 So yesterday, we surveyed knowledge synthesis and we found out what are the common steps to such knowledge synthesis projects and we looked at protocol creation, registration, which is potential for scoping reviews and searching for prior studies, which is hugely important and it's actually one of the steps in your protocol, right? So we also introduced the need for mapping research questions using such tools as Pico, Spyder, Eclipse and other frameworks like that. We will be starting to use the Prisma extension for searches, for reporting searches. That's going to be the basis, the focus of our discussion today. Just as yesterday, we used the protocol documents very much to contextualize why we need to do all kinds of things like doing prior searches. Today, we will really use the Prisma extension to do this, to see which parts need to be reported and why and how to do it. We are going to do a little bit of a review of sensitive and precise searching focused on why you need to be doing this. So I don't know if there'll be enough time or not. Let's try it. And then we'll be talking a lot about subject databases and distinctions between databases, the citation indexes, because that comes into the protocol as well and into the Prisma extension. So what we try to do today is make sure you have the tools that you need to be able to fill out your Prisma extension for reporting searches, which is absolutely necessary for systematic reviews, but it's hugely important for scoping reviews as well, right? And because doing protocols is so important in both, I really recommend that you do a Prisma extension for reporting searches, whether you're doing a systematic or scoping review. And we're going to be talking about database selection and most importantly, the Zora selection, because the Zoras is a very important part of doing your structured searching. And today we're really just doing the search step. Do we really need 15 minutes? Oh boy, we could use the whole day and you'll see in a minute, there's a lot of things and it will go fast. First we need to get started with our Prisma preferred reporting items for systematic reviews and meta analysis hub. So we go to the hub and at that hub, we're going to see that there are extensions. So right now, there are actually a couple more extensions than since I took this screenshot, so I'll have to update the screenshot just a couple of months, right? They've published at least two more extensions, but the extension for searching, which was published in 2020, exactly actually just over a year and a half ago, and which was partly developed by my wonderful colleague, Patricia Ayala, who works at the Gerstein Science Information Centers, Systematic and Scoping Review Collaboration. So this searching extension gives you a table that guides you through what to include when you're documenting your information sources, search methods, search strategies, a peer review of search strategies, and your record management for your search, for your knowledge synthesis project. This is very important. Here is the link to her original paper, but then you see the searching extension also showing up on the Prisma page. You can get the paper that explains the entire process, or you can just work with the table and the table is very much self-explanatory. So the table looks like this, and you think, okay, at least it's shorter than the table for scoping reviews. It's shorter, but the items are very meaty, and you'll see that some are fairly simple like listing every database name that you use for your searches, but some, as you see under item two, search strategies are actually very laborious. So you have to report full search strategies for everything and include all the search strategies for every database and information source, copied and pasted exactly as run. Not just one, not just one search strategy, not just the search strategy that best exemplifies what you'd like to do. No, whatever you run as part of your knowledge synthesis, you have to report in your search strategies report. So this is very, very important, and there are a number of other issues that need to be fully reported. For example, here, I'm pointing out that if you are using, if you have searched any of registries, such as TRIP that we saw yesterday, Johanna Briggs Institutes, Database, Prospero or any of the other databases were protocols deposited as well as full studies, you do need to report those. And what is this multi-database searching? We will be talking about platforms in just a moment. So everything we're covering today speaks directly to an item somewhere in this Prisma search extension. And citation searching is also important. And that is covered at least briefly today in our citation indexes. So I hope so far so good, right? We're working with our Prisma search extension. And that's where we're going to stay for the rest of this session. But of course, we're going to work with databases as well. So we're going to be looking at full search strategies today. So I've highlighted that item prior work we discussed yesterday. It's also important to remember that as you start keeping records of your searches and search strategies, you need to keep the date. Why is that important? Dates are crucial because, of course, databases change their content. Every day, new items are added. And another important thing is that as you'll start working with databases that have the Zorai, some of them, like, let's say, the full bells and whistles of it, Medline, update daily. Some, however, like other versions of Medline, update weekly. So if you don't have a date and if you don't keep track of which versions of databases you're using, you might not be able to reproduce your searches. And remember, systematic and scoping reviews and all knowledge synthesis are all about reproducibility. That's why dates are also important. So let's take a quick look at this search structure. So we have three lines in this. And by the way, you'll see these search structures exactly like this. In fact, a lot uglier in many other, in many, in many searches. So we have the first part of the search here. Social