Digital Health Ops: A Playbook for Every Growth Stage | Rahul Agarwal, COO Medplum

On this episode of Ops I Did it Again, Rahul Agarwal (COO of Medplum) joins Danielle to share his digital health operations playbook. The “hero’s journey” playbook breaks down pro tips and gotchas throughout every stage of scale: from pre-seed all the way to scaling to 50 states.
Hosted By:
Danielle Poreh
Rahul Agarwal

Show Notes

On this episode of Ops I Did it Again, Rahul Agarwal (COO of Medplum) joins Danielle to share his digital health operations playbook. The “hero’s journey” playbook breaks down pro tips and gotchas throughout every stage of scale: from pre-seed all the way to scaling to 50 states.

If you are building a clinical workforce, scaling your digital health operations or looking into the future on how you should be building, this episode will give you several tactical tips to implement.

This episode is sponsored by Out of Pocket, because no one is prouder than us than us:

To register for the upcoming Healthcare Call Center 101 crash course course visit:; Use code: ANSWERS for $100 off; Next cohort starts 4/16- 5/2

To register for the upcoming Healthcare 101 crash course course visit:; Use code: IBELIEVEINME for $100 off; Next cohort starts 4/23-5/9


Danielle Poreh (

Nikhil Krishnan (twitter:


Rahul Agarwal (


(00:00) Intro

(02:23) Rahul explains his playbook

(02:58)Tips for early stage operators

(06:32) Leaving the jungle: finding product-market fit

(08:55) Operational playbook for scaling digital health services

(13:45)  Courses by Out Of Pocket!

(15:20) How to define encounters

(24:10) How to develop care plans/clinical pathways

(30:59) Optimizing provider recruiting and workload with EHR design

(33:30) Standardize your metrics and avoid metric soup

(35:50) Dos and don’ts when scaling to 50 states

(39:43) Practical experiments for orgs to try

Podcast Transcript

Danielle: [00:00:00] All right. I am rolling solo on this pod. Nikhil's busy enlightening everybody on healthcare 101 and all the inner workings of it, but we're going into ops like 205. And in this episode, I'm joined by Rahul, who is the COO of Medplum. He's also a fellow nerd and operator, but in addition to how he thinks through problems is that he comes from the software development world as an engineer. So we've got like this engineer turned operator coming in and he has seen a ton of builds of EHRs. Now he's working on an EHR itself, but this episode is all about. Helping you see around corners at different stages of growth and things you can tactically do today as you're scaling through the early stages all the way through like mega hyper scale to make sure that your systems like on the back end and the [00:01:00] way you're thinking through operational decision making is sound and sets you up for scale.

Danielle: So this is gonna be a fun one, a little bit different of an episode. Where we'll be learning Rahul's playbook, we'll also be getting into how you can scale through all 50 states, which I think so many different companies are trying to get through. And of course, we'll top off with an experiment section for everyone.

Danielle: So, Rahul, thank you for being on the podcast today. 

Rahul: I'm excited to be here. How are you feeling? I'm great. I'm great. I'm nervous, but I'm excited to share what I can. 

Danielle: Tell us a little bit about this playbook that you've written up and why you think it's important and who should, who's going to benefit from learning all about this playbook. 

Rahul: Yeah, so at Medplum, like you said, we make headless EHR platform, and as such, I've had the benefit of working with a lot of digital health companies kind of on their journey from startup to scale up to serving tens of thousands, hundreds of thousands of patients. In the course of just working with people, we saw some common patterns. We saw. [00:02:00] How even though everyone is tackling a different clinical niche, you know, some people are doing pediatric mental health. Some folks are doing men's health online. Some folks are doing maternal fetal medicine. Really, there's surprisingly common set of challenges that everyone faces in the scaling and operations side of the business. So as we work with customers, we've developed this mental playbook. Want to share that with your audience. 

Danielle: What are like the fundamental building blocks of this playbook or like, what are those common problems that you've seen? 

Rahul: The way I broke down this playbook is by stage, I call it the digital health heroes journey, right? You're starting off in the jungle. You're trying to learn about your patients. You're really, you're passionate about a space and then you're moving all the way up to like the grand finale. You've got a 50 state, 2 million. Patient practice and you're in employee benefits programs. You're in value based care. You're building Medicaid and Medicare Not just you know cash pay for a certain patient population. That's kind of the bookends of where we want to go.

Danielle: All right. Let's zoom in then on the [00:03:00] jungle stage.

Rahul: Yeah.

Danielle: Let's go there.

Rahul: So you get dropped off. You're in the middle of the jungle. You're like, I really care about this problem. Let's take, for example, last trimester, the third and fourth trimester health, maternal health, for example, third trimester's what we call before the baby's delivered, fourth trimester, first month of life. This is an important problem. It's underserved. And you want to support your patients through that. At this point, you're optimizing on learning. learning about the patient population, what kind of things you can actually serve digitally. What the mix of in person, digital, synchronous, asynchronous care is, this is very specific to your company. My playbook at this stage doesn't have a lot, except for one. Specific piece of guidance, which is to familiarize yourself with the standard terminology that the health system uses. And when I say terminology here, I'm not just saying like the lingo or the jargon. I'm specifically talking about like the way different practices interact with each other via code systems you know, ICD [00:04:00] 10, CPT, NDC for drugs. These kind of, it's alphabet soup. But really, the point of this is, this is not something you want to reinvent the wheel on here. And if you Started thinking in these dimensions up front, you're not gonna have to redo it when you have thousands of patients. 

