Forty sites entering patient data once, with every change logged.

What was going wrong

A coordinator at a hospital in Vijayawada fills in a paper form at the bedside. It is scanned and emailed, or sometimes carried. Somebody in Hyderabad types it into a spreadsheet. A monitor later finds a value that cannot be right, emails the site, waits, and eventually gets a correction on a fresh scan. Now two versions of that patient’s record exist, and the reason for the change lives in an email thread.

Multiply that by forty sites in six cities. Data arrived in every form a human can send it in. Checking one record meant chasing three people, and the chase was the job: coordinators spent their days on email rather than on patients. Nobody could say with certainty how many questions were open on a trial at any moment.

The part that worried the organisation most was not speed. It was that the record of what changed, who changed it and why was spread across scans, spreadsheets and inboxes. An audit meant reassembling that by hand and hoping nothing had been missed.

What we built

A platform where the site enters the data once, at the site, on the screen the coordinator already has open. The form is the record. Checks run at the point of entry rather than after the fact: a value outside its expected range, a date that cannot follow another date, a field left blank that the protocol requires, all are raised while the coordinator is still looking at the patient’s notes and can answer them.

Every change is logged. Who made it, when, what it was before, and the reason they gave, kept with the record rather than in an inbox. Nothing is overwritten and nothing is deleted, so the history of a value is part of the value.

Queries stopped being email. A question raised by a monitor appears against the exact field it concerns, assigned to a named person at the site, visible to both sides until it is answered. Anyone can see how many are open on a trial, at which sites, and how long they have been waiting, which turned a standing argument into a number on a screen.

Access is by role and by site, so a coordinator sees their own site and nothing else, and every login and every export is recorded.

What the owner was left with

Every record captured once, with a full history of every change attached to it. Queries and audits are answered from the platform rather than from three people’s memories. Coordinators got their days back, which was the change the sites noticed first.

The platform runs in the organisation’s own cloud account, with the backups restored in front of them rather than assumed, because a trial’s data cannot be the sort of thing anybody hopes about. Site staff were trained site by site, and the written guides live with the organisation.

The handover review said plainly what was still manual: a handful of laboratory results still arrive as files from instruments that speak nobody’s language, and those are still checked by a person. That is the piece we would look at next, and the note explains how.

This sits under business systems, with the hosting and the restores treated as part of the job rather than as somebody else’s problem.