# Cancellations are an information problem

> Most day-of-surgery cancellations are caused by information that existed weeks before the operation but reached the anaesthetist too late to act on. Because the limiting factor is timing rather than clinical thoroughness, cancellation rates respond to moving assessment to the point of booking far more than to assessing more carefully at the clinic.

*Source: https://getpathways.ai/insights/why-day-of-surgery-cancellations-are-an-information-problem*

A case is cancelled at 07:12. The patient has been fasted since midnight, is
gowned, and has arranged for a family member to take the day off work. The chart
is opened properly for the first time and it shows uncontrolled diabetes and an
SGLT2 inhibitor that was never withheld. The list loses a slot it cannot
backfill at that notice.

Nothing in that sequence is a clinical failure. The anaesthetist made exactly
the right call with the information available at 07:12. The problem is that the
same information was available on the day the patient was booked, nineteen days
earlier, when it could have changed the outcome instead of ending it.

## The pattern generalises

Look at a run of cancellations and the specifics vary while the shape does not.
Four causes account for most of them:

- **Medication management that was never planned.** SGLT2 inhibitors are the
  standing example, because withholding must begin days ahead and because
  patients describe them as "a tablet for sugar".
- **Comorbidity that was never optimised.** Poor glycaemic control, uncontrolled
  hypertension, anaemia — all modifiable, all requiring a window.
- **Risk that was never screened.** Undiagnosed obstructive sleep apnoea changes
  the anaesthetic plan, but only if the eight STOP-Bang questions were asked.
- **Logistics that were never confirmed.** No escort home, fasting instructions
  misunderstood, no discharge support.

In each case the determining fact existed well before the morning of surgery. It
was not discovered late because it was hidden. It was discovered late because
nobody looked until the point at which looking no longer helps.

## Why "assess more carefully" does not fix it

The instinctive response to a cancellation is to tighten the pre-admission
process — a longer questionnaire, a more thorough clinic visit, another
checklist. This improves the quality of assessment for the patients who reach
the clinic, and does nothing for the patients who do not.

That is the constraint. A pre-admission clinic booked out three weeks ahead is
not short of clinical capability; it is short of a mechanism for deciding who
should be in it. Slots fill in booking order, which reliably produces a clinic
of healthy ASA 1 and 2 patients while the frail 82-year-old with poor functional
capacity is handled by a fifteen-minute phone call. Assessing the people in the
room more carefully cannot reach the people who were never in it.

## The lever is timing

If the determining information is knowable at booking, then the question worth
asking is not how thorough the assessment is but how early it happens.

Moving assessment to the point of booking changes what a finding *is*. A
patient-reported HbA1c of "around nine" discovered nineteen days out is an
optimisation window. The same finding discovered at 07:12 on the day is a
cancellation. The clinical content is identical; only the remaining time
differs.

This is also why the intervention has to cover every booked patient rather than
the ones who look high-risk. The patients who cancel lists are, by definition,
the ones whose risk was not visible in advance — if it had been visible, they
would have been assessed. Screening the obviously complex patients catches the
cases you were already going to catch.

## What this asks of a service

Screening every patient at booking is not realistic as a manual process. It
means contacting several hundred patients a month, taking a structured history
from each, reconciling their actual medications against a referral letter that
may be months old, and applying consistent screening criteria across all of it —
work that scales linearly with list volume and that no pre-admission clinic has
the staffing to absorb.

That is the specific problem Pathways was built for: a voice agent that takes
the structured history from home, medication capture that reads the boxes rather
than trusting recall, and deterministic screening that applies the same
guideline-referenced criteria to every patient. What it produces is not a better
clinic visit. It is a queue that is ordered by risk, early enough for the
ordering to matter.

The measure of whether it works is not how good the assessments are. It is
whether the fact that would have cancelled the case arrives while there is still
time to act on it.