The Morning Round in 90 Minutes

A morning round at a tertiary government hospital in India is allocated about 90 minutes, on paper. The unit has 28 in-patients, give or take, depending on the previous night's admissions.

That is 3.2 minutes per patient if every minute counts. It doesn't.

Of the 90 minutes:

• 12–15 minutes for the senior to assemble the team and walk to the ward
• 8–10 minutes for shift handover (the night resident still finishing)
• 4–6 minutes spent in transit between bed groups
• 3–5 minutes for the nurse to flag urgent issues
• 2–3 minutes for trainees to find a chart that has been moved

Net time per patient: 1.8–2.2 minutes. Less than a single sentence per bed, on average.

What 1.8 minutes buys

What gets done in 1.8 minutes per patient on a ward round, by the resident who carries the list for the day, is approximately this:

• Look at the bed number on the list
• Verify the patient is the patient on the list
• Glance at the vitals chart
• Ask the patient one direct question
• Listen to one answer
• Note one change

That is it. There is no room in 1.8 minutes for chart review, lab review, drug history, or any kind of considered reasoning. The reasoning is deferred to outside the round — either before, in the resident's pre-round review, or after, when the resident is alone with the chart at the nurses' station.

The round is not a data-gathering exercise

Once you accept the arithmetic, a lot of received wisdom about ward rounds stops making sense. People talk about the round as though it were where the clinical thinking happens. It cannot be. You cannot think about twenty-eight patients in ninety minutes while walking.

What the round actually is, structurally, is a ratification meeting. The resident arrives with a proposed plan for every bed, formed earlier. The consultant samples — genuinely engaging with perhaps four or five patients, accepting the resident's framing on the rest — and either ratifies or overturns. The round's real output is not information; it is authority. It converts twenty-eight provisional plans into twenty-eight sanctioned ones.

That reframing has a sharp consequence. If the round is a sampling process, then everything depends on whether the right four or five patients get sampled. And the mechanism that decides which patients get the consultant's attention is not clinical at all. It is whatever the resident happens to raise, which is a function of what the resident happened to notice, which is a function of what the resident happened to look at before 8 a.m.

The real work is the pre-round

The "real" work of the ward round is not the round itself. It is the preparation that makes the round possible. A bad pre-round means a bad round. A good pre-round means the resident already knows which beds need 30 seconds and which need eight minutes.

The tool implication: the surface that holds the patient list and the recent vitals and the current orders is the pre-round tool, not the round tool. If a workspace makes the pre-round faster — by surfacing the patients whose state has changed since yesterday — the entire round is better.

What "changed since yesterday" has to mean

That phrase is doing a lot of work, and it is worth being concrete, because a naive implementation is worse than none. "Changed" cannot mean "any row was written." Every patient has a row written every day. Sorting by last-updated timestamp returns the list in the order the nurse charted, which is to say, in no useful order at all.

The changes that should pull a patient upward are the ones a resident would want to be told about before walking:

• A result that crossed a threshold — not "a new potassium exists," but a potassium that moved out of range, or moved a lot while staying in range.
• A vitals slope, not a vitals value. Four sets going the wrong direction matter more than one abnormal reading, and are much easier to miss. This is the CCF case in a single sort key.
• An order that has been awaiting a result too long — the gas sent at 11 p.m. that nobody has looked at.
• A medication change made overnight by someone who is now off shift.
• A note from another team — the surgical review that happened at 4 p.m. and changes the plan.
• Discharge readiness, which is not an alarm but is the highest-value thing a consultant can ratify quickly.

Six signals, each cheap to compute, and together they are the difference between a list and a briefing.

The failure mode of sorting by change

There is an honest caveat here, and it is the reason this cannot be the only view.

A patient who is quietly, steadily bad — unchanged for four days, on a trajectory nobody has re-examined since admission — generates no change signal at all, and sinks to the bottom of a change-sorted list every single morning. Sorting by novelty systematically deprioritises the patient whose problem is that nothing is happening.

So the changed-first ordering is a default, not a replacement for the full list, and "days since this patient was last discussed properly" is itself a signal worth surfacing. A ward round should be uncomfortable about the bed nobody has mentioned in a week.

Why the list order is the product

Medows opens the ward list with the changed-since-yesterday patients sorted to the top by default. That is the most important affordance in the app. Not the AI consult. Not the handover. The ordering of the list.

It looks like a convenience feature and it is not. In a system where attention is the binding constraint, the order of the list is the allocation of attention. A list in bed-number order tells the resident that Bed 1 and Bed 27 have equal claim on the next ninety seconds, which is false every single morning, and quietly leaves the ranking to whichever patient the resident happens to remember being worried about.

Sorting is the cheapest clinical intervention in the whole product. It changes what gets looked at, and on a ninety-minute round for twenty-eight patients, what gets looked at is very nearly the whole of the medicine.