indicated that he views the role of the Director of Special Populations as "the eyes and ears" on these matters. The committee believes that the director also needs to have a strong voice in priority-setting activities affecting special populations, as well as clear institutional guidance for identifying high-priority objectives for research on such populations.
Moreover, the Office of Special Populations Research currently lacks many of the institutional advantages that would ensure that an NCI commitment to research among special populations has a chance to be successful. It has no independent resources to fund a separate portfolio of initiatives for special populations research, has no clear criteria for recommending the priorities for such initiatives that are dependent upon the resources of other parts of NCI or means of guiding such initiatives, and holds no official position on any of the NCI advisory committees responsible for setting major intramural or extramural priorities.
In response to committee questions regarding the appropriateness of focusing research on special populations on the basis of burden of disease, Klausner indicated that the recommendations of the Special Action Committee on Minority/Special Populations did not represent current NCI policy. Although no specific recommendation of the Special Action Committee was repudiated in Klausner's testimony, the committee believes that many of the recommendations merit careful consideration. Among its useful recommendations were proposals to improve the input of staff and other persons knowledgeable about the needs of special populations, an acknowledgment of the importance of better incidence and mortality data among these groups, and the need for support for a data driven research agenda taking into account the top five cancers affecting each minority population (National Cancer Institute, 1996a).
The committee generally agrees, however, with the conclusion of the NIH priority-setting pamphlet (National Institutes of Health, 1997) regarding the difficulty and multiple ambiguities of the concept of burden of disease and how it is measured. The recommendations of the Special Action Committee Reports of 1992 and 1996 were silent on the issue of how those top five most burdensome cancers should be conceptualized, but this omission need not be fatal to NCI efforts to focus on neglected areas of research on cancers affecting each of the special populations. An increased burden can be detected in many ways, ranging from increased mortality to an increased incidence of particular cancers. No one-size-fits-all approach should be applied to each population. A more fruitful strategy will require the assessment of the unmet research needs of each population with whatever data are available regarding the type of burden specific to that subpopulation, in combination with an appraisal of the scientific opportunities. In short, no simple formula for addressing each group's top