Danielle: This one, to me, in a way, seems It's obvious only because I come from like a very deep medical device background where codes like ruled my undergraduate education, but I know folks have sometimes right. They start with the problem and then they sort of try to backtrack into codes that exist and or start cash pay and then later on realized there was no formal infrastructure of coding. I guess like in your world and. In mind, as I interpret this, is that the point of this exercise is like, no, where you're going to plug into and make money over time? Whether there is an existing revenue stream for you to tap into 

Rahul: exactly like right now, you'll probably be doing [00:05:00] cash pay. So it's not really about billing, but it's about just like learning to speak that language, building that muscle in your company of like when I offer a new service line, like I should go look up what the codes are like medical billing and medical coding is complicated. You can get a two year associate's degree in actually just. Medical billing and coding, which is about how do you associate what you did? you're not going to get it right on the first try. And even the definition of right changes as health payers change their reimbursements and stuff. It's really about not getting in the, what we call the bad habit of. Making up your own things. You can either say if someone came in for a rash or they came in for like, I'm making up a code here. Sorry. One three red spot on the bump. building that practice of specificity. Again, my playbook here, we're not at scale right now we're at learning about our patients, but this is one core element that I want. people to roll into their learning. 

Danielle: Tactically beyond diagnostic codes, what other billing lingo or general lingo should you be keeping an [00:06:00] eye out at that stage? 

Rahul: Diagnostic codes was ICD 10 procedure codes, which is CPT. There's a lot of different kinds of codes out there, but in the US. at, when you get to billing, you're going to need CPT anyway. For most people starting off, it'll probably just be the visit code. So 99215, which is like a 45 minute visit, it's not rocket science and they're pretty general purpose there. And then drugs, if you're prescribing and LOINC for, if you're doing labs.

Danielle: So take, Take us to the next step in the journey, the heroine's journey of leaving the jungle. What does that look like? Who's there? 

Rahul: So this is kind of leaving the jungle, you're starting to see some signs of product market fit, which services, which care models resonate with patients. Maybe you've dropped some things that you thought were going to work, but didn't maybe beyond your, just your. Chief medical officer, you've started to hire some other clinicians and MD, maybe some nurse practitioners and stuff too, or like, it's starting to feel like a real practice, not just like one doctor and some entrepreneurs doing [00:07:00] something together.

Danielle: Shit is like starting to go on fire. So, you know, something's working. All right. 

Rahul: Exactly. And this is probably the time you're like, "Oh, we should build an app," 

Danielle: Red flag! 

Rahul: Like spreadsheets, it's just getting a little harder, and we can't just do everything like super manually or we, we need run books, right? We can't just like know the right thing to do in their head. This is kind of where we're at here or 

Danielle: playbookizing. 

Rahul: Yeah. And in that part, I think, the big thing I've seen some folks get upside down is okay, I mentioned the apps as kind of a joke, but really it's, like, okay, patients love us and they love our little mobile app where instead of using MyChart or just phone call system, they can just do everything digitally. My big kind of announcement to them is the app is not the product. The app is a service to your product The The product is the operation itself, that patient care, the turnaround time with getting an appointment, the the level of followup that people get, that is the actual product you're selling. The app is just an enabler, just like the EHR is an enabler. All these things are just tools for that patient [00:08:00] experience. 

Danielle: Do people just get really excited from like a product perspective or like a marketing perspective around just like the facade of it and what it, what it delivers versus they under, they under invest in the upside of it?

Danielle: Is that kind of the overarching thing you see? 

Rahul: Yeah, 

Rahul: I think, you know, I, the brand is important, but we see a lot of folks get like turned upside down on the brand, especially cause some of this stuff is very consumer facing, right? Like you're saying, for sure, focusing on this very specific clinical niche cash pay. So I'm basically in a consumer business, right? Brand is how I stand out. 

Danielle: Hmm. 

Rahul: And. But brand is a very all encompassing idea. People can get locked into, like, how does it look? What is the number of clicks? Which are important things, but these are all in service of the end to end patient experience between when they find you and when they get care.

Danielle: Yeah, yeah. So what is the ops playbook for behind the scenes, making sure you're [00:09:00] executing right there? 

