The Business of Health Care

Why Can’t the ICU Be More Like a Cockpit?

In the world of patient safety, we’re constantly reinforcing the importance of teamwork and communication, both among clinicians and with patients. That’s because we know that patient harm so often occurs when vital information about a patient’s care is omitted, miscommunicated or ignored.

Yet for all we do to improve how humans work together, clinicians compete against an environment in which there is very little teamwork or communication among the technologies that they need to care for patients. And there’s little that clinicians or hospitals alone can do about it.

Take, for example, the plethora of alarms from cardiac monitors and other devices that compete for clinicians’ attention. Vendors act as if we are in an alarm race, with each making their devices’ beeps more annoying but no clear prioritizing of the most important alarms. A study on one 15-bed Hopkins Hospital unit a few years ago found that a critical alarm sounded every 92 seconds. As a result, nurses waste their precious time chasing an ever-growing number of false alarms—or becoming desensitized to false alarms and ignoring them. Across the country, this has had tragic consequences, as patients have died while their alarms went unheeded. (Read a 2011 Boston Globe series about this issue.)

In most other high-risk industries, such as aviation and nuclear power, technologies are integrated. They talk to each other, and they automatically adjust based on feedback. Indeed, because of systems integration, pilots fly a small amount of a flight, and even in some treacherous situations, they hand over the reins to the autopilot. Although Southwest Airlines or the U.S. Air Force can buy a working plane, you cannot buy a working hospital or ICU. You must put it together yourself.

There are many other examples of how health care is grossly under-engineered. Consider these:

  • The main therapy to reduce mortality in patients with acute lung injury is to program the breathing machine to deliver “small breaths” and low pressure. However, patients receive that treatment only about 20 percent of the time. The electronic health record does not “tell” the breathing machine that you have acute lung injury or alert doctors to provide this potentially life-saving therapy.
  • A primary intervention to reduce mortality in patients with sepsis is to give the right antibiotics fast, generally within an hour of when sepsis begins. Because sepsis is diagnosed with several tests taken at different times rather than a single test, the diagnosis is often delayed by many hours. The diagnosis could be automated—and the need for antibiotics displayed—if the monitors and electronic health record communicated with one another.
  • To prevent pneumonia in patients on ventilators, one of the main interventions is to elevate the head of the bed to 30 degrees or more. Yet we often measure the angle by looking at it and guess wrong more often than we guess right. There is typically not a highly visual display of the angle, let alone remote monitoring of it.

Health care organizations are making major investments in electronic health records (EHR) in the belief that it will solve their problems. But today’s EHRs have crude decision support, rudimentary analytics and, most concerning, limited examples of sharing data with other technologies such as monitors, infusion pumps, and ventilators.

No one group has the ability to fix this. What we need is an unprecedented level of teamwork and communication among the industry, clinicians, researchers and other groups to make the many devices and information technologies work together.

We’ve taken what we hope will be the first step toward realizing that vision. On Feb. 2 in Baltimore, the Johns Hopkins Medicine Armstrong Institute for Patient Safety and Quality hosted a conference attended by several select companies, a foundation, and researchers from across Johns Hopkins University, including the Applied Physics Laboratory, who believed that we are stronger together than alone and that we can make ICU care safer.

Attendees included representatives from systems integrators (Lockheed Martin) as well as manufacturers of noninvasive monitoring technologies (Masimo Corporation), devices (CareFusion), pharmaceuticals (Sanofi-Aventis), space and furniture solutions (Nurture by Steelcase), and innovative infection control products (Cantel Medical). The Gordon and Betty Moore Foundation participated as well. Design firm IDEO facilitated the event.

The idea was risky. Before the conference, I wasn’t sure that the different groups would want to work together. But in the end, the day was magical.

