Use of health insurance data to identify and quantify the prevalence of main comorbidities in lung cancer patients.

BACKGROUND

Identifying comorbidities in lung cancer patients is a complex process in population-based studies and no gold standard exists. The current study aims to identify and measure the main comorbidities using administrative health insurance data, which were available on a population-based level.

METHOD

A literature search was conducted to identify comorbidities in lung cancer patients and to select Anatomical Therapeutic Chemical codes to measure them. For each patient, the volume of delivered relevant drugs for each comorbidity in the year preceding the diagnosis of lung cancer was computed, based on the Defined Daily Doses reimbursed. Case definition rules were set by comparing the identification of comorbidities via health insurance data with the reporting of them in the medical files in a sample of hospitals.

RESULTS

Four comorbidities were identified: chronic respiratory diseases, chronic cardiovascular diseases, diabetes mellitus and renal diseases. A very good to moderate agreement between the prevalence based on medical files versus health insurance data was obtained for diabetes mellitus (kappa = 0.83), chronic cardiovascular diseases (kappa = 0.64), chronic respiratory diseases (kappa = 0.48) but not for renal diseases (kappa = 0.22). Because only 27% of patients having renal diseases recorded in the medical files were identified using health insurance data, this comorbidity was not withheld. Among 12,839 lung cancer patients diagnosed in 2010-2011 in Belgium, 29.7% had chronic respiratory diseases, 57.5% had chronic cardiovascular diseases and 14.1% had diabetes mellitus.

DISCUSSION

This study showed that it was possible to capture three major comorbidities in lung cancer patients using administrative health data, namely, diabetes mellitus, chronic cardiovascular diseases, and chronic respiratory diseases. However, the agreement was only moderate for the last one. A prerequisite for using this methodology is that administrative health data are available for all patients.