Proton Therapy Operations Get Measured in Minutes, Pauses, and Planning Delays
A three-room proton therapy center has translated day-to-day operational strain into a structured performance framework, offering a rare look at how a shared beam delivery system behaves under real clinical pressure. The analysis focused on calendar year 2024 and organized work into three domains: beam performance, treatment planning and pre-treatment workflow, and treatment room operations. In a setting where even short interruptions can ripple across multiple patients, the value of a stable KPI structure is not abstract. It is operational survival.
The center’s aggregate site uptime reached 95.3% when weighted across clinical operating time in the three treatment rooms. Individual room uptime ranged from 91.8% to 97.2%, a spread that reflects how a multi-room platform can mask local variability when performance is averaged across the whole site. Beam pauses were frequent, but most were short. Across the year, those interruptions accumulated to 2,238 minutes of lost treatment time, which represented about 0.5% of total clinical operating time.
Beam Uptime Stayed High Even as Small Interruptions Added Up
For a proton facility, uptime is only part of the story. Pencil beam scanning depends on coordinated hardware, software, and clinical flow, so brief pauses can carry outsized scheduling consequences even when they do not dominate total operating time. The annual lost time remained small in percentage terms, but the raw minute count shows how often beam delivery had to stop and restart. In a center delivering complex therapy across multiple rooms, that pattern matters because interruptions do not occur in isolation; they compete with downstream simulation schedules, physics checks, and patient throughput.
The use of a commercial analytics platform linked to radio-frequency identification tracking and the oncology information system gave this analysis a practical edge. It allowed automatic capture of room-level events and workflow timing without relying on manual reconstruction after the fact. That kind of data architecture is essential for proton centers, where operational drift can hide inside apparently acceptable averages.
Planning Time Exposed the Bottleneck Before the Patient Reached the Room
Treatment planning took an annual mean of 8.5 business days from CT simulation to patient scheduling, and physician-related activity contributed the greatest variability across disease sites. That finding points to a workflow problem that begins before the first treatment fraction is ever delivered. In proton therapy, planning is not a passive administrative interval; it is where contouring decisions, target complexity, and clinical review shape the downstream pace of care.
Variability at the physician stage suggests that some disease sites demand more iterative review than others, likely because contouring and prescription decisions are more complex or less standardized. The center also separated planning time into physician, dosimetry, and physics components, which makes the framework useful beyond a single aggregate number. If one segment stretches while the others remain stable, the limiting step becomes visible rather than diluted inside a total turnaround time.
Room-Level Timing Stayed Tight, But Long Tails Still Appeared
Treatment room operations were generally well controlled. Monthly average appointment variance ranged from 5 to 11 minutes, and most treatments began within 10 minutes of their scheduled time. That degree of punctuality suggests that the center maintained disciplined patient flow despite the coordination demands of a multi-room proton program. Mean waiting-for-beam time was also low and stable, ranging from 1.5 to 2.3 minutes, which indicates that patients were not spending extended periods idle after reaching the treatment area.
Still, the distribution of timing showed that not every fraction fit neatly into the schedule. Slot utilization remained stable throughout 2024, with 73% of fractions completed within scheduled slots. Mean overrun and underrun durations were modest, but P90 overrun durations exceeded 10 minutes for the full year. Averages can flatter operations. The upper tail tells a different story, one that matters when a handful of long fractions can cascade into room delay, rescheduling, and staff overtime.
That long-tail behavior is where operational KPIs become clinically relevant. In a center with multiple rooms drawing from shared beam resources, a single overrun can create a short chain reaction. A stable mean waiting time can coexist with intermittent schedule stress, and the P90 overrun measure makes that visible in a way the average cannot.
Why These KPIs Matter for Proton Centers
This framework is built around practical questions that proton therapy teams already ask informally: Is the beam available when it is supposed to be available? How long does planning really take? Where does schedule friction begin? By separating beam performance from planning and room operations, the analysis gives administrators and clinical leaders a way to localize inefficiency instead of treating throughput as a single blunt metric.
The pathophysiology of the treatment process is different from the pathophysiology of disease, but the logic is similar: small failures accumulate, and the system eventually shows strain. Shared beam delivery systems, extended treatment hours, and high clinical complexity make multi-room proton centers particularly sensitive to delays that might be trivial in a lower-intensity environment. A KPI framework built from automatically collected operational data offers a way to track those delays before they become entrenched.
The reference data from 2024 also create a baseline for future comparison. A site with 95.3% uptime, 8.5 business days of mean planning time, and 1.5 to 2.3 minutes of mean WFB time can now measure whether a workflow intervention truly improves performance or merely shifts the burden elsewhere. The value of benchmarking lies in specificity. A center can only improve what it can actually measure.
Reference Data, Not a Finish Line
The study offers single-institution operational reference data, not a universal standard. Its strongest contribution is the structure itself: a reproducible set of KPIs that can be monitored longitudinally and compared across sites. Future multi-institutional benchmarking will require similar definitions, consistent data capture, and enough transparency to make room-level differences interpretable across centers with different patient volumes and disease mixes.
The remaining challenge is straightforward to state and hard to solve. Proton centers need shared operational definitions before they can compare performance with confidence. This framework provides one way to get there, but its usefulness will depend on whether other institutions adopt comparable metrics and whether future analyses can show how specific workflow changes alter uptime, planning delay, and room efficiency over time.