What should we be using tech for in healthcare?

Maybe the better question… Where should we be using tech in healthcare?

February’s post is here! The end of the month seemed to come quicker than usual. Oh well… On with the show!

I still have my link up for a survey about where clinical/patient care professionals want to see tech focused in healthcare, 5 short questions if you have a minute!

I’ve previously written about how many terrible ways technology is inserted into healthcare, and how clinicians need to be a bigger part of the design and integration. I think maybe the question to answer now is where in healthcare does technology belong? I decided to take a very long and winding road to describe some of the problems and opportunities I’ve experienced. Hopefully this gives a better understanding of why designing technology around healthcare is hard. But there is still a huge amount of opportunity. Kind of a long one this month, enjoy!

There has been a lot of hype about AI/ML/LLMs lately, and it is really cool, I’ve played around with a bit. I see a lot of places in healthcare that it can be really useful. As you read through, I’m sure you will too. The biggest opportunity in my mind is summarizing information that is scattered throughout patient charts. Not only for the clinical teams, but also for the patients.

Health is personal, for the patient and the clinician.

When we talk about health it’s personal. Deeply personal. It’s who we are. Whether that is physical or mental health, it’s about us as individuals. When seeking treatment we ask people we trust, this is especially true of chronic or serious conditions that require long term treatments. The opposing side of that relationship is also true, clinicians want to help, we want to know that we are making a difference in someone’s life using our knowledge and skills. To do that, we need to know the patient, we need the details to make informed recommendations and provide relevant guidance. The goal of technology systems should be to enable this, to enhance engagement from the patient and make the patient information readily accessible and relevant to make the best use of the clinician’s time.

Healthcare is becoming just another corporate machine, another cable company, another airline company. Does it matter where a doctor or nurse works? If we have to follow the same checklists no matter where we work, we might as well do it for the highest bidder. Or we might as well do it in another industry where we can’t see the impact of our robotic actions. That’s the trouble, we’re using digital tools to turn the clinician into a robot. At that point, we can be replaced by a computer. Maybe that’s the goal? But will that improve people’s health?

So, in writing, I decided to take a deep dive into a relatively straightforward procedure appointment. What happens? Where can we better use technologies to ease the burden on the patient and clinical staff. Even a relatively straight forward procedure has a number of variables that don’t lend themselves to easy decision trees, the trees start looking more like a bowl of spaghetti.

Duration and level of engagement matter.

The level of personal engagement can vary based on the type of relationship. If it’s a transactional relationship like an urgent care visit, the technology can be rather impersonal. It still needs to be easy to use and intuitive, and the information has to be accessible and readable to the clinical staff on the other side. Whether virtual or physical, the proper application of technology can be a huge time saver for staff. Virtual check-in, insurance, allergies, medications, brief history, chief complaint can all be entered in by the patient prior to the visit. Displaying this for the practitioner on the other end is relatively straight forward too.

As the duration of the relationship lengthens, the challenges in being able to update and share information increases. Going to a facility for a procedure can be a medium duration relationship (temporary chronic condition?), one that may or may not have a need for a lot of follow up appointments. Much of the intake can be done with technology, but many times the questions become more detailed and nuanced, something which does not lend itself to simply listing things. Here to though, the relationship is important, patients want to trust and know the people that they will be spending more time with and putting their safety in their hands. Anyone who has done a pre-op admission knows the myriad of checklists that need to be completed. It’s mind-numbing. In trying to streamline these it becomes important to pull relevant information from the history into the checklist, this may have been collected as recently as the pre-op phone call, or part of the larger medical history. This is a challenge, and it’s reflected in the way information is often collected and stored within the EHR. So what happens? We ask the questions again.

Have you ever wondered why you get asked the same questions over and over again? Sometimes it’s for safety (right patient, right procedure, correct limb, etc…), but more often than not the reason is that its faster for us to ask again than it is to find the information in the chart. The opportunity here, and where it has largely been missed, is in gathering relevant and meaningful information from the chart and displaying it a relevant and meaningful way, one that doesn’t require hours of scrolling. This is true for each care team member that the patient interacts with. If the care feels disconnected going from the receptionist, to the nurse, to the doctor, then how would you feel about the quality of care you’re receiving?

Taking a trip down the procedural pathway…

When we look at a procedural environment, where is tech most important to solve the problems and deepen the engagement? Let’s start with the one-off procedure, a patient comes for a procedure only once to solve a problem or have a test completed. Patient’s still want to feel they are being taken care of, that the care is personal and tailored. Clinical staff want to provide this, we want to provide good care, and an engaged patient that wants to get better.

