Problems in dealing with missing data and informative censoring in clinical trials

Acommon problem in clinical trials is the missing data that occurs when patients do not complete the study and drop out without further measurements. Missing data cause the usual statistical analysis of complete or all available data to be subject to bias. There are no universally applicable methods for handling missing data. We recommend the following: (1) Report reasons for dropouts and proportions for each treatment group; (2) Conduct sensitivity analyses to encompass different scenarios of assumptions and discuss consistency or discrepancy among them; (3) Pay attention to minimize the chance of dropouts at the design stage and during trial monitoring; (4) Collect post-dropout data on the primary endpoints, if at all possible; and (5) Consider the dropout event itself an important endpoint in studies with many.

Time to All-cause Treatment Discontinuation as the Primary Outcome in the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) Schizophrenia Study

Time until all-cause treatment discontinuation was the primary outcome of the CATIE trial. We discuss the advantages and disadvantages of this outcome, and evaluate its association with clinical correlates through graphical response profiles. We investigate the characteristics of patients who discontinued for patient decision, including a reclassification of patient decision into other reasons. All-cause discontinuation is compared to a related outcome, time until treatment failure. Patients who discontinued had lower quality of life scores than other patients. Patients discontinuing for lack of efficacy had worsened efficacy scores compared with an improvement for other patients. Those who discontinued for patient decision had lower compliance. Blinded reclassification of discontinuation for patient decision identified 5% of cases as lack of efficacy and 21% as intolerable side effects. Reclassified patients participated in the next study phase at a higher rate than those remaining as patient decision (67% vs. 10%). Treatment group differences for time to discontinuation due to patient decision were attenuated after censoring the reclassified patients, but were still suggestive. Treatment comparisons for time to treatment failure were consistent with all cause discontinuation, although somewhat smaller. All-cause discontinuation is recommended as a simple and comprehensive outcome for pharmaceutical Phase II-IV clinical trials.

The versatility of multi-state models for the analysis of longitudinal data with unobservable features

Multi-state models provide a convenient statistical framework for a wide variety of medical applications characterized by multiple events and longitudinal data. We illustrate this through four examples. The potential value of the incorporation of unobserved or partially observed states is highlighted. In addition, joint modelling of multiple processes is illustrated with application to potentially informative loss to follow-up, mis-measured or missclassified data and causal inference.