203404 Challenges in randomizing highly-connected communities of Latina women

Monday, November 9, 2009

Kimberly Wagoner, DrPH (c), MPH , Department of Social Sciences and Health Policy, Wake Forest University School of Medicine, Winston-Salem, NC
Kristie Long Foley, PhD , Medical Humanities Program, Davidson College, Davidson, NC
Donna Kronner , Div of Public Health Sciences/Dept of Social Sciences & Health Policy, Wake Forest University School of Medicine, Winston-Salem, NC
Janet Tooze, PhD , Department of Biostatistical Sciences, Wake Forest University School of Medicine, Winston-Salem, NC
Scott Rhodes, PhD, MPH, CHES , Div of Public Health Sciences/Dept of Social Sciences & Health Policy, Wake Forest University School of Medicine, Winston-Salem, NC
Introduction: Randomization is the “gold standard” of research design and is highly desirable in establishing intervention efficacy and effectiveness. However, randomizing individuals in well-networked communities presents unique pragmatic and ethical challenges to the scientific community. In this pilot study, we examine patterns of spillover in an intervention targeting recently immigrated Latinas and discuss implications for the science of community-based research.

Methods: Two Latino apartment communities in Winston-Salem, NC, participated in an intervention to improve Pap smear screening and satisfaction rates among uninsured Latinas. Apartments within each complex were randomized to Intervention or Control. Participants meeting inclusion criteria completed a 45-minute psychosocial and behavioral assessment. Those randomized to the Intervention were also invited to the 2-session, lay health advisor-lead intervention in their community, designed to empower women to seek and negotiate Pap smear screening. No one was prevented from attending the classes as the overall goal was to reduce excess morbidity and mortality associated with cervical cancer.

Results: 177 women were recruited to participate. Mean age was 30.8 years with a range of 18 to 62. Among community-class attendees, 52% had been randomized to the Intervention, while 30% were from the Control group. The remaining 18% who attended were not randomized to either condition.

Conclusions: Findings suggest that randomization at the participant level may not be conducive to community-based research due to participant interconnectedness within communities. Spillover due to participant's proximity and networks reduced the study's internal validity. Future studies should consider randomizing at the community level to reduce this threat.

Learning Objectives:
1)Describe randomization at participant and community levels. 2)Identify key challenges to randomizing at the participant level in well-networked communities. 3) List alternative study designs that could be useful for community-based research.

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: Currently, I am a candidate for the DrPH and anticipate graduation in May, 2010. I am a Senior Research Associate at Wake Forest University School of Medicine where I've worked with multidisciplinary teams and various community groups to design, implement, and evaluate interventions focused on substance use prevention, women’s health issues, and childhood obesity.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.