Lengthening survival after cancer due to advances in detection and treatment has resulted in an increasing awareness of treatment complications, known as late effects, arising years after cancer diagnosis and treatment. Identifying, preventing and treating these late effects informs treatment decision-making and optimizes survivors' physical and psychosocial well-being. We recently reported a 2.3 increase in the odds of having a stroke among breast cancer survivors with a history of having undergone chemotherapy versus survivors not undergoing chemotherapy (95% confidence interval 1.4 - 3.8; Geiger et al., J Natl Cancer Inst, 2004). Other reports provide some support for an association between chemotherapy and stroke, and there are pathological mechanisms that could explain an association. We propose using the unique patient and data resources of the healthcare delivery system-based Cancer Research Network to understand further the relationship between chemotherapy and stroke more than one year after cancer diagnosis. Specifically, we aim to use Cox proportional hazards regression modeling to estimate the relative risks of stroke after chemotherapy in over 110,000 ethnically diverse patients diagnosed from 1994 to 2003 with bladder, female breast, colorectal, Hodgkin's lymphoma, adult leukemia, multiple myeloma, non-Hodgkin's lymphoma, ovary and testis cancers, adjusting for age, gender, race/ethnicity, anatomic cancer site, stage at diagnosis, year of diagnosis, and dispensed medications for hypertension, diabetes, anti-coagulants and tamoxifen (for breast cancer only). Our proposed study responds direction to the National Cancer Institute's Program Announcement 04-012, Cancer Surveillance Using Health Claims-Based Data Systems, which recognized both the importance of complications of cancer treatment as a research topic and the unique resource of existing automated healthcare data for addressing the topic. Examining and quantifying the possible association between chemotherapy and risk of the late effect of stroke will provide further evidence for treatment decision-making and long-term follow-up of cancer survivors. In addition, verifying and better characterizing the association will guide determinations of the need for and design of further research to improve our understanding of the impact of specific chemotherapy regimens on stroke. ? ? ?