The Centaur in Healthcare: AI and humans (WIRED)

from WIRED magazine article

https://www.wired.com/story/algorithm-doesnt-replace-doctors-makes-them-better/

In the battle between the future of super-intelligent Artificial Intelligence and the paltry skills of increasingly left-behind human brains, some rays of hope. There are a growing number of projects dedicated to combining the skills of AI and humans to perform better than either alone.

The WIRED article above discusses Dermatology AI and how it improves the performance of physicians in detecting skin cancer. However, it mainly improves resident and primary care physician performance, and not expert dermatologists.

Is this good? Bad?

And, what is a Centaur? A horse-human hybrid from greek mythology.

I think this illustrates Arther C Clarke’s (paraphrased) saying:

Any teacher physician who can be replaced by a machine should be.

This is not necessarily a bad thing. Consider: if we can allow AI to be trained to augment physicians or advance practice providers in every case where the providers’ experience is not expert-level, we could raise the standard of healthcare throughout the country, or the world.

AI’s still can’t hold a hand, counsel patients on complex and competing issues, be compassionate, and create human connection.

We already have our computers helping remind us of the mundane yet critical tasks of doing the right thing for out patients: remembering tetanus and pneumonia vaccines, remembering to screen patients for colon and cervical cancer, remembering to repeat diabetes exams at frequent intervals. Why not allow them to give a second opinion on whether a skin mole is likely to be malignant?

CMIO’s take? More like this please. The co-evolution of AI and human is accelerating. We are finding a way forward.

Access to Telemedicine and disparities (WIRED)

from WIRED magazine article

https://www.wired.com/story/access-telemedicine-is-hardest-those-who-need-it-most/

As we work on telehealth options for patients, it is important to keep in mind the population of patients we serve. This article demonstrates the differences in access our most vulnerable patients have in accessing technology.

We cannot rest. We have miles to go, before we sleep.

Explaining Science using a Horse Metaphor

This is a poster I saw last fall at an AI conference sponsored by University of Colorado School of Medicine. The science is interesting and very well conducted.

HOWEVER! Pay close attention to the middle column. The authors have, in my opinion, created a clear metaphor to explain a very complex concept. Do YOU know how to do this? As scientists, we are increasingly subspecialized. How do we translate our important findings so that others can understand? This is getting harder and harder to do, as even OTHER SCIENTISTS (I count myself among them) have a hard time understanding what SCIENTISTS IN OTHER FIELDS are doing.

CMIO’s take? Take a moment to digest this, and decide for yourself: can you tell a story and explain as well as these scientists did? Why not?

Czernik: Counter-intuitive way to connect with the patient (Annals Internal Med)

My awesome colleague Zuzanna Czernick and collaborators have written a brilliant piece about the EHR. She used a CT scan image with a large pulmonary abscess to get the attention of an otherwise hostile, disengaged hospital patient.

Link to article Annals of Internal Medicine : https://annals.org/aim/fullarticle/2738161

My awesome colleague Zuzanna Czernick and collaborators have written a brilliant piece about the EHR. She used a CT scan image with a large pulmonary abscess to get the attention of an otherwise hostile, disengaged hospital patient.

The EHR, although widely disparaged, is also a wonderful tool to bring medical data alive for the patient. She offers a few guidelines on how to most effectively create the trusted “triangle” of provider-patient-computer:

  • Prepare
  • Setup
  • Educate
  • Chart together
  • Review

There are so many opportunities to connect with our patients; why not bring up a screen to show an image, a result, a graph, a note written by a consultant that illustrates and answers a question.

CMIO’s take? Yes, we need counter-intuitive (and soon perhaps simply intuitive) stories about the benefits of a modern information system in caring for patients. Thanks, Dr. Czernik!

Information Blocking and the End of Secrecy in Healthcare? (a rant, a talk and a uke song)

What is CT looking at? Could it be … his own radiology image in his patient portal? What does this mean about secrecy in healthcare?

What is CT looking at? Could it be … his own radiology image in his patient portal? What does this mean about secrecy in healthcare?

