|Year : 2015 | Volume
| Issue : 2 | Page : 195-197
How to use medical search engines?
Department of Medicine, AIIMS, New Delhi, India
|Date of Web Publication||30-Sep-2015|
Dr. Saurav Khatiwada
Department of Medicine, AIIMS, New Delhi
Source of Support: None, Conflict of Interest: None
In this era of Google search, it is easy for beginners to fancy literature review limited to this popular search engine. Unfortunately, this will miss a vast index of articles which exist within our reach. Some specialized search portals leading to their corresponding databases deal with the tremendous medical literature that has been generated over decades. This article deals with the “what?” and “how to?” of these available databases of the medical articles. This will make your literature review efficient and the confidence in collected evidence - accurate.
Keywords: Engines, medical, research
|How to cite this article:|
Khatiwada S. How to use medical search engines?. J Pract Cardiovasc Sci 2015;1:195-7
| Introduction to Medical Search Engines|| |
Multitude of search engines exists for medical literature. [Table 1] lists the popular ones along with their specialties. These engines are in fact a collection themselves and rather be called databases, unlike engines which mean a medium to database only.
PubMed has been the corner-stone of medical search till date. It is a free search engine developed by the US National Library of Medicine to provide access to MEDLINE (the largest medical database which is the online version of Index Medicus) and other non-MEDLINE journal articles via a systematic search. Most other databases/search engines have overlapping results with PubMed, except the regional databases which might have exclusive contents (INDMED, LILACS, etc.). Since Cochrane database is available freely within the Indian territory, thanks to the national subscription from the Indian Council of Medical Research, it can be a useful database too. Embase (Excerpta Medica database) is an Elsevier-owned biomedical and pharmacological database which is considered next to only to PubMed in volume of data available. It requires a paid subscription though, and results significantly overlap with PubMed.
In addition, identifying ongoing trials, unpublished, but completed studies, and conference proceedings can be considered while doing a comprehensive literature search (as in doing a systematic review). The most useful ones are listed in [Table 2].
|Table 2: Important clinical trial registers (for ongoing trials), registers for conference proceedings, guidelines register and register of on-going systematic reviews|
Click here to view
It is generally fruitful to go through the list of references of any related article to find out the useful ones. Similarly, there are search engines such as ISI web of knowledge (paid subscription) and Google, which list you the studies citing a particular study. Contacting experts and authors on a personal level is also a commonly practiced method of identifying studies in a specific field. Ten years back, studies in foreign languages were hard to identify and evaluate. Now-a-days, we may find at least their abstract in English which help identification. Using translating software like “Google translate” can be a useful tool for such studies. “Google Scholar” is fast growing as a powerful adjunctive search engine, but it is not an alternative to the ones listed above as yet.
Evidence-based, peer-reviewed online resources like up-to-date (paid), dynamed (paid), Medscape (free), etc., is upcoming resources which may serve as starting points for literature search with quick information on useful studies on the topic.
| Method for Literature Search|| |
Core principles of literature search
Use of keywords
Literature search is not done by typing a whole question or phrase, unless you are planning to retrieve the single article with the same words in tandem. We specifically write down the important words that are sure to come up in the expected articles. For example, to search for articles concerning percutaneous coronary intervention (PCI) instable angina. We may use the keywords “PCI,” “coronary intervention,” “angina,” etc.
Use of Boolean operators
AND, NOT, OR (sometimes AND NOT) are Boolean operators that help us combine search keywords so that an article set with the desired combination is available to us. The outcome of each of these operators is listed below in [Table 3].
A working search is possible with the above principles. But to hugely enhance the outcome of search the following useful tips should be considered:
Useful tips for literature search
Framing a search strategy
The first step of a search is planning a search. First write down your clinical question. Then try and identify the population, intervention, control, and/or outcome (PICO) in the question along with their keywords and related words. After this pick, the appropriate combination of words. Then, use a combination of these keywords for search.
