Assessing the usefulness of health data linkage in obtaining adverse event data in a randomised controlled trial of oral and implant naltrexone in the treatment of heroin dependence

Abstract

The completeness of self-reported serious adverse events (SAEs) in clinical trials can be reduced by inaccuracies in subject reporting and lost to follow-up. This study assesses the usefulness of a health data linkage system in obtaining SAE data in a randomised controlled study of oral and implant naltrexone. SAEs were collected from 68 heroin-dependent subjects during a randomised controlled trial of oral and implant naltrexone with follow-up to 26 weeks. Patient self-report data were cross-matched against hospital and emergency department (ED) attendances for the same period using a health data linkage system. A total of 29 hospital admissions and 74 ED attendances were identified using health data linkage. Of these, 12 (41.4%) hospital admissions and 50 (67.7%) of ED attendances had not been reported as SAE in the randomised controlled trial. In subjects participating in the trial at the time of the event, there was a 1.25-fold increase in the number of hospital admissions and a 2.25-fold increase in the number of ED attendances recorded using data linkage. Overall (including withdrawn subjects or those lost to follow-up), there was a 1.71-fold increase in hospital admission and a 3.09-fold increase in ED attendance recorded. The use of data linkage should not be used as a replacement for thorough follow-up, as the datasets can take substantial periods to update, making them a poor substitute for real-time follow-up. Additionally, some SAEs such as life-threatening events that do not involve ED or hospital attendance may be overlooked as would SAEs that occurred outside the dataset's range, for example, interstate or overseas. Health data linkage can be used to effectively reduce the extent of missing health data in a clinical trial.