Colonoscopy Recall: Why Reminders Are Not the Fix
Ask a GI practice how its colonoscopy recall system works and you will usually hear about outreach. A letter, a portal message, a call from someone at the desk with a printed list. The assumption underneath is that recall is a reminder problem, and that a practice which reminds harder will recall better.
The evidence for that assumption is thinner than most operators expect. And while practices invest in reminding, the part that quietly decides everything, whether the list is right in the first place, usually goes unexamined.
One note before we start. Surveillance intervals are clinical decisions. Physicians set them, physicians change them, and nothing in this article is medical advice. Everything here is about the administrative system that carries those decisions forward: where the interval is stored, whether it survives, and who is accountable for the queue.
What colonoscopy recall actually is, and who owns it
Recall is the promise a practice makes at the end of a procedure. The physician records an interval, three years, five years, ten, and the practice owes the patient a working path back at the right time.
That promise has an unusual shape compared with everything else the front desk handles. There is no appointment yet. The trigger date sits years in the future, past several staff turnovers and possibly an EHR migration. Nobody is waiting on the phone. Nothing breaks visibly when it fails.
Compare that with a procedure reminder, where the appointment already exists, the date is days away, and failure shows up the same week as an empty room. Practices are good at the second problem because it is loud. Recall is silent, and silence is not the same as success.
The evidence on recall reminders is weaker than most practices assume
Here is the finding that should reset the conversation.
A Gastrointestinal Endoscopy study comparing a physician-initiated recall group with a patient-initiated group found 75 percent compliance in the group that received reminders and 72 percent in the group that did not. The difference was not statistically significant, with a p value of 0.725. The authors' conclusion was blunt: the expense, time, and effort practices put into recall reminders may not deliver benefit over patients initiating their own return. In both groups, patient characteristics such as family history predicted return better than the outreach method did. This study is not recent, which is itself part of the picture. The operational evidence base for recall outreach is thin and dated for a workflow that carries this much weight.
Reminders are not useless. A study in ANZ Journal of Surgery of 816 patients recalled under a graded recall system found 87.6 percent complied within 12 months, so structured systems can perform well. That same study found compliance varied sharply by coverage: 89.0 percent among privately insured patients versus 79.0 percent among publicly funded patients. Which tells you that when recall fails, it does not fail randomly, and outreach volume is not the variable doing the work.
The practical read for an administrator: adding a third reminder to a list that is already wrong buys nothing. Fixing the list changes the denominator for everything downstream.
The problem underneath the reminder
The interval is scattered across documents
In most practices, a single patient's surveillance interval exists in several places at once. The procedure report. The pathology addendum. The letter to the referring physician, which is often the only document that reconciles the two. A structured health maintenance field in the EHR. A note somebody typed.
These do not automatically agree. The structured field is the one the recall report reads, and it is frequently the one that was populated last, fastest, or by default.
The list decays quietly
A 2025 preprint from a large academic centre put a number on this during an EHR migration. Working with a cohort of 118,061 patients, the team found that more than 70 percent of colonoscopy recall intervals would have carried placeholder values longer than the interval the physician had actually recommended. A patient the physician wanted back in three years would have been sitting in the system marked for ten.
That figure comes from one institution's transition and is not a national benchmark. Take it as a warning about a mechanism rather than a statistic about your practice. The mechanism is general: whenever unstructured history meets a structured field, defaults fill the gap, and defaults are longer than the intervals that matter most.
Note what would have happened next. The recall report runs. It looks healthy. Nobody appears overdue, because everyone was quietly moved to ten years. A practice in that position could send perfect reminders on schedule and still miss the patients who most needed to come back.
Nobody owns the queue
Referral intake usually has a name attached to it. Recall often does not. It belongs to whoever runs the report, which means it belongs to nobody in particular, which means it competes with the phone.
That matters more in 2026 than it did five years ago. MGMA's staffing data through 2025 found front-office roles were the most frequently cited turnover hotspot in medical practices, including in practices where turnover was otherwise flat. A multi-year queue held together by one person's habits does not survive that.
What a wrong list costs: a worked example
The following is illustrative math built on stated assumptions. It is not our measured result and not any practice's real data. Substitute your own numbers.
