Propensity based approach for analysis of radiotherapy effect on cardiovascular and pulmonary disease risks in lung and breast cancer patients using SEER-Medicare database

*Igor Akushevich, Center for Population Health and Aging, Duke University 
William F Anderson, BB/DCEG/NCI 

Keywords: Propensity score, radiotherapy, Medicare, cardiovascular and pulmonary diseases

Cardiovascular and pulmonary diseases are the leading causes of non-cancer mortality among cancer patients who has got radiotherapy. The majority of cancer patients are older than 65+, therefore, the Surveillance Epidemiology and End Results registry data linked to Medicare could provide researchers with detailed information on both cancer characteristics and comorbidities in the U.S elderly through the reconstruction of individual disease histories. 176,187 lung cancer patients and 168,337 breast cancer patients were selected for analysis. Cancer diagnoses were collected in 1992-2005 and individuals were followed-up till the end of 2007. Since multiple factors can influence the probability of getting radiation therapy including cancer type (grade, stage, laterality), comorbidity (Charlson index), demographic characteristics (age, gender, race, marital status), and treatment other than radiation therapy (e.g., surgery), the propensity score technique was used to provide a pseudorandomization in compared groups––i.e., with and without radiatiotherapy. Total mortality of cancer patients and cardiovascular and pulmonary disease incidence were analyzed. The most important variables predicted the administration of radiation therapy included comorbidity, cancer type, and surgery performed. Radiation therapy is beneficial for total survival of patients with lung and breast cancers. Patients with lung cancer had higher risk of cardiovascular and pulmonary diseases after radiotherapy performed compared to breast cancer patients. Among both lung and breast cancer patients, the effect of radiotherapy on increased risk of pulmonary disease was higher than for cardiovascular disease. Stage-specific analysis demonstrated less beneficial effect of radiation therapy for I-II stages. Analysis also revealed the importance of the effect of latency period. The propensity-based approach is the most appropriate for analyses of the radiotherapy effect on cardiovascular and pulmonary disease risks in lung and breast cancer patients because of preselection of cancer patients for radiotherapy based on i) cancer characteristics and ii) comorbidities including cardiovascular and pulmonary diseases.