Sociodemographic and Health-Service Precursors of Local-Stage Lung Cancer Diagnosis: A Population-Based Study in New South Wales, Australia

<b>Aims:</b> To investigate sociodemographic associations with medicated health conditions, general practitioner (GP) contacts, and computed tomography (CT) scans of the lung, as 12-month precursors of diagnosis of lung cancer at a local stage (cancers localized to the primary site of bronchus and lung). <b>Methods:</b> Cancer Registry data for New South Wales (NSW) adults diagnosed with lung cancer (ICD-10 C33-34) since the Census of August 2016 (n = 6160) were linked at person level with census and other administrative data. These included residents diagnosed with lung cancer from September 2016 to December 2018. Structural equation modelling indicated adjusted measures of associations with lung cancer, including adjusted odds ratios (aORs), in stepped analyses. <b>Results:</b> The first part of the multivariate modelling showed age, major city residence, and other sociodemographic characteristics that were associated with numbers of medicated conditions. The second part showed the numbers of medicated conditions and other sociodemographic characteristics that were associated with the number of GP consultations. The third part of the modelling showed the numbers of GP consultations and other sociodemographic characteristics that were associated with having CT lung scans. Modelling showed that having CT scans and being female were the main predictors of lung cancer diagnosis at a local stage, with aORs of 2.30, 95%CI 2.01-2.63 and 1.39, and 95%CI 1.23-1.58, respectively. The modelling also showed age, GP consultations, residence in a major city, and other sociodemographic characteristics to be associated with having CT scans. <b>Conclusions:</b> The findings of the study indicate the main precursors of lung cancer diagnosis at a local stage after multivariate adjustment. Irrespective of causal significance, results reveal population-wide characteristics for targeting screening for early detection. They demonstrate the potential value of person-level linkage of cancer-registry data with census and other administrative data for this purpose. Our study of linked cancer-registry and census data revealed broad descriptive features of pathways to early diagnosis of relevance to service screening and planning.