Rahul: So your spreadsheet process is going to start getting messy, right? You've got more patients to keep track of. You've got people who weren't in your founding team that need to know to do the right thing and not have stuff fall through the cracks, kind of a key piece of advice here is one, start figuring out your care models. The head versus the long tail of what you offer, with clinicians, you can't tell them necessarily like you only diagnose these two things coming out of a patient visit, right? They're going to diagnose what they see. You can't say like, "Oh, only prescribe these two drugs," especially in something like uh, I'll go back to like, pediatrics as an example, right? It's like whatever the kid needs, you're going to have to prescribe it. but you kind of want to know it. What is like your 90th percentile of reasons people are coming to you? And really, like, consolidate that versus, and then leave the long tail open, but just put it in the other category. So I'll give some examples there. One of our customers, Summer Health, they're a fantastic team. I think you had them on the pod as well. And they do and I'm also a customer of them [00:10:00] as they are also a customer of me. I love their service. So they are for non acute, non-emergency. Pediatric care over text. I can text them anytime and say, my daughter's eyes got puffy. Does she have an egg allergy? Can you prescribe an EpiPen? I've actually done that real world example, but they kind of know 80 percent of the things that are going to see across the door are a rash. Pig guy is like a common one. So just like they make that its own category fever, cough, a couple other things. They optimize around those, and they have another category too of like other reasons people come through the door, but they optimize their service offering around those common things. So at this stage, it's like, we're going to figure out what are the main reasons people come through the door. This is like our service menu so that we can then put it on rails. We can say like, "what's the standard playbook for runny nose, for everyone?" It's going to serve us better in that next stage when we're dealing with thousands of patients and like a whole, like a team of clinicians that doesn't even know each other, like [00:11:00] they're just logging in. But this is a stage where you have the leeway to actually do the analytics and to know your customers.

Rahul: Another one is we have a customer that does at home diagnostics for men health. They don't do like 100 different blood tests. They do two or three: a testosterone, a metabolic panel, a couple other things in service of their downstream things. And because of that, they can optimize their processes around like, if it's a testosterone panel, these three things need to make sure that they always happen. Like, it's this visit, there's the follow up. this is the interval of when we do a referral or when we do a check in. Things like that. If you have a hundred different items in your service menu, everyone's kind of a, an artist and you'll not get consistency across your clinician population and your patients aren't guaranteed to get the same experience when they sign up 

Danielle: and so this is like pretty much brute force around seeing what is the 90 percent and then behind the scenes getting abundantly clear on those, let's say [00:12:00] five ICD 10 codes and. Mapping out behind the scenes, like everything from like encounter all the way through, finishing up that encounter and super clear what needs to be documented, what intervals you have to follow up in, what is the care protocol and so on and that's sort of the recommendation there. 

Rahul: Yeah, exactly. So I would say, like, you can look at what your clinicians are documenting now at this stage and just like tabulate them, do like group by count. What, what are the ICD 10 codes that they are? Doing and see where there's like potential overlap. are some doing Thirty minute visits some doing 45 minute visits, maybe let's just like combine them. We're all just doing the 45 minute visit. Yeah. Cause like if you, if you go to the straight analytics, sometimes you'll see like, Oh shoot, there's a hundred different codes. But like if people are all just making their own judgments, a lot of them. Talk to the same underlying problem. We just made slightly different choices.

Rahul: So what I'm talking about here is just try to consolidate it. So it's not just about doing less for your patients. It's really about [00:13:00] consolidating what you're doing and try to find the boundaries there. Tactically, look at everything. Something's billing, group by count it, figure out how many things are in the 90th percentile... if you just sort by most common. You want to target up to 10, if you're like, you know, obviously, if you're 11, I'm not going to give you a hard time.

Rahul: If you're like 20...

Danielle: Rahul's gonna find you, you have 11 service menu items!

Rahul: I know where you live! I'm used to HIE's find you. Just kidding! But yeah, if you've got like 20, like try to find where it. Those are, or, you know, you can come back and tell me, like, "we have 20 conditions that we treat. This is core to our business. This is what we do." That's fine. It's really about just knowing and doing that intentionally rather than Everyone's kind of doing their own thing because we weren't looking

Rahul: The following message is brought to you by out of pocket people who have been tasked with building a call center have many questions How did I get this job? What are my options? How will this affect my loved ones? [00:14:00] Will I ever be happy again?

Rahul: You need answers That's why we made a course led by call center veterans who've been there and recovered join nearly 100 others who have not learned the hard way To enroll, call toll free today to reach no one, or go to outofpocket. health. Hear from Stephen, a student who said, quote, This course was hugely validating and packed in many invaluable insights that we would have otherwise had to learn the hard way.

Rahul: Five out of five stars. You can be Stephen. Save money. Start learning. Enroll today for 100 off your seat. Use code answers. That's code answers for 100 off.

Nikhil: This episode of OBS I Did It Again is brought to you by me. Because no one believes in me more than me. I'm teaching a healthcare 101 crash course that starts soon. Uh, you can find it on the out of pocket website at [00:15:00] outofpocket. health. We go over all the major stakeholders in healthcare, payers, providers, pharmacy, all those in between acronyms that you've been too scared to ask people what they mean.

Nikhil: We'll talk about how the money flows, major laws in the industry. It's the fastest way to get up to speed on U. S. healthcare. You can use the code IBELIEVEINME to get 100 off.

Danielle: Are there other guidance points before we jump into the big scale? 