We agreed to cooperate, to be interdependent rather than independent, to be cooperative rather than competitive. The company leaders were genuinely moved and passionate about improving patient safety, quality and value. They recognized the importance of integrating technologies and sharing data. They know they could solve this problem right now if we commit to it. We identified several projects in which we might start collaborating right away, and we hope to select projects and move forward with them.

The sharing and integration of technologies should be the norm rather than the exception. In the future, we should not contract with vendors of EHR or other technologies if they do not agree to integrate their systems. All of these companies have improving patient outcomes and safety in their mission statements, yet they cannot live it if they work in silos.

Our meeting was energizing. We have seen the future. It is not solely in electronic health records. It is in linking EHR with all of our monitors, infusion pumps, beds, and other technologies. It is about engaging patients and their families as active members of the care team. It is to have a system that continuously learns and improves. This is what personalized medicine is all about. It is ensuring that the care we give you takes into account your specific values, genes, medical history and risks. We can do this right now if we agree to work together. We have the technology. The question is, Do we have the will?

Director of the Johns Hopkins Medicine Armstrong Institute for Patient Safety and Quality, Peter Pronovost, MD, PhD is a practicing anesthesiologist and critical care physician who is dedicated to making hospitals and health care safer for patients. Pronovost has chronicled his work in his book, Safe Patients, Smart Hospitals: How One Doctor’s Checklist Can Help Us Change Health Care from the Inside Out. His posts will appear occasionally on THCB and on his own blog, Points from Pronovost.

17 replies »

  1. “They are just mad Obama won” so they don’t agree what he is doing now. Well, that pretty much goes owhtiut saying. You act like it was a popularity contest. They didn’t agree with his views and policies or they would have voted for him. So yeah, obviously, they aren’t going to be in agreement with a large part of the changes he will propose?!And the fact that Oklahoma won’t allow write in’s or 3rd party votes is just fundamentally wrong. We have the freedom to vote for who we want to be president……as long as its one of these two dolts we allow you to choose from? Just asinine.Just a thought, would your father have wanted the government to tell him how much he could/couldn’t sell his medicine for? Or that he had to sell for less to certain people..etc. Those things which would have affected how you grew up (income). What these people are saying is no one should. It is just the choice of a government or a big business run world? How about being responsible for ourselves.

  2. Thought-provoking. I just spent a few weeks as part of my surgery med student rotation in the trauma burn ICU and was surprised to find just how prevalent the sound of alarms were. I recall the sequence of emotions that would go through me every time one went off- first, being shocked and thinking someone was about to die, then being confused as no one seemed to run toward the beeping room or even notice that an alarm was going off there, and finally becoming extensively annoyed that the alarm sound itself was so obnoxious and not worth going off so much if no one was even going to pay attention.

    The idea of an integrated alarm system sounds to be a fine display of ingenuity combined with the obvious- advancement in technology as should be appropriate for the year 2012… When you mention that other fields such as aviation already have such systems, in comparison the world of healthcare technology does seem to be living in the past a bit.

    Definitely food for thought.

  3. Expend a little effort, OK? Potshot one-sentence drive-bys add nothing of value.

    It’s available on Amazon, and I continue to discuss it depth on my REC blog.

    OK, Medicine in Denial, page 37:

    “Policymakers recognize that transformation requires more than technology. Accordingly, certification and “meaningful use” of electronic health records (EHRs) are required to receive subsidies to purchase EHRs under the 2009 economic stimulus legislation. But the requirements for certification and meaningful use as currently conceived are primitive. They fail to incorporate or even consider most elements of the problem-oriented medical record (POMR) standard (the subject of part VI), which became prominent four decades ago. Since that time, the quality of medical records has declined. Use of the POMR standard has receded, and the clinical purpose of the medical record has been compromised.”


    “Patients must be educated in the use of tools such as the problem-oriented record and computerized POMR so that there is some concrete instrument for expressing and capitalizing upon their own motivation. If the patients are not motivated enough to use the tools effectively, then we should get over the illusion that those same patients are accomplishing much with twenty minute visits to providers or that they are complying very precisely with directions from those providers, except in those instances where a normally healthy individual gets specialized care for a self-limited problem from the appropriate specialist, e.g. a broken leg.