In this blog:

  • A rant about Information Blocking, Open Notes, and unintended consequence
  • The slides from my talk at CHIME “Is this the End of Secrecy in Healthcare?”
  • A ukulele song “I Can See Clearly Now, My Sprain is Gone”

THE RANT

Followers of this blog, the Undiscovered Country, know that I’ve written about our Open Notes journey, our innovation to release radiology images to patients via the patient portal, our sharing of patient co-pay for prescriptions, and we are participating in the Our Notes project to have patients co-author their own clinic progress notes.

And now, CMS’s latest regulation on Information Blocking, part of the 21st Century CURES act, has detailed stipulations on what must be released to patients, including Open Notes in clinic, in the emergency department, and for inpatient notes. It turns physician paternalism on its head: we should RELEASE all information to patients UNLESS there is a compelling reason not to.

If your organization is NOT scrambling to get this in front of your providers to discuss: immediate release of progress notes, consult notes, history/physicals, operative reports, discharge summaries, laboratory report narratives, radiology report narratives, pathology report narratives, THEN YOU ARE BEHIND THE EIGHT BALL.

Full disclosure, I was part of a Robert Wood Johnson sponsored event to explain Open Notes to congressional leaders in Washington DC 2 years ago. Little did I know we’d end up here, with the regulations not only catching up to the literature (benefits of Open Notes in ambulatory settings), but surpassing it and requiring Open Notes (patients ability to access their providers notes written about them and their are) for Inpatient Settings. We published our experience with Inpatient Open Notes in 2013. The results: lukewarm. Our providers and nurses were very concerned before the project, somewhat less concerned after. Our patients were underwhelmed with the offer of viewing their notes. Others have written about inpatient Open Notes, that potential challenges with communication, anxiety and increased workload may negate the benefits. Specifically patients and providers were concerned: will providers STOP writing important discussions and debates in progress notes because of fear that the patient or their family will immediately see them (eg: “there are several possible cancer diagnoses we are considering that might cause this”, OR, “be careful when you go in that room, the father can be very aggressive”). Certainly, with more care and thoughtfulness, we can write better notes, but should we require that? Are hospital providers not already working too-long shifts and already burned out from excessive administrative work? I’m uncertain.

I am concerned that MANDATING release of inpatient daily progress notes immediately to the patient may result in significant unintended consequences, with benefits that may not overcome the risks.

But, here we are. The full details and FAQ of Information Blocking (how soon must notes/results be released? how extensively? Retroactive to ALL notes written electronically (for us, back to 2003)? are still pending, and yet the regulation goes into effect in November. 2020. Soon.

We have had rules for built-in time delays to the release of test results to patients that have been in place since 2003. These applied to both outpatients and inpatients. We have been pleased that our release of blood tests to patients has been “immediately” since 2003. However, we do delay complex radiology imaging (CT, MRI, PET) for 7 days and pathology results for 14 days so that potential cancer diagnoses are communicated from the provider rather than “discovered” online.

This will now have to change, and urgently. I am convinced we can get to IMMEDIATE release of all results and notes, but it will take some hard thinking, some hard cultural conversations, some letting-go of old traditions, some problem-solving of potential new problems, lots of anticipatory planning (how to educate patients on what they might be the first to see online), and also (as per the Leading Change principles) to grieve the loss of the “old ways.”

IS THIS THE END OF SECRECY IN HEALTHCARE?

So, this is the slide deck content for the talk I gave at CHIME (College of Healthcare Information Management Executives) last fall, and at a couple of other national venues, detailing the information transparency efforts we are undertaking at UCHealth:

https://www.dropbox.com/s/mzhujn6vqiwsevt/2020-02%20CT%20Lin%20End%20of%20Secrecy%20in%20Healthcare.pdf?dl=0

AND A SONG

Far be it from me to lose an opportunity to sing you a song…

Social Distancing reduces more than just COVID-19: Guest Blog (Dr. Eric Glissmeyer)

Dr. Glissmeyer, informaticist, Utah, notes that emergency department visits plummeted in March and are much slower to rebound. Why?

