Search strategy
A search strategy involves a systematic approach that ensures you find the most suitable information for answering your (research)question. Your final search strategy will depend on various factors, such as the positioning of your question, the source(s) and the objective (e.g. a thesis compared to a systematic review).
Below are some tips on how to design a search strategy for those working on a literature review.
1.Position your question
• What?
• For who?
• Why?
• Main point / side issue?
2. Select your source
• The visible or the deep web
3. Create your search strategy
• Identify the main concepts
• Search for alternative terms, technical jargon and index terms
• Combine the search terms
4. Evaluate your result
• Relevant?
• Adequate?
• Reliable?
Firstly, it’s important to consider if your topic is suitable for a literature review in one or more biomedical databases. Then, you should determine the goal of your research and consequently which type of review you will choose. There are a number of review types (non-limitative list below), depending on different factors:
- Systematic reviews summarise evidence on efficacy, safety, diagnostic test accuracy, etiology, incidence, prevalence and qualitative research. A systematic review attempts to gather empirical evidence from a relatively smaller number of primary studies related to a focused research question following a strict methodology aimed at minimising bias.
- Living systematic reviews are a type of systematic review that is continuously updated to include relevant new evidence as it becomes available.
- Scoping reviews attempt to provide an overview of a potentially large and diverse body of literature related to a broad research question. Various types of evidence are included in this rather descriptive review.
- Rapid reviews are a form of knowledge synthesis in which parts of the systematic review process are simplified or omitted to produce timely information.
- Umbrella overviews summarise the evidence from multiple research syntheses into one accessible and useful document. It is based on reliable high-quality systematic reviews on a specific health problem or topic. It examines the consistency of findings across reviews.
- Realist reviews aim to identify and explain (complex) social interventions or programs and the interactions between context, mechanisms and outcomes for policymakers.
- Integrative reviews / comprehensive reviews / critical overviews focuse not on one specific clinical question but on a complex healthcare problem (often within nursing).
- Narrative reviews provide an educational overview of the current state of knowledge regarding a health-related topic. No specific standard or guideline to minimize bias. Best practices:
- Minimum: a clearly defined topic, literature search, logical structure and interpretation & discussion
- Structure: IMRaD (Introduction, Methods, Results & Discussion)
For more information on these different types of reviews, and which type of review may fit best for your literature review, please see our Blits information session ‘How to choose your review type’ on Ufora.
Formulate a research question in which, if possible, all aspects of your topic are named. To help you consider all aspects, you can use a framework, such as PICO(S/T), SPIDER, SPICE, PCC,…
The PICO framework is a tool to help you formulate a clinical research question. In this model, the patient category (Population/Patient), the intervention (Intervention/Indicator), the control treatment or comparison (Comparison/Comparator/Control) and the outcome (Outcome) are clearly described.
For example: ‘In men, does having a vasectomy (compared to not having one) increase the risk of getting testicular cancer in the future?’
Population/patient = adult males
Intervention/indicator = vasectomy
Comparator/control = no vasectomy
Outcome = testicular cancer
Spin-offs of PICO such as PICOT (including Time) and/or PICOS (including Setting) contain more information than the classic PICO and may therefore be a better alternative. Here is a template for formulating PICOT questions. If you need other types of PICO and spin-offs (e.g. in the context of a systematic review of case control studies), please contact us (kcgg@ugent.be).
PICO(S/T) is the most commonly used framework. There are also others, such as:
- SPIDER (Sample, Phenomenon of Interest, Design, Evaluation, Research type) mainly for qualitative research questions
- PCC (Population, Concept, Context) mainly for scoping reviews
- SPICE (Setting, Population, Intervention, Comparison, Evaluation)
- ECLIPSe (Expectation, Client group, Location, Impact, Professionals, Service)
In addition to formulating a clear research question, it is essential to define selection criteria when writing a systematic review (preferably also per aspect of your research question). Each of the aspects is described in yet more detail in the inclusion and exclusion criteria (e.g. age limit or comorbidities under the description of the Population). Other important criteria to be described are: setting (e.g. Western society), duration of administration of the intervention and/or the follow-up, study types and language of the studies that will be included/excluded.
Choosing a suitable source depends on the type of source and the aim of your question.
When you are looking for background information (subject information and/or public information), it may be useful to conduct a search in Google or consult an encyclopaedia (e.g. Wikipedia) and/or search in MedlinePlus.
