CP metrics that satisfy state reporting

Report median and p90, not average. The median describes the program on a normal day; p90 exposes the referrals that fell into a gap. A program with a good median and a terrible p90 has a coverage problem in a specific territory or protocol, and that is a fixable finding.

Completion rate on its own invites suspicion. Completion rate reported alongside the reasons for every abort, could not reach, patient declined, clinical escalation, reads as an operating program rather than a filtered number.

This is the metric programs avoid and reviewers respect. Escalations are not failures; they are the program finding sick patients. Reporting them with dispositions demonstrates that the pathway is clinically real.

The hospital already owns the data required for this view: its own ADT and registration records. Joining enrolled patients to their own ED visits and readmissions before and after enrollment produces a defensible single-site result without waiting on an HIE feed.

Report the comparison group and the caveats. A number with visible limitations is more persuasive than a number without them.