Malek: Health information technology is a tower of Babel, by intent

Editor’s note: Marvin Malek, MD, is a physician who practices internal medicine at Central Vermont Hospital. This opinion piece was first published in We reprint it here with permission as Dr. Malek's observations are right on target and expose the fundamental flaw with the government's health IT policy. Although specific to Vermont, Dr. Malek's policy recommendations apply to all States and countries seeking to implement an effective unified health system. RAM

On Feb. 12, 2002, Fred Hernden, a retired Vietnam veteran, developed weakness and abdominal pain. He had recently moved to New Mexico from Woodstock, Vt. It was not by accident that the small home he bought was located near the Albuquerque Veterans Affairs Hospital. Within seconds of sitting down with Fred in the examining room at the VA, Dr. Steve Searles, Fred’s new doctor, pulled up Fred’s entire record from the White River VA on his computer, including detailed information about the treatment Fred had received for colon cancer, and the exact dimension of the aortic aneurysm that had been identified six months earlier at the White River VA. Based on his exam that day and review of these previous records, Dr. Searles could develop a diagnostic plan on the day of that first visit.

Even a full decade later, Fred’s experience can only happen at a Veterans Affairs hospital. If Fred were being treated in any private sector hospital, receipt of medical information from other practitioners will require days, sometimes weeks, nearly always be incomplete, tests will be needlessly repeated, treatment may be delayed.

During the 1990s, the VA system was converted from a medical backwater to the forefront of high quality medical care. Ten years earlier, doctors and programmers within the VA system began the pioneering work of developing a software program to be used as a medical record. Ultimately named “Vista,” this well-liked software is now used by every clinical worker in the VA system. This linkage of every VA facility in the country with the same, user-friendly software program was the centerpiece of the effort to improve quality of care throughout the VA system.

This effort paid off. In the largest such study yet performed, VA hospitals outranked a large group of private sector hospitals in every area measured, including preventive care, chronic care, patient satisfaction, and treatment of heart disease, while spending 35 percent less than the private sector hospitals.

In 2001, the Institute of Medicine released an influential series of reports documenting the large number of serious medical errors occurring on a daily basis in hospitals and medical practices across the U.S., and they recommended adoption of electronic health records as a key solution. In early 2004, President George Bush set out a landmark federal policy whose goal was to have every Americans’ health record be in electronic form by 2014.

Bush era policy — and the consequences

But the Institute of Medicine — as well as subsequent federal policy — did not specifically recommend the VA system’s route toward computerization: To use a single, popular software language to connect every hospital and doctor’s office. Instead, under Bush era policy — which has been continued through the Obama years, new agencies and programs were created to promote wider electronic health records adoption. Also, through the federal Medicare program, the federal government adopted strong financial incentives to providers to use any “certified” software program — initially bonus payments, then later financial penalties for failing to do so. The government also provided funding across the country for state-based organizations to encourage hospitals and medical practices in each state to make use of electronic health records, and to provide technical assistance. In Vermont, that organization is called Vermont Information Technology Leaders (VITL).

In response to this policy, over 100 private sector software companies rushed to take advantage of this huge market niche. The rush occurred because it would be key to these companies’ financial success to quickly get an attractive product on the market so that hospitals would opt for their product first. This was important both because hospitals’ decisions would likely influence the decisions of doctors in the hospitals’ service areas, but especially because the costs of transitioning from one software system to another would likely be as great or greater than the cost of initially adopting an electronic health record.

Unsurprisingly, given the rush to quickly gain market share, these software products have been a huge disappointment to providers: They are falling far short of their potential to improve quality of care, and it will require years of use — and substantial upgrades of the software — to even think about the possibility that electronic health records adoption will ever actually reduce health care costs.

If the late Steve Jobs were forced to use any of these products, he would be appalled at the primitive functionality doctors and nurses are contending with.

Babel, by intent

And in addition to being “user-unfriendly,” another disadvantage of the government’s strategy of leaving medical software to the private sector was that none of these programs “speak” with any of the other competing software products.

The 11th chapter of Genesis tells the story of humanity disobeying a commandment of God by building a very tall tower as a symbol of their earthly power. To punish mankind for this disobedience, God “confused the languages of humanity so that they could no longer communicate with each other.” As a consequence, the tower at Babel could not be completed.

In health care, we have created our own “Babel” of incompatible medical software products. And this time, we can’t blame God.

Federal policy not only failed to encourage all providers to adopt a single software system, but the Office of the National Coordinator of Health IT actually requires VITL and all the state-based agencies to highlight a minimum of two software programs as a condition of receiving federal grant funding.

But it’s not exclusively the government’s fault: There is plenty of blame to go around: None of the major hospital or professional organizations pushed for having all hospitals use the same software — in spite of the enormous success the VA and other countries have achieved doing just that.

Recognizing that this laissez-faire approach has primarily benefited the software developers rather than patients, the government has been funding VITL and the other relevant state grantees to work on “inter-operability” — meaning taking all these incompatible languages, and trying to extract key data elements and making them available to those using other programs. Needless to say, even after years of this effort, this interoperability retrofit remains a distant second place to the seamless communication VA providers have with each other through the VISTA system.

So we remain in the non-communicative chaos of Babel. And it is safe to say that our goal — the “towering” need for improved quality and lower cost — will not be achieved until we create a unified information system.