media, forward EXP or social near forward slash two, blah, blah, blah. There's a whole bunch of oars. And if you take a look at the, let's say, the core content of these items, it's media, social media, Twitter, YouTube, Facebook, blogging. OK, so what are we getting at here? What's the topic? What's the theme? It seems like we have synonyms for social media. So the first part of the search, the structured search, expresses the concept of social media and it uses proximity searching, for example, so that near forward slash two is an example of a proximity operator. So the first one's all about social media. The second part here of the search is all about professionalism, professional ethics, professional behavior, right? So it's a second chunk and it has all to do with professional ethics. So you see, we have two chunks. They're not connected. Each of them are separate. Look at chunk number three here. You have number one. So it looks like concept one. And you have and number two and a whole bunch of limits. So here we actually have the final search. So we achieve concept overlap and we introduce some limits. This is in a nutshell, the basic structure of structured searches. And we're going to spend the rest of the session looking at, first of all, databases that have the Zorai that allow you to add these forward slash exp terms because those are terms that come from the Zorai. These happen to be exploded, the Zorai's concept. And these need to be combined using Ores with keywords or what we call sometimes text words that are terms that happen in every other field of the record, because the Zorai's terms or subject headings, that they're curated controlled vocabulary. They only occur in controlled vocabulary parts of every record. So we need to combine those very useful but limited occurrence concepts with concepts that could be happening everywhere else in the record for every article that we are looking for. Plus, we have a few other things that we want to add, like specific limits, sometimes date ranges, and that happens only at the final search stage. So we'll be doing a lot of that today. This is something else that takes us back to our Prisma documentation, and here the Prisma flow diagram shows us what it is that we are doing in the searching stage. In the searching stage, we're going to be identifying through every database a number of records. And then after searching several databases that we select based on their subject coverage and relevance, we then deduplicate the results of these searches and we can use a reference manager for that, something like EndNote, or even just COVID-19 itself is not too bad, although it's much better to use EndNote. And then once we have our final set of articles, then we start screening. So a quick review or introduction of a couple more technical terms has to do with what we want to do, sensitive versus precise searching. So sensitive searching and precise searching are two different search strategies. First of all, what we may want to do is in some cases get a few great articles, but what we're doing in knowledge synthesis we're trying to survey the field. So we're looking for strategies that allow us to survey the field. And I'll ask you to tell me which is the most sensitive search. So in terms of precision of those two, it looks like B is more precise. So per number of total articles found or identified, the relevant ones, the green ones are a greater percentage. Therefore, B is more precise and I actually counted the number of squares, believe it or not, B is definitely almost twice as precise as A in terms of searches, in terms of coverage. Now, what about sensitivity? Could this be a sensitive search? I know I'm not defining sensitivity yet on purpose. Right? So let's just think about it. What about this? Ooh. So it seems to be covering more of the green guys, but there's still green guys outside of the ellipse. What about this ellipse? Is this a sensitive search? Well, how do we decide? We have to have some kind of an understanding of what sensitivity is. And it's a little bit trickier because it is the number of relevant found sources. So the green guys inside the ellipse divided by the total of relevant existing sources. And here's the problem. We don't know how many relevant existing sources there are at any given time. We can only kind of estimate that we expect there to be a lot or not too many. So given that sensitivity is measured against total number of relevant existing sources, do you think C is better or D is better? I actually would like you to think about it for a second because they're a little bit counterintuitive. And this kills researchers when they're starting. This is something that really, really upsets people. Look, C is absolutely better. D is not that much worse. So what does that mean to us in practice? Obviously, we don't have ellipses or circles or these blobs when we release our searches. We have searches where we use our terms and we use Boolean operators and other operators. And of course, we use the Zora. So what does that mean in terms of the search? It means that, unfortunately, wider, broader searches that are more inclusive, that let in more of the not useful stuff that we see here in the D are actually still pretty good. So this is a really big lesson to take. Right? There are two corollaries of this. First is the precision sensitivity tradeoff. Pardon me. You cannot write a search that is both really precise. That only gets you the stuff that you want, like the little blue ellipse and highly sensitive, like our C ellipse at the same time. Right? You can't do it. So if you want to be inclusive, which is what we want to do in knowledge synthesis, we have to write as expansive a search as possible. So that means as many synonyms as possible. That's tough. And of course, there's also the relevance trap that if all your results are highly relevant, oh, boy, you're in trouble because you're definitely missing stuff. So it's really important to keep that in mind. And with knowledge synthesis, a bigger ellipse is always better. So in knowledge synthesis, the circle is better than a very tight little ellipse.