If the 1960s film, The Graduate, were to be made today (Graduate, II: The Stimulus), the iconic scene at the party where a friend of the family takes Dustin Hoffman aside and whispers in his ear, “I have only one word for you, Plastics“) would be transmogrified into, “I have only three words for you, Health Information Technology.” HIT, and its love child with the recently passed stimulus package, the Electronic Medical Record, have all the characteristics of a good idea destined to go bad. One of my vivid memories from my early career was being shown a new kind of quiet printer, called an ink-jet, in the Mass General laboratories of Dr. Octo Barnett, one of the early pioneers of electronic medical information systems. Octo was an evangelist and visionary for the new digital technology at a time when most doctors had never seen a computer, much less used one. He was himself a physician working in a hospital and he had a keen appreciation for the problems in adapting the technology to medical practice. He had no option for making medical practice adapt to the technology, so his knowledge and experience was especially pertinent to making the technology work for medicine and not the other way around. Unfortunately Octo’s experience seems to have been lost somewhere along the way. This is a theme ably and persistently presented by the bloggers over at Health Care Renewal, in particular Dr. S. Silverstein who blogs under the name MedinformaticsMD:
As many have now observed, electronic medical records products on the market are clunky, fail to support the cognitive and workflow needs of clinicians, contribute to clinician cognitive overload and increased clerical work burden, and may be harming patients at a rate far higher than, say, VIOXX, which was summarily withdrawn from the market to be followed by billions of dollars of lawsuits.
Health IT can indeed improve healthcare quality, reduce costs and achieve other benefits claimed of it, but only if it’s done well. There is tremendous underlying social and technical complexity underlying those two words “done well”, and unfortunately in 2009 most health IT is simply not done well. Yet the technology is actually worshipped in what I see as a Bernard Madoff syndrome, an irrational exuberance in IT for reasons not entirely clear to me considering the existing objective evidence. (MedinformaticsMD, Health Care Renewal)
The example of Vioxx seems especially pertinent given recent work purporting to show that the adverse cardiovascular effects of that arthritis drug might have been picked up if there had been a medical information system available to mine. As an Editorial in Nature notes, Obama’s enthusiasm if not infatuation with HIT as a saviour for runaway medical costs is just the next step in a program initiated by his predecessor, George W. Bush in 2004. Bush established the office of a national coordinator for a program to have access to an Electronic Medical Record (EMR) for every American by 2014. The stimulus bill spends 53 or its 407 pages on this and pumps $19 billion into pushing Bush’s plan forward. Ironies abound. Most of the discussion about EMRs has been about their use in improving the quality of clinical practice, although probably it is cost control that will be the lever that makes it happen. Nature calls attention to another feature of HIT that is especially seductive to those of us in research, especially epidemiologists like me: the use of EMRs for population studies, including but not limited to post-market surveillance of drugs and medical devices (the Vioxx case).
MedinformaticsMD and others (see links at the end of his piece here) have called attention to a number of possible undesirable effects of a badly done HIT effort (and in the view of many of us the speed with which it is being urged on us is likely to lead to a bad outcome), but there are also many deatils to be ironed out on the research end, too:
Imagine, for example, that Mrs Smith checks into the hospital for hip-replacement surgery. Her doctors will be able to call up the electronic records of her personal physician, her physical therapist, her pharmacist and everyone else who provides her with health care. But this scenario poses some knotty questions. Even leaving aside the technical challenge of securing all those far-flung databases against attacks by hackers, who has the right to access any given database? And how much are they allowed to see there ? everything, or just those portions of the data they need for their jobs, keeping the rest private? And who decides? Patients? Doctors? Hospitals? Anonymization and pseudonymization, in which identities are masked but all data can still be ascribed to respective individuals, protect individuals while enabling research, but how should such measures best be implemented?
These questions have been contentious even in the United Kingdom, which has deployed an electronic records system through its unified National Health Service. They will be even more contentious in the United States, where the health system is anything but unified, and where privacy concerns run deep.
The scientific questions are just as knotty. For example, will each patient have to give permission each time a researcher wants to access his or her records? That sounds reasonable ? and indeed, this is the spirit of the strict data privacy and security provisions of the federal Health Insurance Portability and Accountability Act. But that would make using the records for, say, routine epidemiology prohibitively cumbersome. There needs to be a clear policy allowing for such uses, as long as the researchers have no access to any aspect of the data that reveal a person’s identity. But at the same time, the health-records system should provide individuals with an audit trail that shows them who has been looking at their records, and what they have seen there.
Or consider another scenario: a research group that wants to do a study aggregating EHR data over a large area. Will they have to get approval from every Institutional Review Board (IRB) ? which oversee all research on human subjects in the United States ? in every hospital and university involved? Common sense suggests that one IRB process is enough ? and then only if the research uses personally identifiable data. But mechanisms to coordinate the oversight will have to be created. (Editorial, Nature)
Don’t get me wrong. As a researcher I look enviously at my colleagues in countries where a national health system provides them with research resources (for an epidemiologist this means data) almost unheard of in the US. But I see a rocky road ahead and I am not especially optimistic. I fear HIT will neither live up to its promise of saving money nor of providing a gold mine of research data. It will fulfill its potential for changing medical practice in unforeseen ways (some of them perhaps even good). This is not a function of government intervention in health care. It is being forced upon us by a way of financing and practicing medicine that has broken the bank and needs to be reined in. If left up to the private sector, as it has been until now, HIT will almost certainly turn out even worse, with incompatible systems allowing no way to exchange data and giving us data mainly geared to what private insurance companies want (and don’t want), not what the patient needs.
The US will likely remain backward among modern industrialized nations when it comes to its medical care system, and that will include its medical information systems. As I said, ironies abound.