The smaller and simpler the system the easier it is to find what you want. It is pretty easy to find pertinent Cochrane systematic reviews from the the 9,000 in the Cochrane database. It is quite a bit harder to find what you want in PubMed with its 30,000,000 articles.
The larger and more complex the system the more time it will take to create an effective search.
The ideal search finds all the articles which answer a question and no more. In real life no search ever does that. Time and effort spent correlates with how close you get to the ideal.
Precision and recall are often used in information science to describe how well a search functions. You can compare them to sensitivity and specificity. A very precise search returns a small number of useful results and is likely to miss other relevant articles. A search that has high recall finds nearly everything about a topic. Much of what is found is irrelevant. A high recall search requires more time and effort to sort results. Take aways? There is a tradeoff between precision and recall in searches. There is no such thing as a perfect search. It is necessary to consult multiple sources to lower the chances of missing important papers. Examining related articles and reference lists of pertinent articles will identify additional relevant information.
One of the cornerstones of Evidence-Based Practice (EBP also called Evidence-Based Medicine or EBM) is the PICO model. PICO is a framework which helps questioners frame a clinical question.
The rest of this guide discusses methods to create effective searches for databases or search engines.
The search process is iterative. It is rare to create a good search on a new topic in your first attempt. Trial and error will lead to an improved search
Formulate the Search: To use a database effectively, think before you type!
1. After some background reading your original question may need to be reformulated; it may begin to look like several questions or you may answer it with background information. What is a problem with not looking at foreground sources?
2. Create a simple, searchable question. This is the single most important (and difficult) step!
What does “simple searchable” mean? Avoid unnecessary detail. Carefully choose your words. Be willing to change words if they do not seem to work when you do your search. To build a simple searchable question try out the following 4 construction methods:
a. Make a simple sentence of your question. Use the subject and object as search terms, avoid the verbs as search engines are usually no good with those. Search engines do not understand cause and effect or time relationships.
b. Write down a list of all the concepts within your topic. Select the most important two or three as your search terms.
c. Most time consuming, organized and effective method: Create a table of concepts. Select two or three. Think of synonyms for each. String all that together and use that as your search.
d. Use the PICO model. This approach is derived from Evidence-Based Medicine (EBM), or more properly EBP (EB Practice). It is probably most useful for treatment questions.
What these various methods look like follow below. At the bottom of the box is an example of each method in action.
a. One searchable sentence:
b. Concept List: Examine your question, choose the most important 3-4 concepts:
c. Concept/synonym table: