In Brazil, a country of continental proportions, law requires that health institutions directly involved in the cancer treatment in the country have to register and consolidate cancer patients' data. This information is important for the management and evaluation of the cancer treatment, and for the definition of public policies. In this work we propose the use of a high-performance computational infrastructure, based on grid computing, to ease and accelerate the consolidation of the remotely distributed patients' records in these institutions. In particular, we present a case study of implementation of a tool for patients' survival estimation based on the Kaplan-Meier method. We also discuss the issues related to architecture, implementation and integration, as well as the benefits and prospects for the use of this infrastructure on the establishment of a national health information system.