Large-scale linked datasets combining cancer registry information with administrative health data have become essential infrastructure for cancer health services research and policy. These resources enable comprehensive examination of the cancer care continuum, from diagnostic pathways and treatment patterns through to survivorship, costs, and long-term outcomes. They can also help generate evidence to inform clinical practice guidelines, health technology assessment, health economics, and health system planning.

Internationally, linked cancer datasets combining registry information with administrative health data have been established across diverse healthcare systems. Countries with unique personal identifiers and centralised data systems, such as Denmark [1] and Taiwan [2, 3], achieve near-universal linkage across cancer registrations, hospital care, and prescription data. In the United Kingdom, national cancer registry linkages to Hospital Episode Statistics and the Systemic Anti-Cancer Therapy (SACT) dataset provide whole-of-population coverage within a single-payer system [4]. SEER-Medicare in the United States of America, established in 1991, combines cancer registry data with Medicare claims but is restricted to adults aged 65 years and older in fee-for-service Medicare [5, 6]. Countries with federated healthcare systems, including Canada and Australia, face additional challenges in achieving cross-jurisdictional data harmonisation, with provincial and state-level registries maintaining varying levels of treatment data linkage [7].

Australia has a universal healthcare coverage system providing publicly funded access to prescription medicines through the national Pharmaceutical Benefits Scheme (PBS), medical services through the Medicare Benefits Schedule (MBS), and hospital care, for which extensive administrative data collections are maintained to provide reimbursement for these services. Despite recent advancements in Australian health data linkage capabilities, such as the National Health Data Hub [8] or Person Level Integrated Data Asset [9], there remains no established infrastructure for large-scale linkage of cancer registry data with treatment and outcome information at a national level that includes all Australian jurisdictions. As a result, there are significant gaps in understanding diagnostic pathways, quantifying the number of treatments provided to patients, characterising patterns of care beyond first-line settings, examining survivorship and long-term outcomes, and estimating costs of cancer care for patients and payers.

The Transforming Outcomes through Research in Cancer Healthcare in Victoria (TORCH-VIC) cohort was established in 2022 to address this gap within Victoria, Australia’s second most populous state. Victoria is located in southeastern Australia, with an estimated population of 7 million representing approximately 25% of the national population [10]. The state capital, Melbourne, is Australia’s second-largest city (5.3 million people), and home to approximately 75% of Victoria’s population, with the remainder in regional and rural areas [10]. Victoria has a large population base, with a mix of urban and regional communities, socioeconomic, and cultural diversity, which makes it well-suited for cancer health services research, including for examining equity in access to care and outcomes.

The primary aims of the TORCH-VIC cohort are 1) to determine the type and frequency of diagnostic tests and treatments used in contemporary Victorian clinical practice, 2) to estimate the costs of cancer care, 3) to identify disparities in access and provision of care, and 4) to study adverse events and multi morbidity in the cancer setting.

The foundation of TORCH-VIC is the Victorian Cancer Registry (VCR), a population-based registry responsible for accurate and timely reporting of cancer incidence, mortality and survival in Victoria [11]. Under the Victorian legislation Improving Cancer Outcomes Act 2014 and its related regulations, all Victorian hospitals, pathology, and radiotherapy services are required to report to the VCR the details of all persons with a cancer diagnosis [12]. This mandatory reporting ensures whole-of-population coverage of all people diagnosed with cancer in Victoria. The VCR is complemented by comprehensive state-based administrative datasets including hospital admissions, emergency presentations, radiotherapy treatments, elective surgery, and death records, alongside national datasets capturing prescription medicines and medical services. TORCH-VIC connects these data sources through cross-jurisdictional linkage, bridging state and national systems to maximise the public benefit derived from these comprehensive data collections.

TORCH-VIC was initially developed to support the Predicting the population health economic impact of current and new cancer treatments (PRIMCAT) research program [13], which addressed national Health Technology Assessment (HTA) policy challenges and data gaps. PRIMCAT aimed to support Australian HTA agencies with validated epidemiological data and models [14] to support subsidy decision making and streamline access to new cancer treatments for Australians affected by cancer [15]. Consumer representatives participated in the initial research prioritisation and contributed to TORCH-VIC data development and governance, ensuring that the research agenda reflects their perspectives and community benefit. The seed funding for the TORCH-VIC cohort (also known as the PRIMCAT dataset) came from the PRIMCAT research program, funded by the Medical Research Future Fund (grant number 2020/MRF1199701).

TORCH-VIC is a population-based linked data resource covering all adults (aged ≥ 18 years) diagnosed with colorectal, lung, melanoma, prostate, breast cancer, and lymphoid leukaemia in Victoria, Australia, between 1 January 2010 and 31 December 2021. Specifically, these include the five most commonly diagnosed cancers in Australia, which collectively account for over half of all cancer diagnoses, and lymphoid leukaemia to provide representation of haematological malignancies. These cancer types encompass diverse treatment modalities and have experienced substantial therapeutic advances, making them priorities for evidence generation from routinely collected health data. Data collections for linkage were selected based on availability within the Australian health data infrastructure, coverage of key domains across the cancer care continuum, and alignment with the study aims. Cohort members were identified through the Victorian Cancer Registry (VCR). The cohort comprises individual-level administrative and registry data from three national and eight Victorian (state-based) data collections. All data sources linked in TORCH-VIC are described in Table 1, which also presents a data coverage timeline. Cohort identification is based on VCR cancer diagnoses from January 2010 to December 2021, however, the temporal coverage of linked data collections extends beyond this period. PBS and MBS data are available from January 2008, enabling analysis of pre-diagnosis healthcare utilisation, while most other collections extend to mid-2023, providing continued follow-up of outcomes.