During many winter seasons, pediatric hospitals are bursting at the seams. RSV, Human Metapneumovirus, and other respiratory viruses like non-SARS CoV-2 Coronavirus cause significant disease burden sending pediatric specialists scrambling to find space to admit children with bronchiolitis who need supplemental oxygen and other forms of respiratory support. Patients with the same viruses will “double bunk” in single rooms to receive life-saving care. 

2020 has been very different. COVID-19 was announced to have arrived in Utah March 6, 2020. March 13 2020 Utah schools announced that beginning March 16 online home learning would begin and Saturday March 14 the first case community spread of COVID-19 was confirmed. March 16 the Utah Department of Health issued a public health emergency limiting some services and businesses and Intermountain Healthcare and University of Utah Health announced elective and non-emergent surgeries and many non-urgent ambulatory services would be canceled. March 27 the Governor issued a “Stay Safe, Stay Home” directive. Social distancing during these months, via economic and public gathering restriction, was the only public directive. Mask wearing in Utah did not become widely encouraged until July 2020.

As a result of these social distancing measures, we have witnessed a dramatic decrease in infectious diseases. The following data are from germwatch.org and contain data of common infectious disease prevalence in Utah, as identified by testing performed at and sent to Intermountain Healthcare labs, clinics, and hospitals.

Utah disease prevalence, tests performed at Intermountain Healthcare labs
Group A Strep dropoff quicker in Spring 2020

We have seen a drop in Emergency Department census that is unprecedented. We attribute this change to the decrease in circulating viruses, commonly spread bacterial pathogens and different healthcare consumer choices. In over 15 years, we have not seen ED volumes in the low ranges we are consistently seeing them now. 

Even as many economic restrictions have lifted in recent months, ED census remains lower than previous. At this point, we are uncertain which of the following influencers are playing roles, if all, or others?

  • Health care consumer choices (avoiding health care)
  • Social distancing reducing disease transmission
  • Mask wearing reducing disease transmission as social distancing/economic restriction has begun to lift

Emergency departments are a clinical service entirely dependent upon what is referred or self-referred to them. Yet they are a critical part of the healthcare system for unexpected, emergent care and as a venue for coordinating complex care.

Unprecedented dip in ED visit volumes, Spring 2020

We are seeing an apparent, but slow, increase in ED census over the past 3 months. Much slower than the stock market rebound 😉

July 2020 Daily Census Range: 57-92 July 2019 Daily Census Range: 79-120

Our hospital and others around the country have been bursting at the seams with seasonally variable infectious diseases like RSV and influenza. We now know that social distancing behaviors that decrease disease transmission can have a significantly decrease disease transmission. Data from the Southern hemisphere indicate that influenza season may be better than previous years, perhaps because of societal behavioral changes.

-Eric Glissmeyer, MD
Associate Professor, Department of Pediatrics, University of Utah
Division of Pediatric Emergency Medicine, University of Utah
Medical Director, Care Transformation Information Services, Intermountain Healthcare

EHR v Covid-19. Patient portal trends, 6 months into pandemic

EHR Patient Portal Advocates! I need your brains.

EHR Patient Portal Advocates! I need your brains.

This is the curve of percent of patient appointments each month for the past 3 years, with patient portal accounts. I blogged a version of this back in March. Back then, if you drew a straight line from 2017 through FEB 2020, it looks pretty linear. Then, BOOM, a sharp uptick in March and April. The percentage increases from 71% to 77% in 2 months. Then it has stabilized and flattened in the 76-77% range since then.

On the one hand, one could hope that the increase in patient portal sign up would continue until we got to 80 or 90%. We know that 90% of healthcare ORGANIZATIONS offer a patient portal and that 52% of patients in an ONC survey have a patient portal account (2017 data).

So, in 2017 we were around 56%, right in the ballpark.

I think we understand the bump in portal usage in March and April: our dramatic upscaling of telehealth by 200x from 20 per day to 4000. In order to schedule patients for a telehealth visit, we required the patient to register for and log in to our My Health Connection (Epic’s MyChart) patient portal at UCHealth. Nothing like a pandemic and an available well-oiled telehealth service to bump patient portal stats.

