Statistics from Altmetric.com
Khalid S Khan, Regina Kunz, Jos Kleijnen, Gerd Antes. (Pp 136; £17.50.) Royal Society of Medicine Press, 2003. ISBN 1-85315-525-X.
This is a didactic paperback of a 136 pages which is devoted to the worthy topic of avoiding subjective conclusions in appraising medical publications.
The first half is given over to discussing the structure for reviewing medical literature under a five step structure (1. Framing questions for review. 2. Identifying relevant literature. 3. Assessing the quality of the literature. 4. Summarising the evidence. 5. Interpreting the findings).
The second half is devoted to four case studies tackling different topics. 1. Identifying and appraising systematic reviews (drug treatment for recent onset schizophrenia). 2. Reviewing evidence on safety for public health intervention (safety of public water fluoridation). 3. Reviewing evidence on effectiveness of therapy (antimicrobial therapy for chronic wounds). 4. Reviewing evidence on test accuracy (ultrasound scan tests for postmenopausal women with vaginal bleeding).
As the authors emphasise proper objective reviewing depends very heavily on information technology to ensure comprehensive coverage of all relevant material.
The now bewildering number of sources include Medline and Cochrane collaboration information. Any really systematic trawl of literature will now necessarily depend on help from information technologists librarians and epidemiologists. The days when eminent clinical authors could assemble favourite references to support their own prejudice should be long passed.
There is detailed discussion on various types of publication bias, including the fact that US and European literature will be different, and that negative results tend to be published in local foreign language journals whereas positive ones tend to appear in English language and international journals. The pitfalls of meta-analysis and the need for rigour in assessing the quality of evidence are well emphasised.
It is useful to remember that an adequate review involves study of full length papers, particularly scrutinising references to try and identify further important publications not found on original literature searches. Perhaps because medicine and clinical research depend on a high degree of trust the topic of fraud is not subject to detailed discussion. This pernicious factor cannot be ignored since there are well documented instances of papers and authors whose conclusions are totally misleading.
This is an informative and useful if rather daunting survey of how to construct significant and useful review articles, and can be recommended to candidate authors unfamiliar with the territory.