The prevalence and influence of self-reported conflicts of interest by editorial authors of phase III cancer trials