The waiting list that grew shorter

A hospital department cut its average wait from fourteen weeks to six without seeing a single extra patient per week. The work was entirely about the order of the queue.

The old list was first in, first out, with urgent cases pulled forward by a consultant reading referral letters each morning. This is a reasonable system and it had a hidden cost: every pulled-forward case pushed everyone else back by one place, and because urgency was judged from letters of varying quality, the pulling was inconsistent. Patients who were referred with a vague letter waited longest regardless of their condition.

The change was to separate the list in two. A short, protected stream handled cases meeting explicit criteria, with a fixed number of slots per week. Everything else ran strictly in order, with no pulling forward at all. The total capacity did not change. What changed was that the ordinary list became predictable, which allowed the department to tell each patient a date rather than a position, and allowed the clinic to plan.

The counter-intuitive part is the second effect. Average wait fell because the variance fell. In a queue where cases are constantly reordered, the clinic cannot batch similar procedures, so the mix each day is random and the slowest case sets the pace. With a stable order, the schedule could group procedures of similar length, and the same staff completed more.

The lesson is not that queue theory beats clinical judgement. It is that clinical judgement applied case by case to an unsorted list produces a system no one has designed, and the patients it treats worst are the ones whose letters were least articulate.