Welcome back to section C. In this section, we're going to talk about elements of the question. We've been talking about how to classify your question so that you know what type of study design you're going to be using in your systematic review. And then we talked about thinking about is it a broad question or narrow question? How do I want to frame it exactly? And now we're going to talk in little bit more detail about that framing process. What are the elements of the question? And by that I mean, what types of patients are we interested in including?What types of interventions or comparisons do we want to make? And what outcome are we interested in? So, here, we're talking about the steps and the practice of evidence-based health care. So, if you're providing health care as a health care practitioner or as a public health worker, it's not enough just to do the research or to synthesize the existing research. You have to take the next step and actually implement what you find. So the first step is to frame the question. First we have to have a question. Then we have to find the best evidence that supports the answer to that question. You want to critically appraise it, and then you want to apply that evidence. Either make a new policy, or encourage people not to do what they're doing already, or something like that. And so that's a very important part of this course. We'll teach you how to do the systematic review, but unless somebody's applying that evidence, it does no good to do all this research. Well, we're just going to be focusing on number one here, framing the question, but if you don't get this right, you can't apply the evidence. So again, I emphasize how important it is that we get this part of the whole process done right. So when you frame the question, you consider the patients and the interventions and comparisons outcomes you want to address. You want to classify the question, and you want to then identify the appropriate study designs for addressing the question. And we've talked about those things so far. Here's a little more detail on those things. Figure out what the question is, so do it in a very, sort of broad way. Sit with your group. Talk about in very sort of. You didn't realize it before this, but in very broad terms. And now you realize that the question is actually harder to develop than just that quick one sentence that you're throwing out. But then think, well, what type of question is this? And so what type of evidence will we look for to address that question? So here's an example. We have a patient who has amblyopia. And you're a doctor, and that's basically with a lazy eye. It's usually a child. Sometimes it can be an adult, and they have a lazy eye. And let's say that patient is coming to you, or the parent's bringing the child to you, or you've seen this detected in a school screening, something like that. You as a health care provider might have many questions going on in your head. What's the cause of this condition? Were there environmental exposures? Is it because of the parents and something the kid inherited? What's the likelihood this amblyopia's going to get worse? What if I do nothing? Maybe it'll get better on its own. What if we put a patch over the eye that's functioning better than the lazy eye? Will that get the lazy eye functioning better if the person is dependent only on that lazy eye? That's a therapy type question. Maybe there are harms of patching, and in fact there are. We know there are harms of patching. If you're a child, and you're wearing a patch on your good eye. Let's call it the good eye. It's not the lazy eye. You might get teased on the schoolyard. That doesn't sound like a harm maybe to those of us who are adults, but when you're a kid, getting teased is really a bad harm. And you don't want that to happen to your child. So, that is a potential harm, and you have to think about, well, maybe we should put skeletons on the patch or something that makes it look cool. And not just like a Band-aid. So, what potential differences in outcome are there if a patient's treated earlier rather than later? Should I just wait for this kid to see if the lazy eye will fix on its own? Should we do screening in preschool to try to catch it much earlier so that we have better outcomes over a larger proportion of the kids? So, that's both a screening study and a prognosis study. You want to know what happens if you don't patch, what happens naturally, a natural history type study. And then also you might have a screening question. If I catch this earlier, if I start doing preschool screening to catch amblyopia early, will this make a difference in the outcome for the children? Will more kids be helped? And what kind of change am I looking for overall once we decide that treatment's necessary? What type of outcome am I looking for? Can these kids be 100% cured, or am I just looking for sort of a middle of the road half lazy eye? What am I expecting? And this encompasses several questions, but it might be the question that you would have as a parent or a doctor. So, as you can see there are a lot of questions in there, and you have to decide which question you're trying to address in your systematic review. You can only address one at a time because they require different types of study designs. So once you have your well-formulated question, and we aren't quite there yet, but we're starting to narrow it down, then this can be used for your whole systematic review process. It will help you determine which criteria you want to select your studies, your eligibility criteria. It will help you to develop your search strategy when you search bibliographic databases when you do hand searching, and it will tell you what types of data do you need to abstract. So, you're not just going to find studies that are relevant to your question. You're going to be abstracting data so that you can combine the data in what we call a meta-analysis. All right, so what are the components of a well-constructed and answerable clinical question? Remember, we want something answerable. And I've mentioned before PICO. People, interventions or exposures, comparisons and outcomes. Here it is. Patients or populations. Interventions or exposures. Comparison groups and Outcome. And that spells PICO. And it could be P-E-C-O if we're talking Exposure. I never knew whether to pronounce it pie-co or pee-co, but since we're talking about epidemiology type studies and exposure as well as intervention, I think pee-co is probably the better pronunciation since it encompasses many different types of studies. And I'll just mention here that you can do a systematic review of any type of study. Systematic reviews aren't only of randomized trials or intervention studies. A systematic review is for any type of question you come up with. However, there will probably be a limitation on which study designs you'll include based on the type question you're asking. So, let's talk about the types of people, or populations, or patients. What we mean is, we wanted to find the condition or the disease that we're interested in, and we want to include explicit diagnostic criteria. So if we are looking at people who had heart attack, how do we define heart attack? Often what you'll see is that systematic reviewers define their population or their disease or condition the way the authors of the original studies that they include in their review have defined it. So, if a study defines heart attack using specific criteria from, let's say, I'm making this up, New York, and you might define a heart attack slightly differently. Say in a different time period. That may be all right if you as the systematic reviewer say, however the authors defined heart attack, we're considering that okay. Or you might say, we're only including studies that defined heart attack this way. So you have to decide again whether to be broad and more inclusive or whether to really define your disease or condition in a specific way so that your audience and the question you actually want addressed are properly addressed. You would probably say at this point what type of setting you're interested in, although some people separate out setting. So, are you looking at people who are based in the community? Are you looking at hospitalized people? Are you just looking at one age group, one race or ethnicity, sex, outpatients, and so forth? So again, if we're talking about secondary prevention of heart attack, do you want people who are still in the hospital when they're given aspirin, or they could be community dwelling? Are you going to accept people who are in nursing homes or intensive care? So, what setting and population are you interested in? When we think about the exposure or interventions, we're also thinking about risk factors. That is, what's the timing of the exposure or the intervention? How long after that first heart attack, for example. What route of administration? What dose are we talking about? And how long should people be treated or exposed? So, for example, if you're looking at the effective exercise on falls. How long should people have to have exercise? That is, is going to exercise class once a week for half an hour okay? Or do you define exercise as half an hour of at least walking every day? So, you have to define your exposure or your intervention very carefully as part of your PICO and ultimately your inclusion criteria. When you're thinking about your comparison group that's a comparison with the intervention of interests or the exposure of interests. So let's say we're interested in people who work at a chicken processing plant and wondering if they have increased adverse outcomes. Let's see, I'm going to make this up, increased infection with certain type of bacteria. And so you'd have to think, well, how do we choose our comparison group? People not working at a chicken plant at all? Or are we interested in just the people who are exposed directly to the chickens? And maybe their comparison group would be people who also work for the company but might work in an office. Or are they the appropriate comparison group? Maybe it's people in the same region who don't work for a chicken processing plant but work manufacturing wire. So, I'm making up these ideas, but you have to think about, just as you would in any epidemiologic study for example, what makes a good control group. And what you'll find is that in systematic reviews of epidemiologic studies, you will often see more than one type of control group or comparison group because you can get different results, as you might imagine, depending on the comparison group used. For clinical trials, comparison groups are typically placebos or standard therapy or no treatment depending on what type of intervention is given. There are ethical issues, of course, for clinical trials because you have to make sure that it's okay that your comparison group either receives no treatment or a placebo. For example, treating people with severe depression with placebo might be considered unethical, and you might have to compare your test intervention to some sort of standard therapy. You'll find more out about this as you do your systematic review. It's extremely challenging especially for epidemiologic or etiology studies. Now let's finally talk about the outcomes that you select. Outcomes are getting harder and harder to define, and I will touch on this a little bit. But the things that you need to think about as you define them is, first of all, is this outcome important? Is it important to the patient or the consumer, or is it just important to doctors? So for example, lab values are viewed as important to doctors, but perhaps less important to patients. Patients might be interested in, how soon can I go back to work? Do I have to leave work early? Do I have nausea? That might be extremely important to a patient and less important to a doctor. Probably everyone's concerned about death though, so that's a good bet that it's patient important as well as important to a caregiver. If you're thinking about outcomes, you might also want to know, when are they measured? Does it make sense to test the effectiveness of a class for teenagers teaching them about using condoms to prevent sexually transmitted diseases and only follow them for two weeks? That probably doesn't make too much sense. You probably want to follow them for about a year to see whether the class did any good at all in getting them to use condoms, but I have seen studies that followed them just for two weeks. They also looked at knowledge retention as an outcome, and I'm not sure that knowledge retention is really an outcome that makes much sense. And this is the type of issue you might want to consider when you're deciding your outcomes. Is knowledge retention as important as say, using a condom? For preventing sexually transmitted diseases? Probably not. And knowledge retention is a typical outcome that you'll see in all different kinds of intervention studies. So you want to know, when was it measured? What was the outcome? Was it important to the person? Does it make sense? What was used as your measure? Did you use mean? Or did you use change from base line? And, what sort of test was used to actually determine your outcome? So if you're doing a vision test, did you use that E chart or did you use some other type of chart for measuring how well somebody sees? And if it's a baby, the E chart sure won't work because they don't know their letters. So, deciding on which type of chart was used is also part of deciding on your outcome and how you're going to measure it. So, here's a well-constructed and answerable clinical question. And I'm interested here in amblyopia. As we mentioned before that is the lazy eye. So, for preschool children with mild to moderate visual acuity impairment, are glasses or spectacles and patching effective in improving visual acuity compared with glasses alone or no treatment? And so this is a well formulated question. You can see that the preschool children with mild to moderate visual impairment is the population. Your intervention is glasses and patching, and your comparison group is glasses alone or no treatment. Your outcome is visual acuity. Now this question says nothing about the timing of measurement of visual acuity or how long the glasses and patching are given, but that would be given in more detail form later on and not in the question itself. And I'll go over that so that you can see exactly what I'm talking about. We'll cover examples a little bit later on, and that's where you can see how we actually define the details of the question even though the question itself is about one sentence long. Well that's it for section C, and next we'll talk in section D about refining the question.