So here's these articles we talked about for Ms. Jones. I printed them out. >> Fantastic, so that's great. We had access, you were able to get the full text for both of them? No, it looks like you only printed out the abstract for the Cochrane, or the summary but you've printed out the full text of the other one. >> Mm-hm. >> And that's good, we'll come back to the Cochrane in awhile. These things, I don't know if you notice how many pages it would've taken off. >> It was very long. >> It was about 160 pages so no need to actually print out 160 pages. The great thing about Cochrane, as we'll come back to, is they do a great job with the summary and you can do that. But you did print out the whole study from the piece of the systematic review. And I think that's a reasonable idea just given that we got to kind of dive into some of this right now. >> Okay. >> So let's talk a little bit about why we picked these too. So both of these are primary or secondary literature. >> Secondary. >> Secondary, okay, and how are you defining secondary literature again? >> It's kind of an upper level view of an assessment of the literature on the certain topic, not a specific study in randomized control trial. >> Good, so its definitely an upper level view. It includes a lot of these randomized control trials. It may include other studies not just randomized control trials. It may include controlled clinical trials where randomization isn't use. Or when there is, it sometimes you can even put in cohorts studies into these as well but you got to be careful with how you pull the data but at a higher level it's looking at multiple studies simultaneously, pulling them looking for quality and then determining what the bottom line conclusions are and implication for practice. That's what these systematic reviews do. It's also systematic in another way that they should be systematic in looking at the literature to decide what studies they're going to include. So what do you know about, what are kind of the big biases when we do systematic reviews? What bias might you be worried about? >> Publication bias? >> Great, fantastic, publication bias is probably the most important bias that you think about. So what is publication bias? >> Articles are often more likely to get published if they find a positive result versus not finding a result. So it's important to look at studies that haven't been published. >> Great, fantastic, that's exactly right in fact most of the studies that are out there show some sort of difference. It's not as common, that if you do did a study that compares one agent versus another agent, or one agent against placebo, if there was no benefit, is it likely going to be published. And the answer is it is less likely. And so systematic reviews also implies not just the way the studies are put together but how you searched for those studies. And you're not doing the searching, I'm not doing the searching but there are some people that are doing the searching and the question is, did they use a good method? And are those the right people to be doing the study? So they should be looking for unpublished studies and that's hard to do. You got to pick up the phone, you got to look at controlled clinical trial registers to see what are all the studies being done out there. And figure out whether they actually made it into publication. You gotta search multiple databases, not just the US database, not just the European database, you gotta search large international databases as well. So, these are kind of clues that you can look at to evaluate the quality of a systematic review. >> Okay. >> That's a lot of work for an individual practitioner to do. And so I'm not going to advocate that you do that, but there maybe some queues, some sort of quick ways to look at a systematic review that might be helpful. I will say that if you're using Cochrane, they do all of this in the background. Its almost like you're going to the car dealership its all what's happening behind the scenes and the question is do you trust that dealership? Do you trust them to do it? And personally the Cochrane collaboration has done such an amazing job of putting these together for a number of decades now but the vast majority of practitioners ad clinicians trust their process. So that takes owners side off of you to have to look into the details so you don't have to look at the pages three through 163. You can just look at pages one through three and that's a nice thing. So we'll do that last. >> Okay. >> Let's do a little bit of diving into the piece here. So what I advocate is to do a couple things. Read the title, make sure it's the right title that you do which you did and was the reason we selected it. Read the abstract, I think it's reasonable to read the abstract. This tend to be more structured than they were in the past and so things are laid out nicely in the abstract. You want to kind of get a sense of where that goes. And then if you don't have much time, I honestly recommend that you read quickly the method section next because that gives you an idea of the quality. And then go to the big tables, look for the tables that's looking at what you're interested in. Remember we made that question at the beginning but we were interested in women, we were interested in topiramate, we were interested in other classes of agents, we were interested in the outcome of frequency, not necessarily duration. So we got to keep that in mind to make sure that this study is actually answering the questions and if it's a high quality one. So, let's look at the methods. Well, we can look at the abstract here. First of all, title looks okay. Feel okay about that? What about the abstract? Is there things in there that make you think that maybe this is along the lines of what we're interested in? >> So talking about their methods, it looks like they did assess the risk of bias. >> Good, yeah. >> Which is good. And. >> And when you look at the results sections here, they're looking at multiple drugs or just topiramate? >> There are 59 drugs. >> 59 different drugs from 14 different classes, so it's