Scientific information can be found in journals and databases.
For clinical, evidence-based information, databases such as UpToDate are suitable.
For each type of source it is useful to know more about the information source (e.g. users in Wikipedia), the search algorithm (e.g. order of results in Google Scholar) and the search options (e.g. Advanced Search option and thesaurus in certain medical databases).
For more information, consult the information sheets about databases.
The Internet has different layers: the free (or visible) web and the deep (or invisible) web. The deep web essentially refers to databases, for which you will find at most the homepage via search engines such as Google, Bing or Yahoo. In order to search through the content, you will have to use the search engine associated with the database in question.
In the Cochrane Library, you can only search for systematic reviews in the Cochrane Database of Systematic Reviews.
Via PubMed you can find systematic reviews using the Clinical Queries filter. You can use systematic[sb] in your search. For example, if you are looking for systematic reviews on lyme disease, then you can type the following in PubMed: systematic[sb] AND lyme disease.
This filter uses a specific search strategy that you can find here on the PubMed website.
Via Embase, you can add [systematic review]]/lim to your search query or select the filters Systematic Review and/or Cochrane Reviews under the EBM (Evidence-Based Medicine) filters.
In the database Scopus you will find that your search query (thus after you have entered terms to search) also shows the link View # patent results in the bar with the number of results. This link is only visible if there are patents that correspond to your search query. The information is checked and retrieved from five patent organisations via LexisNexis.
Look for the most commonly occurring (subject)specific search terms and synonyms for every aspect of your question (see How do I formulate my research question?). The use of the ‘correct’ search terms is crucial for finding the desired information. Wikipedia, Oxford Reference Online and Thesaurus.com are potential sources for English search terms. Always assume that search systems simply check whether the terms you have specified in your search strategy appear in for example title and abstract. In most cases, if your term is written slightly differently or your term only appears in the full text of the article, the information will not be found.
If you are searching in a database with a thesaurus (e.g. PubMed or Embase), then it is best to use relevant keywords/index terms from that thesaurus as well. Below is a table that you may help you search and list relevant search terms.
| Population | search term including singular/plural, used as noun/adjective and verb conjugations, spelling variations (British and American spelling), Latin names, acronym, etc. * | synonym(s) including subject terminology, singular/plural, used as noun/adjective and verb conjugations, spelling variations (British and American spelling), Latin names, acronym(s), etc. * | keyword (e.g. MeSH term for searching via PubMed; Emtree term for searching via Embase) |
| Intervention | search term including singular/plural, used as noun/adjective and verb conjugations, spelling variations (British and American spelling), Latin names, acronym, etc. * | synonym(s) including subject terminology, singular/plural, used as noun/adjective and verb conjugations, spelling variations (British and American spelling), Latin names, acronym(s), etc. * | keyword (e.g. MeSH term for searching via PubMed; Emtree term for searching via Embase) |
| Comparison | search term including singular/plural, used as noun/adjective and verb conjugations, spelling variations (British and American spelling), Latin names, acronym, etc. * | synonym(s) including subject terminology, singular/plural, used as noun/adjective and verb conjugations, spelling variations (British and American spelling), Latin names, acronym(s), etc. * | keyword (e.g. MeSH term for searching via PubMed; Emtree term for searching via Embase) |
| Outcome | search term including singular/plural, used as noun/adjective and verb conjugations, spelling variations (British and American spelling), Latin names, acronym, etc. * | synonym(s) including subject terminology, singular/plural, used as noun/adjective and verb conjugations, spelling variations (British and American spelling), Latin names, acronym(s), etc. * | keyword (e.g. MeSH term for searching via PubMed; Emtree term for searching via Embase) |
*Use single or double quotation marks (depending on the database) for terms that must appear together and in exactly this order (= phrase searching).
Are there any tools I could use to find additional search terms?
Key articles or systematic reviews related to your research question can be useful sources of inspiration for search terms, synonyms and keywords. You can also use tools such as PubReMiner that lists numerous commonly found terms and/or MeSH terms based on your search query. Yale MeSH Analyzer is another tool that gives you an overview of terms that appear in references of PMIDs that are relevant to your research question. Based on your search query or a piece of text, you can also obtain advice about possible relevant MeSH terms and key articles via MeSH On Demand.
