Why We Still Hand Over on Paper

The morning after night-shift, two pieces of paper get folded and slipped into pockets. One is the outgoing resident's. One is the incoming resident's. They are nearly identical: the ward list, with two-letter shorthand and pencil annotations. The handover takes thirteen minutes if the wards are quiet, twenty-three if not.

This is 2025. Almost every other part of clinical work has been progressively softwared. The handover is not.

The evidence is not the problem

The published evidence that this matters is unusually clear. Starmer and colleagues implemented the I-PASS handoff bundle across nine pediatric residency programs (NEJM, 2014). Within a year: 23% reduction in medical errors. 30% reduction in preventable adverse events. Same staff, same patients, same hospitals — just a structured handover.

The Joint Commission's sentinel-event analyses (multiple annual reports) consistently identify communication and handoff failure as a contributing factor in roughly 70–80% of serious adverse events.

Set those two findings side by side and the situation is close to absurd. The single most reliable intervention available — no new drug, no new equipment, no additional staff — is a change in how twelve minutes at the end of a shift are structured. And the medium those twelve minutes run on is a sheet of paper folded into quarters.

So why is the handover still on paper?

What paper is actually good at

Before the three reasons, the honest part: paper is not losing this competition by default. It is winning on merit, on the axes that matter at 8 p.m.

It opens instantly. It never logs you out, never asks for a password with wet hands, never shows a spinner. It works one-handed while the other hand holds a file. It accepts an annotation faster than any input method ever built — a circle around a bed number is a quarter of a second. It survives a dead battery, a dead network, and a basement corridor with no signal. It is trivially shareable by being held up. And it costs nothing, so losing one is a bad afternoon rather than an IT incident.

Any replacement that is worse on those axes will lose, regardless of how much better its handover is in principle. Most clinical software is worse on all of them.

Three reasons

No tool fits the workflow. EHRs were designed for documentation, not for the spoken handover that happens between shifts. The handover note buried in an EHR is rarely the one that gets read. The distinction is between a record made for the future — for audit, billing, medico-legal defence, the discharge summary — and a message made for one specific person in the next ten minutes. Almost every field in an EHR note serves the first purpose. The handover needs the second, and there is nowhere to put it.

Structure is hard to enforce. I-PASS is a structure — Illness severity, Patient summary, Action list, Situation awareness, Synthesis. It works on paper, but only if the resident remembers to fill in every box. Most don't, by the third handover of the day. And when a system tries to make them, adherence decays anyway — enforcement produces filled fields, not better handovers.

The handover composes itself from the shift. A good handover note is not a fresh document. It is a summary of what happened in the last twelve hours — what was ordered, what came back, what's pending, what to watch. The current tools don't have this context.

That third one is the real barrier, and it is a data problem rather than an interface problem. To compose a handover you need a system that knows the order was placed at 11 p.m., that it has not resulted yet, that the vitals moved three times since morning, and that two of the day's tasks are still open. A tool that has only the notes cannot do it, no matter how good its editor is. This is why "add a handover module" to existing systems keeps producing a blank text box with a heading.

The fourth reason, which nobody says out loud

The handover is not only an information transfer. It is a transfer of responsibility, and it is performed verbally because that is how responsibility has always changed hands.

"You have the ward" is a speech act. Something happens when it is said out loud, in front of the person receiving it, that does not happen when a document is sent. The outgoing resident is released; the incoming one is now the person who gets called. A document does not release anybody, and everyone involved knows it, which is why even units with functioning written handovers still do the conversation.

This is worth respecting rather than engineering around. It also explains a failure mode of handover software: a tool that replaces the conversation removes the accountability transfer and leaves people unsure who owns the ward at 8:15. The right target is not to eliminate the twelve minutes. It is to make the twelve minutes be about the five patients that need discussion, rather than about reciting the twenty-three that do not.

What we built

The Medows handover takes the third route. The patient's card already knows what was ordered, what came back, what's pending. The handover is a button that composes a two-line summary per patient from that context: "Now: …" and "Pending: …" The resident reviews, edits, and sends it to the next doctor over WhatsApp.

WhatsApp because that is where handover already goes — not for want of alternatives, but because it wins on the axes paper wins on. Composing it deliberately, minimised to what the receiving doctor needs, is a better answer than a policy asking people to stop.

It is not a perfect handover. It is, however, an honest one — a summary of what actually happened in the shift, rather than a reconstruction at 2 a.m. by a tired resident with a folded sheet.

If the largest published reduction in preventable patient harm in residency literature is a structured handover, the structure should be the path of least resistance. That is the bar.