Ascertainment and Classification of Outcomes

Abstract

Successful completion and valid conclusions from a clinical trial rely on having complete and accurate outcome data. Misclassification of outcome events can introduce systematic error and bias as well as reduce the statistical power of the trial. Outcomes of interest in clinical trials vary and can include major clinically recognized events; asymptomatic subclinical measurements; and/or patient-centered reported outcomes. Final classification of study outcomes often involves use of standardized computer algorithms, processing of materials and review with outcome classification committees, and linkage with electronic data sources. Processes for accomplishing the goals of outcome ascertainment and classification can be designed as centralized systems, de-centralized networks of investigators, or a hybrid of these two methods. Challenges to obtaining valid outcomes include ensuring complete follow-up of study participants, use of standardized event definitions, capture of relevant diagnostic information, establishing protocols for review of potential events, training of clinical review teams, linkage to data sources across various platforms, quality control, and administrative oversite of the process. Designers of clinical trials need to consider carefully their approach for event identification, capture of diagnostic data, utilization of standardized diagnostic algorithms and/or clinical review committees, and mechanisms for maintaining data quality.