Using Parallel Adjudication to Remove Bias from the Clinical Event Committee
A CEC is a group of independent experts tasked with reviewing and adjudicating all the events of interest in the clinical trial. The CEC performs blinded assessments of clinical event data and decide if the event under scrutiny meets pre-specified clinical event definitions. The event definitions are established in advance of trial commencement and are detailed in a CEC charter document that is available for review by regulatory agencies. The CEC results are used purely for trial analyses and are not fed back to the site investigators.
A CEC is made up of a variable number of subject matter experts and a CEC chairperson. Members should have clinical expertise in the endpoints of interest and ideally have CEC experience. If a broad range of endpoints must be assessed, then subject matter experts from varying specialties may be required.
It is most important that the selected chairperson has CEC experience and leadership ability. The chairman’s knowledge of adjudication becomes particularly important during committee meetings where adjudication disagreements are discussed. Trial details must be specified up front to ensure that members are able to commit for the full duration of the study and have sufficient capacity to meet the workload and timeline requirements. Event volume can often be large, so the number of committee members as well as member availability must be taken into consideration, with redundancy built in to account for illness and vacation time. The CEC members must be free of conflicts of interest and cannot act as trial investigators. Selection of CEC members can be performed by the sponsor or, as is more often the case, by a clinical research organization (CRO) or a specialist adjudication organization. The sponsor usually must approve the committee after selection.
There are several different CEC workflows that can be utilized, however, the one most resistant to bias is parallel adjudication. Each outcome event packet is randomly assigned to a pair of adjudicators. Each adjudicator reviews and adjudicates the package independently. If the answers given by each adjudicator are in agreement, then the event is considered adjudicated or “classified.” If there is discord between the reviewer outcomes, then the event is sent to a committee meeting for a group decision to be made. Minor disagreements such as event dates can often be dealt with before they reach full committee meetings (e.g. by both reviewers re-examining the event data and ensuring that the discord is not the result of a simple transcription error).