Inclusion of elderly patients in oncology clinical trials

Since the creation of the International Society of Geriatric Oncology, the proportion of phase III trials reporting at least one analysis dedicated to elderly patients has grown (46.7% between 2011 and 2014 versus 19.3% between 2001 and 2004). However, evidence-based data are mostly extracted from subgroup analyses that can only be considered as preliminary evidence.Physicians need clinical trials assessing benefits and harms of treatments to avoid under-treatment or over-treatment of elderly patients. The main objectives of this report were to examine how data regarding elderly oncology patients were presented in medical literature; and to assess the evolution of this presentation between two time periods. All phases I, II and III trials dedicated to the treatment of cancer among elderly patients published between 2001 and 2004 and between 2011 and 2014 were reviewed. All phase III clinical trials assessing cancer treatments among adults in the same periods were also reviewed to evaluate potential subgroup analyses in elderly patients in these studies. Key characteristics of interest were extracted by two investigators before descriptive and comparative analyses were undertaken. A total of 1084 trials were included: 366 and 718 from the first and second time period, respectively. Twenty-seven and 193 of these trials were phase I and II trials dedicated to elderly or frail patients, respectively. A large proportion of phase III trials published between 2011 and 2014 reported at least one analysis dedicated to elderly patients (46.7%) versus 19.3% during the first time period. The use of subgroup analyses of elderly patients in phase III trials was the most frequent source of information. Subgroup analyses were more frequent among trials with industrial funding, trials published in high impact factor journal, intercontinental trials and trials with large sample size. The age threshold defining the elderly subgroup increased over time. Elderly patients have become a topic of interest during the past decade. However, data available are mostly extracted from subgroup analyses, which can only be regarded as preliminary evidence.