Survival analysis based on cancer registry data is of paramount importance for monitoring the effectiveness of health care. As new methods arise, the compendium of statistical tools applicable to cancer registry data grows. In recent years, machine learning approaches for survival analysis were developed. The aim of this study is to compare the model performance of the well established Cox regression and novel machine learning approaches on a previously unused dataset. The study is based on lung cancer data from the Schleswig-Holstein Cancer Registry. Four survival analysis models are compared: Cox Proportional Hazard Regression (CoxPH) as the most commonly used statistical model, as well as Random Survival Forests (RSF) and two neural network architectures based on the DeepSurv and TabNet approaches. The models are evaluated using the concordance index (C-I), the Brier score and the AUC-ROC score. In addition, to gain more insight in the decision process of the models, we identified the features that have an higher impact on patient survival using permutation feature importance scores and SHAP values. Using a dataset including the cancer stage established by the Union for International Cancer Control (UICC), the best performing model is the CoxPH (C-I: 0.698±0.005), while using a dataset which includes the tumor size, lymph node and metastasis status (TNM) leads to the RSF as best performing model (C-I: 0.703±0.004). The explainability metrics show that the models rely on the combined UICC stage and the metastasis status in the first place, which corresponds to other studies. The studied methods are highly relevant for epidemiological researchers to create more accurate survival models, which can help physicians make informed decisions about appropriate therapies and management of patients with lung cancer, ultimately improving survival and quality of life.

Background: Patient-reported outcome measures (PROMs) and patient-reported experience measures (PREMs) are tools for assessing outcomes of and experiences with health care from the patient’s perspective. In Germany, PROMs are widely used in research for evaluating patient outcomes and quality of care. However, the application of PREMs is rather scant, especially in oncology. The study aimed to assess the feasibility of patient-centred quality evaluation in oncological care in Germany using the German adaptation of the Danish National Cancer Patient Questionnaire. This questionnaire is a PREM/PROM-tool addressing patients of all cancer sites and covering the entire cancer patient pathway. Methods: The Danish National Cancer Patient Questionnaire was translated into German via forward-backward translation. Face-validity was tested among three cancer patients in a conventional pre-test. The German adaptation contains 99 questions. A pilot test was carried out among 245 newly diagnosed breast and colorectal cancer patients in the German federal state Schleswig-Holstein. Patients were recruited via clinics participating in the Oncological Care Registry (12 specialised units in seven hospitals) and contacted six to nine months after diagnosis. Response behaviour and response patterns were compared to the Danish study population (n = 1964). Results: The willingness among clinicians to support patient recruitment as well as the response rate of patients to the questionnaire was high (65%). Moreover, response behaviour and response patterns of German and Danish patients were consistent. Despite the generally good response behaviour of patients to the single items, the authors observed that questions assessing the diagnostic process did not fully capture German pathways. Only 19.3% of the German patients stated that their diagnostic process was initiated by a visit to a general practitioner (GP) in contrast to 52% in Denmark. The assessment of patient experiences in the diagnostic phase heavily focuses on experiences in general practice, which does not seem appropriate in the German health care setting. Conclusion: The translation was successful, and the feasibility of a future large-scale study within existing structures is given. However, some modifications of questions heavily related to the Danish health care system, especially referring to the diagnostic phase, are necessary.

The aim of this study was (i) to update the reporting of colorectal cancer survival differences over time in the German–Danish border region (Schleswig-Holstein, Southern Denmark, and Zealand) and (ii) to assess the extent to which it can be explained by stage and primary treatment. Incident invasive colorectal cancer cases diagnosed from 2004 to 2016 with a follow-up of vital status through 31 December 2017 were extracted from cancer registries. Analyses were conducted by anatomical subsite and for four consecutive periods. Kaplan–Meier curves and log-rank tests were computed. Cox regression models using data from Schleswig-Holstein from 2004 to 2007 as the reference category were run while controlling for age, sex, stage, and treatment. The cox regression models showed decreasing hazard ratios of death for all three regions over time for both anatomical subsites. The improvement was stronger in the Danish regions, and adjustment for age, sex, stage, and treatment attenuated the results only slightly. In 2014–2016, colon cancer survival was similar across regions, while rectal cancer survival was significantly superior in the Danish regions. Regional survival differences can only partially be explained by differing stage distribution and treatment and may be linked additionally to healthcare system reforms and screening efforts.