Your Hospitalists Are Doing Software’s Job at 4 A.M

A few weeks ago I was on a demo call with a hospital medicine team at a Midwest hospital that runs a census around 200. Before we got into the software, I asked them what the worst part of their morning was. 

An advanced practice clinician on the team mentioned that she’d been tapped to build the morning patient list a couple of weeks earlier while the person who usually does it was out. 

She started at 4 a.m. She drove to the hospital to do it. 

And her husband got in the car and followed her in. Not to help. Just to see what on earth she was doing at that hour. 

It was never the sorting that took two hours 

Here’s the thing about the morning list that many people miss. 

It isn’t data entry. If it were data entry, you could hand it to anyone and it would take twenty minutes. It’s judgment — a couple hundred small decisions made before the sun has even risen, most of them before anyone has opened a chart. 

She put it better than I could: 

“Even though it’s a relatively straightforward process, there’s a lot of thought that has to go into it. And it’s a lot of brain power real early in the morning before we even have taken a look at a patient chart in the medical context. So it’s pretty taxing.” 

A hospitalist on the same call, who covers a teaching service, added: 

“I would say the biggest thing is just the time commitment. There’s a lot of nuance to how we do it, and it’s going to vary between institutions.” 

And then the part that turns a long task into an unpredictable one. Their census doesn’t sit still — it swings from about 175 to about 220 — and neither does their staffing: 

“Not only are you just sorting, but you’re figuring out, I have one less provider today, or we have one less unit that closed yesterday that now I have to redistribute those.” 

The most expensive part is that only a few people can do it 

The reason this program went looking for software wasn’t just the clock. It was what happens when the person who knows how to do it isn’t there. The rules for who gets which patient often live in two or three people’s heads. Which patients stay with their provider. Which floor belongs to which team. Who’s already carrying the heavy ones. Who just came back from vacation. Where the long-term care patients go. 

Why one physician ends up with fifty patients who were never really theirs 

On the demo, we talked through a couple specific examples where the pain hits most. 

On this team, overnight admissions tend to pile up under whichever physician is listed on the admission — a doctor who may have accepted a transfer at 3 in the afternoon for a patient who physically arrived at 10 that night, long after his shift ended. Their words: 

“There might be a doc for example, who’s on admissions this week, and he’s listed as the admitting doctor, even though he never admitted the patient. They showed up after his shift ended.” 

So the morning list opens with one provider showing forty or fifty patients. Then somebody has to redistribute all of them by hand, which is exactly the manual work that pushes the start time back to 4 a.m. 

Assign, our automated patient assignment solution, doesn’t do it that way, and the reason is a design decision our product manager Jaclyn Corbett made a point of explaining on the call. The census we build isn’t drawn from the admitting field at all. It’s built from two relationships in the Epic feed: attending and consulting. If anyone in your hospital medicine program is the attending or the consulting provider on a patient, that patient is on your census that day. 

The admitting name is a historical record. It tells you who accepted a patient once. It doesn’t tell you who’s responsible for them today. Build your list on it and you inherit a long list of patients every single morning that you may not need to see. 

Build it on the attending relationship instead. The night hospitalist who took the patient at 2 a.m. shows up as the attending, that patient comes onto the census as someone who hasn’t been seen by a day rounder yet, and the rules distribute them across the day teams before anyone walks in. 

The consult that quietly sinks to the bottom of your list 

The hospitalist on the teaching service asked us about scenarios like this: 

“Maybe the patient was admitted 2 days ago, but we got a consult request overnight. If we filter by the admission time, it’s going to push them way down on the list and we’re going to miss that that’s a new patient that needs [assigning].” 

When you sort by admit time to find your overnight work, a patient who’s been in the building since Tuesday is nowhere near the top — even though a consult was requested on them at 3 a.m. and nobody on your service has laid eyes on them. 

Here’s how we’d handle it: 

The first is that consulting is one of the two relationships that builds the census, so that patient is on the list whether or not a human notices. As Jaclyn put it: “even if you were to miss it, like validating it in the morning, it’s going to be on the list and get assigned.” 

The second is the indicator column, which is a configurable field you can use to tag and group patients however your program thinks. You could have every consult flagged as a consult so you can sort them together and put eyes on all of them at once. Same for anything else you want surfaced — a patient who needs an H&P, a patient coming off a long-term care unit, whatever your version of “don’t let this slip” is. 

Now those “I have to remember to check for this” notes becomes “this is grouped at the top of my screen.” 

