Three numbers, three owners, one patient
In a 2023 survey of 249 ASTRO member radiation oncologists, 93 percent said their practice was experiencing shortages of key clinical staff, 80 percent said the situation was worse than the year before, and 53 percent said those shortages were delaying patient treatment. Respondents reported practice operating costs averaging 23 percent above pre-pandemic levels, and 77 percent named professional staffing as the main factor driving the increase.
In a separate 2024 ASTRO survey of its physician members (n=754, 16 percent response rate), 92 percent said prior authorization causes treatment delays, 68 percent reported delays of five days or longer against 52 percent in ASTRO’s 2020 survey, and 33 percent reported patients abandoning radiation treatment over authorization issues.
In a 2024 JAMA Network Open review of 206 insurance denials at a single academic radiation oncology center, 34.9 percent of the 202 evaluable cases experienced a delay, with a median of 5 days, a mean of 7.8, a standard deviation of 9.1, and a range of 1 to 49 days. Of the 206 denials, 96.6 percent involved commercial payers. Among cases eventually authorized, 27.2 percent required payer-mandated changes to technique or dose, and among the 21 cases where dose was reduced, the median reduction was 24 Gy.
Each of these numbers has a home. The first belongs to whoever runs staffing. The second belongs to the authorization team. The third belongs to revenue cycle. They are reported on different cycles, to different people, in different formats, and they are almost never added together.
The patient experiences the sum.
Delay has a published price
The reason this matters more in radiation oncology than in most service lines is that the association between delay and outcome has been quantified in peer-reviewed literature, repeatedly, for two decades.
Chen and colleagues, in a systematic review of 44 studies published in Radiotherapy and Oncology, pooled 20 high-quality studies of local control and found that longer waiting times were associated with a relative 14 percent increase in local recurrence risk for each month of delay, pooled across disease sites (RR 1.14, 95 percent CI 1.09 to 1.21). Site-specific estimates were higher in head and neck cancer: 1.28 per month for post-operative treatment and 1.15 per month for definitive treatment.
Huang and colleagues, in a 2003 Journal of Clinical Oncology systematic review, found that post-operative radiotherapy for breast cancer delayed beyond eight weeks was associated with a pooled odds ratio of 1.62 for local recurrence (95 percent CI 1.21 to 2.16). For post-operative radiotherapy in head and neck cancer, a delay beyond six weeks after surgery was associated with a pooled odds ratio of 2.89 (95 percent CI 1.60 to 5.21).
Hanna and colleagues, in a BMJ meta-analysis covering 34 studies and 1,272,681 patients across seven cancers, found a four-week delay in radical radiotherapy for head and neck cancer associated with a hazard ratio of 1.09 for mortality (95 percent CI 1.05 to 1.14), and a four-week delay in adjuvant radiotherapy for cervical cancer associated with a hazard ratio of 1.23 (95 percent CI 1.00 to 1.50), an estimate whose confidence interval reaches the null.
These are associations drawn from observational data, not demonstrated causal effects, and the evidence is not uniform. Žumer and colleagues, studying 262 head and neck cancer patients at a single institution, found no significant association between time to treatment initiation and locoregional control. A responsible reading is that delay effects are real, site-dependent, and strongest where tumor repopulation kinetics are fastest. The general point holds: in radiation oncology, days are a clinical variable, not only an operational one.
Where the days actually come from
If days are the output, the useful question is which upstream processes produce them. Three sources dominate, and each has measurable structure.
Source one: planning and coverage capacity. Workforce discussions default to physician supply. ASTRO’s own commissioned supply and demand analysis, published in 2023, projected balance between radiation oncologist supply and demand through 2030. The measurable vacancy and pipeline pressure sits elsewhere, in physics, dosimetry, and therapy.
ASRT’s 2026 radiation therapy staffing survey (n=560, 2.4 percent response rate, margin of error plus or minus 4.1 points) reported a radiation therapist vacancy rate of 11.4 percent against 13.6 percent in its 2024 survey (n=645, plus or minus 3.9 points), and a medical dosimetry vacancy rate of 6.8 percent against 9.6 percent. Both year-over-year changes fall inside the surveys’ combined margins of error, so the honest reading is that vacancy rates are no longer rising, not that they have meaningfully fallen. The 2024 edition also found that 65.6 percent of radiation oncology departments experienced turnover, averaging 3.40 FTE departures per department, with the leading stated reason for departure being leaving the profession entirely.
