Article type
Year
Abstract
Introduction/Objective: Data from more than 50 randomized trials, having started between 1965 and 1993, and comparing a loco-regional treatment to same treatment plus chemotherapy in head and neck cancer were collected to perform a meta-analysis. The aim of this work is to explore the variations between trials of the effect of chemotherapy.
Methods: Updated individual data were obtained for each trial and verified using protocols and publications. Treatment effect was studied using log-rank test stratified by trial. Data on randomization method, publication status, trial size, submission format: paper vs. computerized, and significant imbalance on covariate were recorded.
Results: 76 trials (11 650 patients) were identified. As of January 97, data from 51 trials (74% of the patients) were available for analysis. Data on age, sex and site of tumor were available for 99% of the patients, 98% for stages and 51% for performance status. Among these 51 trials, 50% were randomized by telephone, 62% included less than 150 patients, 36% started before 1980, 25% were received on paper form, 9% were unpublished, 14% published as abstract, 74% published as full paper in English, and 8% had a significant imbalance on one covariate.
Discussion: Meta-analysis based on individual data is feasible in head and neck cancer. The variations of treatment effect according to trial characteristics will be presented at the meeting. These results will be based on the above data and will also include additional data from trials currently under checking. Supported by grants from ARC (6560), PHRC (IDF 95013) & European Commission (Biomed 961321).
Methods: Updated individual data were obtained for each trial and verified using protocols and publications. Treatment effect was studied using log-rank test stratified by trial. Data on randomization method, publication status, trial size, submission format: paper vs. computerized, and significant imbalance on covariate were recorded.
Results: 76 trials (11 650 patients) were identified. As of January 97, data from 51 trials (74% of the patients) were available for analysis. Data on age, sex and site of tumor were available for 99% of the patients, 98% for stages and 51% for performance status. Among these 51 trials, 50% were randomized by telephone, 62% included less than 150 patients, 36% started before 1980, 25% were received on paper form, 9% were unpublished, 14% published as abstract, 74% published as full paper in English, and 8% had a significant imbalance on one covariate.
Discussion: Meta-analysis based on individual data is feasible in head and neck cancer. The variations of treatment effect according to trial characteristics will be presented at the meeting. These results will be based on the above data and will also include additional data from trials currently under checking. Supported by grants from ARC (6560), PHRC (IDF 95013) & European Commission (Biomed 961321).