Rahul: Yes, the other big one is figuring out what is an encounter for you. You know, In the beginning, when we were in the jungle, we're just trying to treat people. We haven't defined the clear boundaries of like. This was one encounter. This was another. Maybe I called this patient three times a day because I like forgot to ask him a question. Well, that's fine. Now we're starting to get a little bit more formal and especially in digital health models, it can be hard to define like what is one encounter versus two like if we, you and I are just texting and I go to sleep and then I continue the text thread in the morning, like when did that encounter start and stop?

Rahul: In Legacy Healthcare, when you're walking into the four [00:16:00] walls of the clinic, an encounter is very easy. It's like when you walked in and when you walked out. the rest of the healthcare ecosystem is still works around that model. It's like, you know, you bill per encounter, you kind of aggregate counters into what we call like a case or an episode of care. There's kind of analytics on an encounter, like how long did the encounter take how many encounters did it take to get to a certain health outcome? And it can be hard to map those to the digital world. 

Danielle: Yeah. In my head. Yeah. The pretty picture of like door in and door out, and then that's like on the left and on the right is like this maze where it has the home and then your office and everything in between, which is where the digital health world lives.

Danielle: And so what's the start and what's the stop there? 

Rahul: The whole point of doing this was that you don't have to go into the clinic, you don't have to walk in and walk out. So we are generally adverse to putting these like, artificial boundaries on things. But my point here is like, you gotta learn how to talk like the rest of the U. S. healthcare system. You're gonna have a much easier time if you've made some decisions around this. Even if [00:17:00] they're not like perfect just being able to package your interactions and being able to share that with your paying partners. When you do like, analytics and clinical quality management. So you're getting certain certifications like clear or something like that, knowing where these encounters were. Being able to tie things to that for instance, like, was this diagnosis made during a certain encounter? Like, which encounter was this diagnosis made where I identified that Rahul has an allergy to, pistachios.

Danielle: There are a lot of allergies going on in your home. There's eggs and pistachios. 

Rahul: It's on my mind. my daughter's at the age there's a, there was this landmark study for allergens where they said, if you introduce. Allergens to six month olds, they are much less likely to be allergic to that thing than life.

Rahul: So now all the yuppie parents are just loading up all their kids with peanut butters, tree nuts.

Rahul: They actually make these powders with just all the allergens that you just sprinkle into their oatmeal. So my daughter's eight months and we're just hardcore on like the aller we're just putting everything into her.

Rahul: We're not even feeding her. We're just mainlining her allergenic foods. [00:18:00] 

Danielle: How is she doing? 

Rahul: She's doing fine, but we did discover she had an egg allergy, which triggered some of those other things that I was alluding to. 

Danielle: Right. That's why you were texting. 

Rahul: That's why I was texting. Summer Health had my back.

Rahul: It's also kind of scary because until you know they're not allergic, they could, the next thing you introduce, they could go into anaphylactic shock. So you're like, hopefully this is okay. 

Danielle: Yeah, that's terrifying. What, what would be the encounter? Let's do that one as an example then. Like what was the encounter start and what was the encounter end of the, of the egg allergy saga?

Rahul: I worked with Summer Health on this when they were designing their system. 

Danielle: you're like, answer me faster. 

Rahul: Well, no, well, if I may answer, like, there's like bad. Okay. And like, The ideal way of doing it. Yeah, the bad way of doing it is you just don't have encounters. And like we, cause I'm texting them, right? So we just have one text that I did between me and Summer Health. That's how it looks on my phone. 

Danielle: me and Nikhil's level of, of how much we measure our encounters. It's [00:19:00] just a long ass text threads about everything. 

Rahul: Just one big, glorious encounter. strong advice not to do that. And that was like, one thing it was like, it wasn't obvious when we started working together, it's like. Oh, we need to like break this up. The okay way to do it is like to use some kind of time-based rule. Like patient has responded in two hours. Like the encounter is over next time they text me. That's a new encounter. great It's passable. We've talked about what happens if they go to sleep, it's like, you're going to be time zone aware. Whose time zone is, is it the clinician's time zone? Is it the patient's time zone? Is it both? Cause one of them might be asleep at any time. What if. They were dealing with the actual anaphylactic shock and like they just didn't have time to text. So it's, it's something, but it's not the ideal.

Rahul: The ideal is actually you just give the physician control. You say like, I am starting an encounter now. And I'm not closing it out until like, I feel like this has been resolved, which kind of feels like a step backwards sometimes we're like, Hey, this is manual, but in our experience, we found that like, there's no automated rule that isn't either crazy [00:20:00] complicated and you're using, like, I don't even think chat GPT can like get enough nuance on this, like, no, like we're not talking about a different thing.

Rahul: Also people interleave conversations. I might. Start a new thing with Summer Health about a different thing and then circle back to the old thing. So it's really using clinician's judgment of like, we are done talking about this topic. I'm closing out this encounter and I'm like going to write up the note for it. That's the way to do it. 

Danielle: Do you ever run into like forgetting though, if it's too long, so what do you do in that case?


