The Eighteen-Minute Handover

The first time I watched a colleague hand over his ward, it was 2:18 a.m. He had a folded sheet of paper in his coat pocket — names, bed numbers, two-letter shorthand. Twenty-seven patients. He spoke at the speed of someone who had done this for years, but I could see him glance at the paper between every patient. The handover took eighteen minutes. By the end, the doctor receiving it was already reaching for his own folded sheet.

That was the moment that made me believe the ward needed software of its own.

Forty seconds per patient

I went home and started counting. Eighteen minutes is about forty seconds per patient. Forty seconds per patient is not enough to convey what is happening on the ward to a doctor who has never seen the patients. Forty seconds is enough to convey the urgent thing about this patient. Anything else gets dropped.

Say it out loud with a stopwatch and the constraint becomes concrete. Forty seconds is bed, age, one-line diagnosis, day of admission, the current active problem, and one instruction. That is a full forty seconds delivered at pace, with no interruption, no question from the receiver, and no pause to check the paper. Add a single clarifying question — "sorry, which bed?" — and you have spent a quarter of the next patient's allocation.

The compression is not a mistake

The handover protocol literature (Starmer et al., NEJM, 2014; Riesenberg et al., Am J Med Qual, 2009) recommends a structure: illness severity, patient summary, action list, situation awareness, synthesis. The recommendation, applied across twenty-seven patients in eighteen minutes, would consume the entire handover in the structure alone, with no time for content.

It is worth being fair to both sides of that gap. The protocols were largely developed and validated on lists where the arithmetic works — a resident handing over eight or ten patients has room for five elements each. Nobody wrote them intending them to be impossible; they were written for a ward that most of the world does not have.

So real handovers strip down to two things per patient: what is happening now, and what is pending. That is the I-PASS bundle when you remove the parts that don't fit in forty seconds. The residents doing this are not failing to follow the protocol. They are executing a rational compression under a constraint the protocol did not anticipate, and the parts they drop are predictable — contingency planning first, the receiver's read-back second.

The second folded sheet

The detail I keep returning to is the one at the end: the receiving doctor reaching for his own sheet.

He was not taking notes. He was reconstructing the list — writing out twenty-seven beds, in his own shorthand, in his own layout, because the outgoing resident's sheet was going home in the outgoing resident's pocket. Every twelve hours, the ward's working state gets destroyed and rebuilt from a spoken account of itself.

Two consequences follow, and both are worse than they sound. The reconstruction is lossy, so the night list is a copy of a copy — and by the third shift it has drifted from anything a chart would confirm. And there is no shared object. Two doctors, four hours apart, are holding two different lists that disagree, and neither can tell which one is wrong because the only reconciliation mechanism is another verbal handover.

A folded sheet in a coat pocket is a beautifully evolved instrument, and it has exactly one fatal property: it cannot be in two pockets at once.

Composed, not remembered

The Medows handover composes itself in exactly that shape: "Now" and "Pending" per patient, drawn from the actual events of the shift. Two lines. The receiving doctor reads them in less than twenty seconds. Twenty seconds per patient × twenty-seven patients = nine minutes of handover. Half the time of paper.

"Drawn from the actual events of the shift" is the load-bearing phrase. The lines are assembled from things the system already knows happened: orders placed, results still outstanding, vitals that moved, medications changed, tasks marked done and not done. None of it depends on the outgoing resident recalling anything at 2:18 a.m., which matters because the patient most likely to be forgotten is the one whose deterioration was gradual enough not to have generated a memorable moment.

What the composition explicitly does not do is form a clinical opinion. It does not decide that a patient is stable. It reports what happened and what is outstanding, and the resident — who edits every draft before it goes — supplies the judgement. A system that wrote "stable" on its own authority would be reproducing precisely the failure that makes handover dangerous.

The nine minutes are not the win

The half-time isn't the point. The point is that the time saved isn't spent on more administrative work. It is spent on the patients that need more than forty seconds — the one that is deteriorating, the one whose family is asking hard questions, the one whose CT just got read.

And the conversation should not disappear. If the composed handover became a document that got emailed and read alone, the read-back would vanish entirely — and the read-back is where the receiving doctor's misunderstanding surfaces while there is still someone in the room to correct it. The draft is meant to be the substrate for the conversation, not a replacement: it carries the twenty-two routine patients so the two doctors can actually talk about the five that matter.

A handover should be a triage of attention. The folded-paper version is a triage of which patient gets mentioned. The auto-composed version is a triage of which patient needs your time. The difference matters at 2:18 a.m. when the next twelve hours are going to decide outcomes.