On the physics side, the constraint is structural. In the 2026 MedPhys Match, 170 therapeutic medical physics residency positions were offered and 168 were filled. Across therapy and imaging, 343 applicants submitted rank order lists, 195 matched, and 148 did not. The bottleneck is training capacity, not interest in the field, and that has a direct implication for planning: a program expecting to hire its way out of a physics gap is competing for a pipeline whose size was fixed years in advance. An article on AAPM’s Career Center describes staff therapeutic physicist searches as typically running three to six months from posting to accepted offer, with chief and director searches exceeding six months. That is editorial guidance rather than survey data, and it should be treated as a working assumption, not a benchmark.
Workload benchmarks exist, though they disagree with one another, which is itself worth knowing. ASTRO’s Safety Is No Accident (2012 edition, Table 2.3) sets minimums of approximately one medical dosimetrist FTE per 250 patients treated annually and one radiation therapist per 90 patients treated annually, while stressing that requirements vary considerably with case mix. The Canadian Organization of Medical Physicists published a weighted staffing algorithm in 2021, built on a baseline of 0.50 physicist FTE per 1,000 treated cases plus additional terms for complex cases, highly specialized cases, and brachytherapy fractions, which across 23 Canadian centres predicted a mean of 276 plus or minus 52 treated cases per physicist FTE. A 2024 AAPM meeting analysis applying the ACR, ASTRO APEX, and IAEA models to a single 8,200-start center produced 42 to 47 physicists, a 12 percent spread, and 24 to 36 dosimetrists, a 50 percent spread, depending on which model was used. There is no settled standard, particularly for dosimetry. A program should pick a published model, state which one it uses, and measure against it consistently.
Source two: prior authorization. Authorization is now a larger and better-documented source of delay than most staffing gaps, and it is almost never counted in the same conversation.
ASTRO’s 2024 member survey found that respondents reported 71 percent of authorization requests approved initially and 73 percent of denials overturned on appeal. Read together, those two self-reported figures describe a process that eventually says yes to the overwhelming majority of requests, after imposing a delay on all of them.
The cost has been measured directly. Bingham, Chennupati, and Osmundson, publishing in JCO Oncology Practice in 2022, used time-driven activity-based costing to price treatment-related prior authorization at $491,989 per year for a single academic radiation oncology practice, and estimated roughly $40.1 million annually across US academic practices. Individual authorization events cost between $28 and $101 and consumed 51 to 95 minutes of staff time, rising at the upper end when a peer-to-peer review was required. The finding that matters operationally: 94 percent of the practice-level cost came from treatments that were ultimately approved.
The clinical consequences reported in ASTRO’s survey were not abstract. Thirty percent of respondents said prior authorization had led to an emergency visit, hospitalization, or permanent disability. Seven percent said it had led or contributed to a patient’s death. Ninety-four percent said it worsened staff burnout, 60 percent had hired additional staff to manage it, and 80 percent had reallocated existing staff. These are physician self-reports from a self-selected 16 percent response rate, and they should be presented as such, but the direction is consistent across every published measure.
Source three: the rework loop. Denials and charge corrections produce delay indirectly, by consuming the clinical and administrative time that would otherwise go to throughput.
Kodiak Solutions’ proprietary revenue cycle dataset, covering more than 2,100 hospitals, showed an initial claim denial rate of 11.81 percent in 2024, up 2.4 percent year over year. In a voluntary survey of 516 Premier member hospitals, approximately 15 percent of 2022 private payer claims were initially denied, more than 54 percent of those denials were ultimately overturned and paid, and 3.2 percent of denied claims had already been pre-approved through prior authorization. A later Premier survey of 280 hospitals priced the provider-side cost of fighting a denied claim at $57.23 in 2023, up from $43.84 in 2022, aggregating to $25.7 billion nationally, of which Premier estimated roughly $18 billion was spent on denials that were ultimately overturned anyway.
On the physician practice side, MGMA’s DataDive figures put approximately 8 percent of claims as denied on first submission, which corresponds to roughly 92 percent first-pass acceptance.
None of this rework produces treatment. All of it consumes the same staff hours that treatment planning and authorization compete for.
Why programs cannot see the total
The three sources arrive on different clocks.
A coverage gap changes planning throughput within days. An authorization pattern changes over weeks. A denial and rework pattern surfaces in the revenue cycle a quarter or more later, by which point the coverage gap that contributed to it may already be closed. The financial symptom outlives its cause, which is what makes root cause analysis feel impossible and makes the finance conversation feel disconnected from the clinical one.
