Dear CMS Administrator Verma

Stephen Soumerai ScD
Ross Koppel PhD, FACMI

Your address to HIMSS acknowledges many of the problems with Healthcare IT, highlighting lack of interoperability, lack of data exchange, and lack of cybersecurity, and suggesting some regulations that could be eliminated. This is a welcome realization of some of EHR’s more obvious limitations and problems.  However, most of your recommendations for improvement of health IT are insufficient, unproven, or have been repeatedly shown to fail.

We applaud your acknowledgement of: 1. The frustration (and often rage) of many clinicians when using the current EHRs’ clunky and inefficient user interfaces; 2. Patients’ frustration and alienation when doctors spend much time entering data into the EHRs rather than listening to them; 3. The need for better cybersecurity; 4. Benefits of increased patient access to their data; and 5.Healthcare systems’ refusal to share patient data with others clinicians (data hoarding).

We are also delighted for your strong support for the Sync for Science program.

However, your solutions to these problems are faulty or have already failed—and thus we are obliged to explain why and how they fail:

  • Belief in the magic of value-based care as a cure for excessive spending
  • Claim that “open APIs” (defined in next sentence) will solve the problems of lack of interoperability. Here, APIs refer to software programs that try to translate different forms and formats of information into a single commonly understood item.
  • Belief that a patient’s personal data store or personal EHR, called “MyHealthEData,” will help solve the problems of patient care. 
  • Confusing patients with customers. We train doctors to make diagnoses, order and interpret tests, and help patients make profoundly complex decisions. Healthcare decisions are not like buying a toaster. 
  • Attributing so many of the problems of EHRs to the regulations created in 2009, the “Meaningful Use” rules.   In fact, meta-analyses of EHRs—both before the Meaningful Use rules, and after–fail to find they reduce costs, mortality, or morbidity.

We briefly consider each in turn. 

Value-based care:   Value based care makes total sense in theory—that we should only pay for outcomes and needed care. Unfortunately, the best studies show these bonuses and penalties to physicians don’t work.  Often these are seen as insulting incentives for work they already do.  Repeated studies also show that Accountable Care Organizations and many of the pay-for-performance plans are likewise unsuccessful. Tiny savings are overshadowed by huge delivery system implementation costs. Perhaps worse, they harm patients by avoiding care for the sick who drag down clinicians’ performance and “health measures.”

Part of the problem is that it is not possible to adequately adjust quality measures for doctors who care for more poor and sick patients. Other factors matter to health, like language, rurality, frailty, abuse, and hospital resources, to name a few. How many of these factors are in insurance company or CMS claims databases? “Risk adjustment” is a phony promise.

APIs as solution for lack of interoperability:  These APIs—in this case, software programs that attempt to translate different forms and formats of information into a single term–can be useful. But they can’t overcome the lack of data standards allowed by the ONC in the past years. EHRs have the same “measure” defined in 40 different ways. For example, one EHR records blood pressure as 120/80, another as “labile” or “under control” or “stable” or “left foot” or “seated, post injection.”  Faith in EHRs was so all encompassing that the federal government refused to require data standards. Without data standards, APIs often become failed workarounds that can’t possibly overcome the differences in the way the data are recorded. Often those differing measures appear in the same EHR. [1] 

Belief in “MyHealthEData” to provide personal efficacy, address lack of interoperability, and improved transparency:  Providing more information from EHRs to patients and their caregivers is a reasonable concept but won’t work as a solution to healthcare’s ills or as a national policy. Those of us who have been working on this for several decades are aware of the problems you may not have considered: 

1. A patient’s several EHR “portals” contain massive amounts of contradictory and confusing information.  Do you think the average patient will understand an increase in BUN (Blood Urea Nitrogen) from 9.2 to 9.8 mg/dL? Moreover many test results only make sense in relation to other test results and/or over a timeline. Can patients do that?

2. The operating instructions will be incomprehensible to many patients, especially poor and sick ones.

3. They will drop their cell phones with their MyHealthEData” and lose all the information.

4. They will have acute medical events, such as a stroke, and will be unconscious, failing to bring the device in the ambulance.

4. Substituting patients for hospitals as the responsible parties for patient safety multiplies the potential problems a thousand fold.

5. Even now, most medication lists include drugs that are dangerous or lethal in combination, or have not been used for years, or are duplicative and lead to double or triple dosages. 

3) There is no such thing as an innocuous medical test result, e.g., a test result saying one does not have STDs may be both comforting and disconcerting to a patient’s family. 

Last, you must be aware that MyHealthEData will be disastrous for the people who need it most: the sick, the disabled, the elderly, the poorly educated, those with less informed caregivers. Those of us who work with such systems call them ideal for the “worried well,” and generally useless for the rest of the population. We can only shift some of the responsibility for patient safety on patients.