    The Power Of The Right Tools:
    Tools extend our muscles, our senses, our memories, and our analytical capacities. Extending our muscles and our senses with automobiles, power tools, telescopes, etc. are commonplace. Extending our basically chemical and electronic minds with electronic computers is becoming more commonplace.

    For patients who, up until now, have had little exposure in school or elsewhere to the use of the medical record as a powerful tool in their own health care, the particular form of this tool will be of little consequence so long as it is clear to them and usable by them. A computerized problem-oriented record will not be any newer or more confusing to them than traditional paper records since they never had either record in the past.

    Physicians, nurses, and other providers have been trained with a whole set of habits and notions about medical records and their availability to patients. It is difficult for some of them to switch to electronic tools that provide specific guidance for solving problems within the context of patients’ other problems. Some not only do not want to switch to an electronic record system, they still do not recognize that the record should be a tool the patient’s use as much as a tool for their own use.

    In health care, patients and very inexpensive paramedical people who are already a permanent part of a community must be taught to use the problem solving guidance in their own records and eventually in computers. After all, rescue squads with remarkable skill in heart and lung disease have been developed all over the country, and people with only a high school education or less have been taught to do sophisticated medical work. Surely we all can learn to deal with many of the less life-threatening disorders such as sore throats and body aches if we have our records and the right guidance tools. Expensively trained medical professionals should be reserved for specialized tasks that we cannot master and cannot do for ourselves. They also should be used to build the guidance in the tools and to monitor occasionally our records and behaviors to make sure that we are behaving in a disciplined and reliable manner…

    …Physicians and other providers often make time the constant and achievement the variable with patients. They try to do everything for the patient themselves and even keep all the records to themselves and instruct the patients hurriedly over a series of timed appointments. They do not have the time or money to give the necessary time to those who need it; on the other hand, they also have patients who return for repeated office visits that are unnecessary because those patients understood their situation at the first visit and can manage their own affairs. In such medical practices the patient is not only being denied his essential role as an informed participant. in his care, he is also being denied the basis to form an accurate judgment about the quality of health care he is purchasing.”[Medicine in Denial, pp 261-2, 264]”


    “The massive scope and intricacy of our increasing knowledge, and its infinitely variable applicability to individuals, make it increasingly obvious that the minds of highly educated physicians cannot be relied upon to recognize the patterns that define unique individuals and their medical needs. In that environment, we will heed Bacon’s warning not to “falsely admire and extol the powers of the human mind,” and we will embrace the use of external tools to empower the mind. Both the mind and external tools use language to reference clinical concepts. Lack of precision and consistency in the use of language has long been recognized as an obstacle to semantic interoperability among disparate health information technologies, particularly electronic health records. Accordingly, major efforts have been underway for many years to develop standardized medical terminology, taxonomies of medical concepts and corresponding coding systems. These efforts, however, valuable as they are, leave unresolved the problem of unstructured clinical judgment by physicians. For example, using standardized terminology to record the results of an initial workup does not assure that the contents of the initial workup will be complete or accurately coupled with medical knowledge. Assuring those goals requires some form of knowledge coupling tools as described above. Standardized terminology and coding is pursued most fruitfully when it is driven by needs that arise in developing knowledge coupling tools and using those tools in medical practice.” [Medicine in Denial, pg 192]

  4. Bobby, just a plain reference to that book, without any further explanation, is about as valuable to the discussion as “nana nana boo boo”.

  5. The safety of HIT systems has been ignored by all. There is not any vetting in the premarket and no after market surveillance. Hospitals always blame the users to protect their financial partnerships wioth the vendors.

    Pronovost is out there and he too, has ignore the lack of safety accountability of the EHRs and their vendors.

