Our study provides a comprehensive and up-to-date overview of real-world treatment patterns among lung cancer patients in Germany. Including more than 42,000 patients diagnosed in 2015–2018, we quantified proportions of different treatments (systemic therapy, surgery, radiotherapy) by age and found marked differences. For example, surgery was conducted in about 30% of patients aged 50–79 vs. 18% of patients aged 80 or older. In patients without surgery, 72% of patients aged 50–69 were treated with systemic anticancer therapy vs. 54% of patients aged 70–79 and 25% of patients aged 80 or older.

To our knowledge, there is only one study from Germany investigating treatment of lung cancer patients according to age (Walter et al. 2019). The authors of this study, which included 13,283 patients, concluded that the likelihood of receiving tumor-directed treatment was lower in patients above 65 as compared to younger patients. However, the included patients were diagnosed in 2009. Since then, treatment options were extended, e.g., due to the rapidly evolving landscape of new therapeutics, especially monoclonal antibodies and tyrosine kinase inhibitors. Furthermore, follow-up was limited to six months in this study.

Studies investigating age-related patterns in the use of tumor-directed therapy from other countries that are comparable to our study (i.e., focus on the whole group of lung cancer patients) are also scarce and only included cancer patients diagnosed before 2012 (Nilssen et al. 2016; Costa et al. 2017). In a study from Norway including 24,324 lung cancer patients diagnosed between 2002 and 2011 the proportion receiving surgery was 23% in patients aged 50–69 years, 18% in patients aged 70–79 years and 5% in patients aged 80 years or older (Nilssen et al. 2016). A study from Brazil including 40,403 lung cancer patients diagnosed between 2000 and 2011 compared treatment in patients < 70 vs. ≥ 70 years. While the proportion receiving surgery was rather similar in both age groups and in those < 70 years lower compared to our study (18% vs. 16%), the proportion receiving systemic anticancer therapy showed an age gradient (62% vs. 49%) (Costa et al. 2017).

In our study, 21% of included lung cancer patients did not receive any cancer-directed treatment. This proportion increased with age: it was 12% in patients aged 50–69 years, 22% in patients 70–79 years and 48% in patients aged 80 years or older. Also in this regard, there are only a few comparative studies that focus on the entire group of lung cancer patients. A study from Australia including 1116 lung cancer patients with a mean age of 72 years diagnosed between 2006 and 2013 as well as a study from Korea including 2,148 lung cancer patients diagnosed between 2009 and 2014 both reported a proportion of 28% untreated patients (Ngo et al. 2022; Choi et al. 2019). When weighing up the potential benefits and harms in older patients, the higher prevalence of comorbidities and poorer performance status as compared to younger patients certainly influence treatment decisions. On the one hand, the physician may assess the risks to be greater than the benefits and therefore advise against treatment. On the other hand, the patient may also decide against treatment for other reasons. A systematic review that summarized studies describing factors that led older adults to accept or decline cancer treatment found considerable variation in the underlying reasons, but the most consistent determinant was physician recommendation. However, the review also emphasized the need for further studies that are based on large, representative samples and examine decision-making taking into account health literacy and comorbidity (Puts et al. 2015).

From the perspective of clinical cancer registration, knowledge on the extent of the age gradient for the different types of treatment as well as the proportion of untreated patients is of high importance. For epidemiological cancer registration, there are various established methods for estimating completeness, i.e. the extent to which all of the incident cancers occurring in the population are included in the registry (Parkin and Bray 2009). To estimate the completeness of treatment information for incident cancers, however, quality indicators are lacking. Without reference values, clinical cancer registration lacks guidance as to whether the data are sufficiently complete to describe cancer care or whether further measures need to be taken due to a lack of completeness. Based on the results of our study, the expected proportions can roughly be estimated for the different types of treatment. It is also possible to take into account the age distribution of the incident patients recorded in the respective registry and combine it with our age-specific values, which allows the expected proportions to be estimated even more accurately. Even though the accuracy of such quality indicators could certainly be further improved, the analyses presented here based on health claims data can be useful as a first pragmatic approach to assessing the completeness of treatment data from clinical cancer registration.

Our study had strengths and limitations. Strength include the large sample size, the sophisticated algorithm used to ensure the valid identification of lung cancer patients in claims data, the continuous follow-up as well as the completeness of information regarding therapy in the inpatient and outpatient setting. Completeness in claims data does not depend on active reporting by physicians as it is the case in cancer registry data, and there is no bias due to non-responder as it is the case in survey data. Unlike cancer registry data, however, claims data lack detailed information on stage at diagnosis and there is no information on histology. Some advanced stages may have been misclassified as non-advanced if the patients died shortly after diagnosis without any treatment that would have prompted the coding of affected lymph nodes or distant metastases. These limitations are inherent to German health claims data but in the absence of other data sources, health claims data are still useful for the purposes described above. A record linkage of cancer registry and health claims data combining the advantages of both databases would be ideal and there are currently legislative prospects in this respect, but it will be a while before this is implemented.