Professional Documents
Culture Documents
Downloaded from https://www.cambridge.org/core. Universitas Indonesia, on 10 Jul 2018 at 06:32:21, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0959259803013212
104 Norman Vetter
the paper but the simplest example might be: level tematic reviews represent ‘the best available evi-
1 (randomized controlled trial, meta-analysis); dence’. Since this study, the Cochrane Collabora-
level 2 (other clinical trial evidence); level 3 (con- tion has taken steps to improve the quality of its
sensus/expert opinion). reviews through more thorough pre-publication
Non-systematic review articles have had a bad refereeing, training and support for reviewers,
press. A paper critically appraised all the clinical and improvements in peer review. Cochrane felt
reviews published in six general medical journals that the use of evidence-based criteria (i.e., the
in 1996.4They used explicit criteria that have been QUOROM ~ t a t e m e n t )for
~ reporting systematic
published and validated, to judge the reviews, and reviews may help further to improve their quality.
found that of 1.58 review articles, only two satis- There is doubt that, when given the same excel-
fied all 1 0 methodologic criteria. Less than a quar- lent data, people will come up with different rec-
ter of the articles described how evidence was ommendations. Good examples of this have been
identified, evaluated, or integrated; 34% addressed seen in a series of disputes between the National
a focused clinical question; and 39% identified gaps Institute of Clinical Excellence (NICE) and the
in existing knowledge. Of the 111 reviews that Drugs and Therapeutics Bulletin, where the two
made treatment recommendations, 48% provided august bodies have come to different conclusions
an estimate of the magnitude of potential benefits about the same treatment.’O What are we practi-
(and 34%, the potential adverse effects) of the tioners to do? Read as many reviews as possible,
treatment options, 45% mentioned randomized I guess, especially those found in Reviews in Clin-
clinical trials to support their recommendations, ical Gerontology, then take a consensus.
but only 6 % made any reference to costs. Another
paper by Bramwell, examining the reviews in the
Search strategy and selection criteria
Journal of Clinical Oncology in the USA, similarly
found poor result^.^ For this review of reviews, I searched the PubMed
Another paper on the subject looked at reviews database by the National Library of Medicine
on the treatment of neck pain,6 and assessed the website for the past 20 years, and, from review
quality, conclusions, and agreement between of the authors, titles, abstract, and source loca-
reviews on the conservative treatment of neck dis- tion, articles in full were selected for further
orders. Computerized bibliographic databases examination. Medline searches were repeated
were searched for reviews published before Janu- with Ovid Technologies, Version 4.4.1, through
ary 1998 that included neck pain and evaluated the BMA Search Site on 9 August 2003. I also
conservative therapies. Only reviews that reported searched the Internet using the Google search
at least one controlled clinical trial were consid- engine. References were selected according to the
ered. Of 2.5 review articles selected, 12 were sys- authors’ identification of relevant topics for the
tematic reviews. Statistical pooling was performed review and did not include non-English language
only in two high-quality systematic reviews. The papers. Articles and their abstracts were initially
evidence was inconclusive for the use of conserv- collected using Reference Manager v.10.
ative interventions or manipulation or traction in
neck pain, but many of the reviews displayed
References
major methodological flaws.
Even Cochrane Reviews can be improved upon. 1 Siwek J, Gourlay ML, Slawson DC, Shaughnessy
A recent study compared Cochrane reviews and AF. How to write an evidence-based clinical
reviews published in paper-based journals. Two review article. Am Fum Physician 2002; 65:
assessment tools were used to collect the data, a 251-58.
23-item checklist developed by Sacks’ and a nine- 2 MacAuley D, McCrum E, Brown C. Randomised
controlled trial of the READER method of
item scale developed by Oxman.8 Cochrane
critical appraisal in general practice. BMJ 1998;
reviews were found to be better at reporting some 316(7138): 1134-37.
items and paper-based reviews at reporting oth- 3 Thornley B, Adams C. Content and quality of
ers. The overall quality was found to be low. The 2000 controlled trials in schizophrenia over 50
authors make the point that this represents a seri- years. BMJ 1998; 317(7167): 1181-84.
ous situation because clinicians, health policy 4 McAlister FA, Clark HD, van Walraven C etal.
makers, and consumers are often told that sys- The medical review article revisited: has the
Downloaded from https://www.cambridge.org/core. Universitas Indonesia, on 10 Jul 2018 at 06:32:21, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0959259803013212
What is a clinical review? 105
Downloaded from https://www.cambridge.org/core. Universitas Indonesia, on 10 Jul 2018 at 06:32:21, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0959259803013212