Consent, participation, and non-response in a registry-based survey of long-term cancer survivors in Germany

BACKGROUND: Long-term cancer survivors constitute a growing population with characteristic health care needs. Survey-based studies are essential to understanding this population, yet low participation raises questions about potential non-response bias. This study analysed response and non-response in a registry-based cross-sectional survey among long-term cancer survivors in North Rhine-Westphalia, Germany. METHODS: A sample of 24 291 adult long-term cancer survivors was drawn from the North Rhine-Westphalian Cancer Registry (diagnosed between 2008 - 2018, ≥5 years post-diagnosis, aged 18-74 at diagnosis). Survivors were contacted by postal letters and asked for consent. Those consenting received a questionnaire. Survey participation was defined as returning a questionnaire (paper-based or online). Age, sex and clinical data (calendar period of diagnosis, tumour stage and ICD-10 group) were available for all invitees and used to compare responders with non-responders. RESULTS: Of 24 291 invited survivors, 3 297 (13.6%) consented and 2 355 (9.7%) answered the questionnaire; 71.4% of those who consented participated. The contact proportion (survivors reached) was 85.8%, and the cooperation proportion (participants among eligible survivors reached) 11.3%. Non-response patterns differed across subgroups: younger survivors were more difficult to contact, whereas older survivors were more likely to refuse participation. Contactability decreased with increasing time since diagnosis. Lower response was observed among survivors with advanced tumour stage and lung cancer, while higher consent was seen among prostate cancer and leukemia survivors. Among consenting survivors, non-participation was mainly due to non-return of the questionnaire, with reported reasons including fatigue and lack of time. CONCLUSIONS: Overall response in this registry-based survey was low and primarily driven by non-consent, whereas participation among those who consented was high. In the two-step recruitment approach used in this study, selection occurred mainly at the initial consent stage, while loss after consent was comparatively limited. Future studies using similar procedures should consider strategies to reduce barriers at the consent stage and account for differential response.