There is a plausible mechanism connecting them, and it is worth stating carefully because the evidence is partial. AAPM Task Group 100 states that “many errors that occur in radiation oncology are not due to failures in devices and software; rather they are failures in workflow and process.” Talcott and colleagues, publishing in Practical Radiation Oncology in 2020, inserted deliberately deficient plans into routine chart rounds and found they were detected 55 percent of the time, with cases presented earlier in a session significantly more likely to be caught than cases presented later. A Bayesian network analysis of 81 radiotherapy incidents at Chinese centers, coded using HFACS, found skill-based errors carried the highest modeled probability, with inattention, mental fatigue, physical fatigue, and time pressure as the strongest associated factors; the authors recommended that departments “rationalize the workload.”
Coverage pressure and workload pressure are the same variable, and the peer review literature shows that review quality degrades under time pressure. Whether that pathway connects a physics vacancy in one quarter to a denial pattern two quarters later has not, to our knowledge, been directly studied. It is the mechanism we would expect, and more usefully, it is testable inside your own data.
What to do about it
The change worth making is structural rather than analytical. Most programs already collect enough data. Very few review it in one place.
Measure one interval, decomposed into three. Track median simulation-to-start, and split it into clinical time, authorization time, and planning queue time. The total blends the three and hides which one is moving. The decomposition tells you, in a single monthly number, which source is currently producing your delay. The planning queue component in particular is a direct read on dosimetry and physics capacity, and it moves months before anything reaches finance.
Benchmark physics and dosimetry workload against a named published model. Pick ASTRO’s Safety Is No Accident, the COMP 2021 algorithm, or the ACR model, name it, and hold to it. Published models disagree by as much as 50 percent on dosimetry staffing, so the specific choice matters less than the consistency. What matters is having a defensible answer when someone asks whether the department is staffed appropriately.
Track appeal overturn rate as an operations metric, not a billing metric. If a large share of denials is overturned on appeal, the appeal work is friction rather than clinical judgment, and it is being paid for in clinical staff hours. That belongs on the operations dashboard beside the delay numbers, not buried in a revenue cycle report.
Plan coverage for absence rather than for hiring. Departures are normal. People take leave, retire, and change roles. With physics searches commonly running three to six months and residency output fixed years ahead, a program that begins searching when someone resigns has already accepted a multi-month capacity reduction. Continuity planning means knowing in advance who covers an absence and whether that person is already credentialed at your institution, because credentialing lead time is what converts a short absence into a long capacity problem.
Be precise about what remote coverage can and cannot do. As of this writing we are aware of no AAPM task group report, practice guideline, or position statement addressing remote medical physics practice specifically. MPPG 10.b defines scope of practice for clinical medical physics and MPPG 7.a addresses supervision at a distance, but neither governs remote performance of clinical physics duties. What does exist is published operational experience. Khan and colleagues, describing four centers across the US, Canada, the UK, and Italy in Radiotherapy and Oncology, documented that plan quality checks, image registration, IMRT and VMAT plan review, SRS and SBRT planning review, HDR planning and review, and treatment planning dosimetry all moved off site successfully, while first-fraction SRS and SBRT setup, HDR brachytherapy delivery, LDR seed implants, and most patient-specific delivery QA required on-site presence. Knutson and colleagues reported a large academic department running rotating remote physics coverage with no treatment cancellations attributable to physics unavailability. Against that, a national survey of 834 Japanese linac facilities, with 487 responding, found that only 10 percent of respondents used remote treatment planning at all, and documented governance gaps among those that did. There is no controlled study demonstrating that remote plan checks are equivalent in quality to on-site checks. The defensible claim is that specific, well-defined physics and dosimetry tasks have been performed remotely at scale without reported incident, and that any remote arrangement needs explicit governance covering scope, security, licensure, and escalation.
The question that changes the meeting
The unproductive version of this conversation asks whether the department is short-staffed. Every department is short-staffed by some definition, and that question has never once changed a decision.
The productive version asks something different: what is the earliest signal in our data that delay is increasing, and who sees it first?
If the answer is the finance report, the signal is arriving two quarters late, and it will arrive as a margin problem rather than as the capacity problem it actually is.
About Team Net Medical
Team Net Medical supports radiation oncology programs across radiation oncologist, medical physics, dosimetry, and radiation therapist coverage in locum, hybrid, remote, and onsite models, alongside authorization support, charge capture, and revenue integrity work, with KPI reporting built to connect clinical capacity to financial performance.
Send us your coverage forecast for the next two quarters and we will model the delay and revenue exposure against it. Call 202-800-6222 or Connect With Us.