Rahul: You ping both. So like, are you done? And eventually, even with Summer Health, clinicians will close it if I haven't responded in a while, but it's like, it is their judgment. And they've decided we've come to a resolution or they can choose, Hey, like this was not resolved.

Rahul: Can I keep helping you? And they will keep reaching out to me rather than it just being an automated thing. 

Danielle: Does that happen often where things just stay open?

Rahul: It does, especially at scale, when you're talking to, you're managing like a hundred clinicians or someone like at some point someone will have forgotten to close it, and that there are definitely not just like, "Hey, this has been open for over a day." This goes back to if [00:21:00] you go to your previous step of like knowing yourself and knowing what you do, Hey, we typically have 45 minute kind of interactions, even if it's tech space versus like oncology is like, maybe we're actually encountering, like kind of talk for hours at a time.

Rahul: Another customer of ours, imagine pediatrics, but they're very different. They focus on really high risk kids with like. Either terminal problems or very severe life altering conditions, and they provide tiered support of they can have clinicians chat with the parents.

Rahul: They can have, like, folks come in for an emergency intervention and send, like, their clinicians to your house and things like that. But they would probably have longer encounters because these are More confusing, more acute issues and they might be like, Hey, can you stay on the line?

Rahul: Like the parents are about to try something and then they'll report back. So I can't say one size fits all, like this is the time for the nudge. But if you've identified what your typical encounter looks like, you can tailor your time based nudges rather than having it be a hard rule to close the encounter. You can use [00:22:00] those timers. To find your nudges and then usually the other nudge you have is a human nudges at like a supervisory level. Maybe you have like a senior like, clinical advisor that handles, all the California clinicians giving them a dashboard. Like, Hey, these people still have encounters that have been open for like 5 days. Like, maybe you should like, check in with them. So like the automated nudges as well as surfacing it up to the next level of supervisor can be helpful.

Danielle: Does the playbook of a physician owning the decision on the encounter ever not work? 

Rahul: Probably.

Danielle: Or the provider. Sorry, I meant to say generally, but you get what I mean. 

Rahul: Yeah. There's kind of different levels of not work. It's like, do they close it too early? And then sometimes the patient will check in again.

Rahul: Yeah, that'll happen. 

Danielle: Then what do you do in that case? 

Rahul: You can give them tools to like reopen it. You can also just say, "Hey, this is a new encounter. That's a follow up to that encounter." There's nothing perfect. And like, I don't want to say this is a silver bullet to anything, but it is much better than not having a notion of encounters, [00:23:00] especially in the asynchronous case, like with synchronous, even if you're telemedicine, it's like pretty easy. Like we're going to record this podcast. The encounter was the start at the end of it. The other thing that kind of having this encounter breakdown helps you with is who was participating, especially in the case where you're doing caregivers. So we have we're working with another company scene health where they do geriatrics. So adult children helping their parents. So it's like, who actually showed up to this? Was it just the. Geriatric patient was the patient and a family member. Oftentimes there's coordination of multiple caregivers. It's not like with parents. It's parents and child. 

Danielle: Yeah. We ran into that a lot in DME. 

Rahul: Right? Right. It's like maybe multiple siblings or coordinating care for their parents. 

Danielle: Right. 

Rahul: So different people showed. And that's all relevant information when you're thinking about, like, the concierge patient experience isn't just for that adult patient, it's for their caregivers. It's about making sure they have all the information about the health of the patient. It's important in the case [00:24:00] of knowing who was not present in certain things, you know, like, who may be missing information. When did that encounter start and stop?

Okay. Done it.

Danielle: All right, let's go. I'm ready for next stage. Sounds juicy. 

Rahul: you have 100,000 patients, and you're also getting to a point where you're hiring more MDs, nurse practitioners, care navigators, all these people to handle the load.

Rahul: And now they're slotting into a system. They're no longer like all friends with each other, not always on the same meetings. You're realizing, oh, shoot, MDs are expensive. They have a day job that you're competing with here, and they're also pretty, pretty particular. about how they want to do certain things.

Rahul: This is usually the point where it's harder to provide the same level of care that you got for free when you were in the jungle. The jungle was hard. I always know what you were doing, but you were there for your patients. You could pick up the phone three times a day because it was you, the founder. you, the CMO. Now, you're hiring people who are, it's a job for them, and it's not that they don't care. But they need to [00:25:00] know what the system is. First kind of major tip here is like, Let's develop some care plans, also known as protocols, also known as clinical pathways. We all kind of use different terms for this. Basically, we're going to say, of the services we offer, we're going to take that, that 90%, the head of your distribution, and we're going to put it on rails in terms of building checklists So that, an MD knows when they come in, just hired off the street, they know like, the railway of doing it. For each kind of step in the checklist, you want to know: who's doing it? Is it the patient? Is it the provider? And if it's the provider who is such a care navigator. Customer service someone with a nurse level credential or like an MD. What is the task that they are responsible for? And are there any timing elements here? Like, does it need to get done within a certain amount of time?