First contact, the fax.

The initial point of contact is usually a referral for the procedure. The first thing is the surprising number of faxes still used to send the order or referral. Yes. Faxes. From practices with EHRs sometimes from practices with the same EHR. This still has not been solved! After the referral, is there a prior authorization needed? Now more back and forth with the referring physician.

What’s the next issue? In order to start the myriad of checklists and documentation behind the scenes that goes on for patients they have to be scheduled for a procedure. This is true of a referral from inside the system or outside the system. Why you may ask? Well… If we get all the documents before an actual procedure is scheduled they just get thrown into the EHR. Once a procedure is scheduled, we need to go back into the EHR and find where all those documents went to associate them with the scheduled visit. The best way to save time and make sure we have all the information we need is to create the procedure and associate all documents with that procedure. If we don’t, and we can’t find one of them, we ask the patient or referring again.

Sounds like a simple work around to just schedule a few months out. Except… We haven’t even talked to the patient yet. Here lies another problem, for patients that actually use their portal with that system, they are now scheduled for a procedure. It shows up on their end, and they may even get an alert. What does a patient do? Call or email, hopefully. Sometimes patients will just assume that the procedure is that day, and make arrangements. A day or so later (if everything is going well) they will get a call asking when they would like to schedule the procedure. Confusing? Yes. Bad patient experience? Yes. Or maybe it’s just an over-eager physician practice that wants to provide an exceptional patient experience, they see a procedure has been scheduled and start calling the patient. Except… Patient knows nothing about it, confusion reins supreme.

Patient’s that have had this experience before will either call or wait for a call to confirm a date and time for the procedure. Hopefully everyone in the system follows this process, and hopefully that’s been the experience of patient’s who have gone elsewhere. Most likely it’s not though. Everyone has customized their workflows to the work around of a process that has been customized for each area since EHRs went live. Makes perfect sense.

In the best case scenario the patient calls and schedules an appointment with a referral and pre-auth in hand (although… You need a procedure date for a pre-auth). The procedure can then be scheduled, and hopefully everything will get done before the procedure, usually that needs to be at least a week out, in the best case scenario.

Why can’t we just create a dummy procedure room that doesn’t appear anywhere online and schedule a patient? Simple enough right? It’s apparently not that easy. Even if it is done, there are a number of problems that we’ll run into, similar to just scheduling a procedure far out.

Either way, we get a scheduling team that is answering calls from referring practices and patients confused about when to schedule their patients, and fewer patients get scheduled due to the scheduling people answering questions about the ‘ghost’ scheduling. Following me yet?

Procedure date is set!

Now the patient is scheduled for the procedure they are set to go! New patients, whether new to the system or new to the individual area, what is needed? H&P within 30 days? Done. Fax it over. <<sigh>> Yes, the fax machine. Can it be transmitted electronically? Ideally yes, often it is, but for some reason that nobody can really explain, more often than not it is faxed, because nobody seems to have figured how to get it to the right spot in the EHR, so it ends up as a PDF in an endless list of files. Labs? Ask the patient where they want to go, submit the order for labs to that location. What if it was sent to the wrong location? Can’t another location just pull it up, they probably can, but have you ever tried to do this? Nobody at the lab location knows how to release the order. Write a new order is the easiest solution. Are the labs resulted? Patient says they had their labs drawn, go in the system and see if we can pull the results in.

At least in the same EHR system (meaning the same health system) we can pretty easily find an H&P. Score one for a closed ecosystem.

Pre-auth expired, too much time elapsed.

Call back the referring to get a new pre-auth.

The Pre-Op call.

A few days before the procedure, patient receives a phone call to review everything for the procedure. Almost there… Scheduling nurse reviews medications, checks history, allergies, reviews dietary restrictions, activity restrictions post-procedure, confirms they have a ride home. Goes through the whole check list and answers questions. Hopefully the patient has a pen and paper handy to write notes or questions. Even here there are a myriad of variables that can take us down a wandering pathway of questions and unique problems. Medications are one, availability to take time off work, availability of having someone available to bring you to/from the procedure, past medical history, chronic conditions that can impact anesthesia/sedation decision, the list goes on and on. People are still better at making these judgment calls than machines, but technology could help shorten these discussions. Hopefully everything is OK, nothing was missed, and the procedure can go on.

Procedure day is here!

