Randomized Cluster Crossover Trials for Reliable, Efficient, Comparative Effectiveness Testing: Design of the Prevention of Arrhythmia Device Infection Trial (PADIT) Journal Articles uri icon

  • Overview
  • Research
  • Identity
  • Additional Document Info
  • View All


  • Randomized clinical trials are a major advance in clinical research methodology. However, there are myriad important questions about the effectiveness of treatments used in daily practice that are not informed by the results of randomized trials. This is in part because of important limitations inherent in the methodology of randomized efficacy trials which are performed with tight control of inclusion, exclusion, treatment, and follow-up. This approach enhances evaluation of clinical efficacy (performance in controlled situations) but increases complexity and is not well suited to test clinical effectiveness (performance under conditions of actual use). The cluster crossover trial is a new concept for efficient comparative effectiveness testing. Deep tissue infection occurs in 2% of patients after arrhythmia device implantation, usually requires system extraction, and increases mortality. There is variation in antibiotic prophylaxis used to reduce implanted device infections. To efficiently evaluate the comparative effectiveness of antibiotic strategies now in use, we designed a cluster crossover clinical trial, which randomized implanting centres to 1 of 2 prophylactic antibiotic strategies, which became the standard care at the centre for 6 months, followed by crossover to the other strategy, rerandomization, and second crossover. This method greatly reduces trial complexity because it aligns study procedures with usual clinical care and increases generalizability. Pilot studies have tested the feasibility and an 10,800-patient trial, funded by the Canadian Institutes of Health Research, is now under way. The cluster crossover randomized trial design is well suited to efficiently test comparative effectiveness of existing treatments where there is variability of practice, clinical equipoise, and minimal risk.


  • Connolly, Stuart
  • Philippon, Francois
  • Longtin, Yves
  • Casanova, Amparo
  • Birnie, David H
  • Exner, Derek V
  • Dorian, Paul
  • Prakash, Ratika
  • Alings, Marco
  • Krahn, Andrew D

publication date

  • June 2013