    Paper checklists work cause they are simple. EHRs and CPOEs kill patients cause they are poorly usable.

  6. Gawande’s and Pronovost’s point re checklists is not that you can replace clinical thinking with checklists (or pilots with computers), but rather that in appropriate instances (procedures like central line insertions) a checklist can prompt appropriate infection control measures to bring down the rate of infections – with well-designed trials to back up those assertions.

    A good analogy between clinicians and information tools like checklists could be in the ‘rifleman’s creed’. You wouldn’t replace soldiers with guns to win a war like WWII, but putting soldiers in unarmed (or with poor guns) should be viewed as being vastly inferior than giving them the right, well-designed tools.

    IMO, for clinicians those tools for the most part simply don’t exist today.

  7. No argument there. And, Lawrence Weed is not optimistic that things will change for the better anytime soon.

  8. Weed in 1968 promoted SOAP notes. We are not allowed to document sensibly today

  9. A person who never went to college can get all his FAA ratings from Private pilot to AirlineTransport in about 6 months. That includes Commercial , IFR, multi-engine. he then can get a tupe rating in a particular aircraft and be eligible for flight crew status.

    A physician must finish college with qualifying grades, take the MCAT nd perform adequately, apply to medical school, interview, get accepted, go to med school for 3-4 years, and do a residency for 3 + years (7 for general surgery ple 3 more for cardiothoracic surgery).

    What makes you think you can possibly standardize care into a apread sheets, check lists and care paths?

  10. Very thought provoking. A controlled structured should be a top priority thought out a hospital setting. One has to remember all of the unknown variables involved in an ICU environment and the positive effects this form of a structure would provide.

  11. Let me play the devil’s advocate here: are most of these suggestions solidly feasible, and deliver good bang for the buck/effort?

    Take the monitor alarms: very likely, the alarm will go off (like most HC technology/arrangements) with high sensitivity but very low specificity (so the maker cannot be sued because the alarm does not go off.

    Now consider something much more complex: “A primary intervention to reduce mortality in patients with sepsis is to give the right antibiotics fast, generally within an hour of when sepsis begins. Because sepsis is diagnosed with several tests taken at different times rather than a single test, the diagnosis is often delayed by many hours. The diagnosis could be automated—and the need for antibiotics displayed—if the monitors and electronic health record communicated with one another.” Does that imply that the diagnosis can be made with absolute reliability? I am no intensivist, but doubt it. An automated system will likely create false alarms (like these monitors mentioned) and decouple alert clinicians and nurses from closely following the patient (I am aware that I am describing the ideal case here).

    The cockpit is IMHO a somewhat misleading analogy: flying is complex but – opposed to medicine – highly standardizable, and plane crashes are more dramatic events, usually multilethal than an ICU patient getting ventilator associated pneumonia. Getting the bed angle fine (and of course, hand hygiene) sounds really good, but I am not YET ready to believe that a semiautomated ICU is worth fighting for.

  12. Sounds like a great start.

    One hopes CMS, private insurers, JCAH, and billing and coding experts were intentionally excluded from your conference. They’re the ones who have rendered our current EMRs clinically useless.

  13. Vendors of EHRs view their unique, proprietary user interfaces as a key means of competitive differentiation. So collaboration is non-existent and different systems can use icons and colors to mean entirely different things. Basic information is displayed in different places and in different ways.

    As many, including the IOM panel on Health IT and Patient Safety, have identified, safety is FAR from a top priority for either EHR vendors or the cheerleaders at ONC. And it is a problem.

    Planes would crash every day if cockpit displays were as varied and chaotic as ICU equipment and EHR software are. Unfortunately, healthcare deaths are less visible than plane crashes (and often attributed to other factors than ill-designed equipment or the predictable errors that occur when humans try to use systems that exceed their cognitive functions).

    Collaboration, research and innovation sadly won’t solve this problem. Only regulation will.