Lab Result Follow-Up Automation That Closes Every Loop

A lab result that arrives but never reaches the patient is two problems at once: a safety event and a liability. It almost never fails at the laboratory. It fails in the handoff, in the hours between a result landing in your system and a person taking responsibility for it.

The published range is wide enough to be uncomfortable. A systematic review of 19 studies found that between 6.8% and 62% of laboratory results went without documented follow-up in ambulatory settings, and between 1.0% and 35.7% of radiology results. A separate study of 5,434 records across 23 practices found that 7.1% of clinically significant abnormal results, about 1 in 14, were never reported to the patient.

This post lays out a closed-loop tracking workflow with clear human ownership: what arrives, who owns it, when the patient hears, and what gets logged. It also marks where automation genuinely helps, where it must stop, and how the Cures Act changed the callback you make.

Why lab result follow-up breaks in the handoff, not the lab

Results rarely go missing at the laboratory. They go missing in your office, during the gap between a result arriving and a named person acknowledging it. When nobody owns that gap, there is no due date, no reminder, and no proof of contact, so an unclosed result stays invisible until a patient asks.

The spread in the research is the finding. That review of 19 studies put failure to follow up laboratory results anywhere from 6.8% to 62%, and radiology from 1.0% to 35.7%. Rates that wide do not describe medicine. They describe process, and process is the part you control.

The most-cited notification benchmark is older but still instructive. In a 2009 study of 5,434 patient records at 23 practices, 135 of 1,889 clinically significant abnormal results were never reported to the patient. The authors also found that having an electronic record did not lower the failure rate on its own, and it sometimes went the other way when no tracking process sat on top of it.

That last point matters more than the percentages. Software does not close a loop. A named owner, a due date, and a logged contact close a loop. Software only makes those three cheaper to maintain.

What a closed-loop result workflow looks like end to end

A closed loop has six stages: the result arrives and is matched to its order, it gets tiered as critical, abnormal, or normal, the ordering clinician reviews and decides, the patient is notified with that interpretation, a follow-up action is scheduled, and the whole thing is logged with a timestamp.

Every stage has an owner and a timestamp. A loop without a logged close is not closed.

Two of those stages are the ones practices skip. The first is matching results back to orders. Malpractice risk guidance is blunt about it: reconcile the tests you ordered against the results you received, and do not treat the patient's next appointment as your reconciliation method. A specimen that never reached the lab produces no result, and therefore no inbox entry to notice.

The second is the close. If the record does not show who acknowledged the result, when the patient was contacted, and what was decided, the loop is open no matter how well the conversation went.

Memory and an inbox are not a tracking system

A shared results inbox tells you what arrived. A tracker tells you what is late. Those are different questions, and only the second one catches a result that never came back at all. If your practice runs on memory plus an inbox, the results most likely to hurt you are the ones nobody has noticed yet.

The difference is not effort. It is whether an unclosed result is visible to anyone.

A working tracker holds eight fields per order: ordered date, expected date, received date, tier, acknowledged by, patient contacted on, action taken, and closed date. Nothing exotic. The value is that any of those fields can sit empty, and an empty field is a query you can run every morning.

The report nobody runs is ordered but never resulted. A specimen lost in transit, a patient who skipped the draw, an imaging order that was never scheduled: none of them generate an inbox item, so an inbox can never surface them.

Route by result type, and name who owns a critical value

Route results by tier, not by whoever opens the inbox first. A critical value needs a phone call to a named clinician within minutes. An abnormal result that is not critical needs a review queue with a due date. A normal result needs a notification that is accurate and quick. Three tiers, three clocks.

The Joint Commission goal on communication requires accredited organizations to report critical results on a timely basis, with written procedures naming who reports them, who receives them, and the acceptable time frame. Your independent practice may never be surveyed against that standard, and it is still the cleanest template available. Borrow it.

