Current Size: 100%
Effect of active follow-up on survival estimates in a sample of cancer registry cases
Background & Introduction
Many population-based cancer registries rely mainly on passive follow-up to estimate survival. This assumes that registered persons not matched to mortality records are alive at the time-point to which all records have been completely collated. The approach enables unbiased estimates provided all deaths are properly recorded and matched and that there is no emigration of cancer patients. We undertook active follow-up on a sample of cases and approximated the impact that this may have on survival estimates.
Materials & Methods
All patients in the National Cancer Registry Ireland (NCRI) with malignant tumours (excluding non-melanoma skin cancers (NMSC)) diagnosed in the years 1994-2005 were selected. 2000 patients were randomly selected from all patients presumed alive with no activity recorded by the registry beyond 31/12/2008. In June 2012, status of these patients was determined by a government social security department source and via letters to general practitioners (GPs). Observed survival and follow-up distributions of the sampled cohort were computed. Bootstrap simulation was used to extrapolate results from the sample to other presumed alive cases and estimate the plausible impact of active follow-up compared to passive.
Results
168,252 patients were identified with malignant (ex-NMSC) tumours registered 1994-2005. Of these 35% (58,693) were presumed alive in June 2012. 42,034 of these patients had no activity beyond 2008. From the 2000 sampled patients an updated status was obtainable for 63%. There were 68 deaths detected by the government department and 52 further deaths via GPs. Some detected deaths would have eventually been reported through normal sources. 12 patients were reported as having emigrated. 1, 5, 10 year observed survival estimates for the sampled cohort were 99, 97, 93%. Simulation estimated observed survival based on follow-up as being 2.5% lower than that based on the passive approach in the wider cohort.
Conclusions
There were a number of unreported deaths detected which result in slightly lower survival estimates. Further work will examine the impact by age and on relative survival. Few patients were detected as having emigrated but this could increase in the future. Active follow-up is a resource intensive process but efforts could be targeted to maximise impact.
The NCRI is funded by Department of Health Ireland and research support is provided by the Heath Research Board.