Wednesday, May 04, 2005

Explode that subject heading

While this sounds a tad violent, I assure you it's not. And to adequately describe what happens when you 'explode' the subject heading we need some illustrations.

Let's look at the tree structure for antibiotics:

The tree shows that within the category of Chemicals and Drugs there is a subcategory of Pharmacologic Actions and a further subcategory of Anti-infective Agents. Our term ‘antibiotics’ mapped to the MeSH descriptor Anti-bacterial Agents which is a more specific subcategory within Anti-infective Agents. [For the purposes of this illustration, this tree has been severely pruned].

If I search on ‘Anti-bacterial Agents’ as a descriptor then only the articles that were assigned the descriptor ‘Anti-bacterial Agents’ will be retrieved. Now we know from the last post that the subject indexers assign the most specific term available. The citations retrieved will discuss ‘Anti-bacterial Agents’ as a group or in general (maybe this set will also include a new type of anti-bacterial agent).

What if our search question is “Which antibiotics are most effective in treating acute otitis media”?

If we use the descriptor ‘Anti-bacterial Agents’ then there’s the potential of missing relevant articles that have been assigned specific drug terms, such as Amoxicillin, and not the broader term ‘Anti-bacterial Agents’. This, finally, is where EXPLODE comes in to play.

Let’s look at some numbers:

1. exp Otitis Media/dt (2956 citations retrieved)
2. Anti-bacterial Agents/ (124687 citations retrieved)
3. exp Anti-bacterial Agents/ (329921 citations retrieved)
4. 1 and 2 (1249 citations retrieved)
5. 1 and 3 (2183 citations retrieved)
6. 5 not 4 (934 citations retrieved)

In search statement (SS) 1, I exploded ‘Otitis Media’ picking up ‘otitis media’ or ‘otitis media, suppurative’ or ‘otitis media with effusion’ or ‘mastoiditis’. And because this is a drug therapy question, I attached the subheading ‘drug therapy’ (more about this later) to the descriptor thereby narrowing the types of citations to only those that discuss drug therapy of otitis media.

SS 2: I looked for ‘anti-bacterial agents’ in the descriptor field and retrieved only those articles with that descriptor assigned.

SS 3: I exploded ‘anti-bacterial agents’ picking up ‘antibacterial agents’ or ‘alamethicin’ or ‘amdinocillin’ or ‘amodinocillin pivoxil’ or ‘amikacin’ or ‘amoxicillin’ or [160 more named substances].

SS 4: I combined my otitis media set and anti bacterial set

SS 5: I combined my otitis media set and the exploded anti bacterial set

SS 6: Using the NOT operator, I looked at what I would have missed if I used only the descriptor ‘anti-bacterial agents’ without exploding and picking up all the named drug terms.

Here’s some titles from the set of citations I would have missed:

  • Primary care based randomized, double blind trial of amoxicillin versus placebo for acute otitis media in children aged under 2 years
  • Five-day versus ten-day treatment of acute otitis media with cefprozil
  • Comparison of five-day cefdinir treatment with ten-day low dose amoxicillin/clavulanate treatment for acute otitis media
  • Use of oral cephalosporins in the treatment of otitis media in children


Post a Comment

<< Home