Systematic Reviews

There are seven elements to a comprehensive search.  These are:

1 The Pyramids of Evidence

2 Getting from A  to B

 3 Covering your Bases

4 Keywords and Subject headings

5 Boolean Logic

6 Building Blocks

7 "The Long Tail". 

These steps are elaborated upon below.

The 'Pyramid of Evidence' examines:

• where the evidence is held
• where evidence on similar topics is aggregated
• the breadth and diversity of the locations where evidence is stored
• which resource may be best to use, depending on need and context.

Some areas focus more on individual pieces/ items of research, and some areas (so therefore some databases and resources) specialise in collating, aggregating, synthesising or analysing the evidence.

Image courtesy: Edith Cowan University Library

Locations where individual pieces of evidence (e.g. journal articles) can be found are in the base tier of the pyramid.

Assessment at this level is suported by the 2nd information pyramid (above).

The content above this base level in the Pyramid of Evidence highlights where efforts to assess, synthesise or summarise across the existence evidence have already taken place.

Use of this information can:

• speed up the process
• ensure that the efforts towards a research question are not 'reinventing the wheel'
• produce a better, more thoroughly 'evidence-based' end result.

A is your starting point. It where you are before starting your search for/ starting to find information.

B is the point of finding the information that you want and/or are searching for.

There are two ways to get from A to B:

1. You know a certain item that you wish to access exists.
   It may be a journal article, book, chapter etc...  You know the details of the item and use the citation to source and find it.
2. You do not know specifically what items exist on a topic.
   However, you know the concepts or subjects you with to find items/ materials/ evidence on.

Databases searching is inherently focused on the 2nd of the above options.

Therefore you need to construct a search that ably finds the concepts or the subjects you seek, while removing extraneous content.

All databeases have the Keyword function. Some also have the Subject Heading function. This may also be named Subject Terms or MeSH Headings,  depending on the database. Maximinsing use of such functionality is necessary to form a robust search strategy.

To cover your bases ensure that you do not miss any potential content:

• Examine a range of databases targeted towards specific subject areas
• Throw the net out wider, for more general content coverage. These databases also cover the topic(s) required

A - There are a range of discipline-focused databases.  The choice of the databases to search will depend on the nature and context of the research question.  Example databases at this level include CINAHL, MedLine, Business Source Complete, PsychInfo, etc.

B – These databases are broader in nature, and tend to aim to cover a range of topics, yet still have as their focus content from the Health Sciences. Example databases at this level include PubMed and the Cochrane Library.

C – Databases at this level cover a much wider base of content.  However, they remain focused on the Science Disciplines.  Example databases at this level include Web of Science, and Scopus.

Clicking on any of the icons below will take you to the relevant database search facility.   Please ensure you are logged into the UHI network before trying to access these (Cochrane and PubMed are open access)

There are two ways in which you can input a search into a database.  

All databases, government sites, institutions and any sites on the internet with a “Search” box all make use of ‘Keyword’ searching.
Some other databases have extra functionality, i.e. ‘Subject Headings.’

Inputting search words requires a combination of both this principle (#4) and #5 below, which go hand in hand in the construction of the search.

For a Keyword search, imagine that each variation on a theme (synonyms, etc.), covers a part of the topic.
 
To cover the range of ways the concept is described, each variation adds to a total ‘pool’ of information on the topic.

Databases with the Subject Heading search function establish an ‘authoritative’ heading for the topic, and pre-emptively put all the various articles, with all the variations of terminology under this title, meaning that they can all be gathered with the one search term.
It can be readily seen from this that it is useful to know that term in order to search it.  However, if you are close to the correct term, most databases will re-direct you to the correct location.  

For e.g., in CINAHL, it will direct you like this:
“For Steroidal Anti-inflammatory Meds Use: Anti-inflammatory Agents, Steroidal.”

Boolean logic underpins all search construction and functionality.  There’s many functions under the banner, however the 2 most critical are “OR” and “AND”.

The “OR” Operator:

Examine all variations on a theme or subject, e.g. ‘Diet,’ ‘Nutrition,’ ‘Food intake,’ etc.  Each variation of wording will individually yield results.  However, if you search diet OR nutrition OR food intake you will return a greater pool of results.  

The “AND” Operator:

As these Venn Diagram images demonstrate, the overlapping parts of a search for ‘X’ AND ‘Y’ produces less ‘quantity’ of results and greater ‘quality,’ i.e. results that are specifically addressing all relevant topics.

As a general principle, the pattern would be to utilise the “OR” function 1st, followed afterwards by the “AND” function.  
The logic is to collect all the potential variations / permutations by which a concept could be described or categorised in the literature.  
Once this is done with all the various concepts the research wishes to explore, then to cross-reference these complete pictures of the individual subjects with the “AND” function.

All the individual variations on a theme can be kept on individual lines in the search.  

By using the “OR” function, i.e. “w OR x OR y OR z”, there can be a ‘collective’ line in the search pattern for complete coverage of the concept.
Developing such clusters allows clear and easy visualisation of coverage of a subject or concept.

All the lines of the collective individual topics can then be cross-referenced against each other to target the search to the requirements of the research.

Not all databases or other locations that  need to be searched have this level of design and functionality available.

The benefits of being able to build a search utilising this level of functionality include:

• Being clear and readable for yourself and any others.  This helps the elements of being demonstrable and replicable for a literature review.
• From this, it can often be easier to edit, change and correct the search pattern.
• It is also important to remember that the literature search is not done in isolation, but rather as a part of a bigger research project.  Constructing a search this way allows individual building blocks/ search strings to be teased out at any stage of the process, allowing the work of the construction of the literature search to speed up and simplify subsequent search requirements.
• 

There is a balance to be struck between being as ‘perfectly’ thorough in the search process as possible, and a ‘diminishing return on investment’ for time and effort.

The individual context of the subject and the search style means that you need to choose a realistic cut-off point for where and how to end the search process.

After having examined the content in the databases, there is a choice to be made about how much further to look, and where.

Government departments, professional bodies and institutes, Royal Colleges, etc. all can provide relevant and useful information not found in the databases.  

This element can be ‘how long is a piece of string’ in nature. Decide on a cut-off point, and explain why.

Clicking on any of the icons below will link you to the relevant resource.