Mortal Coils: Why We Must Stop Tolerating Failing Health Tech

Eric Perakslis, PhDSome books just stick with you. For me, Atul Gawande's Being Mortal: Medicine and What Matters in the End is one of those books. His storytelling about many aspects of end of life resonates in its immediate relevance- after all, many of us have aging parents and friends-as well as in its professional urgency. It's clear that our healthcare system as a whole is really bad at managing patient and family needs toward the end of life. Making matters worse, our current political dysfunction often derails attempts at essential dialogue so badly (care for a dose of misinformation on death panels, anyone?) that many have stopped trying. But the fact that the problem is a difficult one is no excuse to leave it unsolved.

Toward the end of December, my stepfather took another fall, one that required ambulance transport to the local emergency department. The week that followed was all too familiar. My stepdad, a Korean War veteran in his late 80s, has congestive heart failure (CHF) and recently suffered several compression fractures of his vertebrae. He is, in short, exactly the type of patient that end-of-life care tends to serve poorly. He has too many specialists, is too sick for options such as surgical repair, and, although he would never say it because the Merle Haggard generation never complains, he likely experiences life as one series of indignities after another.

Dr. Gawande's book does an excellent job dissecting this phenomenon and its far-reaching effects. Most older folks experience episodic care performed in the absence of an underlying or overarching strategy. Quality-of-life preferences and personal priorities aren't clear or aren't discussed. End-of-life care isn't mentioned until death is imminent. It's complicated, it's personal, it's difficult and it's not going to be solved by this blog. But what I do hope to accomplish is to highlight the source of many of these indignities and push for progress on solutions.

After this ordeal, all I can think to ask is, when will enough be enough when it comes to poor-quality and clearly failed technologies?

The source of many of these issues traces back to those of us who are health technologists and to the health technology industry as a whole. When my stepdad arrived at the ED of a suburban hospital that was part of the Massachusetts-based Beth Israel Health System, nearly everything that could go wrong, did. The ED was overrun and understaffed. My stepdad was placed on a gurney and left to wait with 11 other patients in a similar state, as there was no other room at the inn. Every clinical encounter was a case study in failed healthcare technology delivery. The contact information of his current specialists was unavailable. His "normal" reference values for bloodwork was unavailable. Results of his latest cardiac echo and renal function exams were all unavailable. Each clinician had to assess his current and past statuses as if he had just parachuted down from the moon, because there appeared to be no record of his life on earth despite his recent history, including all diagnostics and inpatient admissions, occurring at this same institution.

My first job was as a student intern in the biomedical engineering shop of Newton-Wellesley hospital the summer of 1985, when I was 19 years old. All the medical records were on paper, but the care being delivered then was better. At least all the records were in one place and, if someone needed care away from home, there were only two places you might have to call: the family physicians' office, or the local hospital. If you did have to call, you would have spoken to a human, whether at an answering/paging service or a human in the actual hospital or doctor's office.

dmitrochenkooleg via Pixabay Network cables - Image credit - dmitrochenkooleg via Pixabay

Today, data are scattered across thousands of database tables within any single electronic medical record (EMR) system, but also across dozens of other systems that hold pharmacy data, imaging data, insurance data, laboratory data, etc. Pretty much none of it is available on demand in any given clinical setting. The inevitable result of this disconnected galaxy of data "black holes" is mistakes, or if not outright mistakes, well-intentioned missteps based on lack of background data within the acute-care setting. In our particular case, my stepdad was given medications that then themselves became the issue to be managed. Care delivery in the absence of adequate context will always be dangerous. Luckily, he is a tough old dude and is bouncing back well. It doesn't make me feel better, though.

After this ordeal, all I can think to ask is, when will enough be enough when it comes to poor-quality and clearly failed technologies? The "EMR-industrial complex" is not getting smaller. Competition is being squeezed out of existence, while costs continue to rise and the outright hatred that clinicians have for these systems makes the news in varying forms almost monthly.

But while the failure of technology is obvious, equally as clear is what remedies won't work. Over the last 30 years I have advised and worked with several of non-governmental organizations (NGOs) on issues surrounding patient data availability and access. Together with colleagues at Johnson & Johnson, I invented one of the first clinical data warehouses that could associate clinical phenotypes and genomics and, instead of selling it to Silicon Valley, we convinced J&J to make it open source and give it away. I've served and invested in dozens of public-private partnerships as an industry leader and spent time as the chief information officer at the US Food and Drug Administration trying to help fix federal IT. Is my stepfather's bad week my fault? I tend to think so. It must be. If not mine, then whose? Even though the one big thing in health tech that I personally haven't done is implement or fix an EMR system, with all the opportunity I've had, how have I not improved any of this?

One of the most poignant messages of Being Mortal is that we're only human. We're imperfect; we struggle with difficult problems; we're likely to make mistakes. Indeed, some mistakes we make more than once. But the book also urges us to wrestle with the most difficult problems, as otherwise they will never be solved.

An obvious place to start is to try to stop making the same mistakes. No matter your good intentions, the problems standing between us and an adequate, fit-for-purpose clinical information technologies system will not be solved by one more set of standards, one more well-intentioned NGO, one more government task force, one more boondoggle federal grant, or the next update from EPIC or Cerner. (Yes, I just named names.) We must understand and admit these things. Not only is doing the same thing over and over and expecting a different result the definition of insanity, it also diverts energy, resources, and opportunity away from new ideas that may work. Further, once any of the above-mentioned approaches get underway, they're just about impossible to stop. Our best efforts to date have simply created new silos and organizations that now crowd the space, consume resources, and make retracing our steps even more difficult.

