Fewer Names on Every Patient’s Care Team: What a High-Quality Assignment is Actually Optimizing For

A hospitalist program director reached out to us for a reason many other hospitalist directors don’t. Her program runs a census around 200 and every morning her night cross-cover physician divides up the new admissions before the day teams arrive. That takes about 30 minutes. 

Thirty minutes is roughly what it would take with our automated patient assignment software running, and she knew it. She called anyway. 

So I asked what she was hoping to get out of this, since it was not time savings. Her answer: “What I’m hoping to gain is the idea of creating teams that will work better together, that long-term should translate into better teamwork.” Then the concrete version — “right now the way we are [assigning] is just based on floor. I would love to be able to do that based on nursing and case management.” 

She had worked out something it takes most programs a lot longer to see. Her morning list is not a clerical document that names a doctor. It is deciding which nurses, which case managers and which advanced practice providers each of her physicians will be coordinating with for the next 24 hours, across her entire census. Every patient on that list is a small decision about who talks to whom. She did not need a faster morning. She needed a better coordinated team. 

Now, for most programs the morning is genuinely painful and I am not going to wave that away. The manual version usually runs two to three hours, with somebody awake at four or five cross-referencing a census against a schedule and working from handwritten sheets turned in by the overnight team. One APP who got tagged in to cover the list while the regular scheduler was on vacation described it as “still taking them two hours plus to do every day. And this is non-billable work where it’s either a physician or an advanced practice provider who’s doing it.” With rules running, that morning is typically 15 to 20 minutes. That saving is real and worth having. 

But it is the floor, not the ceiling – which is exactly why her question is the more interesting one. What actually makes an assignment good? 

In my experience, there are four things that impact a quality provider-patient assignment: continuity, geography, workload, and how many different people each physician has to coordinate with. But the one most directors look at is headcount equity, which may actually be the least important of the four. 

 

The list you build at 4 a.m. is what the whole building runs on 

The assignment is the routing table for the next 24 hours, and a lot of people outside hospital medicine are reading it. As one hospital’s operations lead put it while I was watching her build the morning list on a 597-patient day, “we send this list to everybody, so they know what patients are being seen by which doctor.” 

At a 400-bed hospital where the hospitalist service carries about 200 patients, the morning list gets sent to the nursing units so they can see “who’s seeing what patient without getting into Epic.” so that when a nurse calls Rounder -10 at 2 p.m., it reaches the right person. 

A program director at another academic center described the real payload better than I could: it is “not so much putting the right doctor on the patient but putting the right team name, putting the right covering provider who’s working on that team, whose first call.” 

Which means an error in the morning list does not stay in the morning list. 

 

A good assignment holds four things at once, and they fight each other 

Continuity. The patient keeps the physician who already knows them. For most programs this is the top rule, and physicians feel strongly about it — one director told me her hospitalists are “very particular about not losing their patients if they’re rounding on them the previous day.” 

Geography. The physician’s patients sit near each other, which is what makes multi-disciplinary rounds, huddles and hallway conversations physically possible. 

Workload, not headcount. An even number of patients is not always an even list. As an APP lead at a roughly 350-census hospital observed, when providers are not tied to a floor and “they just kind of get thrown a ton of random patients, it might not be as evenly distributed like heaviness, acuity.” 

Care-team concordance. How many distinct nurses, case managers and APPs one physician has to coordinate with to get through the day. More on that below — it is the newest of the four, and the one nobody is counting. 

Now watch what happens when you try to do all four manually. One large system ranks them explicitly: continuity first, then geographic rounding. Another site puts geographic fidelity above equal patient counts on purpose. A third does the exact reverse — “the geography is secondary to the patient count.” A multi-site system that switched to geographic rounding last year heard “a lot of grumbles because it was taking quite a long time because they were trying to do geographic with keeping continuity, which causes some issues.” 

None of them are doing it wrong. The trade-off is the job. As a coordinator at a 466-census system put it, “it’s not as easy as saying you have this many providers on and you’re just going to divide all the patients evenly amongst them.” 

A person doing this at 4 a.m. is not balancing four constraints.  

A person working against a 7 a.m. deadline can only work the delta — they take the patients admitted since yesterday, place them, and leave everything else where it was until it visibly breaks. The director I opened with described her own version: her team resets a provider’s list geographically at the start of a rounding block, “but obviously over time it breaks apart, and then we reset every time someone rotates off their rounding shift.” Nobody has two hours to reconsider 250 patients, so the census drifts, and the drift stays invisible until a physician is on five floors. Rules do not work that way — every morning they evaluate the whole census against every rule you have given them, not just the overnight admits.  

