Abstract
Treatment of lung cancer depends on the stage of the tumor and the histological type. In recent years, the histological confirmation of lung non-small-cell lung cancer has become crucial since the availability of selective target therapeutic approaches. The aim of the study was to develop a validated procedure to estimate the incidence and prevalence of non-small-cell and small-cell lung cancer from healthcare administrative data. A latent class model for categorical variables was applied. The following observed variables were included in the analysis: ICD-9-CM codes in the Hospital Discharge Registry, ATC codes of medications dispensed present in the Drugs Prescriptions Registry, and the procedure codes in the Outpatient Registry. The proportion of non-small-cell lung cancer diagnoses was estimated to be 85% of the total number of lung cancer on the cohort of incident cases and 89% on the cohort of prevalent cases. External validation on a cohort of 107 patients with a lung cancer diagnosis and histological confirmation showed a sensitivity of 95.6% (95%CI: 89-98.8%) and specificity of 94.1% (95%CI: 71.3-99.9%). The procedure is an easy-to-use tool to design subpopulation-based studies on lung cancer and to better plan resource allocation, which is important since the introduction of new targeted therapies in non-small-cell lung carcinoma.