Article Text

Download PDFPDF

Hierarchical searches

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Archimedes wants to answer clinical questions using a hierarchical search process. What exactly does this mean? And why is it important anyway?

The hierarchical search aims to look for the best quality evidence first, and work downwards if insufficient research is discovered. Secondary sources are looked through in the first instance. These sources have done some of the work of critical appraisal already. An excellent secondary resource is the Cochrane Library. This database contains a large number of systematic reviews of the effects of healthcare interventions, both “Cochrane” reviews, and in collaboration with the NHS York Centre for Reviews and Dissemination, other published systematic reviews. Other secondary sources include the EBM Journal, which takes good quality studies published in other journals, creates an independent abstract of the paper, and is accompanied by a critical analysis and commentary. Some guidelines are excellent sources of information, and the authors may have done the job of collating and appraising all relevant evidence for you.

Searching in secondary sources is relatively easy: the Cochrane library is all in one place,1EBM Journal lives elsewhere,2 and a very large number of guidelines are available at the redeveloped (US) National Guidelines Clearinghouse.3

Where a secondary resource doesn’t provide enough of an answer, Archimedes authors look back to the primary literature and perform a systematic, but limited, review of the studies they find. For interventions, randomised controlled trials (RCTs) are the preferred source of evidence. For medical interventions, there are empirical observations which show that non-randomised studies tend to exaggerate treatment effects.4 If these can’t be found, it may be useful to look for large cohort studies. These have many more problems than RCTs, with difficulty assessing the presence of extraneous (confounding) factors which provide an alternative explanation for the outcomes observed. However, they are more likely to estimate the true effect than most case-control studies. In turn, a case-control study has a jump on the case study (which is, after all, a posh way of saying “anecdote”).

How do you search for these studies though? Typing “heart attack” into Medline may induce one. There are a few ways to assist the practicing clinician in their searching, by filtering out methodologically poor studies and presenting those with a greater chance of success. If you have the opportunity, many health information specialists (a.k.a. librarians) run courses on searching for clinical information. Alternatively you might like to look at the Clinical Queries section of PubMed5 or the superb “SumSearch” engine.6 Good luck.


  • Bob Phillips