A round burgundy crest emblem with simple gold ring detailing, plain and formal in style
A wide horizontal photograph of mountain scenery near Banff with pale turquoise sky, dark evergreen slopes, and a soft muted mood
A round burgundy crest emblem mirroring the first, with plain gold ring detailing and a formal quiet mood

Annual Scientific Meeting of the Canadian Organization of Medical Physicists — July 9–12, 2014, The Banff Centre, Banff, Alberta, Canada

Building a multi-institutional clinical trial in medical physics

A multi-institutional clinical trial in medical physics turns a promising technique into evidence that can influence everyday patient care. Whether the intervention involves adaptive radiotherapy, image-guided treatment, dosimetry, MRI planning, artificial intelligence or radiation protection, the study must work across different scanners, software versions, staffing models and clinical environments.

For Australian researchers, this means designing for a geographically dispersed health system. A trial may connect tertiary hospitals in Sydney, Melbourne, Brisbane, Perth and Adelaide, with long travel distances, different state health policies and varied access to specialist physicists. A reliable protocol needs to accommodate public hospitals, private radiology groups, rural services and the practical realities of Australian working hours.

Define the clinical question and trial population

The strongest studies begin with a specific clinical problem rather than a technology looking for an application. A useful question might ask whether a new treatment-planning workflow reduces organ-at-risk dose while preserving target coverage, or whether automated image registration decreases treatment delays without increasing error rates. The primary endpoint should be clinically meaningful, measurable and relevant to patients, clinicians and health administrators.

Eligibility criteria require careful thought when participating centres serve different communities. Age, tumour site, treatment intent, comorbidities and prior therapy should be described precisely, while unnecessary exclusions should be avoided. Australian sites may draw patients from culturally diverse urban populations, remote communities and Aboriginal and Torres Strait Islander communities, so recruitment materials and consent processes must be respectful, accessible and suitable for local needs.

A protocol should distinguish between the intervention, the comparator and the standard of care at every site. If one hospital uses a particular immobilisation system or planning algorithm, that variation must be recorded rather than hidden. Clear definitions protect the trial from becoming a collection of loosely related local projects.

Create a network that can work consistently

The coordinating centre should identify each site's capabilities before finalising the protocol. A site survey can document treatment machines, imaging systems, treatment-planning software, dose calculation algorithms, quality assurance equipment, staffing levels and expected patient volume. This information reveals whether a common protocol is realistic or whether the study needs approved technology strata.

A central medical physics team can provide reference datasets, test cases, analysis scripts and implementation guidance. Remote review is useful across Australia, where specialists may support several hospitals and regional services. However, video meetings and shared documents do not replace local ownership. Every site needs a principal investigator, a lead physicist, a data manager and a nominated backup contact.

Training should be completed before patient enrolment. Workshops can include phantom measurements, anonymised case reviews, contouring exercises and simulated data transfer. A short competency assessment helps identify misunderstandings early. Version-controlled documents are essential because a protocol amendment, software update or revised endpoint can otherwise produce incomparable results.

Align ethics, regulation and data governance

Clinical research in Australia commonly requires review through a Human Research Ethics Committee, with site-specific governance approval in each state or territory health service. The national statement on ethical conduct in human research provides a foundation, while local requirements determine contracts, indemnity, privacy controls and access to clinical records. Early engagement with governance offices prevents approval delays after the scientific design is complete.

Medical physics trials often involve DICOM images, treatment plans, dose distributions, machine logs and linked clinical outcomes. The data dictionary should define formats, units, naming conventions, missing-value codes and rules for derived variables. De-identification must be tested rather than assumed, particularly when images contain embedded identifiers or dates that could enable re-identification.

The Privacy Act 1988 and the Australian Privacy Principles are relevant when personal information moves between institutions or into cloud platforms. Data-sharing agreements should specify who can access files, where they are stored, how long they are retained and what happens after the study closes. Encryption, audit trails, role-based permissions and a documented breach response are practical requirements, not administrative decoration.

Standardise measurement, quality assurance and safety

A multi-centre dosimetry study needs a common measurement framework. Centres should agree on calibration references, phantom geometry, detector types, tolerance limits, image-registration rules and procedures for resolving out-of-range results. If a site fails a commissioning test, the protocol should state whether it can continue under a corrective action plan or must pause recruitment.

Independent checks can strengthen confidence in the findings. A central review of selected plans may identify systematic contouring or dose-calculation differences, while periodic intercomparison measurements can reveal drift over time. Automated scripts are valuable for checking plan parameters and extracting structured data, provided their outputs are validated against manual review.

Patient safety must remain central when the trial changes clinical workflow. A risk register should cover wrong-patient selection, incorrect treatment intent, software failure, incomplete image transfer, dose overrides and communication breakdowns. The wider professional conversation about patient safety discussions illustrates why incident learning and transparent escalation should be built into the study from the start.

Make the analysis reproducible and clinically useful

The statistical analysis plan should be written before unblinding or outcome review. It should define the primary endpoint, secondary endpoints, sample size, missing-data approach, subgroup analyses and method for handling centre effects. A mixed-effects model may be appropriate when outcomes vary between hospitals, machines or treatment teams.

Investigators should decide whether the study is evaluating technical equivalence, workflow efficiency, safety, patient-reported outcomes or clinical benefit. These are different questions and require different measures. A reduction in planning time, for example, is valuable only if plan quality remains acceptable and the saved time improves access or reduces staff burden.

Results should be reported in a way that allows another group to reproduce the analysis. A locked dataset, versioned code repository and clear record of protocol deviations support credible publication. The economic effects of collaboration also deserve attention: professional meetings can create local value through accommodation, transport and visitor spending, as described in the account of the Banff conference economy.

Put the launch plan into practice

A realistic launch plan gives the network enough time to complete contracting, ethics, training, equipment checks and pilot cases. Australian centres should account for public-hospital procurement cycles, annual leave around Christmas and January, and the time required to move equipment or staff between distant cities. Scheduling a single national meeting may be expensive, so a hybrid model with regional workshops in Sydney, Melbourne or Brisbane can improve participation.

The following actions help convert a research concept into a controlled, operational trial:

  • Write a one-page protocol summary with the clinical question, endpoint, intervention and safety boundaries.
  • Complete a site capability survey covering equipment, software, staffing, patient volume and data systems.
  • Run a phantom or retrospective pilot before enrolling the first participant.
  • Establish a central repository with permissions, naming rules, version control and audit records.
  • Appoint a quality and safety lead with authority to pause implementation when limits are exceeded.
  • Budget for training, data management, independent review, travel, software support and publication.

Professional maintenance should be included in the workforce plan. Physicists may need continuing professional development, documented competencies and time away from routine service work. Information about CCPM renewal guidance reflects the broader importance of maintaining recognised expertise as methods, regulations and technologies change.

A successful trial leaves behind more than a paper. It creates a shared measurement culture, strengthens communication between institutions and gives departments evidence for adopting safe, efficient practice. Begin with a clearly bounded clinical question, appoint accountable leaders at every site, test the workflow on real data and document each decision. That disciplined approach gives an Australian medical physics network the best chance of producing results that clinicians can trust and patients can benefit from.