Small Area Estimation of Being Current with Colorectal Cancer Test Use

Authors:

Zahava Berkowitz (Presenter)
CDC

Xingyou Zhang, U.S. Census Bureau
Thomas Richards, CDC
Marion Nadel, CDC
Lucy Peipins, CDC
James Holt, CDC

Public Health Statement: State-level information about colorectal cancer (CRC) screening can mask substantial variations in counties within the states. The small-area estimation model identified areas in need of intervention plans by local and state agencies to increase the uptake of CRC screening.

Purpose: In 2008, the U.S. Preventive Services Task Force recommended routine screening for colorectal cancer for adults ages 50–75 years: screening with high sensitivity fecal occult blood test (FOBT); sigmoidoscopy every 5 years with FOBT every 3 years; or colonoscopy every 10 years. We present novel small-area estimates (SAE) for being current with these recommendations at the county level to highlight areas in need of targeted interventions

Methods/Approach: Using the 2014 Behavioral Risk Factor Surveillance System (BRFSS) (N=251,360 adults), we constructed and fitted multilevel logistic regression models, and linked them to the U.S. Census population (post stratified) to generate county-level prevalence estimates for being current with CRC testing. We aggregated the county estimates into state and national estimates and mapped them. To evaluate internal consistency, we compared our SAE model state estimates with the BRFSS states direct estimates using the Spearman and Pearson correlation coefficients.

Results: Correlation coefficients indicated high internal consistency (ρ=0.95). National 2014 model-based SAE prevalence was 67.28%, (95% CI: 67.23%, 67.33%). The 2014 BRFSS estimate was 66.24%, (95% CI: 65.83%, 66.65%). Model-based county prevalence estimates ranged from 40.11% in Alaska to 79.76% in Florida. The 20 lowest ranking counties were in Texas, Montana, Alaska, Idaho, South Dakota, North Dakota, and New Mexico. State prevalence ranged from 58.92% in Wyoming to 75.03% in Massachusetts.

Conclusions/Implications: Our SAE analysis provides reliable county-level estimates. We found that state-level information about CRC screening can mask substantial within-state variability. These SAE can provide opportunities for municipal, state and federal public agencies to better identify areas in need of targeted CRC health promotion efforts.