How to calculate a completion rate without inflating it

The completion rate has three pieces and all three get argued over: the numerator, the denominator and the time window. Moving any of the three shifts the result by several points without anything changing in the patient's reality.

The honest numerator counts attended appointments. A booked appointment the patient does not use is recorded as a missed appointment and leaves the numerator, because the exam is still undone and the result still has not reached the treating physician. CMS pushes its numerator one step further out: the measure for closing the referral loop scores a referred patient only when the referring clinician receives a report back from the clinician the patient was sent to.

Is it measured on the booked appointment or the attended one?

1. Fix the window before starting: an order counts as completed if the visit happens within a declared period from when it was written.
2. Freeze the denominator on the same day you fix the window.
3. Count in the numerator only what was recorded as an attended appointment in the provider's own system.
4. Publish the exclusions one by one, with how many cases each of them removed.
5. Compare against a control group running over the same period.

A month-end figure that rises and falls with seasonality is not a comparable rate. Our methodology declares the window, the exclusions and the comparison arm before showing any figure, because the first question a competent buyer asks is what the denominator is.