April 13, 2020

How Will Doctors Make Decisions at the Peak of Covid-19?

How Will Doctors Make Decisions at the Peak of Covid-19?

The world has always expected more than what is possible from doctors. Now more than ever, we are asking them to go all in. Doctors all over the world are the ones who still go to work every single day, putting themselves at risk, not getting paid any better, and perhaps most critically - are being asked to make life and death decisions which will no doubt take its emotional toll on an occupation that is already infamous for burnout rates and mental turmoil. More about this in my pre-covid post here.

Interested in joining the conversation? Drop us an email at omar@asha.health with your background and interest.

As we approach the peak of the pandemic, we are asking ourselves on daily basis: How do we support doctors to make more informed and efficient decisions in face of a tiny body of evidence for Covid-19? What information do they need that can help them at the point of care?

A matter of life and death..

When we speak about decision making and outcomes, the next few weeks will sadly come down to  this:

  • Which of these patients should I give a ventilator to?
  • How can I improve my outcomes in the next wave of patients?
  • How can I increase my preparedness in terms of staff and equipment?

Why do we need a better way for clinical decisions?

About 90 cents of every healthcare dollar spent is still tied back to a decision made by physicians. This has not been an optimal way to go about deciding what patients need and the amount of waste is US healthcare is jaw dropping. This has resulted in a war of decisions between payers and providers which I found myself right at the middle of, by being a Senior Clinical Data Consultant for some of the biggest payers.

We were building Fraud Waste and Abuse AI programs while physicians were buying smart billing solutions that allowed them to outsmart our tools. It took me a few months to realize how futile and incredulous this whole method of clinical decision making was. How would this change if CMS and payers started informing doctors rather than holding a stick for each extra penny that they try to squeeze? It became clear to me that value based care will always be some kind of dream unless this kind of collaboration between payers and providers became a reality first. Human-driven innovation was the only way forward and in the current crises, this can't be more true.

We could go into a long philosophical debate about bias and whether computers will ever make better decisions than doctors and vice versa. While this may provide us with a satisfying intellectual banter, it is not a useful discussion for frontline doctors.

The real question here is this: What information can machines provide doctors with RIGHT NOW to help them make better informed decisions? Equally important is WHEN and HOW should this information be presented?

Here is a break down of how next generation clinical decision may look like:

Clinical Decision Factor Old World Next Gen
Evidence Based Guidlines PDFs and E-texts Robust & User Friendly
Protocols Generic One-Size-Fits-All Contextualized for both provider and patient
Feedback Non Existent (Only Rejected Claims) Dynamic feedback based on outcomes
Documentation Does not create a patient narrative Automatically Generated and Complete History
Likliehood of Improvement Based on RCT Based on what is working for similar patients

Why aren't the good ol' ways of clinical decision making not effective with Covid-19?

Very little evidence - We are in a situation where the evidence is very rudimentary about which drugs work and which symptoms are important. Our ability to predict (both human and machine) is still very rudimentary in face of how little we know. Does Hyroxycholorquinone work? How long after contraction of COVID-19 does Redemisvir still have a good outcome? Should ARBs and ACE Inhibitors be discontinued due to the activation of ACE2 receptors which the SARS-Cov2 virus work on?

The truth is we know very little about this disease and what works for it. The Infectious Disease Society of America guidelines are still all rated at "very weak". We do nee to act now though. So how do we go about this?

Confusing Treatment Thresholds - After a diagnosis has been established, clinicians have to decide who gets treatment and who doesn't based on the likelihood of them improving if they get the treatment or deteriorating if they don't. This get's complicated with CoVid due to our potential resource allocation problem (we addressed this on a previous blog here).

Our Biases are Screaming - As doctors, we feel much better by coming up with a treatment plan and narrowing down our focus  early on a specific disease. It simply feels very satisfying to feel like we have the answer. With the current focus on Corona virus we can't neglect the fact that our bias is very loud and we will tend to think of Covid-19 before any other condition. A real problem rarely mentioned on the news but many of our physicians are voicing is the lack of treatment that patients with other diseases may not get.  Many of them also refrain from going to the hospital due to the fear of contracting Covid-19.

When do we test for covid-19 is a huge moral and logistical dilemma as well. Who needs testing? Should we stick to the highly vulnerable populations only (such as the first responders, healthcare workers, the homeless) or do we take a #BoxItIn approach of testing widely and as frequently as possible?

Intuition and Empathy - We still do not have enough gut feeling about this. Our empathy will still work but our understanding of who is likely to improve with treatment or not is still developing and we need to be aware of that. What happens when an 89 year old patient with heart disease is a confirmed case but so is a 45 years old that is otherwise healthy. What is the likelihood of each surviving the therapy? We need an objective measure to be able to say whether this patient is improving or not and what is the likelihood of them benefiting?

What Would Next Generation Decision Making Tools Look like:

The tools we are building aim to provide the following:

I - Guidelines that are robust: Decisions will be supported by contextual and easy to use guidelines. References from the latest research should be easy to adopt and is pulled up based on the clinical data logged for the patient in front of me. Reading an e-text and having to scroll down  and figure out what applies to my patient is a waste of the doctor's time.  

This robustness has to extend beyond the patient's case, to also consider to the providers knowledge and the size of the facility. This is becoming crucial as more Nurse Practitioners and Physician Assistants join the workforce with various levels of education that is sometimes completely driven by the physician they are working with. Different providers require different kinds of support.

II - Feedback and Information on Past Decisions: Following evidence-based guidelines is important but equally so is following a doctors gut. How can we reverse engineer the guidelines to have the doctor's immediate experience factor back into them?  

Next Generation decision tools will capture a doctor's decisions and provide feedback on how good the outcomes were. What if physicians are presented with data about the likelihood of survival for a patient based on latest published research to help them make that difficult ICU decision and save more lives? What if the system can provide insights to a doctor in California about what has worked for similar patients in New York? These are the kind of informed decisions that will be key to our survival and healthcare improvement.

III. Efficiency, Standardized Documentation & Interoperability: Some doctors want to learn and improve, and others don't. What NO one wants to do though, is be redundant and be asked the same question over and over. This is true for both patient and provider. We envision this clinical decision tool to sit on top of existing EMRs (with a separate user friendly interface that is built for doctors not payers). These tools won't work without this final piece of integration being available. Doctors are already overworked, and frontline physicians need to be faster than ever.

Read more about this in our earlier blog post here.

Why US Health Systems have not been prepared for this pandemic

Health systems have been pushing for improved quality "measures" rather than actually quality and pushing for billing guidelines over clinical support with doctor-friendly tech design. This has stalled our panacea of a real value-based care and Instead we're still doing various forms of fee-for-service.

We are now seeing health systems in New York finally create command centers that are tracking everything from staffing, to PPEs and expected demand of ICU and ventilators. Many are doing a great job and being miraculously resourceful with their frontline staff. Now is the time to put the right tools in the hands of doctors, physician assistants and nurse practitioners to help them rather then penalize them.

What if health systems and payers have been partnering with doctors and had proper decision making feedback loops in place to keep doctors incentivized to make better decisions not just cheaper/more expensive ones?

If this is to truly become our finest hour, then we need to be open to doing things differently, and who knows maybe we'll finally be able to improve the healthcare system.

How do these align to your interests and needs?


Interested in joining the conversation? Drop us an email at omar@asha.health with your background and interest.