Rahul: does it need to get done within a certain amount of time of the last thing? So the kinds of things that will go on the checklist are. fill out this form. Schedule this appointment. Like have the appointment. [00:26:00] Write the note. Send. Lab results or something to the patient verify their identity, verify their insurance is like all pretty tactical things that like we, it's in our heads, but like every single part just needs to be written down and you want to get the timing of like, okay, we always want to follow up, visit at least between two to three weeks after the last one. And this is kind of where I advise folks to make a flowchart of the process for each of these kind of pathways. So each node in this flowchart is going to be one of these steps. It's going to say who's going to do the thing, what is the thing that needs to be done, and kind of how long do they have to do it, and like when does it have to happen in relation to the last step in the flowchart, cause the next thing we're going to talk about is top of license care. Basically every task you want it to be done by the person with the lowest credential who can do it.

Rahul: Basically it's, Big buzzword in hospital administration in general. My wife's actually an MD and I remember when she was in residency, everyone was [00:27:00] really focused on this buzzword, top of license care. The idea is you as a clinician are only doing things That your license requires you to do if there's something that someone with a lower credential could do, they should be doing it because it's supposed to be the most efficient allocation of time resources and time is your scarcest resource across all these folks As an entrepreneur, it also will feed into your headcount resources.

Rahul: MDs are much more expensive than customer service representatives, which are much more expensive than a bot or something automated. And so you want to build your pyramid of, do as much as you can automated, and then as much as you can by customer service folks. And you want your MDs to be probably the fewest, but the most efficient members of your team.

Rahul: That's also how you'll scale. In terms of having the best customers experience, so providing quality medical care. I break down to four levels of licensure that you want to allocate across in this pyramid. One is stuff that's automatable, like a computer can do it doesn't even have to be a human. No judgment calls. For example, [00:28:00] sending a welcome email when someone signs up for your service, right? Doesn't have to be a human. Maybe in your discipline, there's like something about the way you want to provide care that you want to be like a person. I'm not going to say that's wrong, but it doesn't have to be a person per se.

Rahul: The next level is it should be a human, but not someone who with any kind of license. For example, like asking a question about their Billing settings on my account or maybe following up on an insurance question, which might require some level of training and knowledge. Like they might need to know how to navigate the health care system.

Rahul: They don't need a certification that proves that.

Rahul: So I'm not saying these people are unskilled or unspecialized, but they just there's more supply because they don't. Have a certification requirement to do that kind of job. Next level is the clinical folks that are non MDs And don't mean to be elitist here.

Rahul: It's just it's a different kind of certification a different supply profile. and there's a spectrum in here, but that's kind of [00:29:00] your four levels. Automated, no license, clinical, non MD, and then your MDs and you always want to drive all these tasks that you did in your flow chart to the lowest level. as possible. So you don't want your MDs checking driver's licenses or doing the insurance confirmation in traditional healthcare practice.

Rahul: If we mapped, your front desk, people there are not just answering calls are often doing a lot of this insurance checking and kind of admin work. 

Danielle: So the flow chart is really holistic of the entire patient experience. It's not like encounter based flow chart. This is literally of what happens When the moment somebody comes in all the way to whatever the end state is of your business. 

Rahul: Yeah. And so the boundaries of it are like, yeah, patient signs up to like patient receives care. And there might be a little loop in there is, you know, we have an ongoing monthly check in or something. It should be across, like, I gave the suggestion of you have like five. path, clinical pathways or five care plans, mapping to your five to 10 elements of your service menu, the five to 10 diagnoses that people come [00:30:00] in for. So that's why we did that in the last step. It's like, what are people coming in the front door for? Now we're here. We need to get that intake all into like a kind of a railroad process, which will have to exit ramps. This flowchart should be your happy path. Now, it's not going to encompass every single thing that might come up. That medicine doesn't work like that, but you at least want to document your happy path because then across your five pathways, you'll see like, okay, we have to do an insurance check for all of them, so they should converge there. The point of the kind of doing this holistic thing is you'll see what tasks are in common, who's doing them, and therefore you can say, like, is the right person doing them? Or do we move it down the pyramid? And then oh where are the bottlenecks? Like, this step is starting to take a long time. Like, it should take . 45 minutes to do this. In reality, it's taking someone on average three hours. What can we do to address that? 

Danielle: All right. So guidance around. Checklisting and flow charting and then being really nuanced within each one of those nodes. What else do you suggest folks do here?

Rahul: So now [00:31:00] we get a couple more unlocks. Now that we've done the work of knowing what our service menu is knowing what our care plans are, you can actually simplify the provider inputs. And so this is kind of my EHR nerd coming out. But a lot of these digital health companies live and die on their EHR admin set up because you got to recruit clinicians at the end of the day and you talk to any clinician, their least favorite part of the job is charting and like inputting, like they love seeing patients. They love taking care of people. They hate all the paperwork the heck goes along with it. I've known, I can't mention by name, but I know digital health companies who are losing clinicians because they were too frustrated with the system that kind of cobbled together. I also know, I think Summer Health, they told me that they built something that physicians love. They were actually able to convert some people from contractors to full time. cause it was so streamlined for them. The bar you're trying to beat is whatever their day job is.

