From Genomics and Informatics to Medical Practice
Sensible regulation and effective use of information technology are essential to reaping the benefits of this scientific revolution.
Biomedical research is being fundamentally transformed by developments in genomics and informatics, and this transformation will lead inevitably to a revolution in medical practice. Neither academic research institutions nor society at large have adapted adequately to the new environment. If we are to effectively manage the transition to a new era of biomedical research and medical practice, academia, industry, and government will have to develop new types of partnership.
Why are genomics and informatics more important than other recent developments? The spectacular advances in cell and molecular biology and in biotechnology that have occurred in the past two decades have markedly improved the quality of medical research and practice, but they have essentially enabled us only to do better what we were already doing: to respond to problems when we find them. As our knowledge expands, for the first time genomics will provide the power to predict disease susceptibilities and drug sensitivities of individual patients. For motivated patients and forward-looking practitioners, such insights create unprecedented opportunities for lifestyle adaptations, preventive measures, and cost-saving improvements in medical practices.
To illustrate this point, let me tell you about a recent conversation I had with a friend who had successful surgery for colon cancer 10 years ago. My friend recently moved to a new city. He selected the head of gastroenterology at a nearby medical school as his new oncologist. His initial visit to this doctor was a great surprise. The doctor took a very complete history but didn't do any laboratory tests or schedule any other examinations. The doctor simply asked my friend whether the cancerous tissue removed from his colon had been tested for mutations in DNA repair enzymes. It had, and no defects were identified. "If you had defects in your DNA repair enzymes," said the new oncologist, "I'd have asked you to come in for a colonoscopy right away and every six months thereafter. Since you don't have such defects, you don't need another colonoscopy for three years." Colonoscopies cost about $1,000 and between half a day and a day of down time. I calculate that DNA testing saved my friend $5,000 and 2.5 to 5 days of down time. Moreover, my friend's children now know that when they reach age 50 they won't need colonoscopies any more often than the rest of the population. I can't conceive of a bigger change in medical practice than this.
Advances in informatics will make it possible for every individual to have a single, transportable, and accessible cradle-to-grave medical record. Advanced information systems will allow investigators to use the medical records of individual patients for research, physicians to self-assess the quality of their own practices, and medical administrators to assess the quality of care provided by the health care personnel they supervise. And by granting public health authorities even limited access to the data collected, it will be possible for them to assess the health of the public in real time. These are not pie-in-the-sky predictions. All of these things are now technologically feasible. The sequencing of the human genome, coupled with extraordinarily powerful new methods in DNA diagnostics, such as gene chip technologies, allow us to identify relationships between physiological states and gene expression patterns. They allow us to identify gene rearrangements, mutations, and polymorphisms at a rate previously thought impossible.
Information technology is advancing at a phenomenal pace. Given the enormous financial incentives for further advances, it is not a big stretch to predict that the technology required for storing and processing the data from tens of thousands of chip experiments and for storing and analyzing clinical and genomic data on millions of people will be available by 2005. Indeed, it may already be available.
Barriers to progress
What are the impediments to bringing all this to fruition? There are many, but I will focus on a few. The first is the lack of public understanding of genetics. I am surprised by how little my well-educated friends in other fields and professions know about genetics. The state of genetic knowledge among practicing physicians is also of concern. A 1995 study showed that 30 percent of physicians who ordered a commercially available genetic test for familial colon cancer--the same test my friend had--misinterpreted the test's results. In another study, 50 neurologists, internists, geriatricians, geriatric psychiatrists, and family physicians managing patients with dementia were polled for their knowledge of lifetime risk of Alzheimer's disease in patients carrying the apolipoprotein E4 allele. Fewer than half of these physicians correctly estimated the risk of Alzheimer's disease in patients carrying the apo-E4 allele at 25 to 30 percent, and only one-third of those who answered correctly were moderately sure of the correctness of their response.
Life science researchers must alert their colleagues in other disciplines to the impact genomics will have on our understanding of all aspects of human life, from anthropology to zoology, and especially on what we know and think about the human condition. As C. P. Snow argued 42 years ago in The Two Cultures, science is culture. What Snow did not foresee is that genetics would become inextricably intertwined with the politics of everyday life, from genetically engineered crops to stem cell research. If we are to exploit the promise of genomics for the betterment of humankind, we must have a citizenry capable of understanding the rudiments of genetics. The research community can contribute to creating such a citizenry by ensuring that the colleges and universities at which they teach provide courses on genomics that are accessible to nonscience majors.
