Proton therapy operations get a data-driven benchmark
A three-room proton therapy center has laid out a practical way to measure how a complex radiation oncology operation actually runs day to day. The framework separates performance into three domains: beam delivery, treatment planning and pre-treatment workflow, and treatment room operations. In a setting where a single technical interruption can ripple across multiple rooms, that structure matters because it turns anecdotal frustration into measurable operational signals.
The center’s 2024 reference data show a system that was functioning at a high level, but not without friction. Aggregate site uptime reached 95.3% when weighted across clinical operating time, and individual room uptime ranged from 91.8% to 97.2%. Beam pauses were frequent enough to matter, yet most were short; the cumulative loss came to 2,238 minutes over the year, or about 0.5% of total clinical operating time. In practical terms, the machine rarely stopped for long, but the repeated interruptions still accumulated into a meaningful annual burden.
The numbers speak to a familiar reality in proton centers: resilience is not the same as silence.
Beam delivery remained stable, but the cost of small pauses accumulated
The beam performance metrics center on uptime and beam pauses, two endpoints that carry outsized importance in pencil beam scanning proton therapy. Shared beam delivery systems create operational dependencies across rooms, so a pause in one segment can affect downstream scheduling even if the interruption is brief. High uptime therefore does not eliminate the need for continuous monitoring; it simply means the center is recovering from problems quickly enough to avoid large blocks of lost clinical time.
The 95.3% site uptime is reassuring in a multi-room environment, especially one operating with extended hours and technical complexity. Yet the spread between the best-performing room and the least-performing room suggests that room-level conditions still shape throughput. A room that operates near 97% uptime faces a different workflow burden than one closer to 92%, especially when that difference repeats across a full year of high-volume care.
Those gaps are more than housekeeping metrics. In proton therapy, where the treatment modality depends on highly specialized delivery hardware, uptime is inseparable from patient access, staffing pressure, and the ability to preserve daily schedules.
Planning time was driven most by physician-related variability
The pre-treatment workflow data capture the interval from CT simulation to patient scheduling, with the process broken into physician, dosimetry, and physics components. Annual mean treatment planning time was 8.5 business days. Across disease sites, physician-related activity contributed the greatest variability, which suggests that the earliest parts of the planning chain may be the least predictable segment of the workflow.
That pattern is clinically important because planning delays do not stay confined to the back office. They can alter start dates, compress downstream scheduling, and create uneven demand on physics and dosimetry teams. In a proton center, where cases may include pediatric patients requiring anesthesia or patients being re-irradiated, the planning pathway often has to accommodate more complexity than a standard photon workflow. A delay in the physician step can therefore reverberate through a tightly linked system.
The center’s data do not imply failure; they define the operating range. For administrators and clinicians, that range is the benchmark. An 8.5-business-day average is useful only when placed alongside its internal variability, and the study makes clear that variability clustered around physician-driven tasks more than around every other part of the planning chain.
Treatment rooms stayed close to schedule, even when overruns persisted
Room operations were measured with appointment variance, slot utilization, and waiting-for-beam time. Monthly average appointment variance ranged from 5 to 11 minutes, and most treatments began within 10 minutes of the scheduled time. That is a tight distribution for a high-complexity center running multiple rooms and a shared beam system. It also suggests that the scheduling machinery was not routinely drifting far from target, even when the system experienced pauses or room-specific constraints.
Slot utilization remained stable throughout the year, with 73% of fractions completed within the scheduled slot. Mean overrun and underrun durations were modest, but the P90 overrun duration stayed above 10 minutes across the year. That split between average performance and upper-tail delay is the kind of operational signal that can hide inside otherwise acceptable throughput numbers. Most days may look manageable on paper, yet the worst 10% of cases still consume enough time to disturb the downstream schedule.
Waiting-for-beam time was low and consistent, at 1.5 to 2.3 minutes on average. In proton therapy, a short WFB interval suggests that the room is not idling extensively once the patient is ready, which is especially relevant in a setting where beam delivery is a shared resource. The low wait times indicate disciplined coordination between room readiness and beam access, even when the overall operation is absorbing frequent interruptions elsewhere in the system.
Why a KPI framework matters for proton therapy services
The value of this work lies in its structure as much as in the numbers themselves. A proton center needs more than generic throughput reporting. It needs metrics that isolate the beam, the planning pathway, and the room-level experience so leaders can see where inefficiency originates and whether it is improving or drifting over time. Automatic data capture through a commercial analytics platform linked to radio-frequency identification tracking and the oncology information system makes that kind of surveillance possible without relying entirely on manual review.
Because the data come from a single institution, they function best as reference values rather than universal standards. Still, they offer a concrete starting point for other centers trying to build operational dashboards around proton therapy delivery. The framework may also help institutions compare room performance, track changes after workflow redesign, and identify which component of the treatment chain is most responsible when clinic flow deteriorates.
The study is also a reminder that operational quality in radiation oncology is not only about dosimetric precision. In proton therapy, precision means little if schedules collapse, beam access becomes erratic, or planning stalls before the first fraction is delivered. Reference data like these give centers a language for that problem, and they make it possible to measure whether a workflow is merely functioning or functioning consistently enough to support complex care at scale.
The main limitation is straightforward: this is single-institution evidence from one calendar year, so broader benchmarking will require additional centers and more longitudinal data. The next step is not speculation about ideal performance; it is comparison across institutions using the same definitions, the same domains, and the same operational logic.