Now, more of our patients can take advantage of: messaging their provider, refilling their prescriptions online, accessing price transparency estimates of anticipated procedures or services, viewing and requesting changes of the: medication list, diagnosis list, view their provider’s progress notes (Open Notes), view their radiology reports AND actual images (UCHealth is one of the first patient portals in the country to offer this).

However, what explains the flattening since then?

Theory #1: It’s those surgeons!

The theory: well, all surgery clinics completely shut down during most of March, April. And, WE ALL KNOW that surgeons are terrible with patient portals (rampant overgeneralization). The answer: Actually, NO. Once we select ONLY primary care practices and exclude surgeons and medical specialists, the same curve occurs. It is not the opening and closing of surgery clinics.

Theory #2: its those Seniors!

The theory: well, since most older patients have been staying home, and a good number of them don’t have access to smartphones or laptops or PC’s to do telehealth visits, it must be ONLY younger patients coming in, who of course have excellent rates of signing up for the patient portal for telehealth, right?

The answer: Again, NO. Turns out, if you filter out all patients over age 65, the same curve occurs.

Huh?

So, what is happening here? Are we bumping up against the natural limit of Coloradoans who own smartphones and/or computers? Have we indeed registered ALL patients who will ever agree to signing up for a patient portal for any reason?

Or, have we gotten lazy and gone back to “hey lets encourage patients to come back to clinic!” and thus relaxed our vigilance at setting up patients for a free portal account?

Or is there another theory, another slice of the data I haven’t thought of?

CMIO’s take? I’m asking for YOUR help! Send me an idea, a theory I can slice on this data and see if it explains the plateau!

EHR v Covid-19. Telehealth after 6 months at UCHealth

Six months into pandemic, what is happening to telehealth visits at UCHealth? Inquiring minds want to know!

Here we are, data dilettantes, on our long journey into the unknown.

At the prompting of online colleague John Lynn, we look back at telehealth usage at UCHealth in the past few months. The above graph depicts January 1 to present, the curve of in-person visits at UCHealth (purple) and telehealth visits (cyan). You see that telehealth visits temporarily outpaced in-person visits.

First of all, I feel very sophisticated for writing “cyan” instead of my first (caveman-male) instinct “blue-ish”.

Second of all, notice the curve above compared to our evolving curve from March, 5 months ago (remember, those purple divots are from Thanksgiving and Xmas holidays):

Be careful how you extrapolate, right? Based on this original, one would have thought “Holy Smokes! Telehealth is going to rule the world in a few more weeks!”

And one would have been wrong.

So, now it comes upon us Armchair Data Scientists to hypothesize: why? Why did the curve do what it did? Well, our first external data point is: Colorado Governor’s Safer-At-Home order expired on April 26. On the top graph, this corresponds to the day our 2 lines cross in April, with in-person visits rising again. This also corresponds to our surgery clinics opening up again to see patients.

Averaging the last few weeks of data, we are seeing about 8000 telehealth visits vs 60,000 in-person. Or about 13-15% of appointments being conducted by telehealth. REMEMBER, this is unvalidated data, so, take with some salt.

What have we learned? From anecdotal evidence, I have heard from quite a few patients (most of mine are over 65) that they prefer in-person visits when possible, although telehealth has been “acceptable” when fears of contagion are high. Also, much of internal medicine requires blood testing, vital signs monitoring, examination. Also, I’m finding that non-verbal communication, although “acceptable” via telehealth (tone of voice, body language), it is much richer in-person.

Even when we were conducting 2/3 of clinic sessions exclusively by telehealth, our in-person clinic slots were full, and our telehealth clinics routinely had open time slots. Now that we are scheduling 75% in-person, all our in-person slots are full, and our telehealth slots still sometimes are open.

It will take some intrepid ethnographic researcher to pull interesting trends out of this, as I’m hearing from other parts of the country that telehealth visits are preferred to in-person. Is this: geography and distance needed to travel? Is it the rarity of the specialist’s expertise? Is it access to surgeons? Is it the (gasp) lack of skill of the telehealth provider (please, no)?

