The Reading Room Is Carrying Work It Should Never See

1. The Headcount Trap

When backlogs grow, the instinct is to ask for more radiologists. That instinct is understandable. It is also incomplete.

The latest national modeling does not support a simple labor-only answer. Imaging utilization is projected to rise through 2055, driven largely by population growth and aging. Radiologist supply may also grow, depending on residency expansion, but the shortage is still projected to persist. The implication is uncomfortable but important: many systems are trying to solve two different problems with one lever. One problem is a real workforce constraint. The other is a large internal waste problem. ^[1][2][3]

That second problem is where leadership has more control than it thinks. A shortage does not only show up as an empty chair. It also shows up as a radiologist hunting for missing context, taking avoidable calls, fielding protocol questions that should have been settled upstream, or pausing interpretation to act as the cleanup function for someone else's broken handoff.

Hospitals cannot hire their way out of work that should never have reached the expert in the first place. They have to redesign the path that work takes.

2. The Hidden Tax on the Reading Room

A large share of the burden hitting radiologists is not interpretation. It is recovery work.

By that, we mean tasks that reach the reading room only because something upstream did not arrive complete, clear, or correctly routed. A missing prior. A vague indication. A protocol question that was never resolved. A nonurgent consult delivered as a live interruption instead of a managed request.

The scale of this burden is not theoretical. In a Journal of the American College of Radiology study, on-call radiologists received an average of 72 telephone calls during a typical 12-hour overnight shift and spent 108 minutes on the phone. In Academic Radiology, each additional phone call in the hour before a discrepant preliminary report was associated with a 12% increase in the likelihood of resident error. ^[4][5]

The interruption cost is not just the duration of the phone call. Research on interruptions more broadly shows that people often compensate by working faster after being interrupted, but at the cost of higher stress, frustration, time pressure, and effort. ^[14]

This is where the operational loss compounds. One interruption does not just consume 60 seconds. It breaks sequence. It fractures attention. It forces the radiologist to reconstruct context. And when that pattern repeats all day, it converts expert bandwidth into fragmented, lower-yield time.

This is the core idea behind Ghost Capacity. The capacity is often already there. It is just trapped behind rework and interruption load.

3. Why More Software Often Makes This Worse

When leadership sees chaos, the next reflex is usually to buy software. Sometimes that helps. Often it does not. The reason is simple: technology layered onto a broken workflow usually digitizes the failure instead of reducing it.

AHRQ's work on health IT and workflow redesign makes this point directly: health IT should be incorporated into workflow redesign, not treated as a standalone fix. The World Health Organization makes a similar point in its digital health strategy, emphasizing that digital initiatives must integrate organizational, human, financial, and technical resources rather than operate as isolated tools. ^[7][8]

That matters because usability and workflow fit are not soft issues. They are operating issues. In Mayo Clinic Proceedings, physician-rated EHR usability was strongly associated with burnout; each 1-point increase in usability score was associated with 3% lower odds of burnout. ^[9]

The reading room feels this immediately. If a new tool adds clicks, dashboards, or surveillance without reducing interruptions, hunting, and rework, clinicians experience it as one more layer of demand. That is not resistance to change. It is pattern recognition.

The adoption side matters too. Research on psychological safety in healthcare links psychological safety with stronger safety improvement and lower intent to leave. People are more willing to speak up, surface defects, and engage in redesign when they believe the change is there to help them do better work rather than simply measure them harder. ^[10]

That gives us a cleaner rule for radiology operations: do not ask the reading room to absorb a new system until the new model removes real pain from day one.

4. Build the Perimeter

Aviation solved a version of this problem decades ago. The FAA's Sterile Cockpit Rule bars nonessential duties during critical phases of flight. The principle is plain: when the work is high consequence, the environment has to protect attention. ^[11]

Radiology needs the operational equivalent.

Not a slogan. Not another dashboard. A perimeter.

4.1 Quarantine incomplete work upstream

If a study is missing required context, unresolved protocol detail, or key supporting information, it should not quietly drift into the radiologist's flow and become their problem by default.

This is a workflow design issue first, not a heroics issue. AHRQ's workflow redesign work and WHO's digital health guidance both point to the same conclusion: the right move is to redesign the path of work before adding technology to it. ^[7][8]

4.2 Put an interceptor in front of the reading room

One of the clearest studies in this space evaluated a call triage assistant who answered reading-room calls during the busiest on-call hours. Most calls were handled without disturbing the resident, producing a 71% reduction in interruptions. Mean turnaround time fell from 75.2 minutes to 44.3 minutes, while error rates were unchanged. ^[6]

That is the model. Put a person or system in front of the expert whose job is to absorb, sort, and resolve routine disruption before it lands on the radiologist.

4.3 Move nonurgent coordination out of live interruption channels

Synchronous communication is useful for urgent clinical escalation. It is a bad default for routine coordination. Research on hospital communication has shown that synchronous communication creates an interruptive workflow, with 42% of communication events coded as interruptions in one study. ^[15]

Two newer studies point to a better direction. In a general hospital setting, an asynchronous communication platform cut average task completion time by 20.1 minutes, a 58.8% time savings versus traditional synchronous methods. ^[12] In radiology specifically, adoption of the RadConnect communication tool reduced synchronous consult requests by 53%, from 6.1 to 2.9 per day. ^[13]

The lesson is not that every message should be asynchronous. It is that nonurgent coordination should stop arriving as a live tax on expert attention.

4.4 Measure exceptions, not just throughput

Most departments measure turnaround time. Fewer measure why the work got harder before it was late.

A protected reading room needs exception visibility. How many interruptions came from missing history? How many from protocol confusion? How many from priors, scheduling issues, or administrative questions? If leadership cannot see the reasons, it cannot reduce the causes.

This is where most software strategies fail. They count output but do not classify upstream defects.

5. A Practical Starting Framework

The framework below is a synthesis of the evidence above and an operational recommendation, not a finding from a single study.

Step 1: Count interruptions for two weeks

Measure live calls, in-person interruptions, secure messages, and ad hoc consults by hour, role, and source.

Step 2: Tag the reason for each interruption

Use a short defect taxonomy: missing clinical context, missing prior, protocol question, scheduling or admin issue, tech issue, urgent clinical escalation, other.

Step 3: Define what should never hit the radiologist cold

Create a hard list of prerequisites for work to enter the active reading flow. Incomplete work goes to an exception lane, not to the radiologist by default.

Step 4: Staff an interceptor layer during peak periods

This can be a reading-room coordinator, call triage assistant, AI-supported ops role, or blended model. The point is not the title. The point is that someone other than the radiologist owns routine disruption first.

Step 5: Shift nonurgent consults into an asynchronous path

Use a purpose-built communication lane for routine questions and status requests. Keep live interruption for genuinely time-sensitive clinical issues.

Step 6: Track recovered capacity, not just raw volume

Measure phone time, interruption rate, turnaround time, discrepancy rate, staff-reported stress, and the share of interruptions resolved upstream.