Moving hospital scheduling off unsupported software
The board on the corridor wall was not a backup. It was the system.
The program
| Sector | Health |
|---|---|
| System age | 19 years, unpatched for 4 |
| Scope | Scheduling replacement, phased by department |
| Parallel run | 11 weeks, bounded in the plan |
| Duration | 14 months |
| Status | Complete. Board removed |
Every department had a magnetic board on the corridor wall, and every board was more current than the software. That is the finding, and everything else in this program followed from taking it literally rather than treating it as a symptom of poor training.
The boards existed because the software had failed people in a specific way: it could not be updated from where the work happened, so it was updated later, by somebody else, from notes. The board was updated immediately by the person who knew. Once you see it that way, the requirement is not a better scheduling system. The requirement is to earn the board back — to be so obviously faster than a magnetic strip that nobody keeps the strip out of caution.
We phased by department and bounded the parallel run at eleven weeks in the plan, on day one. In week nine somebody asked to extend it, for good reasons, and we said no, also for good reasons: a parallel run with a movable end date is not a parallel run, it is two systems.
The boards came down department by department, and none of them came down because we asked.
They did not want a better system. They wanted to stop keeping a second copy of the truth on a wall.
>
— Yusuf Bramwell, Principal, Migration and Cutover
An unfaded rectangle
What is left is a clean patch of wall where a board hung for two decades, four screw holes, and a tray somebody has not got round to unscrewing.
