How do I set up an annual mammogram recall list when our RIS was never built for it?

You set it up outside the RIS reporting module, because that module was built to answer "what did we read last month," not "who has not come back." The practical method is to pull one year of completed screening exams, anchor each patient to her own exam date rather than to a calendar quarter, deduplicate on medical record number, remove the patients who should not be contacted, and then release the list in weekly slices sized to your open screening capacity.

That is the whole build. The rest is detail that keeps the list from becoming a spreadsheet nobody trusts by March.

The reason a RIS fights you here is structural. It is an order and result system. It knows an exam happened, what was billed, what the radiologist assessed. It does not hold a concept of "this woman is due again and nobody has spoken to her." You construct that from fields it already has.

What counts as a due patient and what the twelve month anchor date should be

Start by defining "due" narrowly, then widen it later. A due patient, for a first build, is a woman who completed a screening mammogram at your center, whose exam resulted in a benign or negative assessment, and whose exam date is now eleven to fourteen months in the past.

Use the exam completion date as the anchor, not the order date and not the date the report was signed. Orders get placed weeks before a patient shows up. Reports get signed late. Only the completion date reflects when the woman was actually in your room.

The eleven month front edge matters more than it looks. Most commercial plans and Medicare pay for an annual screening mammogram once the patient is past a certain interval from the prior exam, and Medicare's screening benefit is written as eleven full months having passed since the month of the last covered screening. If you contact a patient at month ten and she books at month ten and a half, you have created a claim denial and a very unhappy phone call. Anchor at eleven, contact at eleven, book at twelve.

The fourteen month back edge is your own choice, not a rule. It exists so your first list is a manageable cohort rather than every woman who ever walked through the door. Once the weekly rhythm is stable, add an overdue tier at fifteen to twenty four months and work it separately.

Keep reading: What does MQSA actually require us to send patients after a screening mammogram, and by when?

Pulling the cohort: exam codes, results fields and the queries that break

Screening and diagnostic have to be separated at the source, or your recall list will be full of women in active workup. Pull on the procedure codes your center actually bills.

Code family	What it is	Recall treatment
77067	Screening mammography, bilateral, with CAD when performed	Core recall cohort
77063	Screening tomosynthesis, bilateral, add on	Ignore as a standalone; it rides with 77067
77065 / 77066	Diagnostic mammography, unilateral / bilateral	Exclude from the first build, review separately
76641 / 76642	Breast ultrasound, complete / limited	Signal of an open workup, use as a suppression flag

Then layer the assessment. Your RIS almost certainly stores a BI-RADS assessment category on the finalized report. Category 1 and category 2 belong in recall. Category 0 means the screening was incomplete and additional imaging was requested. Category 3 means short interval follow up, usually six months. Categories 4 and 5 mean the patient is somewhere in a diagnostic or biopsy pathway. None of those four belong in an annual recall list.

The three queries that break in practice

First, exams with no finalized assessment. These are usually reports that were dictated and never signed, or exams that were performed and never billed. They will silently drop out of an assessment filtered query, so run a second count on exam completions alone and reconcile the difference. The gap is your data quality problem, and it is worth naming out loud.

Second, patients with two exam records twelve months apart because they came back already. If you filter only on "exam between date A and date B," you will happily recall a woman who was on your table three weeks ago. Filter on most recent screening exam per patient, not all screening exams.

Third, patients whose records live under more than one identifier. That is the next section.

Cleaning duplicates, deceased patients and transferred care before you contact anyone

Deduplicate on medical record number first, then run a second pass on last name plus date of birth plus the last four of the phone number. Marriage, divorce and a misheard name at the front desk are the three most common ways one woman becomes two records in a breast center. A hyphenated surname entered once with the hyphen and once without will defeat an exact match every time.

Screen for deceased status against your practice management system, and against what your front desk staff knows and the database does not. There is no recovering from a recall letter addressed to a woman who died in February. Build a permanent hard suppression flag that survives every future rebuild.

Transferred care is softer. A patient who moved, or whose OB GYN left your referral network, is a lower yield contact, not a bad one. Flag rather than delete, and work those names last.

