From Abnormal Result to Completed Follow-Up: Reviewing Delayed Diagnosis in Primary Care

By Kanwar Partap Singh Gill, MD · Family medicine physician · 5 November 2026

• A loop, not a moment
• Nine events in the results loop
• A fictional loop, link by link
• Acknowledged is not communicated
• Results that arrive by another route
• Hindsight and the visible-in-retrospect problem
• System design and individual responsibility
• The patient's part, assessed fairly
• Causation is a separate question
• What a fair review can conclude
• What a review of this issue can deliver

A loop, not a moment

Delayed-diagnosis claims in primary care are usually framed as a single failure: the doctor missed the cancer. The record often shows more than one failure point. Some cases do turn on a judgment at a visit; this article addresses the others, where the record shows a loop of separate tasks running from the decision to order a test to the moment the patient receives the care the result called for, and that loop passes through laboratories, radiology groups, inboxes, staff pools, telephone lines, scheduling offices and the patient's own life. A review that treats the loop as one event will misplace responsibility, in one direction or the other.

This article describes how a family physician reviews that loop link by link, how hindsight distorts it, how system design and individual judgment divide it, and where the generalist review stops.

Nine events in the results loop

1. Order. A clinician decides a test is needed and orders it, with or without a stated indication and priority.
2. Specimen or performance. The blood is drawn or the study performed. A test ordered but never done is a common, quiet break: the patient never went, the order expired, the outside facility had no record of it.
3. Result. The laboratory or radiologist finalizes the report. A radiology report may contain a specific recommendation ("follow-up CT in three months"), a vague one ("clinical correlation recommended") or none.
4. Routing. The system sends the result somewhere: to the ordering clinician, a covering clinician, a shared pool, or, for an outside study, a fax queue awaiting scanning.
5. Acknowledgement. Someone marks the result as reviewed. This shows that a user took an action on the item, and little more.
6. Interpretation. A clinician decides what the result means for this patient: normal, abnormal but expected, or abnormal and requiring action.
7. Patient notification. The patient is told, by call, letter, portal message or at a visit, in a form the patient can understand.
8. Action. The clinician orders the next step: a repeat test, imaging, a referral, a change in treatment.
9. Completion. The next step actually happens and its result returns to the loop. Until then, the loop is open.

Each event has a different owner, leaves a different trace and fails for different reasons. The review establishes, for each disputed test, which events the record documents, which can be inferred from other evidence, and which remain unknown.

A fictional loop, link by link

Read quickly, this is a covering clinician who saw the result and did nothing. Read link by link, it raises distinct questions. What did coverage of a departing clinician's inbox include, and was the covering clinician expected to interpret results or only to triage critical values? Was last year's normal count visible alongside the new one? Was the template letter generated automatically on acknowledgement? Is there a telephone note or staff message elsewhere in the chart?

The answers can point either way. If a staff message asked the patient to schedule a visit and three call attempts are recorded, the picture is of a loop that the practice tried to close. If no one was assigned to interpret the departing clinician's results and the letter told the patient nothing, the picture is of a process gap. A new iron-deficiency anemia in a man of that age is the kind of finding that ordinarily prompts a search for its source, so the review takes the seven-month interval seriously; it also takes seriously the possibility that the record produced is not the whole record.

Acknowledged is not communicated

The decisive distinctions in a results case usually fall between adjacent links. An acknowledged result is not an interpreted result. An interpreted result is not a communicated one. A communicated result is not an acted-on one, and an ordered follow-up is not a completed one.

Automated features complicate the picture. A result may be released to a patient portal automatically, before or without a clinician's comment; that is not the same as a clinician explaining it. A template letter saying "your results have been reviewed" communicates that something happened, not what it meant. An inbox can be cleared in bulk. None of these is a failure in itself. Each is a reason the reviewer cannot infer one link from the presence of another.

Results that arrive by another route

Some of the hardest cases involve results that never entered the loop as results. An outside imaging report may arrive by fax, be scanned as a document and filed under a general heading, with no rule routing it to any clinician. The review then asks when the report was received, when it was scanned, how it was indexed, whether it generated a task, and whether a later visit offered a natural opportunity to see it. The scanned date printed on the page may be weeks after the report was created and days after it arrived.

