Less pitfalls and variation in population based cancer survival comparisons within the European Union: Lessons from colorectal cancer patients in neighbouring regions in Denmark and Germany - The Fehmarn Belt project

Background: International comparisons of cancer registry based survival are often used as a marker of the performance of health care services on cancer. These are debated for comparability, validity and lack of detail with respect to prognostic factors. With a view to cross border collaborative cancer research and care in the European Union (EU), we used the established routine registration systems for cancer in the neighbouring regions Zealand, Denmark, and Schleswig-Holstein, Germany, to study if available routine registration suffices for the study of diagnosis, treatment and follow-up for colorectal cancer patients. The aim was to assess to which extent differences in survival between the regions could be explained. Methods: Colorectal cancer patients recorded 2004-2010 in Schleswig-Holstein (colon: 10,631; rectum 5683) and region Zealand (colon: 3205; rectum 1747) were studied. Excluding cases aged 90+ or only known from death certificates, one-and four-year relative survival by stage for the periods 2004-2006 and 2007-2009 (one-year) were calculated. Findings: A high proportion of patients in Schleswig-Holstein were known only from death certificates (colon 11%; rectum 6.9%) compared to <0.3% in region Zealand. Colon cancer incidence (asr-e) (men 36; women 28) and mortality (men 12; women 8.7) in Schleswig-Holstein were 10 per 100,000 lower than in Zealand; minor differences were seen for rectum cancer. One and four-year overall survival increased in both regions but was superior in Schleswig-Holstein. For patients with reported stage I-III, negligible differences in survival were seen; with further restriction to patients with treatment reported, even so in stage IV patients 2007-2009. Interpretation: Improved data quality and comparability than presently used in the large international survival studies is needed. If stage and treatment is taken into account, more valid international comparisons of cancer survival are possible. Reporting and follow-up must be improved and cases only known from death certificates (DCO) minimised. A high proportion DCO (excluded from analyses) may produce higher survival, whereas a low the opposite. Comorbidity and socio-economic status should be included alongside other prognostic variables in survival studies. Barriers towards proper follow-up and monitoring of outcome e.g. privacy legislation must be considered with a view to future patient mobility. (C) 2014 Elsevier Ltd. All rights reserved.