What are the disadvantages and advantages of using MeSH comparing search for the keywords in title/abstract in PubMed?
Is it more sensitive or more specific?
MeSH are often applied retrospectively to PubMed records so you run the risk of missing the more recent articles if you search on MeSH alone. MeSH will however tell you what the content of the article is about, conversely your search term may appear in an abstract yet it may not be the main focus of the article. As I'm sure you know MeSH aim to capture differences in author terminology so if you fail to search on one synonym you may miss a relevant article. MeSH sometimes are not always available for all concepts - eg new and emerging technologies. For a SR use both mesh and freetext to minimise the risk of missing anything.
1) compared to free text, Mesh term are more specifics. However, you lose sensitivity if you do not use free text, as Michelle said.
2) using Mesh terms allow researchers to have much more reproducible results. This is important.
3) using Mesh term, you might miss recent articles not yet indexed.
4) you cannot use Mesh term outside Pubmed database (Embase: free text)
5) for a systematic review / meta analysis, I would use both, as Michelle said
Agree with both Michelle and Francois that combining free text and MeSH terms is the best strategy. I have a feeling this was the conclusion of some research in this area by Carol Tenopir in the 19990's but I cannot find the reference (with my faulty memory).
Need to consider the use of thesaural terms in strategies to broaden and narrow searches depending on the results of an initial search. That is one thing the expert searcher can bring, is knowing how to respond and when to use, combine with free text or abandon thesaural terms.
In a recent free text search, only one of my first 100 results had been indexed. There is an ever increasing amount of "ahead of prints", plus I assume a higher demand on fewer NLM indexers, leading to a greater lag in indexing even those in print. Because of this, it is probably best to carry out free text as well. I always start with MeSH though (if there is an appropriate MeSH term available), just to get their terms and context. (A recent tip I received was to use PubMed Reminer to identify other terms, authors on the topic etc. http://hgserver2.amc.nl/cgi-bin/miner/miner2.cgi).
In response to your query, MeSH is considered to be both sensitive and specific, I believe they are more likely to include than exclude if it is in any way related to a topic, but you can restrict to Major term. Also do the free text search as well I think. Some "ahead of prints" remain so for over a year.
PubMed includes Medline, PubMed Central, in press papers from publisher, in process papers for Medlines, etc
If you use MeSH in PubMed:
1. You will retrieve just Medline papers (2 million missed records from PMC etc).
2. You will miss papers related your research question but not assigned by proper MeSH (NLM possible errors in assigning proper MeSH).
Slightly related to this question,
Try this tool, it will give you new insights and help scientific literature search
www.visualpubmed.com
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts