Objective: To assess the rate of reporting of ACs and systematically review the process of adjudication described in reports of RCTs published in high-impact-factor medical journals.
Methods: Design: We searched the MEDLINE and Cochrane databases for reports of RCTs with clinical event endpoints published between January 1, 2004 and December 31, 2005 in the 5 medical journals with the highest impact factors. Main Outcome Measures: Data were extracted by use of a standardized form.
Results: A total of 314 articles were selected. ACs were described in only 33.4% of reports of RCTs with clinical event endpoints. ACs were reported for 81.3% of cardiovascular trials but only 28.6% of neurological trials and 15.4% of hematologic or oncologic trials. The composition of the AC was described for 89% of ACs reported. The method of selecting cases to be adjudicated was reported for most of the ACs described (88.1%) and consisted largely of events identified by site investigators (93.3%). Other methods allowing better estimates of the rate of events, such as use of national registries of death, adjudication of all patients and use of a specific computer algorithm, were described for less than 10% of the ACs reported. A description of information provided to the AC was given for less than half of the ACs (47.4%) and the reviewing process was described for less than one-third (28.8%). Finally ACs were reported as being blinded in only half of the descriptions of ACs.
Conclusions: ACs are widely used in cardiovascular trials but less so in other medical areas. Some important aspects of the functioning of ACs are insufficiently reported or raise methodological issues.