Confusion of Patients as Customers:  Under ideal conditions, customers can make considered judgements about the purchase of products and services. This is seldom the case in healthcare. If one had a heart attack while carrying a Wi-Fi enabled laptop, one could direct the ambulance to the best facility based on sophisticated statistical analysis. Of course, that’s absurd. What would actually happen is the patient would be struggling to stay alive and would go where the ambulance took her. More generally, patients go where their insurance companies send them, where their doctors have connections, and where their family members can visit.  Even more important, while many people can evaluate the benefits of, say toaster “A” vs. toaster “B,” few can understand the relative or absolute value of differing medical treatments or medications.  We notice that in your deeply poignant story of your husband’s illness, you did not pick the medical center that saved him–which happened to be where one of the authors of this document (RK) coincidently works.  

Attributing so many of the problems of EHRs to the regulations enacted in 2009, the “meaningful use” regulations:  You are right that some of the meaningful use regulation are annoying. As you say, doctors complain, and research confirms, EHRs increase the work and time needed for patient care and seldom increases safety or reduces costs.   But the dysfunctions of “meaningful use” regulations pale in comparison to the pains inflected by the documentation requirements of insurance companies and CMS. Worse, private insurance companies are incentivized to deny payment, and thus encourage the worst abuses of copy and paste plus meaningless but legible gobbledygook in the EHR. Another source of clinician frustration–the lack of EHR usability–is probably a worse and separate problem. Here, again, the government refused to institute minimum standards, resulting in bizarre differences among EHRs and confusing displays of patient information. These lack of regulations and uniformity are a major patient safety danger. When we get into a new car we assume the brake is on the left and the gas is on the right. Such uniformity does not exist with EHRs.

You mention that the federal government initially spent $30 billion to support EHRs. That’s a pittance compared to what hospitals, clinics, physician practices and nursing homes are obliged to spend: a sum that far exceeds two trillion dollars.  Moreover, the VA and the DoD are about to spend an additional $14 billion on EHRs. 

Ms. Verma, with the complexities of healthcare and healthcare IT, one is reminded of Mencken’s line: “for every complex problem there is an answer that is clear, simple, and wrong.”  Yes, there are solutions, but they are neither buzzwords, nor simple, nor apps, nor the removal of a few regulations. They require an honest appraisal of our technologies, our policies, and our resources. If CMS is willing to undertake such a journey, there are researchers who would join you. 2

We agree wholeheartedly with you that almost 20% of the GDP for medical care is too much to pay when so many other industrialized countries are getting much more health for half the cost. But, too many of your proposed “solutions” have repeatedly failed tests in situ. Blaming EHR problems on regulations is ironic because so many of their failings result from the government’s refusal to accept regulations that would have established common data terms and acceptable user interfaces. Faith in APIs is premature and not supported by evidence. Value-based care sounds great, but objective scholarship finds consistent failure in real-world settings, even patient harm because unhealthy patients are avoided. Belief that patients are customers is only occasionally true; most of the time it’s a fake metaphor used to fool regulators and those with only primitive economic understanding.

Healthcare policy is littered with buzzwords and blind faith that should have been subjected to rigorous testing before spending billions or trillions. Blind hope in technology or in naïve economic theory is unworthy of Americans’ health and pocketbooks. There are scientific methods to evaluate and guide healthcare policy. Given the vast sums and the lives at risk, we commit to careful analysis before we waste trillions of more dollars and endanger the health of millions.

1. Smith SW, Koppel R. J Am Med Inform Assoc doi:10.1136/amiajnl-2012-001419

2. Ross Koppel, “Great Promises of Healthcare Information Technology Deliver Less,” chapter in C.A. Weaver et al. (eds.), Healthcare Information Management Systems: Cases, Strategies, and Solutions,  Springer International Publishing, Switzerland 2016


Ross Koppel PhD, FACMI is at the University of Pennsylvania where he is a Senior Fellow at Wharton’s Leonard Davis Institute of Healthcare Economics, at Penn’s Center for Public Health Initiatives and at Penn’s Dept. of Biostatistics, Epidemiology and Informatics. He is also adjunct Professor of Sociology at Penn.  rkoppel@sas.upenn.edu

Stephen Soumerai, ScD is Professor of Population Medicine at Harvard Medical School and Founding and Former Director of the Division of Health Policy and Insurance Research     ssoumerai@hms.harvard.edu

Categories: Uncategorized

8 replies »

  1. Glad you make the point that ONC or some agency must enforce data standards and its role in the woes of EHR interoperability. As a PHM platform aggregating from disparate EHRs, about half the data we get must be normalized before it’s of any real use. That’s from within about 400 million data points we process daily. We have constantly – and within our TEFCA proposed rule comment – noted same to ONC.

  2. Good discussion.

    “One” point of exception though – “Confusing patients with customers”. Or why healthcare needs to be more like buying a toaster.

    When buying a toaster what questions does the “customer” ask? Does it work? What is the quality of the toast? How many of these toasters have been sold? What is the rate of repair? What do other people have to say about the quality of this toaster? What is the price/value ratio?

    Which of these questions would be inappropriate in considering healthcare? (Other than the emergency care scenario – where yes, the “person” goes wherever EMS takes them.) But even that scenario, you describe as “absurd” the thought that “one could direct the ambulance to the best facility based on sophisticated analysis…well actually that is already being done. Companies like ESO Solutions – already use outcomes data to direct EMS crews to the best facility based on sophisticated statistical analysis given the care needed. They even go a step further and sync real-time available resources given the need. Maybe they’re not Wi-Fi enabled, but they sure are wirelessly connected.

