I do not stake my hopes for the future of healthcare on new legislation or changes in payment methodologies. Those are necessary but insufficient if we are ever going to respond effectively to the challenge to achieve
…Care better than we’ve seen, health better than we’ve ever known, cost we can afford,…for every person, every time…in settings that support caregiver wellness.
I believe that the challenge of the Triple Aim requires the re engineering of healthcare and the application of continuous improvement science. Old systems that produce harmful medical errors and consume resources that are disproportionate to the benefit they provide must be replaced by systems that work for the patients and providers, and that are economically sustainable. I am impressed by Atul Gawande’s assertion that we are no longer ignorant. We are inept. We have knowledge and resources that we must manage more effectively if we are to achieve the high aspirations of the Triple Aim. We have not accepted what Robert Ebert told us more than 50 years ago when he said
The existing deficiencies in health care cannot be corrected simply by supplying more personnel, more facilities and more money. These problems can only be solved by organizing the personnel, facilities and financing into a conceptual framework and operating system that will provide optimally for the health needs of the population.
We all experience the resistance of the status quo. In his book Mistreated: Why We Think We’re Getting Good Health Care and Why We’re Usually Wrong, Robert Pearl lumps the keepers of the status quo into what he calls the “legacy” systems. The legacy systems in healthcare are happy to accept the continuing profits that our dysfunctional processes produce. Specialty societies, large medical centers, the big five providers of insurance, device manufacturers, and big Pharma constitute Pearl’s “legacy systems”. They like the status quo and the profits that our system of care produces even as all of its customers complain about the price. Paul Batalden taught us that every system perfectly produces the results that it was designed to produce. Systems designed for corporate and provider profits perfectly produce profit.
Writers like Robert Pearl and Elisabeth Rosenthal, the author of American Sickness, beautifully describe how these systems work and how their self interests undermine safer, less expensive, and more easily accessed care for everyone. If we are to capitalize on the insights that Ebert and Batalden have given us and move past the ineptitude that Gawande sees, we must learn how to manage complexity, introduce innovations, and learn how to spread what we have learned and developed. Gawande was surely referring to our inability to efficiently spread and implement our newly acquired knowledge when he talked about our ineptitude.
I have been enthusiastic about the potential benefits of employing IBM’s Watson and other forms of Artificial Intelligence to us help overcome the challenges we face as we try to provide high quality care that is financially sustainable and available to everyone. Dreams are easy. Reality requires work. Perhaps others are also a little disappointed in IBM’s ability to translate Watson’s skills at chess and on game shows like “Jeopardy” into healthcare improvements. It would not be the first time that finding the right application for a new and exciting tool proved to be difficult, or that a new and useful tool ran into the resistance of doing things the “old way.”
Robert Pearl is the CEO of the Permanente Medical Group. He has published a new book, Mistreated: Why We Think We’re Getting Good Health Care and Why We’re Usually Wrong. In the second chapter I read his answer to why Watson was not fulfilling my dreams of its impact on healthcare. I have bolded the big take away at the end of the piece.
…meet IBM’s Watson, the supercomputer that took down Ken Jennings on Jeopardy! Using the power of big data, Watson has the ability to quickly find clinical answers buried in millions of pages of medical records…In the context of how doctors have traditionally seen themselves, Watson isn’t a threat. He’s more like a “Lifeline” from the TV quiz show Who Wants to Be a Millionaire? If you’re stumped, Watson may have a suggestion. If doctors don’t like his answer, they ignore it.
Watson’s capacity and speed are incredible. But he’s not really what doctors need. The problem with Watson isn’t that he can’t do amazing things. His problem is that the amazing things he does rarely add value…It is a rare patient for whom a thorough review of every article ever written would be necessary. Besides, Watson can’t discern which papers should or should not be trusted. Achieving the best outcomes is not a matter of discovering the obscure, but rather providing the right care consistently. The biggest opportunity we have as a nation for saving lives is ensuring that every doctor follows the best available approaches every time.
He continues by celebrating Apple’s Siri as the answer that doctors need. Siri fits your pocket and is capable of correcting the problem of forgetting a step in care which he says is the greatest source of physician error. Siri could enforce protocols and decrease harmful variations in routine care.
Siri can follow doctors from room to room while delivering more than enough memory and computing speed to help physicians diagnose and treat nearly all problems that patients experience…Better still, Siri’s expertise will grow over time. Using information contained in millions of electronic health records, she can be programmed to calculate the probability that a particular patient in the ER is having a stroke or needs a particular cancer treatment.
