# Exploring the potential of German claims data to identify incident lung cancer patients

## TL;DR

This study evaluates a method to identify new lung cancer patients in German healthcare data, finding it effective and yielding plausible results.

## Contribution

The study validates an algorithm's effectiveness in identifying incident lung cancer cases using German claims data.

## Key findings

- Identified approximately 9,500–10,500 incident lung cancer patients annually in Germany.
- Age-standardized incidence rates were 45 per 100,000 in men and 27 per 100,000 in women in 2018.
- Survival rates were higher in women compared to men.

## Abstract

Real-world healthcare databases offer great potential for cancer research, but the valid identification of cancer patients is crucial for the suitability of a database in this regard. We aimed to assess the plausibility of an algorithm to identify incident lung cancer (LC) patients in German claims data.

Using the German Pharmacoepidemiological Research Database (GePaRD; claims data from ∼ 20% of the German population) we applied a previously developed algorithm which identifies incident LC patients and classifies them into advanced and non-advanced. We calculated age-standardized incidence rates (ASIRs) per 100,000 for the years 2013–2018. Further, we assessed the ASIRs stratified by the deprivation index of the district of residence and determined age-standardized five-year absolute and relative survival. We stratified all analyses by sex.

Overall, we identified ∼ 9,500 − 10,500 incident LC patients per year. In 2018, (N = 10,625, mean age: 69.2 years in men) the proportion classified as advanced at diagnosis was 71.4%; the ASIRs of LC were 45 per 100,000 in men (9% lower than in 2013) and 27 per 100,000 persons in women (similar to 2013). ASIRs were lowest in persons living in areas with a low deprivation index. Age-standardized five-year absolute and relative survival rates, respectively, were 31% and 34% in women and 27% and 31% in men.

The algorithm we applied to identify incident LC patients in German claims data yielded plausible results, supporting its validity.

Not applicable.

The online version contains supplementary material available at 10.1186/s12890-025-03740-8.