What AIIMS Taught Me About Residency Software

I spent a week shadowing a PGT at AIIMS Bhubaneswar last year. He gave me three hours a day for five days. He let me sit in the resident room, walk on the round, watch the handover, eat at the canteen.

Three things I did not expect:

The paper list is the most carefully maintained artefact in the building

Every PGT carries one in their coat pocket. It is folded into quarters along precise creases. It is updated in pencil, then re-copied in pen if the day's annotations make the original illegible. Loss of the list is treated as a near-miss event.

Look at one closely and it is not a list at all. It is a dense two-dimensional layout that has been refined over years: bed number in the left margin, name and age abbreviated to the minimum that distinguishes, diagnosis in a shorthand that would not survive contact with a coding system, then a column of pending items with a box beside each. Ticks, arrows for trends, a circled item meaning "chase this before you leave." Some residents rule a vertical line down the page to separate "done" from "pending"; some use the right edge for the numbers they will need on the round.

The re-copying is the part I underestimated. It looks like wasted effort — twenty minutes of transcription that a computer would eliminate. It is not. Re-copying the list is a re-read of every patient, performed daily, at a desk, with no interruptions. The resident who copies out twenty-eight patients in the morning has just thought about twenty-eight patients in sequence. Any tool that removes the transcription without replacing that pass has taken something away.

WhatsApp is the primary clinical communication channel

Not the EHR. Not the hospital pager (those exist but are slow). Lab results from the lab manager go to a unit WhatsApp group. The night-shift senior gets called on WhatsApp. The decision to escalate goes through WhatsApp. The handover, in many units, is a voice note on WhatsApp.

It is worth being honest about why it wins, because the reasons are good ones. It is already installed on every phone in the building. It works on a bad 2G signal in a basement corridor. The read receipt tells you the senior has actually seen it, which no pager does. The group is the unit — add a new PGT to the group and they are onboarded. And a voice note takes fifteen seconds to produce while walking, against two minutes of typing with one thumb.

There is a real cost, which nobody in the building was pretending not to see: clinical information about identifiable patients sitting in a consumer messaging app, on personal phones, backed up to personal cloud accounts, in a group whose membership nobody audits. That is not an argument for a policy telling people to stop. Policies telling people to stop lose to 2G in a basement corridor. It is an argument for the handover leaving the clinical system already composed, already minimised to what the receiving doctor actually needs, so that what lands in WhatsApp is the smallest useful message rather than a screenshot of a chart. Why handover defaults to WhatsApp and not email is a longer version of this argument.

The handover is the most stressful eight minutes of the day

Not the admissions. Not the codes. The handover. Because everyone knows that the things you forget to mention will not be remembered by the receiving doctor. Twenty-eight patients in eight minutes is impossible — and known to be impossible — and done anyway, because there is no slack in the schedule for a longer one.

Seventeen seconds per patient. Watch it happen and you see the compression strategy the residents have evolved: most patients get a status word — "stable," "for discharge," "awaiting ortho" — and the entire information content of the handover is concentrated in the three or four patients who get a full sentence. The skill being exercised is triage of attention, not communication. And it works, mostly, until the patient who needed a sentence got a status word instead. That is the CCF case, and every unit has one.

The fourth thing, which I only noticed afterwards

Nobody complained about documentation. Not once, in five days.

I had arrived expecting the American story — the physician buried in note-writing, clicking through templates, the EHR as the thing standing between the doctor and the patient. That is a real problem in the systems where it is a problem. It was not the problem in that building. The complaint, when it came, was always about holding the state of twenty-eight patients in your head across a shift boundary — which is not a documentation problem at all.

If I had built from my assumptions rather than from the week, I would have built a better note editor. It would have been used by nobody.

What it actually taught me

The existing tools — EHR, paper, WhatsApp — are not failures. They are adaptations. The PGT didn't choose paper because they don't know about EHRs. They chose paper because the EHR has no notion of "the ward list as a working surface" — and so paper does the job better.

Paper wins on the things that matter at 7 a.m.: it opens instantly, it never logs you out, it works with one hand while the other holds a file, it takes annotation faster than any input method, and it survives a dead battery. Any replacement has to beat it on those axes first and be clever second. Most clinical software gets this exactly backwards.

The implication for what we're building: do not try to replace paper. Replace what paper is being forced to do because nothing else fits. The patient list. The handover. The carry-across-the-shift-with-you context.

Three design rules fell out of that week, and we have not broken them since:

1. The list is the product. Not the patient record — the list. Everything else is reachable from it in one tap.
2. It has to be in the same pocket as the list it replaces. Phone-first is not a stylistic choice; the coat pocket is the constraint the paper list was designed around.
3. Optimise for the shift boundary. The eight minutes at 8 p.m. is where the harm concentrates, so it is where the software should be doing its best work.

That is why Medows lives on a phone, syncs in real time, and treats the handover as the most important affordance — not the documentation. The PGTs at AIIMS taught me that. Worth saying out loud.