Other results are ordered by someone else and copied to the primary-care clinician: a chest CT from an emergency visit with an incidental nodule, a laboratory panel from a specialist, a test pending at hospital discharge. Whose follow-up task such a result became is often unclear in the record and sometimes unclear in the organization's own process. The review identifies who ordered it, who received copies, whether the discharge or visit summary flagged it, and whether any clinician accepted responsibility for the next step. "Everyone was copied" can mean that no one owned it.

Hindsight and the visible-in-retrospect problem

Once the diagnosis is known, the earlier result looks like a signpost. When it arrived, it may have been one of dozens of items in a clinician's inbox that morning, marginally abnormal, explained by a known condition, or overshadowed by a more urgent problem. Hindsight bias is not a reason to excuse a missed result. It is a reason to reconstruct what the result looked like when it arrived.

The reconstruction asks concrete questions. Was the value flagged as abnormal, and how far outside the reference range? Was a prior value visible for comparison? Did the report contain an explicit recommendation? Was a benign explanation documented? Was a recheck planned, and was the planned interval clinically reasonable? Watchful waiting with a scheduled recheck is a decision, not a delay, if it was reasonable and the recheck happened.

System design and individual responsibility

Results management is a designed process. Someone decided where results route when the ordering clinician is away, whether outside reports are scanned and to whom, whether pending tests are tracked, who handles results for clinicians who have left, and what happens when a patient cannot be reached. When those decisions are missing or poorly made, an individual clinician may never have had a fair opportunity to act. When they are sound and followed, an individual lapse stands out more clearly.

The review describes, at the clinical level, which parts of the loop belonged to the individual clinician and which to the organization's process, and what evidence supports each. It does not allocate legal responsibility between them; that belongs to counsel and the tribunal. ONC's SAFER Guide on Test Results Reporting and Follow-Up sets out recommended practices for organizational result management, and AHRQ's Tools to Improve Diagnostic Safety include resources for learning from diagnostic safety events and for engaging patients in the diagnostic process. Both describe what a process is meant to do; neither sets the standard of care in an individual case.

The patient's part, assessed fairly

Patients are part of the loop. A patient may have been told and chosen to wait, may have changed telephone numbers, or may not have kept a referral appointment. Documented outreach deserves weight: three recorded call attempts and a mailed letter differ from none. A missed appointment, though, is not an informed decision to decline, and a letter in English is not communication to a patient who reads another language. The distinction is developed in When the Chart Says Refused.

Causation is a separate question

A finding that the loop broke does not establish that the break changed the outcome. Whether a diagnosis made months earlier would have been at an earlier stage, and whether earlier treatment would have altered the course, usually turns on the biology of the disease, its expected progression and the treatments available. Those are questions for oncology, radiology, nephrology, cardiology or whichever specialty fits the condition. A primary-care reviewer can frame that question precisely and identify the specialist needed. He should not answer it.

What a fair review can conclude

• A defensible decision under uncertainty: the result was reviewed, interpreted reasonably on what was known, and a plan was made.
• A documentation gap without a care failure: the follow-up happened, but its record is thin or kept elsewhere.
• A process failure: the result routed to no responsible clinician, or a departing clinician's queue was never reassigned.
• An individual lapse: a result was seen, its significance was apparent on the information then available, and no action followed.
• Indeterminate: the record cannot distinguish among these, and the missing records are named.

Each is a legitimate finding, and more than one may apply to the same loop. The evidence decides which.

What a review of this issue can deliver

A result-loop chronology for each disputed test, with every event marked documented, inferred or unknown; an analysis of which links held and which did not; a statement of where, on the information then available, follow-up was clinically indicated, or why the course taken was reasonable; a list of the records still needed, such as routing configuration, coverage schedules, outreach records and, through a qualified EHR forensic specialist, access data; and identification of the specialist needed for causation. This is the core of Primary Care Decisions and Diagnostic Follow-Up, built on the method in The Medical Record Has More Than One Clock. Organizations examining their own result-management process can use Clinical Governance and Care Delivery Review.