Your rules, run against every patient, every morning 

No matter how many conversations I have with hospital medicine leaders, it’s become pretty evident that many programs are unique – and everyone has their own way of how they like to assign patients. One hospitalist asked me: “When it’s deciding how to take away patients, what algorithm does it use? Does it go geographic first? Does it go by where somebody is more concentrated if they’re on several different units?” 

The answer is that it goes in whatever order matters to you, because the rules are built for your program during implementation — after we’ve shadowed a couple of your actual mornings and watched what you really do. 

Some of our customers care most about continuity: a patient never changes providers during a stay, full stop. Others work across multiple campuses, so geography is prioritized. This program wanted geography, continuity, and workload balance, with a target of about 17 patients per provider and a tolerance of plus or minus two — so 15 to 19 is where they want everyone to land. 

You can encode preferences that sound contradictory until you say them out loud. Keep a patient with the same provider for the whole stay, but if they’re readmitted within 30 days, route them back to the physician who knows them. Or the opposite — if you’d rather that provider get a break from a frequent flyer, route them anywhere else.  

What I’d want a program director to see is the screen after the rules run. You can click into a team and watch the logic show its work: here’s a blue-tower provider whose list is all blue-tower patients, except for one green one, because that patient was theirs yesterday and continuity beat geography for that specific patient on that specific morning. New admissions spread evenly. ICU patients grouped, if that’s what you want. Everything you were holding in your head at 4 a.m., held by something that doesn’t get tired. 

The second shift you didn’t know you were working 

Once the list is complete, the work isn’t over. At many programs, next up they have to update the attending in the EHR one by one. Then you get the list to nursing — printed, faxed, walked down, texted, whatever your building does. 

In Assign, you hit approve. The assignments write back into Epic in real time, so the attending updates land while you’re still sitting there. Distribution to your nursing teams goes out by fax, email, or phone, depending on how your units want to receive it. 

So the morning that used to start at 4 a.m. and end whenever the last name got typed in now looks like this: log in at 6:30, verify the census, check off who’s working today, run the assignment, look it over, approve. It’s a 15-to-20-minute process, and you’re done a little before 7 with nursing already notified. 

So, can anyone run it? 

The programs who get the most out of Assign aren’t the ones who saved the most minutes. They’re the ones who stopped having a single point of failure. Today our customers run this process with a genuine mix of people — administrative staff, APPs, hospitalists, clinical coordinators — because once the rules are in the software, running the list is three guided screens instead of two hundred judgment calls. 

Which means when the person who always does it is on vacation, or out sick, or finally taking a Tuesday off, the list still gets built correctly. Nobody drives in at 4 a.m. Nobody’s spouse has to follow them to the hospital to understand why. 

The hospitalist on that call summed it up perfectly: it would “alleviate a lot of busy work in the morning.” 

FAQs

How long does the morning patient assignment actually take with software? Most hospital medicine programs doing this manually spend around two hours, and often start well before dawn to finish in time for the day teams. With Assign, customers typically begin around 6:30 or 6:45 and complete the process in 15 to 20 minutes, including time to review and approve the list. Larger programs doing extra validation may take a little longer; some take less. 

Why do overnight admissions pile up under one physician? Because most manual processes build the list from the admitting provider field, and that field records who accepted the patient rather than who is responsible for them now. A physician on admissions for the week can be listed on patients who arrived long after his shift ended. Assign builds the census from the attending and consulting relationships instead, so the pile never forms and the overnight work is distributed across the day teams before anyone arrives. 

What happens to a consult requested overnight on a patient admitted days earlier? The consulting relationship is one of the two relationships that build the census, so the patient appears on the list and is assigned whether or not anyone catches them during morning validation. Programs that want those patients visually grouped can use a configurable indicator column to tag and sort them together. 

Can the rules reflect how our program actually works? Yes — the rules are built for each site during implementation, after shadowing the real morning process rather than working from a written procedure. Programs prioritize differently: some treat continuity as inviolable, some running multiple campuses put geographic rounding first, and most want a workload target with a defined tolerance. Rules can also be changed as the program grows, adds units or restructures teams. 

What happens after the list is approved? Assignments write back into your EHR in real time, so attending updates land as soon as the list is approved. The finished list distributes to nursing teams by fax, email or phone, depending on how each unit prefers to receive it — which removes the second round of manual entry and hand distribution that follows a manual assignment. 

Who runs the process day to day? Across our customer base it is a mix of administrative staff, clinical coordinators, APPs and hospitalists. That is the point: once the rules live in the software rather than in a few people’s memory, the daily process is three guided screens, so the list still gets built correctly when the usual person is on vacation or out sick.