Friday, November 26, 2010

The Sound Barrier in Health IT

The October issue of JAMIA sports an article titled “Health information technology: fallacies and sober realities” (thank you “pcp” for bringing it to my attention). The authors, who are obviously concerned about the shortcomings of HIT, are attributing much of the current unsavory situation to widely spread misconceptions regarding HIT design, deployment and utilization. Two to three years ago this article would have been correct in assuming that the salient points it raises are indeed fallacies. Today, you would be hard pressed to find many in the HIT industry, physicians in particular, who would disagree with most of the contents excepting perhaps the final conclusion, but more on that later. First, let’s quickly summarize the points under consideration:
  • Safety must be a consideration when deploying large scale HIT systems and just like other medical devices are tested and supervised to ensure patient safety, EHRs need some form of regulatory oversight.
  • HIT is lacking in usability and cognitive support for clinicians. Both clinicians and usability experts need to be involved in building HIT and extensive research is still needed.
  • Users are not always at fault for mishaps and adverse events.
  • Medicine is a complex field and HIT should reflect that.
  • Paper cannot and should not be completely eliminated.
  • Success of HIT should be measured based on health outcomes, not adoption rates.
  • If you build it, they will not necessarily come unless the benefits accrue to them directly.
  • HIT should be designed for collaboration.
Granted this is a very terse summary, but I believe it is sufficient, considering that much of the above has been hashed, rehashed and widely debated since the advent of Meaningful Use in all its glory. Since the article was authored by health informatics folks, the conclusion is that “Foundational cognitive and human factors engineering research and development are essential to better inform HIT development, deployment, and use”. To emphasize the point, the article draws an analogy to aviation and quotes a 1947 article addressing the complexities found in an airplane cockpit and the need to apply “psychological data and research techniques” and “the necessity for designing the machine to suit the inherent characteristics of the human operators”. And the article goes on to remind us that “In the more than 60 years since that publication, commercial aviation has become very safe. While it may not take 60 years for HIT to become as safe, if we do not change from our current course, it never will be”. Here is where I part ways with the JAMIA authors.

Guess what else happened in 1947? Chuck Yeager piloting his Bell XS-1 recorded a 361 m/s speed and officially broke the Sound Barrier. Less than two years later, the first commercial jet airliner, the British Comet, took to the skies and eleven years, almost to the day, after Captain Yeager’s breakthrough, the Boeing 707, the most enduring jetliner in history, made its first transatlantic service flight. The aviation industry was immersed in a whirlwind of innovation. Computers began their ascent to the clouds during WWII as rudimentary calculators. By the late 1970s, the first “electric jet”, the F-16, came with a completely computerized “fly-by-wire” control system. About the same time, “glass-cockpits” consisting of digital displays began to replace the conventional analog instruments in commercial cockpits. Although the driving reason for glass cockpits was their relative light weight, there were great hopes that computerized displays will increase flight safety and reduce pilot errors.

What does all of this have to do with Health IT? It seems currently fashionable to compare health care to aviation. First the checklists of Dr. Gawande and now improved safety as afforded by the introduction of computers. If we are going to make such comparisons and derive lessons learned, perhaps it would behoove us to understand both aviation and health care equally well*. There is no question that flight safety has improved by leaps and bounds in the last 60 years, but this is mostly due to new technologies, new materials and better design tools for building and testing aircraft. In a similar fashion health care has improved in the last 60 years by technology advances, some of which involved computerization, such as CT scans, laparoscopic devices, pharmaceuticals  and a myriad of other high-tech “tools and die” for physicians. Many diagnoses that spelled certain death 60 years ago are fully curable, or at least manageable, today.