Day of procedure, patient arrives. Nurse welcomes the patient, and starts asking all the same questions again. Why? Again, some of it is for safety, making sure medication instructions were followed, dietary restrictions, and they have a ride home. Assuming the nurse who called and the pre-op nurse have the same checklist, the answers can be front and center on the checklist. Why wouldn’t they have the same checklist you might ask? I have asked that question many times. It takes a lot of effort to fix that, trust me, I know, and it still confuses me. During the Q&A portion of the work-up, the patient states they stopped taking their blood thinner yesterday (that little orange-ish colored one), that’s what was instructed. Nurse looks in the chart, the check box is there that medication instructions were given, except that’s not what the usual protocol is, normally it’s stopped a week before. Who told the patient that? This is where we go to the notes section… It’s the most informative section in the entire chart… Except, there’s a lot of notes. If you’ve ever done a chart audit, you know that’s where the details are, but there’s a lot to go through. The nurse can spend the time scrolling through the notes and trying to find the answer, they can call the scheduling nurse who spoke with the patient (if they’re there that early in the morning), or they can page the doctor and just ask. Path of least resistance? Page the doctor. Who orders a new INR to be drawn, let them know the result, then asks who gave the instruction. <<Shrug>> Not sure, it’s probably in the notes.

So what actually happened? The scheduling nurse gave the correct instruction to the patient, but the patient stated that their doctor said they should not be stopping the warfarin. The scheduling nurse paged one of the procedural doctors, who called the primary doctor, who then called back, the procedural doctor then sent a message to the scheduling nurse about holding the medication the day before and drawing an INR the morning of. The scheduling nurse called the patient back, relayed this information, checked the box, then wrote a note. Why isn’t there a comment section with the check box? Good question, because the comment box would need to contain too many characters. Why not a hyperlink to the note? Because… We can’t allow staff to put hyperlinks all willy-nilly in charts! So, the information gets buried in the notes section. If it was actually the same procedural doctor who talked to the primary doctor, they most likely forgot, because it was a week or more ago.

Is this faster than scrolling through the notes? For the nurse it is, but taken as whole, it’s a huge time waste. The pre-op nurse has another patient coming and doesn’t have time to scroll through the notes, they can complete every other task while waiting for a response. The nurse is doing everything to make sure the procedure can be done safely for the patient, while trying to manage their time appropriately. The procedure start is now delayed by half an hour (at least, we hope). What’s the work around for this? A piece of paper. One for each patient with notes on it about specific things that don’t follow the norm. About 80% of patients will have one thing that doesn’t follow the ‘norm.’ On any given day it is not the same one thing for each patient. Why write it on a piece of paper? Because the charge nurse needs a concise run down of the patients for the day, any issues or notes that can impact their readiness, which can be passed on to the pre-op nurse. Scheduling screens that pull from the EHR can’t/won’t display that detail because of all the scattered places the information is located.

Post procedure. Spoiler alert, procedure went fine.

Patient had a successful procedure, everything went well. They had a great experience with the staff, they were frustrated with the repetition of information, but this was explained away that we double check everything for safety. This is mostly true. The physician finishes the procedure, if the patient is not in the same system, the physician will usually dictate a letter back to the referring with a copy of the procedure report. But wait… Isn’t the procedure report in the patient record, electronically stored for everyone’s convenience? It is. But… There can be a number of different ways this is stored in the EHR, but that doesn’t matter, it’s buried in there. The note is long, 3 pages at least. Full of, details… Lots of details… Details nobody but the insurance company and billing people really care about (everything went well, so we don’t care about the lawyers). The most important part of the note, the first couple of sentences, and the last paragraph. The first couple of sentences describe the patient, in most cases this is how we remember patients. Names if we’ve known them for a long time work, but we remember their story more than their name. Even when we speak to others it’s usually something along the lines of, this is Mrs. Smith, she’s a 67 year old female with intermittent chest pain and a positive stress test who you sent for a left heart cath, history of afib and stroke. That’s why, it’s how we remember the details of the person, the story is the trigger, not the name, not the MRN, the encounter number. That last paragraph describes the outcome and recommendations going forward, recovery, changes to medications and anything else the patient needs to do. The purpose of the letter is to provide an easier format to read, because nobody is really sure how the report is going to get transmitted (if at all), if the referring physician will even get a notification, or notice the notification. But they’ll read a letter. What about the referring doctor who’s in the same system? They might get a nice letter, or they’ll get a message that the procedure was done. At which point they may look through the EHR and read the first few sentences and the last paragraph.

Finally! Discharge!

