Frances Belda Maguire
Published: 2018
Total Pages:
Get eBook
Lung cancer is the second most common cancer and the leading cancer-related cause of death in both men and women. Non-small cell lung cancer (NSCLC) is the most common type of lung cancer, comprising approximately 84% of all lung cancer. Approximately half of patients with NSCLC are diagnosed at an advanced stage when survival rates are very poor (5 year survival of 4.5% for distant disease). Systemic therapies are the primary treatment for patients with advanced-stage disease. In the past two decades, many new systemic therapies have been developed. These treatments have been shown to increase survival in clinical trials, but their use and effectiveness at the population-level is unknown. The present study was based on data collected by the California Cancer Registry (CCR), California’s statewide cancer surveillance system. The CCR collects types of systemic treatment in an unstructured free-text format. In Chapter 1, we manually reviewed treatment text fields in the CCR to determine first-line systemic therapies used in patients with stage IV NSCLC and associations with survival. Chapter 2 explored the relationship between systemic therapy use and source of health insurance. Finally, in Chapter 3, we developed a SAS-based text mining algorithm to extract treatment information from unstructured free-text fields and compared the results to manual review. Our findings indicate that the treatments associated with a significant survival advantage (tyrosine kinase inhibitors and bevacizumab combinations versus platinum doublets) for advanced-stage nonsquamous NSCLC were underutilized among the over 14,000 patients included in our study. Substantial disparities in the use of systemic therapies exist by health insurance type in California. Patients with Medicaid, military, or no insurance have significantly decreased odds of receiving tyrosine kinase inhibitors, bevacizumab combinations, or any systemic treatment at all. Our SAS-based text mining algorithm accurately detected systemic treatments administered to stage IV NSCLC patients, with Kappa ranging from 0.71 to 0.92, and may become a viable alternative to the considerable time involved in manual review of patient records. Our algorithm can be applied to other cancer types, potentially maximizing the utility of extant information in cancer registries for comparative effectiveness research.