Search Strategy

Overview

Develop your search strategy for your review article is among the steps that require knowledge and expertise to achieve with quality. The following will provide the details of understanding and developing a quality search strategy

Develop the keywords

This step will reflect how much accurate your work will be in catching relevant papers without including a lot of unrelevant papers. For example, adding the word "heart" alone to your search when you search for articles related to "heart attack" will add a lot of unrelevant articles without much benefit to your target search. The other aspect here is how to use the keywords of interest, and this include the use of "truncation" (e.g., using hepat* to represent hepatic and hepatology) and choosing the right field to search in (e.g., searching "affiliating country" for the word "Jordan" instead of "all fields").

Choosing the keywords

The following are the three strategies that should be used together to develop the searching keywords and how to combine them:

Strategy 1: Check prior review articles and search filter article

Search filter articles are stand-alone articles that thoroughly test a particular search strategy. They describe how to use and the benefits of using searching keywords and how to use them in different databases. An example of search strategy article would be "Development of an efficient search filter to retrieve systematic reviews from PubMed", where the authors described and tested a particular search strategy that can be used in PubMed:

Strategy 2: Use Medical subject heading (MeSH) to search articles related to a particular topic

MeSH represent unique keywords that each article indexed in PubMed will have to describe it. It can be through of as an "Umbrella" that articles of a certain topic will be grouped under it. It would be a strength point if your search strategy has MeSH terms in it. Adding a MeSH term first require checking if a term is included in the MeSH database. After that, you should use the advanced PubMed search to specify that the term you will input is a MeSH term.

Strategy 3: Consult experts

This is most relevant in specialized fields. Some diseases (e.g., glaucoma) might have multiple unrelated keywords where all reflect the disease and should be considered in the search (e.g. , ocular hypertension).

Develop the search criteria

The most widely used approach to develop search criteria is the PICOS approach, where you specify:

Population or participant, Intervention, Comparison, Outcome, and Study design.

Combine your keywords: Once you have identified your keywords and controlled vocabulary terms, you can combine them using Boolean operators (AND, OR, NOT). For example, if you are searching for articles on the effectiveness of a particular drug in treating a specific condition, you might combine the drug name with the condition name using the AND operator. Use truncation and wildcards: Truncation and wildcard symbols can help you search for variations of your keywords. For example, using the * symbol at the end of a word can help you search for all variations of that word. Similarly, using the ? symbol can help you search for words with different spellings.

Choosing databases to search in

The following is a criteria for a "comprehensive search":

At least two electronic databases had to be searched (such as MEDLINE or EMBASE);

A search method to to identify unpublished trials: searches for (I) conference abstracts, (ii) theses, (iii) trials registers; and (iv) contacts with experts in the field;

A search not restricted to English language;

Literature databases to search in

PubMed: It is the main database for biomedical literature, where it searches articles in the National Library of Medicine (NLM) database. It is freely accessible and hosted by the governmental national institute of health (NIH) in the US. The NLM database can also be accessed through the subscription-based OVID. They differ in the way you de your search in, however, they both include articles from a similar source. PubMed also include articles from sources other than NLM, including the PMC and articles funded by NIH, even if published in journals not indexed in PubMed.

EMBASE: A subscription-based database owned by Elsevier for-profit company. It searches articles published in journals indexed in Scopus. It is not exclusive for biomedical literature.

Scopus: Another subscription-based database owned by Elsevier for-profit company. It searches articles published in journals indexed in Scopus. It is not exclusive for biomedical literature.

Cochrane Central Register of Controlled Trials (CENTRAL): Include reports of randomized and quasi-randomized controlled trials. Most CENTRAL records are taken from bibliographic databases (mainly PubMed and Embase.com), but records are also derived from other published and unpublished sources, including CINAHL, ClinicalTrials.gov and the WHO's International Clinical Trials Registry Platform. CENTRAL first began publication in 1996, but its composite nature means that it does not have an inception (start) date, in the way that other traditional biomedical databases do.

CINAHL: A subscription-based database owned by EBSCO, with literature coverage of nursing, allied health, biomedicine and healthcare.

Clinical trial registry databases

ClinicalTrials.gov: An NIH-hosted database for clinical trial and other form of study design registries.

WHO's International Clinical Trials Registry Platform: WHO-hosted database for clinical trial and other form of study design registries.

Literature search engines

Google Scholar: A search engine owned by Google that crawl articles, rather than actually index them.

Testing and fine-tuning the search strategy

It is important to know how to use and combine the keywords together in a literature database. The following table list the strategies that should be considered during the search in each literature database.

Comparison between strategies that can be used in each of PubMed, Scopus, and Web of Science databases

Example advanced search to search for articles on the impact of olive oil diet on myocardial infarction

https://videopress.com/v/DlL0QHGo?resizeToParent=true&cover=true&preloadContent=metadata&useAverageColor=true

You can use the subtract function in PubMed to make sure that the new word added will yield relevant studies:

Supplementing the search by Citation-Reference search

Google scholar has the ability to crawl most relevant articles to your article of interest, meaning that you can give it an article to double check if relevant existing articles might have been missed.

https://videopress.com/v/aWBwW5EG?resizeToParent=true&cover=true&preloadContent=metadata&useAverageColor=true Citation Reference search in Google Scholar