Two details do most of the work here. Agree with your laboratory and imaging centers that critical results are telephoned first and transmitted electronically second. Then put a person behind that phone number, with a named backup and a documented handoff when the first person cannot be reached. A queue does not answer a phone at 6 p.m.

Under the Cures Act, your patient may read it before you call

Federal information blocking rules have applied since April 5, 2021, and results reach the patient portal as soon as they are available, often before anyone in your office has read them. Your callback workflow should assume the patient already saw the number and has been sitting with it for a day.

The narrow exception people reach for does not do what they hope. The Preventing Harm Exception is scoped to specific, substantial risks and does not authorize a blanket delay on releasing results. What is allowed is honoring a patient's own request to hold their results, provided the delay is time-limited and agreed, per federal guidance updated in September 2025.

States have started legislating the gap. Texas Senate Bill 922 took effect on September 1, 2025 and gives physicians 3 days to review and communicate pathology and radiology reports with a reasonable likelihood of malignancy, plus any result revealing genetic markers, before electronic release. Check your own state before you design around a delay.

Patients are not asking you to slow down. In a 2023 survey of 8,139 patients, 95.7% wanted results released immediately, including those with abnormal findings. Only 7.5% said reading a result before a clinician contacted them increased their worry, though that rose to 16.5% among patients whose results were abnormal, against 5.0% for normal ones.

So the design consequence is a script change, not a delay. Open the callback with what the result means and what happens next, rather than repeating a number the patient has already searched online. The questions that arrive by portal message need their own path, which is the subject of handling the portal questions a released result triggers.

What an open loop costs, and what you can safely automate

An open loop shows up years later as a malpractice claim about a result nobody chased. Communication failures now factor into 40% of malpractice cases, and prevention is cheap by comparison, because the tracking half of this work automates cleanly while the judgment half stays with a licensed clinician.

Published figures with their source and year. Reasons to measure your own loop.

The Candello benchmarking report released in November 2025, covering cases closed between 2014 and 2024, found a communication failure in 40% of malpractice cases, up from 30% a decade earlier. Of those cases, 40% closed with an indemnity payment averaging $338,000, and breakdowns between provider and patient rose from 58% to 63% of communication cases. ECRI had already ranked test result management its number 1 patient safety concern for 2019, after reviewing 2.8 million safety events.

Draw the automation line by asking whether a task requires clinical judgment. Matching results to orders, tiering on the laboratory's own flags, chasing due dates, sending a message a clinician approved, retrying a patient who did not answer, and writing the audit trail are all mechanical. Interpreting a value, deciding acuity, and choosing the next clinical step are not.

Anything that touches a result touches protected health information, so a vendor in that path is a business associate and needs a signed Business Associate Agreement before it sees a single result. Treat "HIPAA compliant" as a question about configuration and contracts rather than a badge, and ignore anyone advertising "HIPAA certified", because no such certification exists. If this category is new to you, what an AI receptionist does and where it stops covers the same boundary on the phone side, and post-visit follow-up that closes the same kind of loop applies the pattern after a procedure.

Close the loop before you scale it

Start with one result type and one clinician for 30 days. Route by tier, escalate critical values to a named person by phone, put every abnormal result in a review queue with a due date, and log the close. Then read the overdue list every morning and repair whatever it shows you.

Route by what the result is, and end every path in an acknowledged, logged close.

Walk the flow once. A critical value leaves the queue immediately and reaches a named clinician by phone. An abnormal result that is not critical goes to the ordering clinician with a review due date. If the patient has already opened it in the portal, they get the explanation a clinician approved rather than a repeat of the number. Everything else gets the routine notification, and all four paths end in the same place.

Then measure two numbers in your own data: the median hours from result received to patient contacted, and the count of results past your target that nobody has acknowledged. Both are countable today, and both tend to move before anything else does.

If you would rather size the gap before you change any workflow, the free Growth Leak Audit works from your own numbers.