Not only is she a powerful source of information, …offering up friendly reminders as doctors go step by step through the treatment process.
Pearl fails to discuss that both technologies must be incorporated into work flows that are designed to reduce the work clinicians and administrators must do everyday. I do agree with him that Watson as currently used does not benefit the routine care of patients in any way that could make a real difference in the cost of care and little difference in safety or efficiency. Siri is also not being used as he describes. All of this must be considered against the historical reality that the IOM and other authorities teach us that it takes seventeen years for any new drug, medical device or change in a practice protocol to be commonly accepted and deployed.
I was recently asked how I would recommend that A.I. be more effectively incorporated into our efforts to continuously improve care. The query was fanciful. If a genie granted me three wishes with Watson, how would I use those wishes? I would use my three wishes to bring the power of Watson to the aid if patients, providers, and those responsible for the business of care delivery.
Patients would directly benefit from the systematic application of Watson to the problem of transparency. They need help in choosing which option of care realistically meets their needs. They need cost data as well as outcome data, and data about the experience of care of other similar patients. Watson has access to the huge quantities of data necessary to answer these questions, but its capability is not utilized because we do not have the workflows that can bring it into the process of care.
Pearl points out that patients ask friends about who the best doctor is and are impressed by advertising in part because there is no presentation or organization of easily understood information to guide them in their choices. If Watson is capable of assisting an oncologist in finding the best study or the best protocol for a patient, it should also be possible for it to help find the best place and best team to provide care at a price that fits the economic realities and personal preferences that are of concern to the patient. For this to happen we will need more than Watson. The project needs sponsors and work flows. The sponsor could be a government agency, a business providing a service, or an insurance company or care delivery system seeking to offer a service that provides it a competitive advantage. Once there is a sponsor, Lean is an excellent way to engineer the reality and move up the continuous improvement ladder from idea to reliably performing asset.
Dr. Paul DeChant has long contended that one of the biggest threats to the Triple Aim is clinician burnout. Paul is correct to note that burnout arises from poorly functioning systems, increased regulatory demands, the need for detailed documentation for optimal reimbursement, and increasing operational overhead that robs resources for practice support. Clinicians struggle with computerized “paperwork” and spend much of their time doing clerical tasks that leave no time for the care of the individual patient, much less the time to exercise the use of the tools of population management.
Watson could “augment the intelligence” of every physician across the spectrum from primary care to the most focused and limited specialty practice. Dr. Pearl thinks that Apple’s Siri can monitor adherence to best practices and identify forgotten steps in usual practice like omitted immunization or the failure to provide adequate prophylaxis for infection or manage through ambiguous emergency presentations. Watson should be able to do all of this well, but for it to help us we must learn how to let it make a difference. I believe that Watson’s greatest potential benefit may be its ability to incorporate unstructured data that it has read or “learned” into clinical analysis. Much of the critical thinking and decision making that we do for patients at the point of care involves weaving structured and unstructured data together. Systems must be developed that integrate human based clinical skills with machine learning. It is time to start the journey.
My third use of Watson would be as the manager of the mundane. Administrators, clinicians and patients would all benefit by effective incorporation of Watson into processes like filling out forms, requesting prior authorization, complying with audits, assuring quality through reporting to payers and regulators, finding effective and efficient referrals and managing the loop between the request and the return of the answer to the clinician who asked the question. Watson can “learn” these tasks as an assistant that will eventually do more and more of the work.
If I could ask for a fourth wish from Watson, it would be as an application that redefines “top of the license” for every healthcare professional. We all fear the cliff that lies ahead. We are about to fall into an abyss of professional shortages. I could argue that even now we are inadequately staffed to provide care to the 85-90% of Americans who currently have some sort of covered access to care. We must change how care is delivered, and by whom, if we do not want access and service to further decline in the future.
I worked for years in a system that valued nurse practitioners and physician assistants. What I realized was that their efficiency was frequently undermined by the same fears and concerns that made me less efficient. They were often asked to solve or manage problems where uncertainty slowed them down and caused them distress. Watson can help. I think that it will not be long until Watson is ready to answer almost any question a clinician might ask. It has long been demonstrated that care plans directed by algorithm get better results than expert clinicians practicing the “art” of medicine.
Maybe you are not ready for a new friend who will augment your intelligence, but I am. I was only given three wishes and asked for four, but I can think of many more. I see Watson as the first of many “new friends” on our healthcare team who are ready to help us move beyond the ineptitude that Gawande complained of toward the dreams of the Triple Aim that persist through our current misery and uncertainty.
