they asked all kinds of different questions and really that's what happens in a lot of medical studies it's a similar type of thing we don't feel awkward about the questions we ask it's not a social issue it's usually that we just don't have access to the data to answer the specific question we want so we always want to make an exercise of asking ourselves what did the researchers want to ask and what did they end up asking so let me give you an example remember that we want to link exposures to outcomes and we'll talk specifically about what that means in just a moment but all studies are trying to figure out what causes something in your health so here's a study looking at the association between gun shows and firearm deaths in california and nevada you can look at the results down here and what you can see is that in nevada but not in california two weeks after a gun show there was a spike in gun violence and firearm deaths the implication here would be that oh maybe the lacks firearm laws and the presence of the gun show loophole in nevada but not california leads to these deaths but what question did the researchers want to ask and what was really asked here well the question that they wanted to ask is is gun ownership dangerous right is the presence of guns in a community dangerous do guns cause death but they didn't really ask that right if we drill down what they were looking at is something much more specific does the time period surrounding a gun show in a state with lax gun show laws associate with the gun related violence rate in that state okay so when i phrase the question that way it may seem somewhat less compelling nevertheless that is what the study looked at so be careful of the spin that suggests that a study is answering a larger question than it really is i told you that we want to link exposures to outcomes so what's an exposure well an exposure can be any number of things it can be something we do to you like we might give you a medication we're exposing you to a medication or a placebo it could be something you are like your age or gender we might look at a study that says that men do things differently than women in that case gender or sex is an exposure or it could be a measurement like your height or weight or cholesterol level we could say oh you were exposed to high cholesterol all of those things are sort of typical exposures in medical studies now there are two different broad categories of exposures varying and non-varying so i'll give an example some exposures can change like your cholesterol level right you can have high cholesterol and we can change it through diet or exercise or certain medications but but some don't like your race or your dna we just don't have the ability to change that yet so studies of exposures that can vary are uniquely powerful because they may suggest therapeutic targets something that i can change about you and that is linked to some important outcome suggests that if i can just change it maybe i'll change the outcome too so this might lead you to ask wait a second wait a second so we shouldn't study race we shouldn't study how race impacts health or anything like that because you can't change race and my answer basically is is no you can't change race it's not a very interesting exposure to me what is very interesting to me is how people respond to race because that is something we can change so if i did a study that showed that black people had higher rates of heart attack i would be much more interested in why that is is it because of some inherent you know difference in cardiac function or is it because there are different treatments that are being applied to that race in other words how doctors are treating people based on their race because that is something i can modify so we talked about exposures let's talk about outcomes so outcomes are the juicy parts of medical studies right an outcome can be something that happens to you like dying right most outcomes in medical studies are not good um dying is one of them right but it's a very important outcome a lot of people want to avoid this outcome as much as possible but it can be something measurable measurable about you too like you're like your weight maybe the outcome of a study is weight loss or something like that now there are two flavors of outcomes too called hard and soft at least in in my sort of breakdown there are only three hard outcomes in the world and they are the rates of birth the rates of death and the quality of life okay so a study that looks at one of those three things we would say has a hard outcome but most studies don't look at those three things most studies look at something else like you know how much money you make or if you become famous or not or what your muscle mass is after a after an exercise intervention or something like that many of us would argue that these outcomes don't matter much except in so far as they affect the hard outcomes right i might want to have more muscle mass not because it's great to have more muscle mass but because it improves my quality of life because maybe i i feel better or i look better or something like that i'd like to have more money not because money is inherently good but because it improves my quality of life so hard outcomes you always want to look for in a medical study very powerful to see a study tied to a hard outcome compared to a soft outcome now this brings us to the concept of conceptual models these are also called causal diagrams and basically the idea here is to literally draw on a piece of paper what relationship is being explored in a study they look in a very simple case like this you have an exposure an arrow leading to an outcome okay so for example smoking leads to lung cancer or maybe there's a study that suggests that drinking coffee leads to success in life i have anecdotal experience that this is is not true because i drink a just obscene amount of coffee um but but you know maybe in aggregate the data suggests that coffee leads to tremendous success okay let's increase the complexity a little bit here we don't have to just connect one thing to another thing we can put things in the middle those are called mediators they're things that fall along the causal pathway so in the case of smoking our causal diagram here smoking leading to lung cancer there's something that smoking does that leads to the lung cancer right there's something that happens in the lung tissue itself lung damage in this case so we can say smoking causes lung damage and the lung damage causes lung cancer perhaps the body's response to the lung damage causes lung cancer we can also use in causal diagrams consider the evidence of a third variable a third factor something called a confounder and we'll talk much more about this in later lectures that links an exposure and outcome so take take a study that suggests that foie gras eating that fatty liver from geese is associated with longevity it's associated with a longer life would you go out and eat a bunch of foie gras well probably not because you might think you know what maybe it's not the foie gras that is leading to a longer life maybe just really rich people eat a lot of foie gras and really rich people also live longer because i don't know they have access to better medical care or something like that so we can put these third factors that link an exposure and outcome and it tells us something it tells us you know what maybe this arrow between foie gras and longevity isn't necessary at all maybe this whole relationship that we observe is just accounted for by the presence of having a lot of money in your bank account and there are statistical techniques to address these things once you've figured out your causal diagram so we're going to play a game right now and we can do this right in in the course of the the lecture here um we call it the exposure outcome game and so to get a sense of whether something is an exposure or an outcome we're going to pop this up on your screen and have you try your best that's very good what you learned here looking at the answers is that there are things that can be either an exposure or an outcome right so we could see something uh like um like cocaine use at the very bottom here that could be an exposure we might do a study where we look at cocaine use and match it to rates of heart attack or something or we might do a study that looks at you know um high school experiences and subsequent cocaine use where cocaine use is the outcome so it's not always hard and fast what is an exposure and what is an outcome but in any given study it should be very clear and it's your job to try to figure that out so a few take home points today we need to be honest about the question we want to ask and the question we're actually asking we want to identify the exposure and the outcome of interest in every study keep ourselves honest hard outcomes are the best outcomes remember birth death quality of life those are the big three and number four one studies exposure may be another study's outcome no hard and fast rules here except probably death and taxes i'll see you next time