Search filters or blocks can provide inspiration for search terms. For more details, see ‘What are search filters and where can I find them’?
Search blocks and filters can be subdivided into methodological filters (e.g. RCT filter, filter for Health Economics, filter for systematic reviews) and content filters (e.g. filter to look for literature about children, or about AIDS, etc.).
Search filters or blocks can provide inspiration for search terms. If the filter is validated, it can even be taken in its entirety and used in your search strategy – providing the source is cited. Always check that the search filter is compiled in the correct syntax. For example, a search filter in Embase may be compiled for the OVID interface or for the Embase.com interface.
On the website Library of Search Strategy Resources, you will find an overview of sources where you can find search filters or blocks.
The table below gives an overview of various search filters or blocks.
| Organisation | Various examples of methodological filters or blocks | Various examples of content filters or blocks |
|---|---|---|
| The Cochrane Collaboration | Randomized Controlled Trials (RCTs) | |
| The InterTASC Information Specialists’ Sub-Group (ISSG) | Observational studies
Epidemiological studies Qualitative research Mixed method studies |
Adverse effects
Aetiology Public Views & Patient Issues |
| Canadian Agency for Drugs and Technologies in Health (CADTH) | Systematic review/meta-analysis/Health Technology Assessment
Guidelines |
Economic Evaluations
Quality of Life |
| McMaster University – Health Information Research Unit | Therapy
Diagnosis Prognosis |
|
| PubMed Search Strategy blog
Via the search box |
A specific pathology, e.g. COPD
A specific target group, e.g. healthcare workers A specific type of study, e.g. animal studies Medication, e.g. NSAIDs (non-steroidal anti-inflammatory drugs) |
|
| BMI van de Koninklijke Nederlandse Vereniging voor Informatieprofessionals | A specific pathology, e.g. Alzheimer’s disease
A specific target group, e.g. patients Medication, e.g. Acyclovir or Zovirax A specific treatment, e.g. bone marrow transplant A specific laboratory test, e.g. blood cell count |
|
| Caresearch Palliative Care Knowledge Network | A specific pathology, e.g. anorexia
A specific target group, e.g. the elderly A specific theme related to palliative care, e.g. advance care planning |
Combine the search terms, synonyms and keywords per concept (e.g. Population) with the OR operator. The use of the OR operator indicates that at least one of the terms must appear in your search results. Then combine the search actions of all various concepts with an AND operator (which indicates that both terms must appear in your search results). Use the advanced search function provided in many databases for this purpose.
When combining terms, it is important to use parentheses correctly. Using parentheses changes the order of performing the AND, OR, and NOT combinations, this is also known as nesting. If no parentheses are used, then PubMed combines the search terms from left to right. An example of correctly nested terms: (child OR kid OR infant OR youth OR toddler) AND kidney disease.
For extensive search queries (e.g. in the context of a systematic review or literature review), it is recommended to use the syntax (≈ code) of the database. An overview of the syntax in PubMed, Embase, CENTRAL, CINAHL, Web of Science and Scopus can be found here. By using the syntax, you indicate where a term can appear, e.g. in the title and abstract when using [TIAB] in PubMed. The syntax enables you to stay in control rather than leaving it to the search engine (e.g. the Automatic Term Mapping functionality in PubMed).
To see why it is best to add both keywords and free text words, please see our session on Ufora here (in Dutch). To identify additional terms, you can perform a number of optimisation steps. Check out a roadmap to optimise your search strategy here on Ufora (in Dutch).
It is also recommended to describe the search strategy in detail (preferably by including the used syntax) when performing a search for a systematic review. It’s advised to list the various sources, search terms and search method (e.g. look at the cited and citing articles of the studies included, also called the ‘snowballing’ method).
When searching in multiple databases (e.g. in the context of a systematic review), you need to adapt or ‘translate’ your search strategy to the database you are going to use. You can use this table (also see How do I build a search strategy? on this page).
This is because each database has its own syntax that you have to adapt each time, as well as look up the relevant keywords/index terms in the thesaurus again (if the database works with a thesaurus, e.g. MeSH in PubMed and Emtree in Embase). You can copy all the free text terms in all the databases provided, provided that you, as mentioned before, adjust the syntax to the specific database that you are going to use that search strategy for.
