Registry-derived stage (RD-Stage) for capturing stage at diagnosis for pancreatic carcinoma in Australia.

ABSTRACT

INTRODUCTION: Stage of pancreatic carcinoma at diagnosis is a strong prognostic indicator of morbidity and mortality, yet is poorly notified to population-based cancer registries ("cancer registries"). Registry-derived stage (RD-Stage) provides a method for cancer registries to use available data sources to compile and record stage in a consistent way. This project describes the development and validation of rules to capture RD-Stage (pancreatic carcinoma) and applies the rules to data currently captured in each Australian cancer registry. MATERIALS AND METHODS: Rules for deriving RD-stage (pancreatic carcinoma) were developed using the American Joint Commission on Cancer (AJCC) Staging Manual 8th edition and endorsed by an Expert Working Group comprising specialists responsible for delivering care to patients diagnosed with pancreatic carcinoma, cancer registry epidemiologists and medical coders. Completeness of data fields required to calculate RD-Stage (pancreatic carcinoma) and an overall proportion of cases for whom RD stage could be assigned was assessed using data collected by each Australian cancer registry, for period 2018-2019. A validation study compared RD-Stage (pancreatic carcinoma) calculated by the Victorian Cancer Registry with clinical stage captured by the Upper Gastro-intestinal Cancer Registry (UGICR). RESULTS: RD-Stage (pancreatic carcinoma) could not be calculated in 4/8 (50%) of cancer registries; one did not collect the required data elements while three used a staging system not compatible with RD-Stage requirements. Of the four cancer registries able to calculate RD-Stage, baseline completeness ranged from 9% to 76%. Validation of RD-Stage (pancreatic carcinoma) with UGICR data indicated that there was insufficient data available in VCR to stage 174/457 (38%) cases and that stage was unknown in 189/457 (41%) cases in the UGICR. Yet, where it could be derived, there was very good concordance at stage level (I, II, III, IV) between the two datasets. (95.2% concordance], Kendall's coefficient = 0.92). CONCLUSION: There is a lack of standardisation of data elements and data sources available to cancer registries at a national level, resulting in poor capacity to currently capture RD-Stage (pancreatic carcinoma). RD-Stage provides an excellent tool to cancer registries to capture stage when data elements required to calculate it are available to cancer registries.