We are also still struggling with CMS (Medicare) regulations that, for example, for home vital signs to be “acceptable” for quality reporting, either the MA or the provider MUST view the actual blood pressure from the display of the machine over the video link, or view a printout from the machine, otherwise it “doesn’t count.” Hmm. I get why administrators want good data provenance (proof of authenticity), but isn’t telehealth hard enough? Why make it even harder for patients and docs? Who is going to be so motivated to PAY their co-pay for a telehealth visit, have that visit, and then LIE about their actual blood pressure reading at home, so that they “look good” for the doc, or the doc can “look good” for the regulators, payors? Ridiculous.

Nevertheless, our pandemic / telehealth story evolves. With the fall approaching, schools reopening, flu season coming, watch this space for what happens next.

Things are briefly, perhaps, not as dire as in March and April, in Colorado.

CMIO’s take? Telehealth was gangbusters in March, April, and is now settling down to 13-20% of total volume of clinic visits. We are back to 95% of original clinic volumes (in person plus telehealth together), so there are still some patients who haven’t returned to see us. And, although we have learned a lot, I think we still haven’t optimized “best practice” on when to use telehealth with patients. I think there are still some adjustments and opportunities out there. Let me know in the comments what you’re seeing!

Book Review: The Map that Changed the World

My favorite getaway: renting a place, bringing a bunch of books (or better yet, discovering new books in the rental home), cooking in the Instant Pot, hiking, and playing the ukulele. I have apparently infected at least one offspring with similar interest in reading.

This week’s book? A story about a man living in England in the 1800’s, who ultimately is credited with what is modern day geology.

What’s cool, is the book cover actually unfolds into a large reproduction of the original map.

William Smith’s life is entirely relatable. He learns math and measurement, and has a keen eye for observation. As a young man, he helps develop a coal mine, and notices that the layers of earth, hundreds and thousands of feet down, seem to have a recurring pattern. Furthermore, as he gains experience digging canals, he sees the same patterns laid out across the region of Bath, where he is employed. Over a number of years, he is employed to dig canals, drain swamps, marshes, farms, all the while collecting rocks, fossils and developing his theory of Stratification (a term he coins).

I note that I am entirely a geologic novice, and Permian and Cenozoic terms come and go without lodging in my brain. However, Smith catalogues and builds ideas, and eventually a map of Bath. He links various strata with geologic eras, with aspirations of mapping all of England.

The trouble is, he’s a working man, traveling and helping companies and individuals. Furthermore, he (imprudently) maintains two offices and a home, that he cannot afford. His marriage to a mentally ill woman does not help, and his ideas lay dormant and unpublished for far too long. He DOES publish a fabulous map, sells 400 copies at “7 guineas each”, a disappointing non-recognition of his thousands of miles of travel and careful analysis.

This map is a massive work: it lays out across England, the layers of rock, coal, sandstone, chalk, etc, and the sequence of layers hidden below. In future years, it will end up launching the coal industry, farming, minerology, and influencing Charles Darwin.

In his adult years, he is fraudulently scooped, with others publishing his work as their own. He is denied admittance to the Royal Geological Society, snubbed because he is an orphan, and not born to high society.

With accumulating debt from his properties and failed business, he ends up going to debtor’s prison, losing everything, and then starting over, nearly penniless, living in rentals and traveling to do survey work.

Based on some chance meetings, he ends up getting recognition for his original solo research and work, FIFTY years later, and is finally recognized and rewarded, in his old age, as the Father of Modern Geology.

I can’t help but think that, so much of our lives are happenstance:

  • Whom you meet and connect with
  • How random chance connects you with a job you flourish in, or fail miserably at
  • What the local culture (class-based snobbery, or open-minded scientific inquiry) encourages or prohibits
  • How you develop useful skills, and work hard
  • How you see that others may not; what do you do with that knowledge
  • How you personally persist past obstacles, or succumb to pressures

CMIO’s take? This could have been a story of any scientist, any informaticist, any CMIO. This could have been MY story. And that’s what the best books are about.