Where Do We Go Next?

If we are to make a meaningful change to the status quo, we must address the beast itself. While I don't advocate returning to a paper environment, we must reassess core needs and how best to address them-and this must be done, at least initially, with a clean sheet of paper. At first, the process will inevitably resemble (metaphorically) the scene in Alien where the Nostromo's crew try to remove the facehugging pupal alien from its first victim. But unavoidable pain and difficulty notwithstanding, I hope we can remain focused on signs of hope. Here is a 5-step plan that healthcare delivery organizations can consider.

Step 1: Get specific, measure, and accomplish.

One think I learned from my larger corporate roles is that complex systems can only be controlled via perturbation and measurement. Organizations need to decide which things are most important for care delivery, then focus on those things and fix them. Such efforts need not comprise dozens of goals: even at the pace of a few well-chosen "wins" per year, change will come. In addition, the resulting "muscle memory" will improve the future potential to effect change. A great example of these principles can be found in Dr. Melinda Ashton's "Getting Rid of Stupid Stuff" from the November 8, 2018 issue of New England Journal of Medicine.

Step 2: Stop waiting for things, especially technologies & standards.

We're often warned against "letting the perfect be the enemy of the good," an admonition that's especially salient in the case of standards for data and technology. The i2B2 tranSMART data standards were released before 2010, yet 10 years later, we continue to hear "lack of standards" offered as a reason for poor interoperability. This does not mean that standards are bad or unimportant. However, waiting for the next one to arrive cannot be an excuse for failing to fix an urgent problem or address some pressing need. At J&J, we achieved clinical and genomic data interoperability across business units more than a decade ago using the imperfect tools available at the time. The system has since been applied across sectors and is still serving patients, clinicians, and researchers and winning awards many years later. There are more and even better standards and technologies available now than there were then. Use them.

 Jeremy Lapak via Unsplash Runner up a hill-Image credit-Jeremy Lapak via Unsplash

Step 3: Think bigger.

Healthcare is complex and so is healthcare technology. We've talked about EMRs, but what about mobile apps? What about artificial intelligence? What about digital sensors and wearables? All of these are already here or coming very soon to your quiet corner of the world-and they are all coming at once.

Most organizations approach problem solving in a reductionist manner, by trying break complex things down into a set of smaller, simpler, and (hopefully) more solvable problems. However, this approach itself creates challenges: a) pilots and innovation efforts seldom achieve full cycle to address the larger changes needed; and as Rose puts it, b) "…ideological reductionism manifests a cascade of errors in method and logic: reification, arbitrary agglomeration, improper quantification, confusion of statistical artefact with biological reality, spurious localization and misplaced causality." Fortunately, there are signs that healthcare is starting to think bigger in response to the deluge of available technology, with the Mayo Platform effort offering a good example.

Step 4: Engage the patient and provider communities…really.

Paternalism in healthcare extends beyond individual clinician-patient relationships and is actually an aspect of institutional culture. We hear ourselves speaking about what we think patients will like, not like, would be willing to accept, and so forth. The problem with this was well described by McKinstry in 1992 in the British Journal of General Practice: "It is concluded that paternalism is rarely justified when treating patients of sound mind and then only where restoration of the patients' autonomy is the main aim." It's past time to stop using our (often flawed) perceptions of others' best interests as excuses for inaction or ineffectiveness. Further, as I recently told a colleague regarding my request to engage a larger group in an activity: "Of course, we need to engage people that actually do the work. By the way, isn't the management team the ones who built the current system in the first place? Why would we expect them to be the best ones to fix it?"

Step 5: Train for time, not for distance.

Back when I used to run marathons, I remember that a frequent debate among runners was whether to train for time or distance. Races are actually defined by distance so training for distance makes the most sense, right? For me, not so much. I agree with Tina Muir, who in her "Running for Real" blog states:

"Running by minutes, you know exactly how long it will take, going faster doesn't help you in any way, which means you are less likely to go faster than your body is ready and recovered to go. You just go whatever pace feels right for that specific run. If it is a hard run, you run as fast as you can for 30 minutes or whatever it will be. If it is an easy run, you run for an hour at whatever speed feels like you are recovering as you should be. You get the same work in, with a more relaxed and confident mindset."-Tina Muir

I think most organizations need to cultivate this mindset. Not just because it makes grappling with change easier, but also because it eases the challenges of execution, which are what many organizations are really worried about. Organizations don't necessarily fear change; they fear failure, and this can only be alleviated by methodology…or by not trying. Healthcare must prioritize acceptable technological capabilities and they must commit to making it happen. It's not a sprint. It's a marathon.

Well, I am off to see Grandpa Fred. He has finally flushed the gabapentin out of him and is sitting up, hungry and fully aware. His pain is manageable, although he is yet to recover from the Patriots' loss on Saturday night. I'm looking forward to having him home soon, but I need to put up more bathroom hand rails before that happens.

I will also probably read Gawande's book again. It's not only a great read, it's solid emotional support.

Happy New Year.

Love and peace, Eric


About the Author

Eric Perakslis, PhD, is a Rubenstein Fellow at Duke University, where his work focuses on collaborative efforts in data science that span medicine, policy, engineering, computer science, information technology, and security, while also contributing to training and mentoring future leaders in the field. Immediately prior to his arrival at Duke, he served as Chief Scientific Advisor at Datavant, Lecturer in the Department of Biomedical Informatics at Harvard Medical School, and Strategic Innovation Advisor to Médecins Sans Frontières. [more...]

This article was first published in the Duke University FORGE and it is reprinted by Open Health News with permission from the author. The original post can be found here.