A woman opens a portal at half past nine in the evening. Two values are marked outside the reference range. There is no explanation, no context and nobody to ask.
By ten she has read four pages describing serious conditions associated with those values. By eleven she is frightened in a way that the result does not warrant and cannot be undone by a reassuring conversation three days later.
Nothing about the result was wrong. The failure was releasing a number into somebody's hands with no structure around it.
Step 1: explain what a reference range is
A range covering most healthy people, which means a proportion of healthy people sit outside it by definition. This single explanation prevents an enormous amount of avoidable fear.
Step 2: explain that a flag is not a diagnosis
An out of range value is a prompt to look further, not a finding. Saying this at the top of every result would change how people experience them.
Step 3: explain that results are read together
A single value means little without the others, the history and the symptoms. People read one number in isolation because that is how it was presented to them.
Step 4: explain the common innocent causes
Recent food, exercise, illness, medication, hydration and the timing of a sample. Naming these gives somebody a plausible explanation to hold while they wait.
Step 5: say who will make contact and when
Whether somebody is reviewing it, by when, and what happens if nothing is heard, which is the ground in turnaround time and results questions.
Step 6: never interpret a result automatically
An automated system states process and never meaning. There is no version of this that is safe to do and the boundary must be absolute.
Step 7: give a route to a person
A review service, a call back or a consultation, with the cost and the timescale stated. Somebody frightened at half past nine needs to know when they can speak to somebody.
Step 8: handle the urgent findings separately
Certain results require prompt contact regardless of the hour, and the process must not depend on somebody checking a portal.
Step 9: consider what is released without context
Releasing everything immediately respects autonomy and causes harm when unaccompanied. Where results are released directly, an explanation should travel with them.
Step 10: explain what a repeat test means
Repeats are routine and are read as ominous. Saying that a repeat is a normal next step rather than a sign of something found is worth stating explicitly.
Step 11: explain the onward routes
Further tests, a referral, a watchful approach or nothing at all. Setting out the range of possible next steps is reassuring because most of them are unremarkable.
Step 12: give people something to do
Bring the result to the requesting clinician, note any symptoms, and avoid searching the value. A practical instruction helps more than reassurance.
Step 13: publish the interpretation policy
Whether review is included, what it covers and what it costs, which is the ground in results timing questions labs should answer up front.
Step 14: be available at half past nine without interpreting anything
An assistant that explains what a reference range is, states that it cannot interpret anything, says who will make contact and when, and takes a request for the review appointment gives somebody at the worst hour exactly the right amount of help. The same explanation of what a test does and does not tell somebody appears in free hearing test questions and what happens, and checking the boundaries hold before launch is covered in how to test your chatbot before going live. See features, start at try, and the sector view sits under healthcare.