Yes, our customers choose Assign because they don’t want to create patient lists manually anymore – and they want to save time. But when speaking with one of my colleagues, Jeremy Smith, he told me that when he configures a site, what he optimizes for is quality, and the rules he ends up building are ones a human could not run in the window available:  

• A hospital he is implementing now handed him a list of primary care physicians and asked that any patient with one of those PCPs route to one of two family-medicine-trained hospitalists.  
• That same site runs a round robin for overnight admissions, so a new admit can land outside a provider’s geography — and the next morning the rules look back at everyone admitted in the last 24 hours and re-sort them geographically so providers stay on their unit.  
• Other programs ask us to spread ICU patients evenly, because a physician who opens their list and sees two ICU patients while a colleague has none is going to say something, or to spread after-midnight admissions evenly, since the nocturnist has already billed those patients but the day rounder still has to round on them. 

The question two health systems asked us independently 

The director I opened with is not asking for a shorter list-building time. She is asking to minimize the number of distinct people involved in any one physician’s panel: “I want to decrease the number of other care team members that a physician works with daily. So ideally for a physician, I can assign patients to limit the number of case managers.”  

Then a completely different academic program, around 250 patients, asked for the same thing from the other direction — for their APPs. In their words: “We want the APP to work with the fewest number of doctors as possible. We want them to be as geographically located as possible. And then we want the team sizes to be relatively even or the workloads to be relatively even between APPs.” Same three criteria, same priority order, arrived at independently. Today they rebuild that APP-to-physician pairing by hand every morning, on top of the physician assignment. 

Two health systems describing the same optimization from opposite ends usually means something real is going on, and this is an active design conversation for us right now. Mechanically the building blocks are familiar — our rules run on attributes, and an attribute can be almost anything we receive about a patient, their visit or their care team, including values grouped together, so one set of case managers can be one attribute and another set can be another. We already keep the full care-team record for every hospitalization. The work that matters is upstream, in the data: the case manager assignment has to arrive from the EHR early enough in the day to be usable, and that roster has to be kept current and mapped the way we maintain provider records. 

 

What I would measure instead of time savings  

• Continuity retention — what share of patients keep the same physician day over day. 

• Geographic concentration — how many units a provider’s panel spans on an average day. 

• Workload variance across teams, after weighting, if you weight. 

• Multidisciplinary round participation — how many providers are geographically able to attend the rounds their patients are on. 

Reclaimed time still counts, it is just second-order, and the interesting question is what the time buys. One operations leader made that point better than any ROI slide: as her program has gotten more complex, the deadline has not moved, so the only thing that gives is how early her coordinators start. What she wants is not a smaller team — it is for “the coordinators at these sites [to have] more time to talk to each other.” 

The payoff, when it arrives, tends to be quiet. A leader at a large academic medical center told me after they went live that her providers “did not notice anything.” The software was invisible, which is exactly right. What they did notice was this: “all my patients are on my MDR, which is not something we have routinely had.” 

If you want to start building the case internally, pull one week of your own lists and count two things: how many units each provider covered, and how many patients changed physicians. You can do that without talking to a vendor, and it will tell you more about your mornings than the stopwatch will. 

 

FAQs 

What makes a patient assignment “high quality”? Four things held at the same time: continuity (the patient keeps the physician who already knows them), geographic concentration (a provider’s patients sit near each other), workload balance weighted for acuity rather than raw headcount, and care-team concordance — how many distinct nurses, case managers and APPs that physician has to coordinate with. They conflict with each other, which is why a manual process tends to satisfy one or two and absorb the rest. 

Isn’t the value of assignment automation just the time it saves? That is part of it. A manual morning list commonly runs two to three hours and automation brings it to 15 to 20 minutes, which matters when the person doing it is a physician or an APP who is not seeing patients during that time. But a person working under that kind of time pressure can only place the overnight admits, so the rest of the census drifts until it visibly breaks. Rules evaluate every patient against every rule every morning, which produces a materially different list — and the quality of that list is what the rest of the care team actually feels. 

How does patient assignment affect multidisciplinary rounds? Multidisciplinary rounds happen at fixed times on fixed units. A provider whose patients are spread across several units physically cannot attend the rounds their patients are on, so at some hospitals those providers are excused entirely — while the case managers and nurses on those units are required to be there for the full hour. Geographic concentration in the morning list is what makes attendance possible. 

Can assignment rules account for case managers and nurses? The rules engine runs on attributes, and an attribute can be almost anything that arrives about a patient, their visit or their care team — including grouped values, so one set of case managers can be treated as one attribute and another set as another. The real dependency is upstream: the case manager assignment has to come from the EHR early enough in the day to be usable, and that roster has to be kept current and mapped the way provider records are. It is an active design conversation with several health systems right now. 

What should a hospital medicine program measure before and after automating assignments? Continuity retention day over day, geographic concentration (units per provider), workload variance across teams after weighting, multidisciplinary round participation, unassigned or missed patients and how long before they are caught, and time from an overnight order to first physician contact the next morning.