Explanation for the choice of comparators Control group

The comparator for this study is continued use of the baseline standard radiology reporting process (i.e. standard reporting) at each of the participating sites (prior to the commencement of synoptic reporting).

Current standard radiology reporting varies, with the level of detail and the style of the report being radiologist dependent. The report is typically dictated by radiologists reviewing scans and usually written as a narrative of varying lengths and complexity, finishing with a conclusion. We will accurately document the current standard reporting process, and later compare it with the synoptic reporting template that is under study.

Intervention description Experimental group

After a site commences the use of synoptic radiology reporting (i.e. synoptic reporting), the site radiologist reviewing CT scans will use a structured reporting tool that was developed in the pilot study [12] to describe the patient’s pancreatic lesion in detail. The synoptic report template collects 63 discrete fields of data that concisely describe the pancreatic cancer mass characteristics, blood vessel, and other adjacent structure involvement, and determines the extent of local disease spread. The radiologist will assess a patient’s resectability status (based on the international consensus guidelines) by completing the synoptic report in a Research Electronic Data Capture (REDCap) application. The synoptic reporting template has been previously described [12]. The median time needed to complete a synoptic report for a patient by a radiologist during the pilot study was 4 min, so it is assumed the time will be similar for this study.

Criteria for discontinuing or modifying allocated interventions

Not applicable. As mentioned above, after a site commences synoptic reporting, all the site radiologist/s will be reporting the CT scans using synoptic reporting rather than standard reporting. For trial patients, there is no “intervention” as such that any of them would need to discontinue.

Strategies to improve adherence to interventions

Training will be provided to the sites’ radiologists by a project data manager and an experienced radiologist who has previously used the synoptic reporting tool. The training session will be delivered online and it will take about 30 min to complete. Training will commence 1 month prior to the start of the synoptic reporting phase of a sites’ participation.

To monitor adherence to the synoptic reporting tool, the data manager in the central project team will ensure data completeness and accuracy by undertaking quality assurance checks and ensuring ongoing training and support to participating radiologists is available.

Relevant concomitant care permitted or prohibited during the trial

Not relevant, this study relates only to the discussion and planning phase of treatment and there will be no impact on concomitant or future care subsequent to the MDM presentation.

Provisions for post-trial care

Not relevant. This study relates only to the discussion and planning phase of treatment.

Outcomes

The primary outcome is the proportion of patients diagnosed with BR PDAC, aiming to detect a change in this proportion before and after introduction of the synoptic report. There are a number of secondary outcomes (see Table 1).

Participant timeline

After obtaining site governance authorisation, randomisation, and sequence allocation (Fig. 2), all patients who undergo presentation of radiological staging of PDAC at each participating institution MDM will be included in the study, and their baseline data and subsequent data will be collected as per the descriptions in Table 3.

Fig. 2

SPIRIT figure—Schedule of enrolment, interventions, and assessments of SCANPatient clinical trial. This is a batched, stepped wedge cluster randomised trial, more details of randomisation and sequencing can be found in Table 2. **For more details about timepoints for hospital allocation, please see Table 2

Table 3 Data required, sources, methods and time to collect themaSample size

Currently 8% of all patients diagnosed with PDAC are categorised with BR disease in the Upper Gastrointestinal Cancer Registry (UGICR; unpublished data) which is operating in most major centres in Victoria and approximately half of the major centres in New South Wales. We hypothesise that with the introduction of a new structured radiology report, leading to increased diagnostic accuracy, this will increase to 15%. This is supported by the final results of the pilot synoptic report study (wherein the rate of BR disease was 13% in the pilot study across 95 patients [12]) and international registry data which have shown that the rate of resection of major vascular structures at the time of surgery for PDAC is between 10 and 15% [14], a surrogate for involvement of those major structures and therefore a pseudo-definition of BR disease. As not all patients with BR PDAC will require formal venous or arterial resection, it is likely the actual rate of BR disease is higher in these larger international and national registries. An increase in the diagnosed rate of BR disease to this level will result in more patients appropriately being given neoadjuvant therapy, and a clinically meaningful change in the standard of care if identified.

The trial was originally designed as a standard, stepped wedge design with 8 hospitals randomised to each sequence of a standard 5-sequence/6-period stepped wedge design (a total of 40 hospitals). With 15 patients per hospital in each 6-month period, assuming a two-sided significance level of 0.05, this design has power ≥ 80% to detect an increase from 8 to 15% in the primary outcome. Power is maintained for a wide range of intracluster correlations and alternative within-cluster correlation structures, e.g. for a discrete time decay within-cluster correlation structure [15], with an intracluster correlation of 0.1 and a cluster autocorrelation of 0.55, power is 84%. Under most within-cluster correlation structure assumptions, power will remain > 76% should hospital recruitment be up to 25% lower and patient recruitment up to 30% lower than expected (i.e. with 10 patients each year in each of 30 hospitals). With the batched variant shown in Table 2, for the extreme scenario considered above (an intracluster correlation of 0.1 and cluster autocorrelation of 0.55), power levels remain > 80%. Sample size calculations were performed using the Shiny CRT calculator [16].

Recruitment

Hospitals were invited to register their interest in participating in this trial via an Expression of Interest (EOI) process. An EOI form was initially distributed through professional networks and organisations such as the Australian & New Zealand Hepatic, Pancreatic and Biliary Association (ANZHPBA), the Abdominal Radiology Group Australia and New Zealand (ARGANZ) of the Royal Australian and New Zealand College of Radiologists (RANZCR), the Australasian Gastro-intestinal Trial Group (AGITG), and the Royal Australasian College of Surgeons (RACS). Eligible hospitals that submitted an EOI and agreed to participate were further contacted with the help and coordination from local surgeons and/or radiologists (many of whom acted as site PIs or AIs). Relevant ethics and research governance applications were prepared and submitted, and after the site governance authorisation has been granted for the trial to be conducted at a specific site, all patients who undergo radiological staging of PDAC at each participating hospital will be included in the study. Individual clinicians will list patients for MDM discussion at the participating sites as per usual care.