Input | Illustrative value |
|---|---|
Colonoscopies performed per year | 6,000 |
Share receiving an interval shorter than ten years | 40 percent, so 2,400 patients per year |
Share whose stored interval is wrong, missing, or defaulted | 10 percent, so 240 patients per year |
Average facility fee per procedure, ASGE illustrative figure | 600 dollars |
Annual administrative value drifting out of the queue | about 144,000 dollars |
Two things about that number. It understates the case, because drift compounds: every year adds a new cohort to a queue that never self-corrects. And it is the smaller of the two costs. The larger cost is the clinical one, and it belongs to the physicians who set those intervals and who are entitled to assume the system carried them forward correctly.
How to audit your colonoscopy recall system in one afternoon
Five checks, in order of what they usually reveal.
Run the overdue report and time yourself. If nobody can produce a current list of patients past their recommended interval inside ten minutes, that is the finding. Stop there.
Pull twenty charts and compare four places. Procedure report, pathology, letter to the referring physician, structured recall field. Count how many of the twenty agree across all four. This is the single most informative hour available to a GI administrator.
Look at the distribution, not the average. Sort your stored intervals and count the tens. If ten years is a suspiciously round majority, you are probably looking at defaults rather than decisions.
Check the seams. Any EHR migration, module rollout, or acquisition of another practice is a point where intervals were re-entered or mapped. Sample charts from either side of that date specifically.
Name an owner and a review cadence. One named person accountable for the overdue queue, with defined cross-coverage, and a report reviewed on a fixed schedule rather than when someone remembers.
What to automate, and what stays human
The automatable layer here is narrow and worth being precise about. Reaching the patient. Handling the inbound call when they ring back. Capturing what they say. Confirming what is missing. Moving the record into the queue the practice's physicians have defined. This is administrative routing. It does not set intervals, does not assess risk, and does not determine urgency. Physicians decide all of that, and a system that follows their rules is only as good as the rules it was given.
Two categories stay human by design. Anything clinical, without exception. And the patients the ANZ data points at, the ones whose barriers are financial, linguistic, or logistical, where a person on the phone is the intervention and no amount of outreach volume substitutes for it.
The goal is not to remind more. It is to know, on any given morning, exactly who is owed a return and to have a working path for each of them.
Frequently asked questions
What is a colonoscopy recall system?
It is the administrative process that carries a physician's recommended surveillance interval forward until the patient is booked. That includes where the interval is stored, how overdue patients are identified, who owns the queue, and how patients are contacted. The interval itself is always a clinical decision.
Do colonoscopy recall reminders actually improve compliance?
The evidence is mixed and thinner than most practices assume. One Gastrointestinal Endoscopy comparison found no significant difference between physician-initiated and patient-initiated recall, 75 percent versus 72 percent. Structured graded recall systems have reported compliance near 88 percent within 12 months. The likeliest read is that system design and patient circumstances matter more than reminder volume.
How do we find out if our recall list is accurate?
Sample twenty charts and compare the recommended interval across the procedure report, pathology, letter to the referring physician, and the structured EHR field. Then check the distribution of your stored intervals for an unusual concentration at ten years, which often indicates default values rather than physician decisions.
Why do recall intervals get lost during an EHR migration?
Historical recommendations usually live in unstructured documents that do not map automatically into new structured fields, so those fields get filled with defaults. A 2025 preprint covering 118,061 patients at one academic centre found more than 70 percent of recall intervals would have carried placeholders longer than what the physician recommended. The new system looks populated and correct, which is what makes it risky.
Should an AI system decide who is overdue?
No. Determining who is due, and when, follows from clinical intervals that physicians set. Administrative systems can surface a list, place and answer calls, capture information, and route records into physician-defined queues. Every clinical judgment stays with the physicians.
A quiet close
We build AI reception and administrative referral routing for independent GI groups, and we work with a 15-physician GI group in British Columbia today. If your overdue list is one of those reports nobody quite trusts, we are glad to compare notes on how yours is built. No pitch required.
Sources
Gastrointestinal Endoscopy, comparison of physician-initiated versus patient-initiated recall in a colonoscopy surveillance program. View source
Gauci et al., ANZ Journal of Surgery, patient compliance with surveillance colonoscopy and graded recall, 816 patients. View source
medRxiv preprint, July 2025, end-to-end AI-driven colonoscopy recall workflow at scale, 118,061 patients, not peer reviewed. View source
ASGE Practical Solutions, Barbara Tauscher, reducing the impact of no-shows and cancellations, including the illustrative facility fee figure. View source
MGMA staffing data on front-office turnover in medical practices. View source
Becker's GI and Endoscopy, March 2026, on consolidation and the independent GI practice. View source
Join 600+ Clinics
