Referrals that start in the chart

Every referral program we have worked on has the same shape of problem. The criteria are written down, the intake team is ready, and the clinicians are willing. Then the numbers come in low, and the instinct is to blame motivation.

It is almost never motivation. It is that nobody knew the program applied to the patient who was in the room ninety seconds ago.

Discovery is the actual bottleneck

A referral needs three things to happen in order: someone notices the patient might qualify, someone checks whether they do, and someone sends the referral. Programs tend to invest in the third step, because it is the one that produces a record. A portal gets built. A form gets designed. Training happens.

But the first two steps happen in a chart, during a visit, under time pressure. If checking eligibility means opening another tab and typing a member id, it will not happen during the visit. It will happen later, for the patients someone happens to remember.

Moving the check to where the chart is

The version that works puts the check in the chart itself. When the patient’s record opens, the program’s criteria run against the data already there: coverage, the problem list, age, the provider of record. No one asks for it. By the time the clinician looks up, the answer exists.

That changes what the clinician is being asked to do. Not “go and find out whether this patient qualifies,” which is work, but “this patient qualifies, do you want to refer,” which is a decision. Those are very different asks at 3pm on a Thursday.

Say what could not be confirmed

The temptation, once you have automated eligibility, is to let the system decide. A patient either qualifies or they do not, and the ones who do not disappear.

This is wrong in a specific and costly way. Chart data is incomplete constantly. Secondary coverage is not recorded. A diagnosis lives in a visit note instead of the problem list. If a missing field reads as a failure, the program quietly excludes exactly the patients a human would have caught.

So missing data should be its own answer. Not “does not qualify,” but “could not be confirmed,” with the referral still available and a note attached explaining why the clinician thinks it is worth a look. The intake team resolves it. That path turns out to carry a meaningful share of real enrollments.

What this looks like in practice

The panel leads with a conclusion, not a form. It shows the criteria as evidence for that conclusion, with the failing ones first. It pre-fills the referral from the chart. And when the referral comes back with a status, it comes back to the same panel, so the referring practice sees what happened without chasing anyone.

None of that is technically hard. The hard part is deciding that the workflow belongs in the chart instead of next to it.