The best aviation comparison for Health IT is the introduction of digital displays in the cockpit. Just like EHRs, glass-cockpits are means to distill vast amounts of information and present them to the user in a way that maximizes efficiency and facilitates better decision support. And just like EHRs, digital instruments from different manufacturers vary drastically in appearance and behavior. Basically, glass-cockpits are to analog flight instruments what EHRs are to paper charts.  Unlike the EHR industry, the aviation folks, and NASA in particular, have engaged in methodic and rigorous human-computer interaction research and applied the results to digital flight controls, although according to the Federal Aviation Administration (FAA), much more is needed:
  • “Not all airplane and avionics designers have considered the pilot-machine interface by using good human factors practices. General aviation needs airplanes that are intuitive to operate, requiring as little training as possible”.
On March 10, 2010, the National Transportation Safety Board (NTSB) published the results of its safety studies regarding glass-cockpits in small aircraft and here are the findings.
  • The statistical analysis found that for 2002–2008, light single-engine aircraft equipped with glass cockpit displays experienced lower total accident rates—but higher fatal accident rates—than the same type of aircraft equipped with conventional analog instrumentation.
  • Accident pilots flying glass cockpit equipped aircraft were found to have higher levels of pilot certification and more total flight experience than those flying conventional aircraft.
  • In addition, the FAA has no specific training requirements for pilots operating glass cockpit-equipped light aircraft. The lack of equipment-specific training requirements from the FAA results in a wide range of initial and recurrent training experiences among pilots of glass cockpit aircraft. With the exception of training provided by airframe manufacturers with the purchase of a new aircraft, pilots must currently seek out and obtain equipment-specific glass cockpit training on their own.
  • The results of this study suggest that the introduction of glass cockpits has not resulted in a measurable improvement in safety when compared to similar aircraft with conventional instruments. The analyses conducted during the study identified safety issues in two areas:
    • The need for pilots to have sufficient equipment-specific knowledge and proficiency to safely operate aircraft equipped with glass cockpit avionics.
    • The need to capture maintenance and operational information in order to assess the reliability of glass cockpit avionics in light aircraft.
Sounds familiar? Turns out health care and aviation are in the same boat when it comes to computerized display/records. This is not to say that human-computer interaction and other cognitive studies are not worthwhile, but the results will only be incremental until the final Sound Barrier is broken. Humans and computers must become able to verbally exchange information. Instead of displaying data on a multitude of colored screens and issue flashing warnings and alarms, the computer should just distill all the digital data and say “You have 10 seconds to pull up or you will crash”, and alternatively, “You are prescribing a lethal dose for this patient. Do you wish to continue?” If you find this slightly delusional, just ask Captain James T. Kirk of the Starship Enterprise.

*Disclosure: I am an Aeronautical Engineer by education and Health IT professional by personal choice.


  1. I enjoyed reading your post and comparisons of HIT with aviation changes. Your comparisons reminded me of a post I did last year which included my written testimony before NCVHS on the future of PHRs which included a discussion of the future of PHRs/HIEs in relation to the history of transportation. I thought you might want to scan the post. Here is a link:

    Also, I was intrigued to see your mention of Chuck Yeager, a fellow West Virginian.

    Bob Coffield
    Health Care Law Blog

  2. Thanks, Bob.

    Great testimony to NCVHS!! I think not just PHRs are at the Model T stage right now, but EMRs/EHRs in general still come in any color you want as long as it's black.
    Seems to me that Health IT must be indeed undergoing a serious transformation, since it is so often compared to motion driven industries.

    I had no idea that Chuck Yeager was from West Virginia, but as yet another weird coincidence, I've been doing lots of reading about Virginians lately (before there was a West Virginia)... :-)

  3. As recently noted by Kevin Pho, MD, “Too much data can overwhelm physicians and harm patients.” (

    Perhaps this JAMIA article and Margalit’s insightful analysis as an expert in both the healthcare IT and aeronautical engineering domains will encourage designers of EHRs, PHRs and HIEs to apply good human factors practices to redesigning the physician-machine interface that these platforms currently use to report the results of the approximately 6,500 different patient diagnostic tests.

    The problem is that infinitely variable formats have been used to display cumulative test results as fragmented data, which is incomplete and hard to read, since the inception of clinical computing in the 1960s. For time-constrained 21st century clinicians, the daily daunting workflow challenge is scrolling through an ocean of unorganized test results data both quickly and accurately.

    The practical and durable solution is using a standard reporting format that provides a meaningful view of clinically integrated, complete information that can be efficiently viewed and shared on up to 80 percent fewer screens.

    Providing physicians and patients with a standardized format for viewing and sharing lifetime test results is analogous to EHRs, PHRs and HIEs using the CCD document standard for efficiently viewing and sharing patient care summaries. This is one vital, but overlooked clinical data area where both patient safety and physician workflow can be improved by “designing the machine to suit the inherent characteristics of the human operators.”

  4. Dr. Coli,
    I do recall the KevinMD post and the distiction someone there made between data and information, particularly useful information.
    I completely agree with that distinction and with a need for intelligent software to collect the data and distill the information and most importantly to present it in a useful and unified way to the user at the exact point where it is needed, and it may require different presentations for different situations.

    The human factor design comes in at the presentation layer of course, but I believe that when we discuss cognitive support, there are "human factors" that need to be considered when developing the machine intelligence which performs the analytics and transforms data into information.
    Basically the computer should "know" WHAT information the user needs and it should also "know" HOW the user needs to see it in order to process it most efficiently.

    I did look at your website and I really liked the unified look of the results. Unfortunately, there are currently as many different displays of test results, as there are EHRs, much like the avionics displays in various aircraft. Not that it makes things any better, but we are not alone on this quest....