PURPOSE: In the area of anticancer drugs, the legitimate search for effective interventions can be jeopardized by the strong pressure for accelerated approval, which may hinder the full assessment of their benefit-risk profile. We aimed to produce drug-specific recommendations using an explicit approach that separates the judgments on quality of evidence from the judgment about strength of recommendations. MATERIALS AND METHODS: We used the GRADE (Grades of Recommendation, Assessment, Development, and Evaluation) system to develop recommendations for the use of specific anticancer drugs/regimens; 12 clinical questions relevant to adjuvant treatment of breast (three), colorectal (four) and lung (five) cancer have been assessed by multidisciplinary panels supported by a group of methodologists. RESULTS: For nine of 12 questions, recommendations were produced (one strong and six weak in favor and one weak and one strong against the index treatment); for the remaining three questions no specific course of action could be recommended. The perceived benefits to risk balance of the treatment was the most important and statistically significant (P < .01) predictor of panels' recommendations and of their strength, whereas panelists' personal (age, sex) and professional (specialty) characteristics were not statistically associated. CONCLUSION: Because the GRADE system sets out an explicit process going from evaluation of the quality of evidence and benefit-risk profile to the judgment of the strength of recommendations, in this experience, it proved very useful to combine methodologic rigor with the interdisciplinary participation that is important in the definition of evidence based clinical policies.
PURPOSE: In the area of anticancer drugs, the legitimate search for effective interventions can be jeopardized by the strong pressure for accelerated approval, which may hinder the full assessment of their benefit-risk profile. We aimed to produce drug-specific recommendations using an explicit approach that separates the judgments on quality of evidence from the judgment about strength of recommendations. MATERIALS AND METHODS: We used the GRADE (Grades of Recommendation, Assessment, Development, and Evaluation) system to develop recommendations for the use of specific anticancer drugs/regimens; 12 clinical questions relevant to adjuvant treatment of breast (three), colorectal (four) and lung (five) cancer have been assessed by multidisciplinary panels supported by a group of methodologists. RESULTS: For nine of 12 questions, recommendations were produced (one strong and six weak in favor and one weak and one strong against the index treatment); for the remaining three questions no specific course of action could be recommended. The perceived benefits to risk balance of the treatment was the most important and statistically significant (P < .01) predictor of panels' recommendations and of their strength, whereas panelists' personal (age, sex) and professional (specialty) characteristics were not statistically associated. CONCLUSION: Because the GRADE system sets out an explicit process going from evaluation of the quality of evidence and benefit-risk profile to the judgment of the strength of recommendations, in this experience, it proved very useful to combine methodologic rigor with the interdisciplinary participation that is important in the definition of evidence based clinical policies.