Informatics for Integrating Biology and the Bedside (I2B2) is a multi-year program that aims to learn about the genetic underpinnings of several common conditions (including asthma, hypertension, and diabetes) by mining the medical records of several million patients in the Partner HealthCare system.
Using electronic health records for medical research is still in its infancy, but will someday provide a powerful boost to the acceleration of medical discovery. The ability of Kaiser Permanente to identify that there was a problem with Vioxx
early-on by performing adverse reaction
epidemiology using the health insurer’s electronic patient data is an illustrative example of how electronic data on a population scale might contribute to consumer health (even in near real-time).
With projects like I2B2 it may be possible for consumers to contribute to the health of future generations, or even see returns on health in their own lifetimes, with minimal engagement — compared to pariticipating in a clinical trial for example — as long as they choose to share their personal information.
”If we could use routine clinical care to generate new findings
without having to do multimillion-dollar studies, that would be a true
change in the way medical discovery is done," said Dr. Isaac Kohane, an
associate professor at Harvard Medical School who is one of the
project’s directors. ”We want to use the healthcare system as a living
laboratory…Ultimately…the public will have to decide: Do they want research done this way or not?"
We have reached a point where genotypic data is much less an issue than in the past with the advent of new sequencing technologies and the rapid decline in costs per base pair. The rapid decline in the cost of sequencing should be celebrated (and nourished further), but the reality is that sequence data alone is far from even good. To provide real insights that will improve human health, this data needs to be tied to health information, i.e. phenotypic data. The real bottleneck and financial hurdle now is to get good phenotypic data. And lots of it for lots of different types of people.
So, the solution is simple right? Set-up a website, pass out colored ribbons, print t-shirts, host an awareness campaign (marathons, fund-raisers, etc) and just ask consumers to volunteer their personal health information. It is, after all, for the benefit of human health the world-over. As a sweetner (as if human health isn’t cause enough), researchers might promise volunteers complete confidentiality and anonymity. While most people could care less if their height, age, and serum iron levels were, in some form, public knowledge, there are other types of personal health information that can carry stigma (think HIV, STD, OCD) or might compromise the privacy of family members (i.e. genetic information).
There’s the rub. It is increasingly clear that health information confidentiality and anonymity are promises that researchers will be unable to uphold, even when the information is de-identified. So, when researchers call for volunteers to submit their personal health history to projects like I2B2, they can’t make guarantees that this information will remain totally private.
Isaac Kohane and colleauge Russ Altman proposed a solution in a recent article in the New England Journal of Medicine: health information altruists. From the article:
…large-scale studies of genotype and phenotype should specifically seek out volunteers who are information altruists. The guarantees made to these subjects about the risks of re-identification can then be more realistic. The potential damages can be outlined, but the subjects presumably will elect to take the risk in the hope of helping to address human disease…
In the same paper, the authors outline three steps to make projects like I2B2 more practical:
- "rules could be implemented to make it illegal to link health information contained in research databases to other data resources, so as to prevent the inference of individual information outside the scope of the original informed consent…"
- "researchers who curate genetic databases should have some protection for their activities, provided that they follow an agreed-on set of operating guidelines…"
- "most important, patients should be granted explicit control over the disclosure process. They should be able to indicate the types of users who can see their data, and they should be able to request lists of those who have seen it…"
The next step is to set-up pilot studies and see just how many people will volunteer provided fewer promises of privacy.
My take: Tapping all the health information altruists out there is great way — if not the only way — to get started at the moment. I think there are lots of people to keep projects like I2B2 or the Personal Genome Project busy in the near term. Informed consent will be somewhat tricky. For example, we know that James Watson is definitely qualified to make an informed decision about his personal genetic information and he recently declined to learn about his genetic predisposition to Alzheimer’s disease (a la ApoE). Obviously, many consumers at-large (and many practicing physicians too) have a poor understaning of genetics. We’re all new to the game for the most part.
Exactly how many people will volunteer? Hard to tell. The more the merrier. Perhaps this will be true not only for the gains that can be had toward human health, but also because with large numbers of participants any misappropriation or abuse of the information (by insurers or employers or corporations) would cause such a backlash that would-be wrongdoers may be deterred. Of course, the-more-the-merrier argument can go the other way too.
