Development of a claims-based algorithm to identify colorectal cancer recurrence
Anjali D Deshpande1, Mario Schootman2, Allese Mayer2
1Division of General Medical Sciences, Department of Medicine, School of Medicine, Washington University in St. Louis, St. Louis, MO.
Annals of Epidemiology
|March 22, 2015
Summary
Claims data can accurately identify colorectal cancer (CRC) recurrence, improving survival estimates. This method enhances the study of CRC patient outcomes using large databases.
Area of Science:
- Oncology
- Health Informatics
- Epidemiology
Background:
- Accurate identification of colorectal cancer (CRC) recurrence is crucial for understanding patient survival.
- Population-based administrative databases are valuable for large-scale health research but require validation for specific endpoints like cancer recurrence.
Purpose of the Study:
- To assess the accuracy of claims data in identifying colorectal cancer (CRC) recurrence.
- To quantify how misclassification of recurrence status in claims data impacts overall survival estimates.
Main Methods:
- Validated claims data against medical records from a tertiary hospital to identify CRC recurrence.
- Applied validated findings to linked Surveillance, Epidemiology, and End Results-Medicare data to estimate survival effects.
Main Results:
- Claims data demonstrated 81% sensitivity and 99% specificity for CRC recurrence detection, with near-perfect agreement (kappa: 0.86).
- Uncorrected survival models showed recurrence increased mortality risk 3.04-fold; corrected models revealed a 3.47-fold increased risk.
Conclusions:
- Claims data offer a feasible method for identifying colorectal cancer (CRC) recurrence with acceptable accuracy.
- The developed algorithm can be utilized with SEER-Medicare data for future research on CRC treatment patterns and outcomes.