30 pages of discharge instructions… That might be a slight exaggeration. It might not. In those discharge instructions there will be worthwhile things, there will also be a lot of fluff. If the discharge nurse is on top of their game, they will highlight the important things in the novella they just handed to the patient. In the move toward same day procedures, following instructions post-procedure is something that is pretty important since the patient is at home without immediate access to a medical professional. But most instructions end the same… Call your doctor. <<sigh>> 80% of discharge instructions are the same for a given procedure. Most are confusing though, we write them as best we can, but it’s hard to write something for those who don’t know and do it so that it is relevant, short and easy to read format, while also trying to describe variations that might occur. I know… Look how long it’s taken me to explain a procedure without explaining any procedure. Also, I’ve written standardized discharge instructions. It’s super hard. The part that varies from patient to patient, most often, is the medication. We send the patient off with some new prescriptions, a quick explanation of the new meds, and… Call the doctor if you have any questions. If there is extensive follow up required, like rehab, then at least the patient has the opportunity to ask questions at the appointment, hopefully relatively soon after the procedure. If not, and the patient doesn’t understand, then the patient may be non-compliant. It’s a great catch-all term that we have a check box for. The check box that captures none of the details or nuance of why, that’s probably buried in a bunch of different notes.

How do we follow-up with a patient from a same day discharge? Phone call usually. There are variations, text messaging is helpful, although if not done right, it’s fairly useless. Most patients will end up with a phone call though. Which is good, it gives an opportunity for open-ended questions.

If there is a lot of follow-up required, virtual appointments are great. As long as they don’t require an in-person physical exam, then they can provide a great opportunity to follow-up with a patient and not require them to come back in to the office. Here again is another challenge. Following up with the referring physician there can be pages and pages of notes that are largely useless, that they don’t really care about. The patient goes back to the primary care doctor, and they ask, “how’d it go?” “how are you coming with [rehab]?” “Why did they change this medication?” Seriously, it’s all in the pages of notes written for billing, didn’t you see it in there?! No they didn’t, it’s easier and quicker just to ask.

Opportunities to improve…

At every point along the interaction of scheduling and following up for a procedure there is an opportunity to utilize technology, not to replace the clinician, but to deepen their understanding of the patient, ask more pointed questions to save time, and encourage engagement from the patient.

Let’s start with scheduling.

There are opportunities to allow a patient to self-schedule, if we can get the referral/ordering done right… Do we allow patients to schedule then seek out referral? Or do we need a referral and then allow patients to schedule? More likely the latter. That’s challenging, there can be a lot of variables to this depending on the situation, but the opportunity exists. We can use an app for that! NO. I mean you can, but people won’t use it. Seriously, it’s a challenge just to get people to log on to existing apps and look at medications or upcoming appointments. How about an email with a link to the scheduling website? A QR code from the referring printed on their discharge instructions? Once that is done we can also use that to direct people to a pre-op checklist of activities they need to do. Pick where you want to get blood work, read through/update your medications, complete the STOP-BANG questionnaire, make sure you have a ride from the procedure, emergency contact info, contact info for a person the day of the procedure, and many other questions that can be answered ahead of time. A personal call is still needed, but pointed questions or details can be requested based on the form. The form also has to be easily accessible by the scheduling nurse too. Getting the information to the right person in a way that is in a human readable format.

Coordinating the teams…

Messaging doctors and other clinical staff. This needs a lot of improvement, just among the clinical teams. Thinking back to the coumadin example above. If the nurse writing the initial the message has the time, they will write something along the lines of… “Patient Barbara Smith, 64 y/o F, MRN 1234567, referred for a LHC d/t intermittent angina and +stress, on coumadin 5mg QD for 6 months, referring Dr. Jones wants her to continue w/o interruption h/o risk of afib & stroke, how do you want to proceed?” That’s a lot of typing, and often they’re in a rush to get to the next patient, so the detail is omitted. Why write all that to begin with? It gives context and meaning to the question that can prevent endless messages back and forth or an unnecessary phone call. It will also allow the physician to contact the referring immediately with the relevant context and meaning about the patient so they don’t have to look it up. Forward that message to the referring, with a suggestion on how to proceed and you’ll likely get a quicker response since the referring no longer has to look up the patient to jog their memory. It’s about who the patient is, not just their name. Why can’t there be a patient brief at the beginning of a message? Oh yeah… That Barbara Smith… Quick details, sentence or two, as well as a message string, or hyperlinks.

Day of procedure.

The poor pre-op nurse who is struggling to get the patient ready in the required amount of time, lest their manager come visit them and ask why it takes them 15 minutes longer to work up a patient than nurse Cindy. Yes this happens. I’ve been the one who it’s been done to, and despite swearing I would never do it, I’ve done it as a manager. We all turn into our parents eventually. Some nurses are better with computers than others. But even nurses that are proficient waste a lot of time seeking out details hidden away in the chart. Time that can be saved if we can get all that information to the pre-op nurse in a logical and meaningful way. Checklists are fine, but we still need the details of what happened before, especially when something doesn’t follow protocol that was known ahead of time. Just getting the basic information from the pre-op call to the pre-op nurse can save 10 minutes. Imagine the further time savings if we can get the details for those 80% of patients who have one thing that’s different from the “norm.”

