Alternative Approaches to Data Access
In the interest of investigating ways to provide greater access to research data, the workshop invited representatives of various organizations to describe several alternative approaches. In order of presentation, these approaches were described by Robert O’Keefe, Health Effects Institute; E. William Colglazier, National Research Council; Barry S. Kramer, Office of Medical Applications of Research, NIH; and Kenneth W. Harris, National Center for Health Statistics.
Health Effects Institute. As discussed in Chapter 2, the HEI agreed to accept the data from the Six Cities Study and undertake a reevaluation. The Harvard researchers agreed to open their voluminous files to them. The mission of the HEI, said Dr. O’Keefe, is to perform and evaluate the health effects of mobile source emissions and to inform regulatory decision making. The Institute’s procedures are specifically designed to maintain objectivity in controversial regulatory situations. HEI has a separate board of directors, standing research committee, and peer-review committee, with no direct sponsor participation.
The HEI reanalysis model is not a complete substitute for public access to data. If not carefully considered, a reanalysis may be used as a tool to delay action. It also requires the participation of the original investigators. However, it may have a role in testing major studies that bear on important policy or health questions.
HEI also has another alternative approach to data sharing that flows from its National Morbidity and Mortality Air Pollution Study (NMMAPS). The goal of the new program, called “The Internet Health and Air Pollu
tion Surveillance System,” is to provide public access to data and statistical software and methods that are used in the NMMAPS. The system is regularly updated as new data are made available and investigators publish new work.
Phase I will create an Internet site to disseminate the NMMAPS data and provide statistical software. Phase II, funds permitting, will create an interactive system so that users can download analytical tools and manipulate data on their own. This approach is designed to provide easy access to data and to facilitate interaction with complex public data sets, including an EPA air database, mortality data from the NCHS, weather data, and Census data. Among the benefits of this approach is to open access to those with small budgets.
National Research Council. The National Research Council of the National Academies provides a different mechanism for data sharing and data assessment through independent studies by panels of experts. The NRC was chartered during the administration of Abraham Lincoln to provide advice to the government about scientific and technical matters. About 80 percent of its work is paid for by government agencies, but it remains administratively independent from the government as a private non-profit organization.
The studies conducted by the NRC, said Dr. Colglazier, are advisory in nature, not regulatory, although many of them influence regulatory policy. Typically, the NRC is asked to look at not only the underlying data, but also the experimental design, the methods of analysis and the execution of a study (studies). In addition to the work of the expert committee and National Academies staff, an independent group of experts reviews the finished report before release.
The National Academies work under an amendment to the Federal Advisory Committee Act to provide certain kinds of public access to their studies. When the National Academies receive information from outside parties, they do so in an open meeting. That information is put in a public access file that is subject to FOIA exemptions. The law also allows the National Academies to close their meetings when committees are deliberating on their final conclusions and recommendations. Draft reports and reviewers’ comments do not go into the public access file. The findings are made public when the report is publicly released.
NIH Office of Medical Applications of Research. The Office of Medical Applications of Research, said Dr. Kramer, is the focal point for medically based assessments of medical practices and state of the science. It is not directly advisory to the NIH; instead, it appoints a panel of experts to lead a “consensus development conference.” The panel is asked to form an independent judgment based on available evidence. This report is not vetted by the NIH and stands as an independent report.
The issue for a conference must meet several criteria: it must have public health importance; affect or broadly apply to a significant number of people; reflect a gap between current knowledge and practice; and draw primarily on available scientific information. Additional elements of importance are the impact on health care costs and the degree of public or congressional interest.
The consensus development conference is a relatively new tool. It begins with a solicitation of topics followed by the formation of a planning committee that nominates a panel of independent experts and speakers and drafts questions. Panel members cannot have made a public statement about the conference topic and cannot be speakers. A memorandum of understanding is made with the Agency for Health Care Research and Quality (AHCRQ), which arranges for a systematic review of the literature. The panel then meets to study the issue, using rigid criteria for filtering and rating the quality of evidence.
The conference itself features a presentation of the literature review, speakers, and public input. The Evidence-Based Practice Center, commissioned by the AHCRQ, helps formulate questions so that they are searchable through electronic databases. After the conference, the panel drafts its final assessment.
National Center for Health Statistics (NCHS). The National Center for Health Statistics, said Dr. Harris, is the federal agency primarily responsible for the collection and dissemination of data related to the health of the United States population. The data are collected through national surveys and vital registration systems, including the National Health Interview Survey, the National Health and Nutrition Examination Survey, and the National Survey of Family Growth.
Despite the value of its data collection, the NCHS has been hampered by researchers’ inability to make maximum use of it. For example, the NCHS cannot release data files to the public that contain detailed information about the subjects that could facilitate identification; nor can geographic places of fewer than 100,000 people be identified. Such restrictions severely limit the ability of researchers to address urgent health-care questions. Accordingly, about 2 years ago the Research Data Center was created within the NCHS to help researchers gain access to important data. They do this by ensuring confidentiality and monitoring use of the data.
Costs of Alternative Approaches. Each of the previously mentioned approaches has benefits and shortcomings. Of note is the cost associated with these mechanisms. The HEI reanalysis of the Six Cities Study was moderately expensive and time consuming; costing approximately $1 million over 2.5 years. A full committee report of the NRC at the National Academies costs several hundred thousand dollars to produce. An NIH
consensus development conference costs about $500,000 and takes approximately 1 year. The NCHS charges for access to its research data in several ways, with remote access costs at $500 per month for fields of under 130,000 records, and on-site access costs at $200 per day, plus a $500 set-up fee to retrieve the data files.