Data Visualization of Completion Rate for PRO Objectives in Oncology Clinical Trials Supporting PRO Estimands

OBJECTIVES: Longitudinal analysis of PRO data in oncology clinical trials is widely used to assess the effect of treatment on patient-reported quality of life, functioning and disease symptoms. The introduction of the ICH E9(R1) addendum estimand framework and its application in oncology clinical trials has resulted in development of PRO estimands where discontinuation of treatment, disease progression and death are identified as intercurrent events. To align with planned PRO estimands, which may focus on-treatment period or prior to disease progression, it is important to distinguish at each visit that PRO data are “not expected” (e.g. after disease progression or death events) from PRO data which is expected but missing for other reasons. Our goal was to develop a data visualization solution, which reflected patient disposition and PRO completion rate to identify missing data and intercurrent events over time.

METHODS: Both SISAQoL recommendations and recent FDA guidance clarify that available data and completion rate should be reported and reasons for missing data should be summarized. We developed a data visualization to simplify presentation of this information to a reflect patients’ journeys through the study, including timing of death, implemented in R and SAS.

RESULTS: We present a novel divergent bar chart to illustrate both PRO completion rate and proportion of missing data over time by treatment arm, together with the cumulative occurrence of intercurrent events leading to data “not expected”, including disease progression and death.

CONCLUSIONS: Data visualization of available PRO data and completion rate together with clinical events in a patients’ journey enables clearer quantification of the potential impact of death and progression on the availability of PRO data, as well as the amount of missing data. This improves the ability to interpret PRO data over time in context of the defined PRO estimand.