Introduction: High quality reviews are a critical resource for the clinician. Mulrow (Ann. Int. Med. 1987; 106:485) showed reviews in 4 medical journals did not routinely use scientific methods to identify, assess and synthesise data.

Objective: To perform a qualitative analysis of reviews in JCO.

Methods: Hand search of JCO, Jan 1983-Dec 1995, for reviews and meta-analyses of cancer treatment. Quality was assessed using 8 criteria proposed by Mulrow, rated independently by 2 medical oncologists as: specified (S), unclear (U), not specified (N).

Results: 122 (69%) of 176 reviews dealt with treatment aspects. Meta-analyses (total 12) appeared from 1991, structured abstracts from 1992.

Detailed table on website
http://www.update-software.com/ccweb/cochrane/abpos23.htm

Authors rarely gave information on literature searches, criteria for study inclusion and quality assessment. Quantitative summation was confined to meta-analysis. Results analysed by publication year show recent improvement, especially since 1991, consequent upon inclusion of meta-analyses and, possibly structured abstracts.

Discussion: Reviews of cancer treatment in JCO were not systematic, and methodology needs improvements. Meta-analyses proved an exception. The Cochrane Cancer Network aims to improve the scientific rigour of clinical reviews. It will also prepare, maintain and disseminate systematic reviews of randomised clinical trials in cancer.