A second problem is the widespread public concern about the privacy of medical information, especially genetic information. In response to this public anxiety, Congress tried to develop legislation to protect the public against adverse uses of this information by insurers and employers, but it was unable to put together a majority in support of any of the proposals that attempted to find the right balance between the competing interests of individual privacy and the compelling public benefits to be derived from the use of medical information to further biomedical, behavioral, epidemiological, and health services research. As a result, it fell to the Clinton administration to write health information privacy regulations. These regulations were announced with much fanfare in the closing days of that administration and implemented by the Bush administration in April 2001.
Comprising more than 1,600 pages in the Federal Register, they contained plenty that the various constituencies could take issue with. The health insurance industry and the hospitals complained loudly that they were costly and unworkable. More quietly, the medical schools warned that they could be potentially damaging to medical research and education. According to an analysis by David Korn and Jennifer Kulynych of the Association of American Medical Colleges (AAMC), these privacy regulations provide powerful disincentives for health care providers to cooperate in medical research, because they impose heavy new administrative, accounting, and legal burdens, including fines and criminal penalties; and because they are ambiguous in defining permissible and impermissible uses of protected health information. This is of great concern when viewed in the context of the opportunities for discoveries in medicine and for improvements in health care that could arise from large-scale comparisons of genomic data with clinical records.
The capacity to link genomic data on polymorphisms and mutations of specific genes with family histories and disease phenotypes has enabled medical scientists to identify the genes responsible for monogenic diseases such as cystic fibrosis, Duchenne's muscular dystrophy, and familial hypercholesterolemia. Such analyses will be even more important in identifying genes that contribute to polygenic diseases such as adult onset diabetes, atherosclerosis, manic-depressive illness, various forms of cancer, and schizophrenia. The AAMC study revealed that the proposed regulations could slow this progress. Consider one example.
The regulations require that all individual identifiers be stripped from archived medical records and samples before they are made accessible to researchers. At first glance, that seems reasonable. But as one digs deeper, it becomes apparent that how one de-identifies these records is critical. De-identification must be simple, sensible, and geared to the motivations and capabilities of health researchers, not to those of advanced computer scientists who believe that the public will be best served by encrypting medical data so that even the CIA would have difficulty tracing them back to the individual to whom they relate.
The definition of identifiable medical information should be limited to information that directly identifies an individual. The AAMC describes this approach to de-identification as proportionality. It recommends that the burden of preparing de-identified medical information be proportional to the interests, needs, capabilities, and motivations of the health researchers who require access to it. AAMC says that the bar for de-identification has been set at too high a level in the new privacy regulations.
For example, these regulations require that "a person with appropriate knowledge of and experience with generally accepted statistical and scientific principles and methods for rendering information individually identifiable" must certify that the risk is very small that information in a medical record could be used alone or in combination with other generally available information to link that record to an identifiable person. This certification must include documentation of the methods and the results of the analysis that justifies this determination.
Alternatively, the rules specify 18 elements that must be removed from each record. These include Zip codes and most chronological data. But removal of these data would render the resulting information useless for much epidemiological, environmental, occupational, and other types of population-based research. The regulations also require that device identifiers and serial numbers must be removed from medical records before they can be shared with researchers. This would make it difficult for researchers to use these records for postmarketing studies of the effectiveness of medical devices.
The AAMC argues, and I agree, that sound public policy in this area should encourage to the greatest extent possible the use of de-identified medical information for all types of health research. The AAMC has urged the secretary of Health and Human Services to rethink the approach to de-identification and to create standards that more appropriately reflect the realities of health research, not the exaggerated fears of encryption experts.
Individual and societal rights
This is a classic confrontation between individual and societal rights. Since Hippocrates there has been widespread agreement that an individual's medical history and problems should be held in confidence. At the same time, there is equally widespread agreement that societies have legitimate interests in ascertaining the health status of their citizens, the incidence of specific diseases, and the efficacy of treatments for these diseases. The new regulations give too much weight to individual rights. We need to go back to the drawing board to try to get this balance right. With some creativity, we can satisfy both sides.
