In radiology the order arrives incomplete and the slot gets booked anyway

What does the queue in a radiology unit look like?

Almost nothing radiology books was ordered by radiology. Orthopaedics, gynaecology, internal medicine, the emergency department and chronic-care follow-ups write the orders, and they reach the unit as loose exam orders: a sheet of paper the patient keeps, a line written into another service's medical record, a PDF attached to a referral.

That provenance is the whole problem. The unit inherits the wording of dozens of clinicians who share no nomenclature, and much of it arrives as free text: a brand name, an abbreviation, the anatomical region written by hand. The scheduling system looks for a catalogue service. The order offers a sentence.

A radiology schedule that works looks boring: slots fill with the right protocol loaded before the patient arrives, and the day slips by only a few minutes. One that does not work looks just as full and yields half as much, with slots taken by the wrong exam, gaps that open at eleven, and a rebooking list nobody counts.

Why does the order say one thing and the schedule reserve another?

The clean case is contrast media. With contrast and without contrast are two different exams in machine time, in preparation and in risk, and the difference is usually a tick mark made by hand on the order. The booking code is identical, so the unit reserves capacity before it knows which exam it is about to perform.

An MRI and a CT scan concentrate the defect, because they carry the widest spread of duration inside a single code. A breast ultrasound shows another version of it: it usually depends on a prior mammogram within a window the site defines, and that condition does not travel with the code either.

What falls through here that does not fall through next door?

The equipment is the bottleneck and it cannot be divided. A site has one MRI scanner, and that scanner yields according to how many correct protocols enter the day. When the protocol gets corrected with the patient already changed and waiting, the exam runs long and the list slides: the unit loses the rest of the morning.

• First-hour slots the system offers before the site is actually running.
• Bone densitometry and carotid Doppler studies that drop to the bottom of a patient's list because the scarce slot went to the MRI.
• Orders the patient put in a drawer because nobody followed up, resurfacing months later expired.

The last item is the least measured and the heaviest. An order that never entered the schedule appears in no report: radiology dashboards count completed exams and lost slot time, and an unmanaged referral occupies neither column.

What preparation burden lands on the patient, and why is it logged as a no-show?

Most imaging exams demand something from the patient before arrival: fasting, a current lab result, premedication hours ahead, removing metal, bringing a companion. Each requirement is set by the ordering site and changes between locations, so this page does not publish any of them.

A patient who misses one of them is turned away, and the slot is recorded as a no-show. The cause showed up much earlier: the full instruction existed in a document at the site and arrived late, arrived partial, or went to the wrong phone number. Exam preparation is an information-logistics problem measured in days, and that is how our no-show analysis treats it.

The gap is easier to size from outside the unit. Per Chile's CASEN 2022 survey, 32.3% of Chilean women aged 50 to 69 had not had a mammogram in the previous three years. No radiology dashboard shows that number, because the denominator sits outside the system.

What should the unit measure?

Exams performed per month does not answer the question. The unit needs to know what share of the imaging orders it received ended in a completed exam, how long that took, and where the rest stopped.

1. The denominator: every imaging order issued in the period, including those that never reached a schedule.
2. The drop point: never contacted, contacted but not booked, booked but not attended, rebooked but not closed.
3. Time from order date to completed exam, split by exam type.
4. The completion rate measured against a control group left in the usual flow.

Without the fourth item, any improvement can be attributed to seasonality. The full methodology, with the denominator written down before the work starts, is in how we measure.

What changes when the orders get read and followed up?

The moment of discovery moves. If someone reads the whole order before offering a time, the protocol question gets resolved in conversation with the patient, the slot opens for the right exam, and the front desk stops being where clinical decisions get made.

What enters the queue changes too. Orders that reach no schedule today become workable: the patient is contacted, an appointment is booked, or the patient's decision to decline is recorded. All three are legitimate outcomes and all three can be counted.

None of this requires replacing the unit's scheduling system. Surface works on top of what the institution already has, including the case where a formal integration does not exist yet, and it executes the site's protocols exactly as written. Without an API describes the real starting point.