Using Innovative Algorithm for CRC Screening
CPOS Research Feature: John Bian


John Bian, PhD, Associate Professor, CPOS, is a health economist who has extensive hands-on experiences in managing large observational data (e.g., linked Surveillance, Epidemiology and End Results (SEER) and Medicare data) and in applying appropriate modeling techniques (e.g., hierarchical models) for causal inferences in observational data analysis. His main research interests are to understand determinants of changes in health care delivery systems and the impact of the changes on health care quality and outcomes, particularly cancer outcomes research. 

A recently funded 4-year VA Merit Award provided him and his  colleagues with the opportunity to develop and publish two studies that have demonstrated the feasibility of an innovative and improved claims-data-based algorithm method for more accurately measuring colorectal cancer (CRC) screening adherence. “This new measurement algorithm has 3 unique features,” explains John: (1) it focuses on average-risk populations. Because CRC screening guidelines vary by the individual’s risk level, it is important to distinguish average-risk from higher-risk; (2) this algorithm allows researchers to examine not only screening procedures (e.g., FOBT) utilized in a single year but also screening adherence (e.g., screening colonoscopy) in a given year by retrospectively examining screening procedures (e.g., colonoscopy) prior to the given year; and (3) it may distinguish among different screening adherence modalities.  

John and his team are now using this algorithm to better understand comparative effectiveness of CRC screening adherence, including in a newly awarded pilot grant by the MUSC Hollings Cancer Center to look at Comparative Effectiveness of Colorectal Cancer Screening Adherence among Average-Risk Elderly.

“Colorectal cancer (CRC) screening is an effective tool for reducing CRC deaths,” explains John.  “Although the recommended CRC screening guidelines have been adopted widely, there are still significant gaps in evidence on real-world, comparative effectiveness of CRC screening adherence,” he adds. “First, to study comparative effectiveness of CRC screening adherence, it is critical that adherence can be appropriately measured. Second, although the recommended CRC screening guidelines have been adopted widely, there are still significant gaps in evidence on comparative effectiveness of CRC screening adherence. Thus, comparative effectiveness of available competing screening modalities still remains largely unknown in real-word practice.”

The main goal of this pilot project is to examine comparative effectiveness of CRC screening adherence among the elderly diagnosed with incident CRC, but considered at average risk for CRC prior to diagnosis. To do this, John will conduct a retrospective, observational study using the 2004-2013 Surveillance, Epidemiology, and End Results (SEER)-Medicare linked data to address two questions:  1) Does a higher proportion of years adherent to CRC screening (regardless of modality) prior to incident diagnosis of CRC decrease the likelihood of detection at a late stage? and 2) What is the pairwise relative effectiveness CRC screening adherence by 4 modality in terms of detection of early stage of CRC, conditional on being adherent during at least one year prior to incident diagnosis of CRC?