Keep reading: Should we text, call or mail our annual recall reminders, and how do the three compare on cost?

Suppression rules: active diagnostic workup, short interval follow up, opted out

Write your suppression rules down as a document, not as filter settings inside a query someone will change. This is the section your compliance officer and your lead technologist should both sign.

• Active workup. Any diagnostic mammogram, breast ultrasound, breast MRI or breast biopsy in the last one hundred eighty days suppresses the patient from annual recall.
• Short interval follow up. A BI-RADS 3 assessment moves the patient to a six month follow up queue, which is a different list with a different script and a much shorter tolerance for silence.
• Known cancer history under surveillance. These patients are usually on a protocol driven by their oncologist. Suppress from generic recall, route to a named person.
• Opted out. Any patient who has asked not to be contacted, in any channel, for any reason. One flag, honored everywhere.
• Bad contact. Phone disconnected and mail returned. Suppress from outreach, route to an address hygiene task.

Every suppression should carry a reason code and a date. A suppression with no expiration quietly removes patients from your recall population forever, which is how a center wakes up two years later with a list that is half the size it should be.

Staging outreach so recall volume matches your open screening slots

This is where most recall programs fail, and it is arithmetic rather than judgment. Work it in front of the owner.

Assume, and these are assumptions you should replace with your own numbers, that you run two mammography units, each doing eight screening exams a day, five days a week. That is eighty screening slots a week. Assume you already fill sixty of them from referrals and self scheduling, leaving twenty open slots a week that recall needs to fill.

Now assume a conversion rate. If one in four contacted patients books within three weeks, you need to contact eighty patients a week to fill twenty slots. If your conversion is one in six, you need one hundred twenty. Measure your real rate after four weeks and rewrite this number.

Then check the cohort against the tempo. If your annual screening volume is four thousand exams, your twelve month recall cohort is roughly four thousand patients, or about seventy seven a week spread evenly. That is close to the eighty you need, which is the tidy case. If your cohort clusters, because October is always your heaviest month, the even spread is a fiction and you will need to pull forward or hold back at the edges of the eleven to fourteen month window.

Release the list weekly. Never release a quarter at once. A quarterly dump produces a burst of calls, a burst of bookings you cannot seat, and then eleven quiet weeks.

See how MammoLoop handles this for breast imaging and women's screening centers

Handing the list to the front desk without creating a second untracked worklist

The failure mode is a printed sheet worked with a pen, where the outcome of each attempt lives in a drawer. Two weeks later nobody can answer "did we already call her twice."

Whatever tool holds the list has to hold the attempt history in the same place: date contacted, channel used, who touched it, what happened. Three outcomes are enough to start with. Booked. Reached and declined or deferred, with a callback date. Not reached, attempt counted.

Set an attempt ceiling before you start, so nobody has to decide in the moment. Three outreach touches across a channel mix, then the patient rests until the next cycle. A woman contacted seven times about the same exam stops reading anything you send.

Measuring the list weekly instead of rebuilding it every quarter

Four numbers, every Monday, on one line.

1. Cohort released. How many patients entered outreach last week.
2. Contact rate. Of those, how many were actually reached, not just messaged.
3. Booking rate. Of those reached, how many now hold an appointment.
4. Show rate. Of recall bookings that came due, how many arrived.

The fourth number is the one owners forget and the one that decides whether recall pays. A booking rate that looks excellent alongside a show rate of sixty percent means you are filling slots with air. Track recall bookings separately from referral bookings so you can see the difference.

If a number moves you know which stage to open. Contact rate means phone and address data. Booking rate means script or slot availability. Show rate means the reminder sequence between booking and exam day.

Where to go from here

You can build all of this in a spreadsheet, and plenty of centers have. What a spreadsheet cannot do is age a callback date, count attempts across channels, or stop when a patient has booked.

MammoLoop is built for exactly this loop: it holds the annual recall cohort, stages text and letter reminders on the schedule you set, tracks every callback and attempt in one place, and moves no shows into a recovery queue instead of losing them. If you already know who is due and just need the follow through to happen without sixty calls a week, that is the part it takes off your desk.