Welcome back. This is the final section of the lecture on finding the evidence, searching principles and I'm going to be talking now about documenting your search and just a few final words in conclusion. There are two main areas that we're going to talk about related to documentation. The first is you've been asked to purchase and used Endnote. This becomes your database of records of all the citations that you collect from the various sources. Endnote is one of the pieces of bibliographic management software that's available to you. Some of you may be also familiar with RefWorks, which is also available free at Johns Hopkins or Reference Manager. There's another tool called QUOSA that is also available to you, at least while you're at Hopkins free to download. QUOSA's advantage is that it allows you to do automatic, full text retrieval for at least some of the citations. Any of these sources have fields that you can use to store key information about the search that you've ran, whether you've included or excluded a citation. And that brings me to a review of what it is that you need to record to document the search that you're doing. You need to record when you did the search, the day and the month and the year, which sources you used, which databases. The exact strategy that you used. So in PubMed, for example, you would record something that looked like what you saw in the search details field that I showed you some slides ago. You need to identify, which trials registers you searched and the search strategies you used for them. Any communications you've had, any bibliographies that you've searched or citation tracking process that you used in the web of science, for example. You also need to be able to document what you've found specifically. PRISMA is a standard for transparent reporting of systematic reviews and meta-analysis. It's an excellent source. It's a nice tool that you can use to sort of get an overview of what's required to do a high quality rigorous review. I've given you the website here. There are various articles that present this information. There's a 27 step checklist available and also a flowchart. There are now over 125 journal editors that have signed on to the PRISMA standards and if you wanted to publish a systematic review or meta-analysis with them, they would expect to see certain features. The flowchart that's shown here, shows you the various pieces of information you need to have recorded. On the left at the top, number of records identified through database searching. So you would add up the total number of hits that you found in PubMed, Endbase, Cochrane, any of the subject databases and put that in the box on the left. Box on the right, you would include the number of records that you found through any of the review of references lists, tables of contents, other snowballing techniques. Many of these databases include the same journal and there's quite a bit of overlap, for example, between PubMed and Endbase. So you're going to be removing duplicates as a next step and you need to keep track of how many duplicates you remove, because you're going to need to report that, then you'll be doing a title abstract screen. And you'll have the number that you screened, the number that was excluded at that level, the number you needed to review as a full-text. And even at that point, often there are additional citations that are excluded from your final analysis. I started this presentation back in section one with a story about the impact that doing a high quality search can have on your research and even the life and health of your subjects. This table is taken from another study that I find quite compelling. The researchers search the literature for randomized control trials that study the effectiveness of aprotinin, a blood substitute used in cardiac surgery. I know it's kind of hard to read, but what this figure represents is all of the studies that they found. The oldest is at the top and the newest one that they found is at the bottom. The intention here is to represent how well the various authors did at citing the older literature. If they cited an article, the square is in black. The sort of half grey articles down the diagonal are ones that you could reasonably expect the authors not to cite, because they were published within a year of their study. So at the top, the first study that ever came out doesn't have any other studies to cite. As you go down, you would expect to see more and more black. But as you can tell, there's a surprisingly low citation rate. The most frequency cited study is that original study. The study that has the most in route subjects number 17 is only cited 7 times. Why is this important? Every single one of these published studies had positive findings. They all showed that aprotinin was more effective than a comparative treatment. Researchers did a meta-analysis of these results and after study number 12, the results were unequivocal. Between that time and the last study cited here, over 4,000 people were randomized to prove the same point. That means that at least half of them did not get what would be know to be best treatment had the researchers actually relied on the literature that had already been published. In conclusion, thank you very much for your time and attention to this lecture. We have covered the various components that go into creating a comprehensive, sensitive search to support your systematic review and meta-analysis. That includes developing search strategies for large, electronic, bibliographic databases. We've introduced an approach that includes breaking your research question into components using PICO. We've covered the variety of electronic and sources for peer reviewed literature and gray literature and the other key point, starting from the beginning is document what you do when you do it and you will be well on your way to producing a great systematic review. Thank you for listening and talk to you soon. [MUSIC]