Its exciting to see projects like this starting to take form.
Gareth Cook "Harvard project to scan millions of medical files" Boston Globe. July 3, 2005.
Isaac S. Kohane, and Russ B. Altman. "Health-Information Altruists — A Potentially Critical Resource" NEJM 353:2074-2077, Nov 10 2005. (sorry subscribers only)
PHI liquidity is a good term, in the conference description it is defined as:
"the ability of that information to move around, relatively friction-free, to where it is most useful and relevant"
A really exciting topic. Much of medical progress depends on our ability to aggregate and plumb health information. The possibilities that are available when this information is all hooked together becomes really interesting…and not just for scientists, epidemiologists, drug-makers, physicians, (oh yeah and insurers), but also for regular information consumers.
We’ve got access to general population level statistics now, much of it is irrelevant, outdated, or just impossible to use. Information is always more interesting when it is about you. This remains true at a population level too. Imagine being able to view real-time, population-level health information filtered by your personal health record, including such things as age, pharamaceutical regimen, location, genotype, medical history, family history, weight, diet, etc.
IBM has a new technology called the Genomic Messaging System. Barry Robson, a research scientist at IBM’s Watson Research Lab, is the lead author of a recent paper on GMS, published in the Journal of Proteome Research. Here is snippet from Robson’s paper ($):
…Genomic Messaging System (GMS)…may be considered as a proposed specification for an approach with an emphasis on a specific language for embedding supporting information and management functions in streams of DNA data. Naturally, the details may evolve, but the concept of adding human and computer-generated content (such as annotation) into the DNA sequences, in a more general and powerful way, appears to be a useful one. The proof-of-concept code employs Perl 5 with capabilities for XML management although, as described below, the approach is not confined to XML-based records. Components have been also been recently encoded in Java. The description also includes the Clinical Laboratory Messaging System (CLaMS) as a straightforward modification in which the background genomic default usage is switched off so that it occurs at same rank level as other clinical data. GMS represents technology involved in the domain of information management of patient genomic information and associated clinical information. In particular, the current implementation is concerned with the compression, encryption, and transmission aspects of clinical and genomic data, including bringing data together such as the clinical record and the patient DNA results from the sequencing laboratory. The form of information transmitted is versatile and is capable of storing and transmitting an entire Integrated Medical Record (IMR)."
The GMS paper:
Barry Robson and Richard Mushlin. "Genomic Messaging System and DNA Mark-Up Language for Information-Based Personalized Medicine with Clinical and Proteome Research Applications" J. Proteome Res. ASAP Web Release Date: 22-Jul-2004
By the way, IBM Healthcare and Life Sciences just release another newsletter. Although I can’t find any version of it online, you can get many if not all of the same papers here.
Is there any interest in by medical consumers to read their own medical record? EHR vendors take note, a recent survey of 4500 adults suggests that they do. Of those that responded, 36% were "very interested" and 43% were "somewhat interested."
"Duke University’s Fuqua School of Business on Tuesday announced it is launching an initiative to gain widespread consumer support for a national electronic health information network.
The goal of the initiative, named the Health Data Exchange, is to improve the quality of health care while reducing costs, said Kevin Schulman, M.D., director of the Health Sector Management program at Fuqua.
The initiative represents a new approach to the development of an electronic health information network.
"Instead of focusing initially on the technology — a ‘build it and they will come’ approach — we have outlined a consumer-first strategy," Schulman said. "Through public outreach and education, Health Data Exchange plans to build consumer understanding and acceptance of a national health network as well as strong motivation to participate in the exchange via voluntary submission of personal health information."
A national electronic health information network would make patient medical information available online to medical providers, insurance companies, pharmaceutical companies and other segments of the health care industry. Patients would provide their medical histories to the network through an Internet site. Updates to each patient’s medical records would occur automatically each time the patient seeks medical care or fills a prescription.
Although several groups would have access to anonymous medical histories, only those with specific authorization from individual patients would be able to access a patient’s medical record with his or her name attached."
