Methodological quality of an article on the treatment of gastric cancer adopted as protocol by some Chilean hospitals

• Published in: Revista medíca de Chile 2006-07, Vol.134 (7), p.920-926
• Main Authors: Manterola, Carlos, Torres, Rodrigo, Burgos, Luis, Vial, Manuel, Pineda, Viviana
• Format: Article
• Language: Spanish
• Subjects: Antineoplastic Combined Chemotherapy Protocols - standards; Biomedical Research - standards; Chile; Combined Modality Therapy - standards; Data Interpretation, Statistical; Evidence-Based Medicine - standards; Humans; Practice Guidelines as Topic; Quality Control; Randomized Controlled Trials as Topic - standards; Research Design - standards; Stomach Neoplasms - surgery; Stomach Neoplasms - therapy
• ISSN: 0034-9887

Surgery is a curative treatment for gastric cancer (GC). As relapse is frequent, adjuvant therapies such as postoperative chemo radiotherapy have been tried. In Chile, some hospitals adopted Macdonald's study as a protocol for the treatment of GC. To determine methodological quality and internal and external validity of the Macdonald study. Three instruments were applied that assess methodological quality. A critical appraisal was done and the internal and external validity of the methodological quality was analyzed with two scales: MINCIR (Methodology and Research in Surgery), valid for therapy studies and CONSORT (Consolidated Standards of Reporting Trials), valid for randomized controlled trials (RCT). Guides and scales were applied by 5 researchers with training in clinical epidemiology. The reader's guide verified that the Macdonald study was not directed to answer a clearly defined question. There was random assignment, but the method used is not described and the patients were not considered until the end of the study (36% of the group with surgery plus chemo radiotherapy did not complete treatment). MINCIR scale confirmed a multicentric RCT, not blinded, with an unclear randomized sequence, erroneous sample size estimation, vague objectives and no exclusion criteria. CONSORT system proved the lack of working hypothesis and specific objectives as well as an absence of exclusion criteria and identification of the primary variable, an imprecise estimation of sample size, ambiguities in the randomization process, no blinding, an absence of statistical adjustment and the omission of a subgroup analysis. The instruments applied demonstrated methodological shortcomings that compromise the internal and external validity of the.