What should be done now: federal policy

How could we have done better? By the year 2000 or so, when the VA’s accomplishment was already crystal clear, the federal Department of Health and Human Services should have acted promptly to have the VISTA system adopted at hospitals throughout the country. To their credit, they did open up VISTA as public access, downloadable software available for free to all potential users. But this was not enough: For adoption outside the VA system, VISTA would have required technical support staff widely available to train new users, to troubleshoot problems and glitches that crop up, especially for hospitals that are just starting to adopt the system.

The VA is in the midst of a major upgrade of their electronic health records. A new Web-based software system (currently referred to as “”Aviva”) is being designed with superior data-sharing capacity with the major private sector commercial electronic health records systems.

But a single upgrade is not enough. With all the computer-related talent in the U.S., it’s not hard to imagine teams of programmers working with physicians in all specialties and other health workers as well to continually upgrade and improve the software.

Federal policy should also begin to incorporate a comprehensive cost control strategy into their HIT policy. Neglecting the cost issue has led electronic health records adoption in the U.S. to have taken the most expensive possible route. So long as the electronic health records policy remains balkanized in the private sector, electronic health records costs will remain extremely high.

What should Vermont do?

The state of Vermont has thus far conformed to the U.S. model of HIT policy, promoting the least effective and most costly avenue toward the adoption of electronic health records. Even within our small state, dozens of incompatible software systems are being used by providers of care, and are generally disgruntled with the quality of the software they are using.

Policy recommendation #1: Conduct a survey of all institutions and a random sampling of Vermont health-care workers to determine the number of electronic health records software programs currently in use in Vermont, and also user satisfaction with these software applications (comparing user friendliness to software used for other functions, such as spreadsheet, word processing, etc).

The intent of the recently enacted Act 48 health reform legislation is to create a “universal, unified” health care system that provides affordable care to all residents of our state. Those who crafted the legislation envisioned a single reimbursement system, and this is clearly an integral element of reform. But failing to unify the medical information systems providers are using is ludicrous. If the various providers who are involved in the care of a patient are all using incompatible software languages, the essential goal of providing seamless, coordinated care will not be realized.

Policy recommendation #2: Vermont’s health policy leaders — both in government (e.g. Green Mountain Care Board) and in the private sector — should promote the adoption of a single, user-friendly software system. Appropriate incentives should be put in place for its adoption. A task force should be convened which would include all major providers in the state — including representatives from the Veterans Administration and both university centers — whose goal would be to realize this vision. If appropriate technical support can be obtained to support the transition, having every provider in the state of Vermont adopt the VA’s software system would be preferable, since it is well-liked by users and available free of charge in the public domain. Since this effort will likely take at least two to three years, it seems reasonable to try to transition to the VA’s upgraded “Aviva” system. The state-of-the-art data sharing capacity being built into the new system will facilitate this transition.

Vermonters seeking care in other states in the U.S. will not benefit from this proposal. Regrettably, it is not realistic to expect that the rest of the U.S. will attempt to create this obvious underpinning for the transformation to an efficient, unified system of care. Fortunately, even if this transition is limited to Vermont providers, a big step forward will still be accomplished for Vermont.

Until now, policy makers in Vermont have not regarded the cost of electronic health records adoption as one of the areas we could look to in the effort to achieve control over our health care costs. Remarkably, when the state of Vermont released an annual report on health cost inflation last year, they even went so far as to completely exclude HIT costs from their analysis of hospital cost inflation (see here). There is no justification for walling off HIT costs into an untouchable, sacrosanct category.

Policy recommendation #3: HIT costs should be on the table. The huge expenditures made to private sector software companies each year, the needlessly high training costs and lost productivity associated with the use of user-unfriendly software, and the cost of scrambling to receive linguistically incompatible records from other providers are all highly relevant to Vermont’s health cost control effort.

The task force I am proposing should take into consideration both short- and long-term HIT costs. The hospitals in Vermont are paying extremely high charges to software vendors annually, and will be doing so for years to come. While transitioning to a different software system will be costly for these hospitals, this initial cost would be paid off over time if the cost to hospitals of the new system were significantly lower.

Numerous national foundations are interested in improving the poor performance of the U.S. health care system both in the domains of cost and quality. The health information system is the linchpin of the effort to improve quality and control cost. It therefore seems likely that Vermont could receive grant support from one or more national foundations for this effort.

As is the case with most of the other key aspects of Vermont’s reform process, resistance to change and to subsuming parochial interests to the greater interest of the state as a whole will form a major obstacle to its implementation. But if policymakers and leaders in Vermont’s health care community set out the goal of creating a unified, cost-effective HIT system for the state of Vermont, this goal can be achieved. 


Very insightful comment

As one of the original software architects of the VA VistA system, I agree with your comments. The very notion of representing the full complexity of health informatics in a single, centrally controlled top down "systems" approach is foundational to this. Is it even possible to control a system as complex as this through centralized fiat?

Imagine the Dept. of Education deciding to centrally manage knowledge by initiating a $20 billion program to make knowledge on line. The would invite all of the stakeholders to form standards groups, panels, and review boards. Encyclopedia Britannica, World Book, and Collier's would all be represented, and they would work out a standardized interface; expert committees would be formed to "harmonize" the articles according to the federally-approved hierarchy of terms, and a consumer empowerment subcommittee would be established to insure that consumers would not be overcharged for "portability" between sites.

The notion that we could have an open, peer produced wikipedia-based approach, of course, would have seem heretical to these folks.

here's a video discussion I had with Ward Cunningham, inventor of the wiki, at Health Camp Oregon 2011 VistA shares many similar characteristics with WikiPedia, which may help explain why it is not is not more accepted by the committees, agencies, stakeholders, and "consumer empowerment" committees...