In the previous tutorial, I showed you how to create a concept table for your research question, which includes all the potential search terms you've identified by your preliminary scoping and pearl growing searches. I've also showed you this table, which lists all the databases you've decided to search for the project. Well the next step is to take the terms you've identified, and search with them in the databases. Unfortunately, even though their general concepts of searching tend to be similar, no two database products can be searched in exactly the same way. There are always nuances and idiosyncrasies in terms of their user interfaces, features, and search query syntax. Sometimes even the same database, if implemented on different platforms, can be very different - such as PubMed and OvidSP MEDLINE. Because of these complexities, mistakes are constantly made in search strategies, even for published articles and reviews. This is a report on the electronic search strategies from the Canadian Agency for Drugs and Technologies in Health. This report lists some of the common mistakes made in search strategies. It's a pretty long document - I just wanted to draw your attention to the table of contents. Take a look at these headings in Chapter 6. These are some of the common mistakes such as spelling errors, logical operator errors, missing subject headings, missing natural language terms, missed spelling variants and truncation, et cetera. If these mistakes were made in systematic reviews, they will almost certainly detract from the validity and reliability of the review's conclusions because, as we know, the quality of systematic reviews depends on rigorous search strategies. In these videos, I will introduce some basic concepts in building good search strategies, and demonstrate them using a couple of popular databases. I cannot possibly cover every single database you may end up searching, but with the concepts introduced in these tutorials, and a good help document from the database itself, you should be able to figure out what the best search strategy is for your topic and your databases. And don't forget, your medical librarians are always available for help. When in doubt, don't hesitate to use them as a resource. In this video, I will mostly use PubMed as an example to demonstrate some searching concepts. So here's the PubMed interface. I know most of us can't wait to start our search by simply typing terms in the search box here - and it usually does a decent job returning relevant results. However, I just wanted to show you that PubMed, just as many other databases, implements an automatic user input interpretation layer, in this case known as the "Automatic Term Mapping" mechanism, which takes your search terms, analyses them, creates a query that it considers the best for your search, and sends it to the database. So, for example, in our search for "heart attack", if I scroll down to the search details section, you'll find that this is the actual query that was sent to the database. So even though we only typed in "heart attack", it knows to map the term to the standard medical subject heading "myocardial infarction". All of this is good, except that, it is all computer generated, and is therefore not a hundred percent reliable, and certainly not as controllable as what a search for a systematic review typically requires. Since we're demonstrating searches for systematic reviews, we're not going to simply feed the search box with terms. Instead, we're going to tell it to do specific searches with specific instructions. To do that, we need to understand a little bit of query syntax. The simplest query syntax you can apply in almost any database is "field qualification", that is, how do you limit your search to a specific field in the database, such as title, author, abstract, et cetera. PubMed's field names are well documented in its help file. So click on "Help" on any PubMed page. This is a very long page, but fortunately there is a convenient table of contents on the right, so I'm going to click on the "Search Field Descriptions and Tags" - this is a list of all the field names and their abbreviations you can use to limit your searches to specific fields. So, if I want to search for articles by an author called Smith JJ, I can type in Smith JJ square bracket author and square bracket. You can either use the full field name, or its abbreviation, so this Smith JJ square bracket AU square bracket would run exactly the same search. One thing to mention here is that with the automatic term mapping layer, PubMed is getting more and more forgiving to imperfect query syntax. Your query is now case- insensitive, and it doesn't matter if you have a space or not between your search term and the field qualification, the opening square bracket. What's worth special attention here are these MeSH specific field names. MeSH, or Medical Subject Headings, is the controlled vocabulary system used to describe the subject of each item in MEDLINE. As we discussed in previous tutorials, if a database supports a controlled vocabulary system, it helps to solve the "synonym problem". So in PubMed you can just type in myocardial infarction square bracket MeSH Terms square bracket or simply MH, and it will return articles that are indexed under the MeSH term "myocardial infarction", which, theoretically, includes all the articles that uses the synonymous term "heart attack". Now, every one of the articles in MEDLINE is indexed with a couple of "major terms", which describe the primary topics of the article. So this will distinguish articles that are primarily about myocardial infarction from those that could just incidentally mention myocardial infarction. In PubMed, we can quickly limit our search to articles that have our term as a major MeSH term by using the MeSH Major Topic qualifier or simply Major, M-A-J-R. Again, the automatic term mapping mechanism will try to help, even if you type the word "major", M-A- J-O-R in the square brackets - it knows which field you are actually referring to. Another thing to know about MeSH is that they are typically assigned at the most specific level possible. So if an article can be described by the term "renal hypertension", it will not be assigned the more general term, "hypertension". And articles that do get assigned the term "hypertension", are actually on the general topic of hypertension. This often requires that when we run our searches with higher level terms, such as "hypertension", we also include all of its lower terms, such as "renal hypertension". This process is know as "exploding" in some databases. In PubMed, exploding is automatic with a MeSH term search. So when you search for hypertension MeSH terms, it already includes renal hypertension MeSH terms. If you don't want that to happen, you have to specifically say hypertension mh no exp. Each MeSH term may have multiple facets known as "subheadings", and articles may be indexed with specific subheadings. For example, an article may be specifically about the diet therapy of hypertension, so we type in hypertension slash diet therapy. PubMed is smart enough to figure out that the part after the slash is the subheading, and the preceding part is the MeSH heading. In this case, you don't have to specify that this is a MeSH term search. So all of this assumes that you already know that "hypertension" and "myocardial infarction" are standard medical subject headings. What if you don't know? PubMed comes with a MeSH query builder, that will map any term you put into it, to an existing MeSH term if possible. So we can change the database drop down here from "PubMed" to MeSH", and type in any term you want. So if I type in "heart attack", it'll be mapped to MeSH term "myocardial infarction". Here you also have a chance to choose a subheading, restricting the search to major topic, and disable exploding. You can now click "Add to Search Builder", and you can see that our search query is constructed with the proper syntax. Now if I click on "Search PubMed", this search will run in PubMed.