OK, so you have a research problem that you want to solve or answer using evidence based in the literature. You need to find the right literature and capture it by spreading your net wide, and in the right places. You need a strategy for searching the literature….a Search strategy if you will. I hope that these 5 steps will get you to where you want to be.
What is a search strategy?
1.A structured organisation of terms used to search a database
2.A document that shows how terms combine to retrieve the best results
3.Something that must be adapted for each database you use
4.Something which is tailored to the question you are trying to answer
5.A good Search Strategy is something that takes time to refine
Different ways to search the literature
1.Electronically
2.Manually
3.Snowballing of the literature (going from reference list to reference list to find what you need.
4.We can do a rapid review of the literature or an exhaustive one
5.We can see what other published literature reviews have done and how they have found their literature for ideas.
6.Ask! (Librarians, authors etc.)… authors of great papers often know of other papers you may be looking for…why not ask them?
Step one: Define your research question or ‘problem’.
First…we will use this as an example: Does hand washing among midwives reduce postnatal infections?
| Example: | |
| P (Problem or Patient or Population) | postnatal infections |
| I (intervention/indicator) | hand washing |
| C (comparison) | no hand washing; other solution; masks |
| O (outcome of interest) | reduced infection |
Whilst PICO can sometimes be seen as the go to tool for formulating your question..don’t be limited by it. Here are a few other tools to help your formulate your perfect research question…
Methodology e.g. questionnaires
Issues e.g. ethical decision-making
Participants e.g. midwives or patients
—————————————————
Setting – Where? What is the context?
Perspective – For who?
Intervention (Subject of Interest)– What?
Comparison – What else?
Evaluation – What results?
————————————————————
Sample
P I Phenomenon of Interest
Design
Evaluation
Research type
————————————————————
Client – who is the service aimed at?
Location – where?
Improvement – what do you want to find out?
Professional – who is involved in providing/improving the service?
——————————————–
Context
Intervention
Mechanism
Outcome
———————————————-
Expectation—What do you want the information for?
Client Group.
Location.
Impact— What change are you looking for? How is this being measured?
Professionals.
Service—For example, community services, birth centres or accident and emergency.
Step two: choose which databases you will search
Different search databases should be searched separately as they each have their own dictionaries of terms and keywords. Each database is tailored toward a particular topic of interest. The following set of databases relate to healthcare topics.
1.Web of Science (strong coverage which goes back to 1990 and most of its journals written in English)
2.Scopus (Covers a superior number of journals but with lower impact and limited to recent articles)
3.CINHAL (Prime source of nursing and allied health literature)
4.Pubmed & MEDLINE (Great starting point for any health or medical literature search.)
5.Cochrane (The source of systematic reviews)
6.NHS Evidence & The TRIP database (Search a limited number of high quality sources)
7.PsychINFO (Prime source for psychology and psychiatry literature.)
8.AMED (Allied and Complementary Medicine Database)
9.HMIC (Health Information Management Consortium – great information from DoH and Kings Fund)
There are no strict rules as to how many databases you should search. That would depend on how thorough you are trying to be. Also, many databases will pick up duplicates for you…which you will later need to delete.
Step three: Identify and map your key concepts
A concept map is a visual representation of concepts within your research question or ‘problem’ and their relationships to each other.
To create a concept map:
- Write down the main concepts which relate to your research question and circle them on a blank page.
- Write down other words/concepts and ideas which relate to each of your concepts in groups. Draw lines between concepts to show how they are related.
Step four: Identify your key words
Some of these you may already have found in your concept mapping work, however, you really need to grab every keyword you can to get the best results…sometimes your databases will already have predefined keywords for you to use….helpful
To identify your own keywords, you will need to break down your own research question. I will go back to using our example.
| Does | Hand washing | among | midwives | reduce | Postnatal infections |
| ‘OR’ | ‘AND’ | ‘OR’ | ‘AND’ | ‘OR’ | |
| Hand hygiene | Midwifery staff | Postpartum infections | |||
| ‘OR’ Clean hands | ‘OR’ Midwif* | ‘OR’ After birth | |||
| ‘OR’ Washed hands |
Step five: Build your concepts and keywords into a search strategy
Sounds easy right? Well let me show you how to do this using an example from one of my published systematic literature reviews.
The questions relating to this review were…
1) What interventions have been developed to support midwives and/or student midwives in work-related psychological distress? and 2) What are the outcomes and experiences associated with the use of these interventions?
Key concepts have been underlined.
Below is a search strategy I built to answer these research questions. This was used to search one database only.
As you can see, the search starts right from the bottom with the first concept ‘midwives’… the ‘population’. Each concept moving forward is grouped together with keywords combined with the boolean operator ‘OR’. When I need to combine concepts nearer the top, I combine them using the boolean operator ‘AND’. See below…
This search strategy also uses truncations, where I have entered the root of the word and then a (*) at the end. When you do this, the database will then return any ending of the root word. Another example of this would be ….child* = child, childs, children, childrens, childhood.
If a word you want to find is spelled in different ways, wildcards can also be used to substitute a symbol for one letter of a word. Examples of how you might use this may be
wom!n = woman, women
colo?r = color, colour
(Credits to https://libraries.mit.edu/experts/)
I hope you find this ‘How to’ guide useful. I now look forward to you all going forth to develop and share your own search strategies with me. I can’t wait to see what problems you will solve
If you would like to follow the progress of my work going forward..
Follow me via @SallyPezaro; The Academic Midwife; This blog
Until next time…Look after yourselves and each other ❤