lots of drugs that they're looking at. Did they look at other? Yeah, sorry. You were going to say something. >> But they included topiramate and propranolol, which we've talked about. >> Yep. >> And other beta blockers. >> Great, so that may be helpful for us as well. Do they include, you'd mentioned tricyclics. Does it look like tricyclics are in here anywhere? >> No. >> No. >> Not really. No, don't see anything that's a tricyclic. So then you have to make a decision, am I done or am I not done? And is this good enough if it doesn't include tricyclics, what do you think? Should we keep reading? >> I think we should keep reading. >> Yeah, we spent all this time and got beta blockers and we've got topiramate in there, so that seems to be reasonable. Great, so let's look at the results of this table because it really does look like we've got the two lines that we're interested in. So here's the relative risk, you can see that the relative risk is two for both of them. >> Okay. >> Now, a relative risk of two doesn't mean a relative risk increase. It means that the number of people who are going to have a reduction in frequency by 50% is double in those who are on the drug versus placebo. So two is actually a good thing. Sometimes you gotta think about relative risk if it's a bad thing if it's higher, but in this case it's a good thing. But it's the same for both topiramate and beta blockers. If you look at the absolute risk difference, this is the exact subtraction of the risk of one versus the risk of another, that is pretty similar as well. 29%, in topiramate 22% in beta blocker. Remember, they're comparing the drug versus placebo in both of those. >> Okay. >> So then we get to the million dollar question, what is the number needed to treat? And in other words, how many people do you need to put on a beta blocker or topiramate in order to reduce their frequency by 50% of having migraines. And so what is that number for topiramate and beta blocker, what do you see here? >> Three versus four. >> Three versus four, so what does that sound like to you, good number, bad number? >> Sounds like a pretty reasonable number based on some of the other numbers I've seen for other agents. >> Yeah, good. So yeah, three and four are pretty low numbers. You want to low number, you want to be able to treat a low number of people in order to prevent one patient from having or at least have one patient number reduction than the frequency of their migraines. And if you look at the confidence interval, that's three to six on topiramate, three to seven on beta blockers. If you just compare those two, do they seem similar to each other or different to each other? Three versus four. >> Pretty similar. >> And why would you say that? I was always taught that three and four are different numbers. >> So the confidence intervals are relapsed so they are more similar than you might think. That's good, that's a good way to look at it. Now it doesn't mean statistically they're not similar but it does imply that the confidence intervals being three to six and three to seven that they're pretty close to being similar. So that would then tell us that is beta blockade versus using Topamax different in its efficacy at preventing migraines? What do you think? >> It doesn't look too much different based on these data. >> Okay. >> They look pretty similar. They both seem to result and the outcome, we're interested in it similar rates. >> Okay, good. All right, I'm seeing that my, our patient is really kind of waiting so we gotta make a decision here. >> Okay. >> This is really quick so sorry I'm going to get a little kind of focused here, this was nice to be able to go through all of this but we gotta decide now. We've gone through all of this, she wanted topiramate, beta-blockade was something that you had raised as a potential, now we've got evidence that shows boy, using a huge systematical view of the literature that may not be that different from each other. What are we going to do, Amy? What do we do about this patient? >> Yeah, I think we have to think about what she wants. But then also side effects of the different drugs, cost of the different drugs. >> Okay. >> Those kind of weigh into the decision. If they able to seem to be a good option in terms of what we're looking to do. >> Okay, so I'm going to ask you, on a scale of one to ten, ten being very comfortable, one being not at all comfortable. I'm going to ask you to tell me how comfortable would you be if I walked in there and said, we're going to start Topamax or if I'm going to walk in there and say, we're going to start Inderal, a beta blocker. So where would you be on Topamax in terms of comfort with me saying that? >> Very comfortable, I'd say. I would say I know Topamax has a few more side effects, so maybe in like an eight. >> Okay. >> because there would be the question of, how she would tolerate it. >> Okay, and then if I walked in there and said I know you want to do topamax but inderal is pretty equally efficacious. We're going to recommend starting that. >> Probably around the same maybe a little bit higher but then again she's more leaning towards topamax. You have to weigh that into the equation as well. >> So she would need to be explained and I think at either decision what's really important about all of this is how we process this decision with the patient. At the end of the day it's honestly not our decision. It's the decision of the patient. Because we can say we would like this but if the patient goes and doesn't take it then what are you going to do? So I really do think that it's really incumbent on us to both walk into that room, explain the data on both sides, and give her a sense of where we're leaning but allow her to make this decision. Full side effect profile, full transparency on cost. Does that sound like a reasonable approach? >> Sounds great. >> Great. Amy, you did a great job. You really walked through the literature and really had a great approach to thinking about this very complicated question and hopefully you learned a little bit. >> Yeah, this was very helpful. >> Great.