In any case, we advise you to finalise your search strategy in one database, and only translate the search to another database after you have received feedback (e.g. from a KCGG information specialist, promotor, supervisor, colleague, etc.).
To clarify, an example: dialysis[tiab] in PubMed will show you articles where dialysis appears in title, abstract or author keywords. If you search for dialysis[tiab] in Embase, then this will not work. In Embase, you should use dialysis:ti,ab,kw to search for dialysis in the same fields as you did in PubMed (title, abstract and author keywords).
There’s a short clip here on Ufora (in Dutch) on how to translate your search strategy from PubMed to Embase, Web of Science and Scopus.
You can find an overview of the syntax in PubMed, Embase, CENTRAL, CINAHL, Web of Science and Scopus here.
The PRESS (Peer Review of Electronic Search Strategies) checklist is an evidence-based guideline for assessing search strategies for e.g. systematic reviews. You can also contact us (kcgg@ugent.be) for more information on this.
Rayyan for systematic reviews is a free webbased tool for researchers performing a systematic review or any other form of evidence synthesis. This tool facilitates the screening process by title and abstract and by full text. Users of this tool indicate a few advantages over other tools, namely that it is easy to use, saves time, availability of blinded screening and easily sharing reasons for exclusion.
DistillerSR for systematic reviews is a tool that facilitates the screening, data extraction and reporting processes of your systematic review. This tool makes tools like EndNote (for record deduplication) and Rayyan (for screening) redundant. It also saves time as the entire review is available in one tool. Through its built-in artificial intelligence, this tool can also help detect screening errors. This tool can be provided by KCGG for a fee. You can ask for a quote without any further obligation via KCGG (kcgg@ugent.be).
Covidence is a user-friendly online tool mainly to facilitate the selection process of articles in a systematic review. The online screening is free for 1 review, with a maximum of 2 reviewers and up to 500 references.
Other tools can be searched for via the SR Toolbox website.
You can already deduce something about the reliability of a source or website from the URL and the digital information certificate.
| URL extension | Type of organisation | Example | Objectivity, reliability |
|---|---|---|---|
| .com | Commercial | http://www.dokteronline.com | ? |
| .net | Network | http://www.cancer.net | ? |
| .gov | Government (USA) | https://www.nlm.nih.gov | + |
| .gov(.be) | Government | https://www.ehealth.fgov.be | + |
| .edu | Educational (USA) | http://library.medicine.yale.edu | + |
| .org | International non-profit | https://nl.wikipedia.org/ | +/- |
| .be / .nl / … | Country code | https://www.kcgg.be | ? |
The use of language (e.g. correct, objective language), the proprietor (e.g. Universitair Ziekenhuis) and the extent to which the website is up-to-date will also provide some insight into the reliability. In any case, check whether you can verify the information provided in other reliable sources.
In addition to assessing the quality of the source, it is also important to check the relevance and validity of your information. Is the study recent? Was the appropriate methodology used when conducting the study? (See also Processing > Evaluating)
The aim of a systematic review is to provide an objective overview of all literature relating to a certain research question. The quality of a study, even if it is low, is therefore not a reason for excluding it. In this type of literature research, it is common practice to allocate a quality level to each study. There are various tools available for this purpose that are often divided by study type. For example, the Cochrane Risk of Bias (RoB 2) tool for Randomised Controlled Trials (RCTs) and the Risk of Bias in Non-randomized Studies – of Interventions ROBINS-I for non-Randomised Controlled Trials (non-RCTs).
The RoB 2 and ROBINS-I tools assess potential biases in study design, randomisation, blinding, and data handling, helping you to evaluate the study’s validity and reliability. Critical appraisal tools, like those from the Joanna Briggs Institute or Critical Appraisal Skills Programme (CASP), go beyond bias evaluation, considering the study’s overall methodology and credibility. The choice of tool depends on the study design and research question.
See also Processing > Evaluating (“How do I evaluate the quality of a clinical study?”) and Writings.
If you would like more information or if you have any questions, you can contact our information specialists Nele Pauwels or Muguet Koobasi by mailing to kcgg@ugent.be.
Do you have a question related to your literature review or do you need help with searching, processing or publishing of your literature/data?
Then please fill in our form to give us a clear understanding of your question and so that we can advise you efficiently. Alternatively, you could email us at kcgg@ugent.be.