番茄社区app

Give us enough resources, and the right resources, and we can do just about anything -- and just about everything.

Recently, a colleague of mine who is a super specialized subspecialist announced they were retiring, and a number of patients sent messages through the portal asking me who is going to take over the care of the medical problem that this colleague had been managing.

First of all, I know all of the other subspecialists in our institution who manage these types of medical problems, and all of them are overextended and have no capacity to take on new patients. Secondly, for the most part, these were stable patients who were on one or two medications for a single chronic medical problem, and they had somehow ended up with this specialist.

So, who should be managing these patients? Um, me?

In an ideal world, the general internist and primary care physicians in other fields should be managing the bread-and-butter problems that fall within our purview. Everything from diabetes, to hypertension, to heart failure, to asthma and COPD, to depression and anxiety, to osteoarthritis and osteoporosis -- all of this should be ours. Unfortunately, in the modern healthcare system, patients either elect to see a subspecialist on their own, or their primary care doctors are too busy to add on yet another problem to manage.

Each new medical complaint, each new symptom, each new medical diagnosis, and each new medication comes with a lot of effort and an expenditure of time -- going through evaluations, trying different medications, doing follow-up appointments, answering questions on the portal, interpreting remote patient monitoring -- that add to the workload of the already overburdened primary care physician.

So, after a doctor has already dealt with five, six, 10, or 12 problems in a visit and they have their hand on the doorknob, if the patient says, "Oh yes, I forgot to tell you: I've been having these terrible headaches," I can't blame that doctor for making a referral. At that point, it takes 30 seconds to put in an order for a neurology consult, versus another 30 minutes to do a decent job of truly evaluating someone with a new headache syndrome.

But if we built an equitable system of healthcare, where we had the resources commensurate with the amount of care we are providing, we could probably do a lot more. We'd be able to manage more chronic medical problems, and have the bandwidth for acute issues as well, whenever they arise. If we had the support we needed -- be it social workers, therapists, nurses, technicians, and other members of the team -- we could accomplish pre-visit planning, scheduling routine visits, making sure all the healthcare maintenance items were up to date, and ensuring adequate and timely follow-up.

Lately, we've been hearing about efforts to bring artificial intelligence tools to bear in medicine, and it's starting to trickle down to primary care. At this point, we've been testing some systems for ambient listening and artificial intelligence (AI) using voice recognition to capture our office visits and generate a complete medical record note. And soon, new technology tools are coming to take a first pass at replying to patient portal messages, which everyone seems to think is going to save us a lot of time and free us up to manage our patients' in-office clinical issues.

While I've written before about the pitfalls and shortcomings of the systems that generate notes by listening in, the portal message "first responses" are newer to us and are going to be available in the very near future. In taking a look at how they work, it seems that some AI system reads the patient message, and offers up to the provider a suggested response. Right now, these systems are specifically designed not to be providing anything remotely resembling medical advice.

Studies have shown that these systems are more empathic in the wording used to craft a message reply than an overworked, burned-out primary care doctor who at the end of their day finds themselves with 20 or 30 or more messages from patients, each of which may contain multiple items that the patient wants addressed in a "quick and easy" email response from their medical provider.

I think that if we really want to free us up, let these systems take a first pass at coming up with some actual medical advice, and then present that to the doctor to judge whether this is a safe and reasonable response. After we've edited for content and medical soundness, then let the AI system loose to make it more empathic.

If the system gets a message from a patient that says, "My knee hurts; what should I do about it?" and all the system does for me is say, "Gee, I'm sorry to hear that you have knee pain. There are lots of things that could cause knee pain; you should probably schedule an appointment to see your doctor," then they really have not saved me much time, effort, or intellectual bandwidth.

Maybe we need to wait until these systems are more advanced, connected to a patient's chart and aware of their past medical history and recent labs, and then they might be able to review a photo and take a stab at, for example, what that rash might be. Shingles? Poison ivy? Lupus? Measles?

Free me up on the front end, free me up on the back end, and make my visits more efficient with all the resources we need, and we'll be able to manage almost everything, freeing up our super subspecialized specialist colleagues for the cases that really require their services.

Comment