For example: To identify the utility of steroid in pericardial effusion due to tuberculosis, we may list the PICO format of the question along with related words as follows in [Table 4].
The asterisk sign used in some keywords act as wildcards that is, any letter may follow the word and the word variations will still be picked up. For example: “Tubercul*” will pick up tuberculosis, tubercular, tuberculid, etc.
The related words are an attempt to identify the words that an author might have used in the study to mean our descriptor. Sensitivity of a search can be increased by writing multiple synonyms and related words as shown above, combined with Boolean operators(esp. OR).
After the PICO question and a list of keywords are prepared, we may want to build a search. It is not necessary to write all descriptors for a search. For example, in the above question, we will just pick population and intervention as these are the most important determinants of the search strategy.
“Tubercul*,” “pericardial effusion,” “mycobacter*,” “pericarditis”
“Prednisolone,” “hydrocortisone,” “dexamethasone,” “steroid”
Executing the search strategy
This section will deal with conduct of a search, taking PubMed as an example. Searches in other engines are similar.
After we have identified the keywords, we go to the search engine and write down our keywords and combine them with Boolean operators. It is good to build a search step-wise, rather than a single-go search since the former allows for adjustments.
In PubMed, execute the following search first [Figure 1].
This step-wise search serves for the adjustment. If you get very few studies with the first set of keywords, there must be at least one explanation among these:
- Keywords are not valid
- Expected results are scarce.
If the former is the case, you need to find other keywords, while if the latter is correct, then, you need not add the second set of keywords with Boolean operator AND.
You will notice that the above search [Figure 1] yielded a huge number of entries. The search strategy has itself gone to the history section as search #1 after you ran the search.
Do the same for the second set of search keywords. After that, combine search #1 and #2 with Boolean operator AND. You may notice that each set of keywords had over 1 lakh entries, but, by combining with AND, we got down to <2 thousand entries [Figure 2].
If the entries are still too many to handle, use filters. Filters can be specific parameters of the articles, for example, date of publication, type of article (review, meta-analysis, case report, etc.). Applying filter will further reduce your entries to manageable limits. The text highlighted on the left side of the [Figure 3] are filters, which may be appropriately selected to get the desired entries. For examples, if you require only systematic reviews among the searched entries, click on article types, and select systematic review from the drop down list.
|Figure 3: PubMed entries to be sorted by filters highlighted on the left side.|
Click here to view
Using MeSH terms
Another important concept useful to get to articles in search engines is the concept of MeSH terms. In simple terms, typing these keywords will bring to you a ready-made list of articles classified by the search engine to be pertaining to the keyword and its synonyms. They are specific keywords that have been used by the search engine hosts to categorize arrays of articles related to them, their synonyms and related words.
For example, in PubMed, searching with:
Data type: MeSH terms with Keywords: Tuberculosis.
A drop down menu appears before you complete the word tuberculosis so that you get a ready-made/built-in keyword (the MeSH term) to choose. This chosen MeSH term “tuberculosis” would mean a combination of:
“Tuberculosis” or “TB” or “tuberculid” or “tubercular” or many other variants or words related words to tuberculosis.
Hence, whenever available, the MeSH terms should be used to make a search more sensitive. Similar is the concept of EMTREE in Embase.
There are a lot more features one may learn from the free tutorials that bibliographic databases provide. One for the PubMed is available at https://www.youtube.com/watch?v=dncRQ1cobdc and feature=relmfu.
Different bibliographic databases use slightly different methods for search which are easy to decipher using the tutorials at their respective sites or available freely in https://www.youtube.com
After a literature search, the identified studies require a reference manager in your computer for proper indexing and use. This software is called a bibliographic manager. The most popular ones are Zotero (freely available) and EndNote (paid subscription required). The use of these softwares can also be learnt similarly with online tutorials.
The steps in literature search are preparing a PICO question, identifying major search databases, preparing a search strategy and executing them using operators, filters, and MeSH terms. Mastering PubMed search is the first and major step to keeping up-to-date with the evolving medical literature.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3], [Table 4]