Radiology Staffing: What Your RIS Data Can and Cannot Tell You

The questions your RIS data can answer

Most imaging operations already hold everything needed to answer the operational half of the staffing question. It is rarely looked at, because getting it out requires somebody to ask a specific question rather than run a standard report.

When does demand actually arrive?

Exam volume by hour of day and day of week, per modality. Almost every center discovers its shift pattern was set years ago against a demand curve that has since moved, usually because the roster was built once and then inherited.

How long does each step really take?

Not the scheduled slot length, the measured interval: patient arrival to scan start, scan start to scan complete, complete to ready for reading. The gap between the scheduled and the measured figure is where capacity is hiding.

Where does the work stop?

The longest wait between two timestamps is your constraint. Adding staff anywhere else changes nothing, which is why staffing decisions made without this number so often produce no improvement at all. Find the queue before you fund it, because the step that feels busiest is frequently not the one holding everything up.

Which rooms are actually busy?

Utilization per room per hour, not per site. Site-level numbers routinely conceal one room at capacity and another idle, and the average of the two looks acceptable enough that nobody investigates. The hourly view is what exposes a room that is saturated every weekday morning and empty every afternoon.

The number that misleads most often

The culprit is average exam time. It is the figure everyone has and the one that causes the most bad rostering, because staffing is not determined by the average day. It is determined by how bad the bad days are.

Use the spread instead. If your average CT takes twenty minutes but a quarter of them take thirty-five, the twenty is not the number your roster should be built on. Look at the distribution, decide which percentile you intend to cover, and staff to that deliberately rather than discovering it by accident.

The questions your RIS data cannot answer

This is the part usually left out, and leaving it out is what makes staffing articles useless.

Your RIS cannot tell you what the local market pays, why your technologists are leaving, or whether your shift pattern is tolerable to work. It cannot tell you whether a supervisor is the reason a team is short, or whether the role you are recruiting for is attractive at the salary you are offering.

Those are frequently the actual problem. An operation losing staff faster than it hires cannot roster its way out. Better data makes the remaining shifts better organized, which is worth having and is not a solution to attrition. If your turnover is the constraint, the work is in pay, progression and management rather than in scheduling software, and we would rather say so than sell you the wrong thing.

The labor market, which no internal system can see

Your RIS holds your demand. It holds nothing about whether the person who would meet that demand can be hired at all, and for radiology that gap has been widening for a decade.

A study in the American Journal of Neuroradiology compared radiologist job listings against the training pipeline feeding them. Between 2014 and 2023 it counted 31,825 jobs listed in the ACR Career Center against 10,180 anticipated diagnostic radiology residency graduates, a ten-year cumulative shortfall of 21,645.

Read that number carefully, because most articles quoting it do not. These are job listings on a single board, not unique unfilled posts: one vacancy advertised across three years counts three times, and postings that were filled internally still appear. The figure does not measure a headcount deficit. What it does measure, reliably, is direction, and the direction has been one way for ten years.

The operational consequence is the part worth acting on. When the constraint is supply rather than budget, the lever that works is the one that reduces how many people a given volume needs. That means removing re-keying, cutting the steps between order and report, and making the shifts you can staff tolerable enough that you keep the people already working them.

Staff utilization is not room utilization

Most reporting answers the room question, because the room question is the one the schedule already knows: what proportion of available scanner time was booked. It is a useful number and it is not a staffing number.

Two things it hides:

• A fully booked room can be an understaffed one. If the scanner runs at ninety-five percent because a technologist skipped a break and stayed late, the utilization figure records a success and the retention risk goes unrecorded.

• Staff are not attached to one room. The person covering two modalities on a quiet afternoon looks idle in both, and looks fine in neither when both get busy at once.

The measure that behaves better is the share of a shift spent on work only that role can do. A technologist doing registration is fully occupied and badly deployed, and no room utilization report will ever show it. That number takes deliberate work to produce, which is why almost nobody has it, and it is the one that tells you whether you need another person or a different process.

The honest summary: most centers staff to the average by default rather than by decision, then treat the resulting overtime as an anomaly rather than as the predictable cost of that choice.

How to work the operational half of radiology staffing

1. 1

   Pull volume by hour, day and modality for a full year

   A year, so seasonality is visible. By hour, because that is the resolution a roster is built at. If your reporting cannot produce this, that is the first finding.

2. 2

   Measure the intervals, do not assume them

   Arrival to scan start, scan start to complete, complete to ready for reading. Compare each against what the schedule assumes. The difference is either recoverable capacity or an over-optimistic schedule.

3. 3

   Find the longest wait and staff that first

   The longest interval between two timestamps is the constraint. It is often the step nobody had flagged, and it is the only place additional staff changes the total.

4. 4

   Decide explicitly which percentile you cover

   Pick it, write it down and roster to it. The point is that it becomes a decision with a known cost rather than a default with an unexplained overtime bill.

5. 5

   Check whether you have a staffing problem or a retention problem

   If people are leaving faster than you hire, steps one to four will not fix it. Establish that before investing in the scheduling work.