Infrastructure requirements for secondary data sources in comparative effectiveness research.

Abstract

The growing interest in comparative effectiveness research (CER) has re-ignited the debate about the inadequacy of data from randomized controlled trials (RCTs) to address patient-centered decision making. Despite their well-known internal validity and use as the gold standard for regulatory decision making, the limitations of RCTs are widely recognized. In addition to their lack of statistical power due to inadequate sample size to address certain research hypotheses, practical and ethical considerations may preclude their viability. A case in point is the ethical dilemma in conducting an RCT to establish whether a diet high in fat content may be a risk factor for dementia, which might produce useful public health information but would not be acceptable in terms of protection of human subjects. Frequently, RCTs provide substantial information regarding the efficacy of drugs and other medical interventions, yet leave large gaps in evidence that would be relevant for medical decision making. Even when RCTs are carried out with this intent, they may not necessarily reflect “real-world” experience and, therefore may not provide sufficient evidence to guide patient-centered care. The substantial investment in CER and the broad objective implied in the American Recovery and Reinvestment Act of 2009 have necessitated the need to seek alternative sources of data to meet the emerging health care questions. The stated requirements include an “.... assessment of a comprehensive array of health-related outcomes for diverse patient populations and sub-groups” as well as “... a wide range of interventions;” and “[d]evelopment, expansion, and use of a variety of data sources and methods to assess comparative effectiveness.”1-2 Contending with the changes in health care policy and delivery clearly requires doing things differently, as discussed at a recent workshop sponsored by the Institute of Medicine.3 The workshop summary highlighted the dependence on clinical trials as the “sole source of evidence on the constantly accelerating flow of diagnostic and treatment challenges is unfeasible.” The need for a “learning healthcare system” with “real-time learning from the clinical experience and seamless application of the lessons in the care process” was emphasized. Secondary data, such as registries and retrospective databases, can be used to complement RCTs, since they are less costly and can be used to incorporate real world experience to answer questions. Further, important questions, such as adherence, treatment patterns, and burden of disease, can be answered in retrospective analyses of databases. However, effective use of secondary data requires addressing major methodological and infrastructural issues that may be related to, but often go beyond, those encountered with most RCTbased work. Infrastructural issues, such as tools to efficiently access and correctly analyze the data, need to be developed for effective use of such data sources. Guidelines need to be formulated and data standards established using RCTs as a role model. Data warehouses are also required to be established that respect the privacy and confidentiality of patients. In this paper, we discuss the infrastructural requirements for secondary data utilization in CER, and identify gaps that must be filled to address the underlying issues, with emphasis on data standards, data quality assurance, data warehouses, computing environment, and protection of privacy and confidentiality.