    At a time when a generation looks at healthcare as a combination of retail clinics, urgent care clinics and if a need arises – straight to the specialist, it is time to look at the “person” as a consumer of health services. A consumer that should be asking questions.

  3. Not to mention the lack of any traditionally respected evidence for the answer to any of those questions.

  4. There is something about the analog character of hand-written communication that begets a larger scan of thought than the digital process. Gertrude Stein said it best, long ago: “Writing is the only thing that, when I do it, I don’t feel I should be doing something else.”

  5. Thank you, gentlemen.

    We threw a couple trillion and rising at EHRs before we addressed the most basic questions:

    What problems in health care do we want them to improve?

    What data do we want in them?

    How does that data get in them?

    It seems that the answers have worked out to be a Palinesque “All of them/all of it” to the first two questions, and “Through the manual labor of the most highly trained and qualified people in the organization.”

    Every day I’m thankful that our office has stayed with paper (non-Luddite – wrote first programs in 1968 at age 12 at University of Virginia).

  6. The challenge we face — and I don’t see much discussion along these lines — is that there is a difference between what can be called “basic medical care” and the many other alternatives now available by modern medical advances. Be gentle with me for what comes next. I’m thinking out loud.

    Basic care (or “essential” as ACA would have it) is a baseline for everybody, and includes mental health, contraceptives and preventive medicine. Beyond that, however, are endless cosmetic and advanced sports-related “super care,” high-end dental and reproductive technologies and more. There is ample opportunity for profitable medical practices, simply because there will always be a subset of the population well-off enough to have anything money can buy — from cosmetic dentistry, hair transplants, fertility assistance and all kinds of body enhancements — to a personal physician on call 24/7. Most people, however, need access to less-dramatic (and hopefully more prevention-oriented) care. Ours is a public/private/blended hybrid system.

    The goal of ACA was to adjust the insurance-supported segment of health care by defining what is “essential” and how best to distribute the risks. Two tax-supported delivery systems — Medicare and Medicaid — together with the VA & active duty military service systems, are categorically different from the insurance/private sector medical care.

    Seen from a distance the medical care landscape becomes a variety of different delivery systems. Even the tax-supported systems (Medicare, Medicaid, VA and active duty military) all have different funding streams and interfaces with the private sector. Neither Medicare nor Medicaid, both of which are solely administrative, employs medical professionals. The VA and service systems, on the other hand, have physicians, surgeons, nurses, labs, imaging departments etc. on staff, compensated in accordance with “market-based” professional scales — nearly all tax-supported. But even then, there are sometimes cases requiring private-sector referrals.

    From a funding standpoint the whole system is a convoluted can of worms. So there will never be a single one-size-fits-all system for everyone. It’s possible that during a single lifetime, someone may very well matriculate through the system from one end to the other. It’s not unlikely that some people growing up in poor circumstances may be cared for by Medicaid, become active duty military for a few years, get discharged to the private sector and expected to pay their own way (with or without some kind of insurance, HSA, MSA or combination thereof), perhaps qualify for VA benefits, graduate to Medicare, and finally “spending down” once again to become Medicaid dependent in a nursing home.

    Clearly the role of CMS in this universe has serious limitations. The time may have come to have more than one level of record-keeping, employing more than one level of understanding — a kind of healthcare Flesch-Kinkaid scoring system. with multiple access points. In the same way that other government agencies employ several levels of security (from public to “super top secret”) health care records can also be sifted through for limited access.

  7. Health spending represented 5.0% of our nation’s economy (its gross domestic product) in 1960. In 2016, it was 18.0%. The change represented an increase of 5.0% compounded annually, as adjusted for economic growth and inflation. The annual change has worsened since the implementation of ACA 2010 in 2013. (see Altarum health sector spending reports). Meanwhile, we have encountered continued worsening incidence of our nation’s maternal mortality, worsening mortality from chemical addiction, worsening mass shootings, worsening adolescent deaths from suicide and homicide, and the mid-life complicated chronic illnesses from obesity. And now, the most important person to assume a responsibility for communicating an awareness of all this, basically has NONE.

    At the same time, the folks at the Academy of Medicine (aka Institute of Medicine), AAMC (Association of American Medical Colleges), AMA (American Medical Association), APHA (American Public Health Association), and RWJF (Robert-Wood-Johnson-Foundation) exhibit a deafening lack of commonality about any means to build a “Shoot the Moon” strategy to fix the cost and quality problems of our nation’s healthcare. For all of their efforts over the last 20 years, it would seem that the Academy of Medicine folks would have the best vantage point. Obviously, their reconciliation of our current KNOWLEDGE base about HEALTH is disrupted by the distance separating its scientific and humanitarian realms of evidence. And so, we are all left in a state of Paradigm Paralysis. My diagnosis: institutional codependency. Yes, its real. The two words appeared in a WSJ Editorial last week.