Discharge instructions.

This. All of it. Trash! The biggest impact for LLMs in patient care could possibly be writing discharge instructions that are relevant and meaningful. When nurses have adequate time, they will explain the instructions in detail, write notes on them, highlight them, and make them relevant to the patient. This represents a huge opportunity for improvement, customized discharge instructions. Not only can it help patient satisfaction, but potentially improved outcomes, and decreased phone calls. There is also opportunity here to link for further information. New medication? QR code or website, links to information on the medication or maybe an instructional video for a lovenox injection. We could even go crazy and email a summary of links and resources to the patient in the pre-op phase so they can learn before, and have the information available after.

Follow-up Care.

I’ll talk about a short term follow-up, longer term follow-up like a rehab program typically looks more like managing a chronic condition. I’ll save that for next month, that’s another area that is rife for tech that supports the clinical staff and the patient as well, and an overly descriptive post that I happen to specialize in. For a one-off procedure, or even frequent procedures, something as simple as automated text messages can help focus questions for an in-person call that be much shorter and more relevant than just a call the day after. Going back to our patient, Ms. Smith, the 67 y/o F with a h/o of angina, +stress, who had an LHC the other day. Imagine a simple text string:

Text Bot: Good morning Ms. Smith, checking in after your procedure, did you have any bleeding from your procedure site over night?

Patient: No.

If yes, then that can be a new text string or, “The NP will give you a call shortly at this number to ask some further questions.” Why not ask for more details? You’re likely to get a wide variety of responses, that may not fall easily into an algorithm. It can be more quickly resolved with a phone call than it can with an endless text string asking ever more detailed questions until it figures out what to do.

Text Bot: Great! Are you having any severe pain? On a scale of 1 – 10, how would you rate it?

Patient: 3.

Text Bot: Sorry to hear that, have you been taking your prescribed [Tylenol]?

Patient: Yes.

Established pain ranges are helpful in determining the next course, but often there is more detail needed. If the pain is severe, there might be a problem that needs more in depth information, but a higher rating can just lead to a response that an NP will call the patient.

Text Bot: Would you like to talk to an NP about your pain level?

Patient: No.

Text Bot: OK, couple more questions. Were you able to fill your Lovenox Prescription?

Patient: Yes.

A no response should illicit a call from the NP to find out why, and make sure the patient is aware of how important the blood thinner is.

Text Bot: That’s good, do you have questions about how or when to inject it?

Patient: Yes.

Text Bot: Here’s a link to a video about Lovenox and step by step instructions on proper injection technique. https://www.lovenox.com/patient-self-injection-video We will also have our NP contact you shortly to answer any other questions you may have.

The final message can still be that an NP will do a follow-up phone call, but now they have better information to guide the conversation and make sure the patient understands what to do next. Most complications from “non-compliance” arise from non-understanding, not an unwillingness to comply. Automated messages can be useful, especially if follow-up is needed for several days in a row. But… There’s always a but… They need to be relatively short and lend themselves to simple response (yes/no, 1-10) otherwise the variation gets tricky. Limiting the amount of texts is also important, too many and it seems like I’m interacting with a robot, the questions may be just generic enough to be suspicious. It would be something akin to the endless phone menus when you call customer support. You know, the one where you start screaming “REPRESENTATIVE!” into the phone.

This is it! Almost done!

There are many other opportunities to utilize different technologies to provide a better more engaging patient experience. There are many other opportunities to provide better information for the clinical teams, to provide better sharing of information among the clinical teams. Why were they so disconnected in the first place? Nobody spent time to understand the workflows, why we did what we did. We were given 2 hours with the EHR teams, and that was it. No collaboration.

At least the billing gets done more efficiently.

What other ways can you see a better utilization of technology in this example? In your practice? In your experience as a patient? There are so many opportunities that are squandered by poor execution and a failure to understand the clinical environment. You can create a tech-enabled healthcare environment that clinical staff and patients will use, but it has to be tech-enabled, not tech-replaced.

Thanks for reading and taking a trip down procedural scheduling hell with me! Wasn’t that fun?! As a manager it’s only one of the many challenges you’ll get to tackle… I should probably write recruiting posters instead.

Feel free to comment and/or connect with me on LinkedIn!