Integrating Data From Two Surveys To Estimate The Prevalence Of Cervical Cancer Screening And Its Associated Factors In The U.S.-Mexico Border Region

• Activity Number: 250
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Ruben Smith*+ and Dyanne Herrera and Emily Schiefelbein and Jill McDonald and Gita Mirchandani
• Companies: Centers for Disease Control and Prevention and Centers for Disease Control and Prevention and Texas Department of State Health Services and Centers for Disease Control and Prevention and Texas Department of State Health Services
• Keywords: Complex survey data ; data pooling ; subpopulation analysis ; logistic regression

Abstract

A binational surveillance system for the U.S.-Mexico border does not exist and reliable estimates for cervical cancer screening are not available for the region. We used comparable 2006 data from the BRFSS, a state based telephone survey, and the ENSANut, a state representative area probability sample and face-to-face survey, to estimate the prevalence of cervical cancer screening, and to describe associated factors in women aged 20-70 years without previous hysterectomy living in the border. The two surveys are independent and the target subpopulation for this study is contained in the combination of the populations targeted for the surveys. We considered each survey target population as a super stratum and then the data from the surveys were pooled together and domain type analyses were performed using SUDAAN. Among women aged 20-70 years without previous hysterectomy, 47% (95%CI=44.5-50.2) reported having a cervical cancer screening within the last year. Health insurance (AOR=1.9, 95%CI=1.5-2.4), marriage (AOR=1.8; 95%CI=1.4-2.3) and living on the US side of the border (AOR=3.5, 95%CI=2.8-4.4) were positively associated with having a cervical cancer screening.