Missing Referral or Intake Backlog? Trace the Fax Before Resending

The trace sheet below is an original editorial method for administrative investigation. Its examples are synthetic. It does not set clinical priorities or require staff to delay a time-sensitive escalation under their organization's existing policy.

Start with the reported referral and the available send evidence

Begin with what the reporting party knows. A referral coordinator calling about a "missing" packet typically has some combination of a patient name, a send date, a destination fax number and possibly a transmission confirmation page. Capture each item as stated, and note which ones are documented versus remembered.

The send evidence question is narrow: does the sender hold a confirmation that the transmission completed to your number? A confirmation page with a date, time, page count and destination number is useful evidence. A verbal "it went out Tuesday" is a reported event to record and corroborate. Ask for the number the packet was sent to and compare it against the number your intake monitors. Misdirected sends to a retired or departmental number are a plausible failure to rule out early, though how common that is in your organization is something only your own trace log can tell you.

Record the sender's evidence on the trace sheet before you look at anything on your side. Doing it in this order keeps the investigation honest; you are checking whether your records match a stated event, not building a story backward from the first thing you find.

Trace transmission, receipt and intake processing separately

The core of the method is refusing to collapse four checkpoints into one yes-or-no answer.

Checkpoint 1: Transmission

Evidence lives with the sender: confirmation report, destination number, page count, time. Outcome options: confirmed complete, confirmed failed, or no evidence.

Checkpoint 2: Receipt

Look for an inbound record in the systems your organization uses. Time, page count and originating number can narrow the search, but inspect the packet under your approved access rules before treating it as the referral in question. Outcome options: matching packet found and usable, found but incomplete or unreadable, or receipt unconfirmed. An unavailable log is different from a searched log with no match.

Checkpoint 3: Inbox processing

Evidence is whatever your intake workflow produces when a document is opened, classified, attached to a patient or routed to a queue. Outcome options: processed and routed, processed but misfiled or misclassified, or sitting unopened.

Checkpoint 4: Staff action

Evidence is the task, scheduling attempt, outreach note or referral record that should exist if the packet was worked. Outcome options: action taken, action assigned but not started, or no action and no assignment.

The AHRQ Workflow Assessment for Health IT Toolkit frames health IT as something that affects both clinical and administrative workflow and that can, once implemented, supply information to help reorganize that workflow (AHRQ). That source supports examining workflow. It does not validate this fax-specific checklist. For the proposed trace, record which checkpoints are observable and which remain unknown.

Transmission-to-work trace sheet

Use one row set per investigation in an approved internal system. Add the search range, systems checked and unresolved questions. When reviewing patterns, distinguish confirmed failures from unobservable steps; a small convenience sample does not establish your organization's overall failure rate.

Distinguish an unreceived packet from a packet awaiting staff action

The following cases show three possible evidence patterns, not an exhaustive diagnosis list. The examples below are synthetic and illustrative only. They are not Linear customer experiences or measured results.

Example A: Destination mismatch (synthetic)

Sender reports a send on the 3rd with a confirmation page. Your inbound log shows nothing from that originating number on the 3rd or 4th. No matching processing or action record has yet been found. The confirmation page lists a destination number that was retired when a department moved. Finding: the destination on the report differs from the monitored line, and receipt on the monitored line is unconfirmed. Check whether the retired number forwards elsewhere. Provide the verified intake number, arrange a replacement if needed, and ask the sender to correct its directory. Keep the original route investigation open until its status is understood.

Example B: Received, unworked (synthetic)

Sender has no confirmation page but gives a send date. Your inbound log shows a 14-page document from that number within the stated window. An authorized staff member confirms it is the reported packet and checks its completeness. Processing shows it was classified as "records" rather than "referral" and routed to a general scanning queue. No task exists. Diagnosis: the packet arrived and stalled at processing. A resend would create a second copy in the same misrouted queue. Correct response: reclassify, route to the referral queue, assign an owner, and note the classification error for pattern review.

Example C: Received, assigned, not started (synthetic)

All checkpoints through processing look fine. A task exists, assigned to a coordinator eight days ago, untouched. Diagnosis: this is an intake backlog item, not a transmission problem. The referring office experienced it as "missing" because nobody called the patient. The next action is to assign recovery of the pending work and record it on the trace. Our companion piece on reducing referral backlog covers that side.

Use the last verified checkpoint to direct the next check. A matching but unreadable packet still needs document recovery. A complete packet without an assigned task needs workflow recovery. A send report without a matching receipt leaves the cause unresolved; it does not prove a line or routing fault.

Document the investigation before requesting another copy

Before or alongside a replacement request, record who reported the issue, the evidence at each checkpoint, the last verified step, remaining uncertainty and the reason for requesting another copy. If you do request a resend, note the request time and the number given, so the second packet can be matched on arrival and the first one, if it later surfaces, can be recognized as a duplicate.

Keep both the original reported send time and the replacement request time. A resend should not overwrite the age of the unresolved request. Assign one owner to reconcile any later-arriving copy with the active work.

If you are reviewing software for this work, see how Linear Health handles referral coordination, then book a demo and ask to review a timestamped receipt-to-action trace on a synthetic sample packet.

Fix the observed failure and check for duplicate follow-up work

Each diagnosis has its own fix, and each fix should be followed by a duplicate check.

• Verified destination mismatch: correct the number at the source, check any forwarding path, and watch for the replacement.
• Misclassification or misrouting: fix the specific document, then look at whether the classification rule, template or staff habit that caused it will recur.
• Assigned but unstarted: move it into normal backlog handling with a visible owner and due point.

The duplicate check matters because referral problems attract parallel effort. While the trace was running, a front-desk colleague may already have called the patient, the referring office may have resent to a second number, or a second task may have been created from the resend. Search the patient record for open tasks, recent outreach notes and more than one inbound document from the same originating number before closing the case. Closing a trace with two live tasks on the same referral is how a missing-packet problem becomes a double-contact problem.

How to evaluate whether your tools can support this trace

The trace method only works if each checkpoint produces inspectable evidence. That is a tooling question as much as a process question. Linear Health's Operational AI Evaluation Kit uses a four-level scale for evidence: not addressed, claimed, demonstrated and verified, and it asks vendors to show one end-to-end trace on a workflow that matches real operating conditions and to define completion in observable terms, including the final status and where it is recorded. The kit also flags a demonstration that ends with a message, draft or task instead of a confirmed outcome as a red flag. Those questions map directly onto the four checkpoints above.

As editorial guidance, not a measured finding, apply the same standard to your current stack and to any candidate replacement:

1. Can you see an inbound receipt record with originating number, time and page count? Score it: claimed, demonstrated or verified in your own environment.
2. Can you see when a document was classified, by whom or what, and where it was routed?
3. Can you see the task or action that followed, with an owner and timestamp?
4. Can you run the trace on an ordinary packet with a missing page or an ambiguous patient match, not only on a clean example?

The companion article on automating fax processing in a medical office covers the broader workflow. For this investigation, the purchase question is narrower: can the proposed setup show which packet arrived, what happened next and who owns an unresolved step? Book a demo with a synthetic packet and ask the team to show the available records and identify any step that still depends on staff.

Frequently asked questions

Sources

Sources accessed November 3, 2026.

• AHRQ, Workflow Assessment for Health IT Toolkit.
• Linear Health, Operational AI Evaluation Kit.