Rahul: Some examples we saw there is like, Summer Health. We always mentioned they knew what their service menu was when they're charting an encounter, you got six big buttons. You can use it with your thumbs. Came in for a rash, red eye, [00:32:00] things like that. Like simplifying the amount of typing. Fewer dropdowns, more buttons, all of these things just streamline the thing along. The more you can do for that also pre population is a big thing. So I got two examples. One is a EHR, a pediatric EHR that was built on top of MedForm called DEVELO. They know that in pediatrics, when insurance claims get denied, usually it's not because of an argument about medical necessity. It's usually an argument about, Oh, was the paperwork filled in properly?

Rahul: So given that they know their service menu, they can have really rigorous supplements about like just. Making sure all the right data is pre populated. You don't have to re type in the patient's name every time when you're charting an encounter. You can just pull all that there. So it's less error prone, takes less time, and you're going to get less denials at the end of the day.

Danielle: My takeaway in, in the idea of designing better EHRs or better physician workload is all around Reducing the cognitive [00:33:00] load and coming back to like maximizing time and maximizing correct throughput. If you work on both of those, you actually just make a better product experience for everybody. yeah. Okay. That makes a lot of sense. 

Rahul: Yeah. And I do want to emphasize the recruiting aspect too, because your ability to like service patients will be related to what your capacity is from a physician staffing level and your ability to recruit physicians will affect how many physicians you have to service your patients. 

Rahul: Last thing, kind of, we already talked about identifying your bottlenecks. It's also, we talked about tracking your quality. What's your turnaround time to get an answer to your patient? How many people How many times are people not following the protocol or is like a certain follow up not happening? Things like that. People do this, generally it's not a thing that is a blind spot at this scale, but my big push to folks is start off with the well known metrics that everyone else cares about. So like HEDIS or CMS measures or risk adjustment factors for value based [00:34:00] care. People sometimes go wild with like metric soup, like. Every business unit, like the product team wants one set of business metrics, business development team wants one set of metrics, the clinical staff and the MDs want a different set of metrics, and then you have like 50 metrics, and they're all different shades of the same color.

Rahul: If you align around some of these standard ones, like HEDIS is a big one for performance based reimbursement. from payers, not that everything is about payers, but it's like, it's a thing that other people are going to report anyway. So you might as well start there, see what it doesn't address for your organization, then augment them.

Rahul: Rather than starting from whole cloth and then having to like kind of wedge. your metrics into these standards that eventually you're going to report. So HEDIS, CMS measures, then risk adjustment factors for value based care. 

Danielle: How many, like was like, is there like a special number you've seen be a good amount of, clinical measures or anything 

Rahul: like that?

Rahul: You'll probably get pushed to 10. I think there's always [00:35:00] upward pressure on these with like some teams like not getting quite the visibility they want from your. Initial five. 

Danielle: Yeah, of course. 

Rahul: So I'd say if you aim for five, you'll end up with ten.

Danielle: Love them. Awesome. Okay. Do you want to add anything else to like the moon or the stage of growth?

Rahul: I mean, this is the fun stage because it feels like you're on to something and patients like you, but it also can be incredibly frustrating because it starts to feel like the wheels are coming off sometimes. Or it's like, this is the stuff we used to get for free. Like our patients never complained. Now we're getting some complaints and you just, I want to help you get ahead of that. 

Danielle: 100 Percent 

Rahul: and it's not that this is going to fix everything, but it'll get you, I use the term a lot, fighting fit, it'll get you to have good hygiene so you'll be in a good place to fix the things you need to fix.

Danielle: After the moon phase, is there any other phase afterwards? 

Rahul: Typically, these kind of go hand in hand, but I just broke it out was going 50 state, sometimes you can get a big patient population across like your first three states, New York, [00:36:00] California, you know, Washington state, something like that. a mass a loyal following there now when you're really spreading out geographically, it's exciting, but there's just a couple more challenges that come along with it. And the biggest ones is licensure, making sure what you, when you had 15. MDs who are feeling good because they were hyper efficient, but they're not licensed in all 50 states. Like who's going to cover Mississippi? And so there's just an extra level of organization that I want you all to add on top of that. And my big push here is like, try to organize your practices kind of like a physical hospital, just virtual, which oxymoronic, but what I mean, there's don't try to get cute with terms. Call a group of specialties, a department. Or a ward. That's what your clinicians are familiar with. Don't call them like a um

Danielle: a think boy team

Rahul: think boy team. I try to think a battalion, you know, like,

Danielle: or like the onco care team or like some, some abbreviated thing that you think is like, branded internally some way. 

