The small automation worth a full-time job

A community pharmacy chain that runs seasonal vaccination clinics across its branches. Every autumn and winter the clinics are busy, and every vaccination has to be recorded in the pharmacy’s NHS services portal after the event.

Recording a vaccination is not clinical work, but it was being done by pharmacists because the record forms part of the clinical documentation. For each patient someone had to open the portal, find the patient, click through several screens, enter the vaccination details and save, then start again. Three to four minutes a time, and a great many clicks.

The pharmacy had already improved the first half of the process with a structured spreadsheet that made capturing patient details at the clinic quick. The second half was the bottleneck. Someone still had to read the spreadsheet and type it all into the portal. In a busy season that duplicated effort scaled with every jab given, and it was consuming the time of the most qualified people in the building.

We could not change the portal, which belongs to someone else and offers no convenient way for other systems to talk to it. So we automated the work the way a person does it.

The pharmacy’s spreadsheet became the source of truth and the work queue. Once a clinic has finished and the sheet is checked, a member of staff opens a simple control panel, signs in to the portal for that run and starts the batch. A browser running on a server we manage then works through the sheet: it finds each patient, moves through the screens, enters the details, selects the right options and saves. Each row’s outcome is written back into the same sheet, so staff can see what succeeded, what was skipped and what needs a human look, without reading a technical log.

Three design choices mattered. The batch is validated before it starts, because automation makes a wrong input faster too. One problem record does not stop the run; it is marked and the run continues. And every entry is saved for review rather than submitted, so a pharmacist still checks and completes each record. The automation handles the clicks. The pharmacist keeps the clinical responsibility.

We started with flu, deliberately, to prove the workflow in a narrow setting, and built it so that other services could follow the same pattern. Running the browser centrally meant nothing had to be installed or kept up to date on branch computers.

Pharmacists stopped typing vaccination records. The pharmacy reports that over the vaccination season the automation released capacity equivalent to a full-time pharmacist, time that went back to patients and to the work that actually needs a pharmacist.

It is a small piece of software next to some of our other work, and that is rather the point. Some problems need a new platform. This one needed one repetitive task removed, and that was worth a whole role.