Through linkage to the PBS and MBS, TORCH-VIC captures comprehensive data and generates real-world evidence on treatment patterns and therapeutic advancements that have been shown to significantly enhance survival outcomes for many cancers [35]. The longitudinal design of the cohort enables detailed tracking of cancer care trajectories over time, from initial diagnosis through various treatment stages. This approach is essential not only for understanding the direct impacts of treatments but also for assessing associated costs and resource utilisation. Regular updates to TORCH-VIC with new diagnoses and additional follow-up data keep the study relevant and reflective of current practices. TORCH-VIC data can inform the Australian and Victorian cancer plans [36, 37], as well as support healthcare professionals and policymakers to evaluate cancer care outcomes and associated costs in Victoria.

TORCH-VIC has already been used in several projects. These projects include forecasting the number of individuals diagnosed with cancer who are eligible for novel medicines [15, 38], analysing treatment patterns and their associated outcomes, reporting on overall cancer care costs [39], quantifying travel distance and time to radiotherapy treatment centres across Victoria [40, 41], and quantifying immune-related adverse events for Victorians who have received immunotherapy treatment [42].

TORCH-VIC enables diverse research applications. The linked data support evaluation of treatment effectiveness and safety profiles in routine clinical practice, analysis of time to treatment initiation and its relationship with outcomes, and investigation of comorbidity patterns and their impact on treatment decisions and outcomes. The linkage to PBS captures all dispensed medicines across therapeutic classes, enabling research on comorbidity management, polypharmacy, and supportive care alongside cancer treatment, all increasingly recognised as central to cancer outcomes. The population-based design also enables examination of inequities in cancer care access and outcomes by geographic location and socioeconomic status, assessment of adherence to clinical guidelines, optimal care pathways, estimation of the financial burden of cancer care, and modelling the impact of new therapies on population health outcomes.

The increasing incidence of cancer in younger adults represents an emerging public health challenge. TORCH-VIC captures approximately 14,000 individuals diagnosed before age 50 who may face decades of survivorship. The cohort’s linkage infrastructure enables tracking of late effects, fertility impacts, secondary cancers, mental health outcomes, and healthcare utilisation patterns over the life course. This is particularly relevant for understanding treatment-related morbidity that may emerge years or decades post-diagnosis, informing survivorship care models, and quantifying the long-term societal impact of cancer in younger populations. As these individuals age, TORCH-VIC can capture their complete health trajectories, supporting evidence-based survivorship guidelines and health service planning for the growing population of long-term cancer survivors.

TORCH-VIC will integrate additional cancer types and data collections in 2026, to increase its coverage and scope. These include the National Cancer Screening Register, the Australian Immunisation Register, the Australian and New Zealand Intensive Care Society, the Transmission and Response Epidemiology Victoria, the Community Health Minimum Data Set, and the Home and Community Care dataset. The cohort will also expand to include female-specific cancers (vulva, cervix, uterus, ovary), additional solid cancers (brain, liver, bladder, pancreas, kidney, head and neck, oesophagus, stomach), and blood cancers (Hodgkin lymphoma, non-Hodgkin lymphoma, multiple myeloma, and additional leukaemia types).

TORCH-VIC has several limitations. The study focuses on the Victorian population, which could limit the broader applicability of its findings to other regions or across the entire Australian population. While the VAED dataset is robust in capturing hospital services accessed in Victoria, if people seek hospital care outside of Victoria, this will not be reflected in the dataset (inpatient stay). However, all data reported in PBS and MBS are federal and thus captured regardless of the location of care across Australia. Additionally, TORCH-VIC captures only subsidised treatments and services through PBS and MBS. Medicines accessed through clinical trials, compassionate access programs, or private funding are not captured, which may affect the completeness of treatment data particularly for novel therapies. Stage at diagnosis completeness varies across cancer types, with limited reporting for some cancers such as lung, although treatment-based proxies derived from PBS dispensing data and other data collections can be used to infer advanced disease where stage data are incomplete.

Further, current data collections rely on administrative records, which does not include pathology, imaging, test results, or individual-level socio-economic data such as those captured through Census data. The absence of specific lifestyle data and patient-reported outcomes limits the ability to fully understand cancer aetiology, predict treatment outcomes, and assess impacts on quality of life and patient preferences. Finally, observed variations in healthcare utilisation identified through TORCH-VIC may reflect differences in regional care delivery models, patient preferences, and access barriers, which cannot be ascertained using administrative data alone.

TORCH-VIC adds to a growing international landscape of cancer data linkage resources. Australia’s universal PBS and MBS coverage enables capture of all subsidised medicines and medical services across all adult age groups, providing a complementary perspective to these existing data assets. The availability of such resources across diverse healthcare systems creates opportunities for international comparisons of cancer care patterns, treatment outcomes, and health system performance.

TORCH-VIC provides critical infrastructure for comprehensive real-world evidence on cancer care in Victoria, Australia. Through its comprehensive linkage of population-based cancer diagnoses with prescription medicines, medical services, and healthcare utilisation data, the cohort enables detailed examination of treatment patterns, costs, and outcomes across the cancer care continuum. The methodological framework established through TORCH-VIC can serve as a blueprint for similar initiatives in other Australian states [43, 44], supporting evidence-based policy and practice improvements in cancer care.

FF, KT, BD, SA, SAP, and MIJ were involved in the development of the original protocol document on which this manuscript was based. SA also provided critical consumer perspective into the data collections for linkage. FF, KT, SAP and MIJ acquired the data. FF prepared and analysed the data. All authors interpreted the data. FF drafted the manuscript. All authors critically reviewed and approved the final manuscript.