Individually Randomized Group Treatment (IRGT) Trials in HIV/AIDS: How Often Is the Intraclass Correlation Ignored in Design and Analysis?

• Activity Number: 455
• Type: Topic Contributed
• Date/Time: Wednesday, August 4, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Sherri L. Pals*+ and Ryan E. Wiegand and David Murray
• Companies: CDC and CDC and The Ohio State University
• Keywords: IRGT trials ; intraclass correlation ; HIV/AIDS ; group interventions

Abstract

Interventions in HIV/AIDS are often delivered in groups, creating the potential for correlation among observations. Trials testing the effectiveness of such interventions need to take this correlation into account in design and analysis. Previous reviews have shown that the intraclass correlation coefficient (ICC) in individually randomized group treatment (IRGT) trials is often ignored, and this omission has the potential to increase the type I error rate substantially. We reviewed IRGT trials in 7 HIV/AIDS and public health journals from 2005-2009, identifying 25 eligible articles. Of these, none reported sample size calculations taking the ICC into account, and 23 (92%) reported analytic methods ignoring the potential for correlation within groups. We hope to increase awareness of design and analytic challenges of IRGT trials and encourage investigators to use appropriate methods.