1
Introduction
Evidence of what is effective in clinical practice, especially evidence of what is appropriate for specific individual patients, is often lacking. In addition, if such evidence is available, it is often not translated rapidly into standard clinical practice, nor is it followed uniformly across healthcare practices. Our current healthcare system is plagued by overuse, underuse, and misuse, leading a recent Institute of Medicine (IOM) committee to conclude there is an urgent need to “know what works” (IOM, 2008). This is problematic and challenging given the rapidity with which medical advances render standard care obsolete. A delay in translation or inappropriate care can shorten the life span of patients with life-threatening diseases. Regrettably, much of the information that could improve care is not currently collected or distributed at the point of care, despite recent advances in information technology that make this possible.
Opportunities for rapid learning, incorporating electronic health records (EHRs) and large datasets, were first identified by Etheredge and others (Eddy, 2007; Etheredge, 2007; Kupersmith et al., 2007; Liang, 2007; Lumpkin, 2007; Neumann, 2007; Pawlson, 2007; Perlin and Kupersmith, 2007; Platt, 2007; Slutsky, 2007; Stewart et al., 2007; Tunis et al., 2007; Wallace, 2007). Recognizing the importance and potential impact of rapid learning and the need to build knowledge development and application into each stage of the healthcare delivery process, the IOM Roundtable on Value & Science-Driven Health Care has conducted a series of workshops on the
“Learning Healthcare System” (IOM, 2007). These workshops have focused on various cross-cutting issues important for improving the development and application of practice-based evidence in healthcare decision making. The National Cancer Policy Forum of the IOM decided to apply the theoretical notion of a rapid learning healthcare system (RLHS) discussed in a broad sense in these workshops to cancer care, in specific.
The forum focused on cancer as a model for development of a RLHS because of the high prevalence of the disease, rising incidence in the aging population, complexity, variable outcomes, and high burden of care, as well as cancer’s life-threatening nature that highly motivates patients, their families, and healthcare providers to seek information to improve care. Despite decades of investment in the war on cancer, there is still a major unmet need in measurement of the clinical effectiveness of many cancer treatments. The rapid marketplace entry of new and often very costly healthcare technologies and treatments for cancer, which have not been evaluated completely for clinical effectiveness, has spurred a compelling public interest in advancing the evidence base for the comparative effectiveness of cancer care as rapidly as possible (IOM, 2009b). Fortunately, many of the foundational elements of a RLHS, such as cancer registries, cancer clinical trials, computer systems, academic and community cancer centers and networks, and evidence-based practice guidelines are already in place for cancer and thus offer another reason to examine cancer as a model for the development of a RLHS. There also is a long-standing successful model of a learning system in pediatric oncology, consistently trying to learn as much as possible from every patient, with standardized protocol-based treatments and systematic collection of clinical data and outcomes, resulting in a “virtuous cycle” of incorporation of what is learned into new treatment protocols that successively improve survival rates.
Although it is widely acknowledged that randomized clinical trials are the gold standard for development of clinical guidelines and that clinical and translational cancer research is essential to expanding the knowledge base in oncology, it is also recognized that dependence on expensive, time-consuming trials as the sole source of evidence is unfeasible. Moreover we need a better understanding of how diverse patient populations respond to cancer treatments in typical clinical settings. A RLHS for cancer would take full advantage of private and public sector databases and emerging information technology (IT), including EHRs, to generate and apply the evidence needed to deliver the best care for each individual cancer patient as rapidly as possible. In light of the current substantial public investments in health
information technology and comparative effectiveness research,1 the notion of a RLHS for cancer is both timely and topical. The potential for personalized cancer medicine and targeted therapies for cancer adds further urgency to foster development of rapid learning systems to know what works and deliver high-value cancer care.
The National Cancer Policy Forum held a workshop in Washington, DC, on October 5 and 6, 2009, titled A Foundation for Evidence-Based Medicine: A Rapid Learning System for Cancer Care, with the aim of horizon-scanning to describe the current landscape2 for rapid learning in cancer and to assess what policy and other measures are needed to foster progress and overcome obstacles to more fully develop a RLHS. This document is a summary of the conference proceedings. The views expressed in this summary are those of the speakers and discussants, as attributed to them, and are not the consensus views of workshop participants or members of the National Cancer Policy Forum.3