An operational key performance indicator framework and reference data from a multi-room pencil beam scanning proton therapy center

A retrospective operational study at the Fred Hutchinson Cancer Center Proton Therapy Center found that aggregate site uptime was 95.3% during calendar year 2024, while planning and treatment-room workflow indicators varied by disease site, room, and clinical component. The study evaluated operational data from a three-room pencil beam scanning proton therapy center rather than measuring clinical outcomes or treatment effectiveness.

Researchers assessed routinely collected key performance indicators, or KPIs, in three domains: beam performance, treatment planning and pre-treatment workflow, and treatment-room operations. Beam metrics included room-specific and site-wide uptime and beam pauses. Planning metrics covered the interval from CT simulation to patient scheduling, with time attributed to physicians, dosimetry, physics, and milling. Room metrics included appointment variance, slot utilization, and waiting-for-beam time.

The analysis covered calendar year 2024 at the center in Seattle, which operated a fixed horizontal beamline room, an inclined beamline room, and a full 360-degree rotating gantry room. All three rooms used pencil beam scanning. Most KPI data were extracted from the Vantage analytics platform, which integrated radio-frequency identification patient-tracking records, data from the proton system’s universal monitoring interface, and the Mosaiq oncology information system. The analyses used de-identified, aggregated operational metrics without identifiable patient-level information.

Annual uptime was 97.2% for the fixed horizontal beamline room, 91.8% for the inclined beamline room, and 96.5% for the rotating gantry room. The 95.3% site figure was weighted by clinical operating time across all three rooms rather than calculated as a simple average. The inclined beamline room had the greatest month-to-month variability and was the only room with monthly uptime below 95% in five months, compared with two months for the fixed horizontal room and three for the gantry room.

Beam pauses were recorded automatically. Across the year, these interruptions totaled 2,238 minutes, or 37.3 hours, equivalent to approximately 0.5% of the center’s total clinical operating time. Monthly counts ranged from about 450 to 1,100 pauses, with approximately 95 to 295 minutes lost per month. The highest losses occurred in April, June, and December. Elevated pausing was also observed after major cyclotron maintenance or intervention periods, including an event near the end of February and another in early September.

The annual mean time from CT simulation to patient scheduling was 8.5 business days. On average, physicians contributed 42% of planning time, dosimetry 41% and physics 17%. Physics time ranged from 1.3 to 1.6 business days during the year, while physician time ranged from 2.2 to 4.3 days and dosimetry time from 2.9 to 3.7 days. By disease site, total planning time ranged from 6.4 business days for craniospinal irradiation to 10.0 days for head-and-neck cases. Physician time varied most across sites, ranging from 1.7 days for craniospinal irradiation to 5.5 days for head-and-neck cases.

Monthly mean appointment variance ranged from approximately five to 11 minutes, with the highest averages in September and October. The fixed horizontal beamline room had the largest proportion of on-time entries, at 57%, followed by the inclined beamline room at 45% and the rotating gantry room at 39%. Across all rooms, most treatments began within 10 minutes of their scheduled time, while delays longer than 30 minutes accounted for approximately 6% of treatments. Higher treatment volume did not consistently coincide with greater average appointment variance.

Slot utilization was relatively stable during the year. Underruns occurred more often than overruns, with monthly rates of approximately 64% to 79% and 21% to 36%, respectively. Mean overruns ranged from 5.4 to 7.4 minutes, and mean underruns ranged from 4.8 to 6.3 minutes. Although most deviations were modest, the 90th percentile for overrun duration exceeded 10 minutes in every month, indicating a smaller group of long overruns.

Mean monthly waiting-for-beam time ranged from 1.5 to 2.3 minutes, while the 90th percentile remained below five minutes throughout the year. Waiting times were higher in the early morning and declined toward late afternoon and early evening. Overall, 73% of fractions were completed within their scheduled slots. The authors emphasized that the reported values are single-institution reference data rather than universal benchmarks, because room geometry, case mix, staffing, scheduling, data capture, shared-beam constraints and KPI definitions can differ among proton therapy centers.

Subject of Research: Operational performance indicators in a three-room pencil beam scanning proton therapy center

Article Title: An operational key performance indicator framework and reference data from a multi-room pencil beam scanning proton therapy center

Article References: Saini, J., Wong, T., Schwarz, M., McClure, R., Shepherd, A., Zeng, J., & Rengan, R. (2026). An operational key performance indicator framework and reference data from a multi-room pencil beam scanning proton therapy center. PLOS One, 21(10), e0360926. https://doi.org/10.1371/journal.pone.0360926

Image Credits: AI Generated

DOI: 10.1371/journal.pone.0360926

Keywords: proton therapy, pencil beam scanning, operational KPIs, treatment center uptime, beam pauses, treatment planning, appointment variance, slot utilization, waiting-for-beam time, operational, performance, indicator