So far, the science community has not been involved in the privacy issue. I believe that this is the time for university researchers to join with the AAMC and others to ensure that the privacy regulations are changed so that all members of our society can benefit from our investment in medical and health research. Such information is needed now more than ever.
Although improved privacy regulations are essential, they will not reassure everyone. Toward that end, the scientific community can provide leadership in three ways: First, in the genomic era many, perhaps most, individuals will have genetic tests. Therefore, we must educate our faculty, staff, students, and the public about the benefits and complexities of the new genetics. Second, we must train faculty, staff, and health professions students to obtain informed consent from patients for use of historical and phenotypic data in conjunction with blood and tissue samples for research. And third, we must implement existing technologies and develop better ones to ensure the accessibility and security of medical records.
Implicit throughout this discussion is the need for widespread implementation of electronic medical records, which are as important to researchers as they are to physicians. Electronic medical records will facilitate communication among all health professionals caring for a patient, permit public health officials to assess the health of the public in real time, expand opportunities for self-assessment by individual professionals, and provide better methods for ensuring the quality and safety of medical practice.
In addition, there are special reasons for medical scientists to take an interest in this matter. The most straightforward is that without electronic medical records the process of de-identification will be hopelessly complex, time-consuming, and costly. But even if de-identification of paper records could magically be made simple and cheap, paper records will still be inadequate for genomic research. Genomic research requires the capacity to link specific genes and gene polymorphisms that contribute to disease with people who have that disease. Large-scale studies of this type will be markedly facilitated by the capacity to electronically scan the medical records of tens of thousands of patients.
The Institute of Medicine has issued several reports on the electronic medical record. However, progress has been slow. The reasons for this are many, including the complexity of capturing in standardized formats the presentations and courses of human diseases, the high cost of development and implementation of such systems, and the difficulties inherent in inducing health care professionals to use them. Yet without electronic medical records it will be extremely difficult for teaching and research hospitals to make full use of contemporary methods to screen and identify associations between genes and diseases.
The promise of genomics gives our teaching and research hospitals a new incentive for implementing electronic medical records, and industry and government should recognize that they have incentives for helping them do so. Our teaching and research hospitals have the clinical investigators and the access to patients needed to link genes and diseases. Industry has the capacity for high-throughput screening and the information systems needed to efficiently process these data to identify mutations and polymorphisms. And government, acting on behalf of society at large, has an interest in fostering such collaborations between the not-for-profit and the for-profit sectors. However, the problems, as I see them, are several.
First, there is at present no widespread consensus that the issues are as I have stated them. Second, the teaching and research hospitals have not yet recognized that they will have great difficulties in creating and implementing the necessary information systems without major assistance from government and/or industry. Third, industry has not yet recognized the magnitude of the task ahead and has not determined that the profits to be earned in this area are more likely to come from drug discovery than they are from finding gene targets. Fourth, with respect to intellectual property ownership, in the area of genetics the not-for-profit and for-profit sectors are in head-to-head competition.
We need to find win-win avenues for cooperation between academia and industry and for academia and industry to appeal jointly to government for assistance in catalyzing cooperative ventures with money and with appropriate legislation. The catalytic effect of the Bayh-Dole Act on the development of the biotechnology industry should alert us to the positive effect creative legislation can exert in this area. As I see it, academic medical centers have the patients, the clinical workforce to care for the patients, and the confidence of the public. Industry has already put into place many of the requisite technologies. The challenge before all of us is to see whether we can reach consensus on specific problems that impede cooperation between industry and academia in the area of human genetic research and to find avenues through which the enlightened self-interests of both academia and industry can be united for the benefit of the public.
For the reasons outlined above, I believe that the use of electronic medical records is a key ingredient in speeding progress in all types of medical research in this country and of genomics in particular. I believe that an academic-industrial-government partnership to create and implement a national system of electronic medical records is a feasible and desirable goal. It is one that will facilitate cooperation between academia and industry, speed discovery of linkages between genes and diseases, and at the same time contribute to the improvement of health care delivery in the United States.
The human genome belongs to every human being. The public has provided the resources to characterize and sequence it, and it has entrusted us with the responsibility to use what we have learned about it for the benefit of humankind.
Samuel C. Silverstein is the John C. Dalton Professor and chairman of the Department of Physiology and Cellular Biophysics and professor of medicine at the College of Physicians and Surgeons of Columbia University in New York.