STATEMENT FOR THE RECORD
RICHARD N. MILLER, M.D., M.P.H.
DIRECTOR, MEDICAL FOLLOW-UP AGENCY
INSTITUTE OF MEDICINE
NATIONAL ACADEMY OF SCIENCES
SUBCOMMITTEE ON HUMAN RESOURCES
COMMITTEE ON GOVERNMENT REFORM AND OVERSIGHT
U.S. HOUSE OF REPRESENTATIVES
May 14, 1998
Mr. Chairman, Members of the Subcommittee:
I am Dr. Richard Miller, Director of the Medical Follow-up Agency, at the National Academy of Sciences. Our organization has been carrying out research on veteran’s health issues for more than 50 years, since our founding by Dr. Michael DeBakey in 1946. I am pleased that Dr. Marie Swanson, who participated in the preparation of this testimony, could be present with me today. Dr. Swanson is Professor of Medicine and Director of the Cancer Center at Michigan State University. She is a cancer epidemiologist in charge of one of the ten SEER programs mentioned in the GAO report. She is also a member of our board of advisors and has served on two Institute of Medicine committees studying Persian Gulf War Illness. However, we are speaking for ourselves today and not for the Institute of Medicine, since there has been no official IOM review of the report, Gulf War Veterans: Incidence of Tumors Cannot Be Reliably Determined from Available Data.
In my opinion, the report is generally correct in its conclusion that, on the basis of methods currently employed, the incidence of tumors cannot be readily, routinely, and reliably determined in Persian Gulf War Veterans. The report’s descriptions of the strengths and limitations of studies completed and existing data sources are for the most part correct. One error of fact crept into the report concerning the Surveillance, Epidemiology, and End Results (SEER) Program. While it is true that the data are reported to the National Cancer Institute in aggregated, anonymous form, each registry site does have identifier information, which can be linked to individuals. Thus it is possible to link all 697,000 Persian Gulf War Veterans and Persian Gulf War era controls at each of the SEER registry sites, permitting comparisons of the incidence of tumors of Persian Gulf War Veterans with the incidence in era controls and with comparable U.S. populations.
Obtaining data of this type is a complex undertaking, fraught with potential errors. If it is decided that long term evaluation of the incidence of tumors in Persian Gulf War Veterans is necessary – in spite of the fact that to date there are no data that convincingly indicate excess risk of tumors in Persian Gulf War Veterans – the effort needs to be done very well to avoid errors that may have far reaching implications. Expert advice from outside the government may be required if credible results are to be produced.
The objectives of a system to address the question of whether Persian Gulf War Veterans are experiencing excess cancer incidence and mortality are:
• to provide alerts if tumor rates exceed population norms in the entire Persian Gulf War Veteran group or subpopulations that may be at greater risk
• to conduct targeted (cancer specific, exposure specific) studies when indicated by alerts or by new exposure information
• to have a credible system in which Persian Gulf War Veterans have confidence
• to keep costs within reasonable limits
The combination of a low cost sentinel system with a more definitive, elaborate, and expensive studies to follow up on alerts from the sentinel system would seem to meet all of the objectives if done correctly. The sentinel system would need to have adequate sensitivity to detect possible departures from expected tumor rates, would use existing or projected databases, and would require periodic reassessment of cancer incidence and mortality.
For general cancer incidence surveillance, linkage of the 697,000 Persian Gulf War veterans’ files with regional, population-based cancer registries can be carried out periodically, e.g., every five years. Cancer mortality surveillance would be a useful complement to the incidence system and is being done by the Environmental Epidemiology Service of the Department of Veterans Affairs using the BIRLS records, supplemented with data from the National Death Index. Mortality surveillance is more efficient and less expensive than incidence surveillance and is a reasonable approximation of incidence for certain forms of cancer that are known to have chemical and environmental etiologies, such as lung cancer and liver cancer. Five-year survival is just 3% for liver cancers and 14% for lung cancers. Therefore, for certain forms of cancer, 5-year mortality studies, which can be carried out more rapidly than the incidence studies, will provide more rapid surveillance. Mortality surveillance would be best conducted as a supplement to incidence surveillance, rather than as the sole endpoint evaluated. These methods would provide accurate assessment of general patterns of cancer incidence and mortality among Persian Gulf War veterans and the general population. It also would be appropriate to consider comparison with a comparison group selected to remove the “healthy soldier” effect from the analyses.
To have more than a general surveillance system, specific exposures that are known to increase the risk of specific forms of cancer must be identified and linked to well-defined subsets of veterans who are thought to have had these exposures. Similarly, should the surveillance systems identify subsets of veterans with high risks of specific cancers, studies could be designed to assess relevant exposures, or at least to improve the monitoring systems during future deployments.
I would reiterate that to date there are no data that convincingly indicate excess risk of tumors in Persian Gulf War Veterans. If, however, it is decided that determinations of cancer incidence are desirable, we suggest that the subcommittee consider having an independent expert panel formed to develop a master plan to:
Conduct mortality surveillance, including cancer, on a schedule, such as every five years – now being done by the VA.
Conduct cancer incidence surveillance utilizing other inexpensive, records based studies that can be repeated routinely as part of a sentinel system.
Assess the organization best qualified to conduct the incidence and mortality studies. For example, the VA could contract with individual regional cancer registries.
Determine which population-based cancer registries should be selected to supplement the SEER sites in order to have the most complete cancer surveillance system.
Evaluate changes in the medical information and health surveillance systems underway for both active and veteran populations to determine whether other records systems may be useful or preferred for cancer surveillance in the future.
Recommend a process for updating cancer surveillance activities over time, as new sources of information become available.
Recommend general approaches to cancer surveillance that can be used to monitor cancer occurrence in any past or future deployed populations.
Mr. Chairman, Dr. Swanson and I will be happy to answer questions.