Rahul: Yeah. Yeah. [00:37:00] It's tempting, but it's just like, again, you're hiring folks who don't know your company that well. Again, this might be their. side job. So having them have that mental model of this is how I slot in. One big, uh, I got a couple of do's and don'ts here. So do call them department wardens who kind of map it to a general hospital and then do organize along two axes. One is. The department or specialty that people are working in. So like you're OBs versus your allergist versus, endocrinologist, whatever. And geography is the second one, where are they licensed? Where are they allowed to practice? Don't organize them along your product offerings, which is a common mistake I see. It's like maybe within gynecology, I offer, STDs birth control, a couple of other things. Those are customer facing products, but they're not how you should organize your clinical. Teams, there's going to be. Foreseen, but unplanned circumstances like your doctor's going on vacation. You might need someone on call at weird hours. Someone might be sick. we had, you know, big practice work with us. They're [00:38:00] a 1 million patients, 50 safe practice. They had multiple different service lines, but they'd mapped everything really along their user facing products, partially because they had grown through acquisition. So they used to be all different companies. We kind of went through. Process with them to like, let's map this to like, how would look at a physical hospital? So departments and then they had a leaf note there, which is like care pods, like really tight teams of a couple of nurses. 1 doctor all works together. That's great. And then we map those care pods to like, what geographies can they cover? Like this care pod can cover. These states. And then we could see like the heat map of like, which states have high coverage, low coverage. And they had a written process as a flowchart, but like they just did it as bullets of like, if someone is out, what is the search process to find their replacement? Like first look in their care pod and look in their department for someone of the same credential with the same license, then like, go across to a different one. 

Danielle: On our episode with Nate, he [00:39:00] also broke down the pod structure around complex care management and geographical alignment of that. So that's something that we've heard before to the importance of geography and pods. 

Rahul: Yeah. Again, it's about understanding, having clinicians come to slot in and not having it be a huge cognitive load for them of, okay, in my day job, I do this, but, in. Digital health Company A this is how I think about my job. 

Danielle: Keep it, keep it simple.

Rahul: Talk like the healthcare system. 

Danielle: Talk like the healthcare system. All comes back to that. All right, Rahul, we're cutting to the end of our time, but we like to end every podcast episode recently. This is a new experiment with three experiments folks should try inside of their orgs. So what are three, three experiments you want to encourage folks to try out in their orgs? Ideally, they've worked for you before. 

Rahul: Yeah, The first one is do a ride along with a clinician for a single patient's journey, like from patient intake to care delivery. Like, just [00:40:00] do a ride along and see what are the bottlenecks? What are the things that they maybe haven't complained about? But you're like, Oh, why are you doing that in terms of like data entry? Something like that is like, Oh, like that kind of sucks. Big kind of things to look out for there is like, are they typing in redundant information? This sounds silly, but it's frustrating. And also it's error prone. The cost of fixing a mistake, if they mistype a patient's name, or if they have an off by one error on a date of birth, it's a huge remediation process once the toothpaste is out of the tube. So figuring out, does that happen? How often does that happen? Also look at how many tabs they have open. Oftentimes these clinicians to do their jobs, they're like, they're locked in a few different systems. Maybe they have like their own, like notes, like in a word doc somewhere and it's like just understanding, like how many things they need to do, get all the information for one patient care journey. And, and we'll be pretty insightful in terms of understanding, building that flow chart that I mentioned. 

Danielle: I call it tab city in my course happens to everybody in the startup, but yeah, anybody [00:41:00] delivering care. Definitely clinicians though. 

Rahul: And then it's like, do you buy, do you get the extension that manages your tabs or do you enforce discipline? Or like, I'm just going to close things. 

Danielle: Only time tells, I don't know. I don't have a good answer for that. 

Rahul: I've avoided the tab thing. I was like, I'm trying to go the discipline route.

Rahul: We'll see how it goes. The next experiment, I'll say it's like lists all the tasks for a single care plan. This is related, but like once you've done that ride along, like take one of your care plans, just write down all the tasks that you said, who did it, who does it, how long should it take from the last thing that they did, or like kind of what's the timing constraints there and then ask like, is everyone working top of license on all that tasks, just for one care journey. One thing that your company does, is there anything automatable, anything that can be pushed to a lower license. Last thing is just like take stock of your service menu. The first two were kind of narrow. One care journey, one patient. This one is more broad. Like, collect all the ICD or CPD codes that you're performing right now. Just do a group by count. Look at the distribution. What, like, how, how tight is your [00:42:00] head and how long is your tail? If you do your 90th percentile, is that five codes? Is that 50 codes? And then look, first thing is like, look, is there anything that be deduplicated or is there anything in the head that's surprising? It's like, Oh, I didn't know we were doing that. I didn't know we were treating this and see if you can find at least some opportunities to bring it closer to like five to 10 marker or like have a reason thesis on why we do 50. 

Danielle: It's a lot, 50. Yeesh. And people surprise me every day. Rahul, this was awesome. This was super, super tactical. I, I learned a whole lot about how to think through and orchestrate a lot of change in very simple ways. And I know folks are going to gain a lot out of this episode. So thank you for distilling your wisdom built through the many years of seeing lots of builds so that others can benefit from it. That's the point of the pod and we're grateful that you got to share all this today. Thank you. 

Rahul: I'm grateful to have been on it. Thank you. 

Danielle: Awesome.


Let's Keep In Touch

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
search icon