A waiting list is not a demand problem. It is a flow problem.

The list reaches a number that becomes politically visible — to a board, a regulator, a ministry — and the response is almost always numerical. Add sessions. Add lists. Add a weekend programme. Sometimes add a partner facility.

For a quarter, it works. The number comes down and there is a slide showing it. Then, somewhere between month five and month nine, the line bends back upward. Not to where it started, usually, but far enough that the improvement gets redescribed as "sustained pressure" rather than a fix.

I have watched this cycle run three times in the same institution under three different executive teams. Each team was competent. Each intervention was defensible. And each one treated the list as the disease when the list was the symptom.

A backlog is an output. It is what a system produces when work enters faster than work resolves, sustained over time.

Adding capacity raises the resolution rate temporarily. It does nothing to the entry rate, the rework rate, or the structural constraints that convert a booked case into a cancelled one. So the diagnostic question is not "how much more capacity do we need?" It is: what in this system is manufacturing new delay every week?

That question leads somewhere different. It leads to the referral that arrives incomplete and generates a second outpatient visit. To the pre-assessment scheduled after the theatre slot rather than before it, so that a predictable share of lists cancel on the morning. To the scheduling rule written to protect surgeon utilisation which, entirely logically, produces idle beds downstream. To the pooled-versus-named-list decision nobody has revisited in six years because it is politically expensive.

None of these are capacity problems. All of them are flow problems. And all of them will regenerate a backlog no matter how many additional sessions are funded.

In one mandate, a surgical backlog of approximately 4,000 cases was eliminated in under twelve months — not by adding sessions, but by redesigning the pathway that kept regenerating it. Once the system stopped producing new delay, the historical backlog became finite, and a finite backlog is a project rather than a condition.

The underlying behaviour is not specific to surgery, and it is not mysterious. Any queueing system operating close to its capacity ceiling responds non-linearly: as utilisation rises toward full, small increases in variability produce disproportionately large increases in waiting. This is a structural property of queues, not a feature of healthcare. The practical consequence is that reducing variability is frequently a stronger lever than adding capacity — and it is almost always cheaper.

Which is why the theatre list that starts late for a predictable reason is worth more management attention than the theatre list that is simply full.