A new Slop University paper links a blood donation network's "attended" booking flag against its own phlebotomy outcome log, finding a fifth of appointments logged as attended were early-stopped or self-withdrawn draws that still triggered the same next-eligible reminder as a completed donation.
A 46-centre audit links the flag that starts a donor's next-eligible text to what actually happened in the chair, across 142,387 appointments
A blood donation appointment produces two records of what happened in the chair, and only one of them tells the donor when they can come back. A new Slop University paper puts those two records side by side across an entire network, and finds they agree only moderately.
A reminder clock doesn’t know it’s reading a field built for something else. It just keeps counting, exactly as configured, long after the thing it was counting stopped meaning what everyone assumed.
— Professor Verity Marris, Director, Trajectory Analytics Group, School of Emergent Priorities
The School of Emergent Priorities team linked the booking system’s “attended” flag, the sole trigger for the next-eligible-to-donate text, against a separate phlebotomy outcome log across 46 centres and seven months of appointments. A fifth of the visits logged as attended were early-stopped or self-withdrawn draws that never reached full volume, and every one still received the same standard-interval reminder as a completed donation; incomplete-draw donors went on to return well past the interval the reminder had assumed for them. A supplementary “volume confirmed” toggle, already piloted at fourteen centres, left the underlying mismatch statistically unchanged.
“What surprised us wasn’t the mismatch itself,” said Dr Anneke Tolan, Senior Research Fellow. “It was how ordinary the toggle’s non-effect turned out to be — an optional field, added in good faith, absorbed by exactly the practice it was meant to interrupt.”
The University reads the result as one more instance of a question its measurement schools keep returning to: a field logged as evidence something happened isn’t always evidence of what it is later trusted to mean.
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.dwyux3.