Closing the loop: results inboxes, critical values and referral tracking

Clinical follow-up tracking means every test result is reviewed and acknowledged by a named clinician, critical values reach someone within a defined time with escalation if they don't, and every referral or procedure decision is tracked until it is booked. The question to ask of any system is simple: who owns this result, and how would we know if nobody did?

This guide is for clinicians, medical directors and operations managers. It uses the results inbox and the Clinical Decisions Hub in the Helix EMR as examples.

Why loops stay open

• Results arrive after the patient has gone home, and the ordering doctor is off that day.
• A locum or visiting consultant ordered the test and has since left.
• The result arrives as a PDF by email, outside the record.
• Abnormal but non-critical results sit unread because nothing flags them.
• A referral is advised verbally and never created.
• A referral is created, but nobody calls the patient to book it.

None of these are unusual. They happen in every busy clinic, and they sit behind many complaints about delayed or missed diagnoses.

Two cases deserve special attention. The first is results from outside laboratories and imaging centers, which often arrive by email or on paper. Someone must attach each one to the right patient and route it to the right clinician, and the safest rule is that an external result is treated exactly like an internal one once it is in the record. The second is incidental findings on imaging reports: a nodule mentioned in the last line of a report requested for something else. These need an explicit decision and a follow-up plan, not just an acknowledgement.

Part 1: A results inbox with ownership

Results should come to the clinician, not wait to be found. A results inbox should:

• Collect lab, imaging and pathology results across all patients in one list.
• Filter by type, status and urgency, with separate counts for critical and overdue results.
• Move each result through clear statuses. In Helix these are New, Reviewed, Acknowledged and Actioned.
• Record who acknowledged each result, and when.
• Open the patient's chart from the result for context.

Results from your own laboratory are the easiest case. If your clinic runs an in-house laboratory information system on the same record, results post straight into the inbox with no scanning and no retyping.

Coverage and handover

Decide in advance who covers a doctor's inbox during leave. Name covering clinicians, and make overdue results visible to the medical director. Software can show overdue items; the rota decides who acts on them.

Part 2: Critical values that can't be missed

Critical values need a different process from routine results. Laboratories define critical limits for tests such as potassium, glucose and hemoglobin, and are expected to communicate them promptly to a responsible clinician, often with read-back to confirm the message was understood.

A good system:

1. Detects values beyond the facility's critical thresholds automatically.
2. Alerts the ordering clinician immediately.
3. Records notification attempts and read-back confirmation.
4. Escalates to a covering clinician or department head if the value isn't acknowledged in time.
5. Keeps an audit trail for quality review.

Helix supports facility-defined critical thresholds, read-back and tiered escalation. Unacknowledged critical results also appear as clinical flags in the pre-visit brief the next time the patient is seen.

Part 3: Tracking decisions until they are booked

A referral, a surgery decision or a 'see again in six weeks' is only useful if it happens. The receiving doctor's inbox handles one part: accept, decline, mark as seen, complete. But between accepted and seen there is often a phone call nobody makes.

Helix's Clinical Decisions Hub tracks referrals, surgery bookings and procedure orders from the moment the doctor decides until they are booked. It gives the patient relations team a board with these lanes:

Each card carries an urgency clock (emergency 4 hours, urgent 24 hours, routine 72 hours), and every contact attempt is logged. Surgery cards are due on the surgery date rather than on a countdown. Once booked, the visit sits in the normal clinic appointment scheduling calendar.

Part 4: Tell the patient

A result is not fully closed until the patient knows what it means and what happens next. Clinics lose trust, and sometimes patients, when normal results are never communicated and patients assume the worst, or when abnormal results are communicated without a plan.

• Agree which results can be shared directly with the patient and which need a conversation first.
• For normal results, a secure link or message is often enough. Helix's patient portal can send lab results through a secure link, with no app to install.
• For abnormal results, record who spoke to the patient, what was agreed and the follow-up booked.
• Record failed contact attempts, too, so the next person knows to try again.

A simple results policy

Software supports the process, but the process has to be written down. A one-page policy should cover:

1. Who is responsible for a result: the ordering clinician by default, with named cover during absence.
2. Target times for reviewing routine, abnormal and critical results.
3. The critical value list, thresholds and escalation chain, agreed with the laboratory.
4. How and when patients are told, by result type.
5. What happens to results for patients seen by locums or visiting consultants.
6. Which indicators are reviewed, by whom, and how often.

Part 5: Measure the loop

What gets measured gets closed. Useful indicators to review each month:

• Results not acknowledged within your target time, by clinician.
• Critical values: time from result to acknowledgement, and how many escalated.
• Referrals and decisions overdue against their clock.
• Decisions closed as lost, and why.

Treat these as internal quality measures and discuss them without blame. The aim is to find gaps in the process, not in the people.

Where AI helps, and where it shouldn't

AI can summarize the chart before a visit and flag what is outstanding. It should not acknowledge results or close referrals on a clinician's behalf. In Helix, the pre-visit brief surfaces flags such as unacknowledged critical results, and every action stays with a person. We look at documentation AI in AI medical scribes: how ambient documentation works.

For polyclinics, where referrals cross departments every day, see running a multi-specialty clinic on one EMR.