Read the whole press release here.
Oxford University announced in a press release the Integrative Biology Project. Excerpt:
"[The project] aims to build a giant computing grid to support the modelling of complex biological systems, which will use advanced computer simulations to help understand and treat disease. The international, multi-institutional and cross-disciplinary project team includes world experts on the modelling of heart disease and cancer. The intention is to use computing power to bring together knowledge at all the different levels of biological understanding – from gene function to physiology – in order to provide a coherent theory of biology which can be applied to disease."
The Integrative Biology Project homepage is here.
Further description of the project can be found in this Vnunet article.
Bill Briggs, Working Together: I.T. and Evidence-based Medicine. Health Data Management, Jan 15 2004.
"Until very recently, the most reliable source clinicians had for current best practices was the medical textbook…
"Today…clinicians striving to practice evidence-based medicine gain instant access to a mountain of information on best practices through information technology. And provider organizations are using I.T. to mine their own data and access external clinical evidence-based guidelines from a range of data sources. Their goal is to get current information into the hands of clinicians where they need it most: at the point of care, where patient treatment decisions are made. The payoffs, many experts say, include reduced medical errors, greater operating efficiency and improved patient care. Sooner or later, higher reimbursement levels might be tied to evidence-based medicine as well.
"…Data on best practices and other research is being mined internally and accessed externally via the Internet from many sites. The challenge for provider organizations is how to put global data in context and apply it to the individual patient."
"…Evidence-based medicine requires merging data from external sources along with that of patient-specific data residing locally in multiple information systems, including electronic medical records and computerized physician order entry. All are necessary to feed information on a given situation to hardwired or mobile computers to support clinicians’ decisions at—or near—the point of care."
The article looks at a couple early adopters of evidence-based medicine, including Vanderbilt University Medical Center. William Stead, M.D., the associate vice chancellor for health affairs and director of the Informatics Center at Vanderbilt is quoted toward the end of the article: “I don’t think the average physician here or elsewhere really knows when they are following the evidence and when they are not…Nor are they aware of the frequency at which they are not following the evidence. I don’t think Vanderbilt or anyone else is practicing evidence-based medicine to the degree we need to.”
UPDATE: A little back-and-forth on IT and Evidenced-Based Medicine (EBM) over at Matthew Holt.
Nancy Weil, New IBM Unit Integrates Drug Research, Healthcare. HealthIT World. Jan 13 2004.
"IBM has started a new business unit focused on medicine within its life science division, promising to help customers integrate healthcare data, including research, genetic tests, patient medical records, clinical trials records, and medical images…IBM has coined the phrase ‘information-based medicine’ to describe its approach to integrating, mining, analyzing, and managing medical data from a variety of sources."
"…’What we see happening is the convergence between life sciences, healthcare, and IT,’ [VP Mike] Svinte said. ‘The thought is that convergence really has the ability to change the way healthcare is delivered, to change the diagnosis and treatment.’"
"…The field of information-based medicine is so new that a successful business model has not yet been clearly defined…The key question is who will pay for such data-integration systems."
Jeff Berman, Micohealth Remotely Connects Patients and Providers, Health-IT World.
Micohealth, a Web-based application from Medmanager Interactive, is helping patients with diabetes and heart problems manage their conditions and keep clinicians better informed.
Micohealth provides a way for patients to collect vital information, monitor symptoms, and communicate with clinicians. By tracking daily activity, including exercise, and food and medication intake, patients and clinicians are better equipped to assess how effectively the disease is being managed.
UPDATE: The California Healthcare Foundation website recently added the following guide by The First Consulting Group: "Online Patient-Provider Communication Tools: An Overview" (pdf warning!), November 2003. (Thanks to The Informatics-Review)
Mark Hagland, Arrested Development: Early medical management improves the outcome of diseases, even chronic ones. Healthcare Informatics, November 2003.
"[Many] are aware of the challenges ahead for managed healthcare as the incidence of chronic illness continues to rise, taking utilization and costs with it. They also know that they need to use IT-powered medical management strategies, including disease management, to improve the health of plan members and cut costs by intervening earlier, better and smarter."