Temporal trends in relative survival following percutaneous coronary intervention

OBJECTIVE: Percutaneous coronary intervention (PCI) has seen substantial shifts in patient selection in recent years that have increased baseline patient mortality risk. It is unclear to what extent observed changes in mortality are attributable to background mortality risk or the indication and selection for PCI itself. PCI-attributable mortality can be estimated using relative survival, which adjusts observed mortality by that seen in a matched control population. We report relative survival ratios and compare these across different time periods. METHODS: National Health Service PCI activity in England and Wales from 2007 to 2014 is considered using data from the British Cardiovascular Intervention Society PCI Registry. Background mortality is as reported in Office for National Statistics life tables. Relative survival ratios up to 1 year are estimated, matching on patient age, sex and procedure date. Estimates are stratified by indication for PCI, sex and procedure date. RESULTS: 549 305 procedures were studied after exclusions for missing age, sex, indication and mortality status. Comparing from 2007 to 2008 to 2013-2014, differences in crude survival at 1 year were consistently lower in later years across all strata. For relative survival, these differences remained but were smaller, suggesting poorer survival in later years is partly due to demographic characteristics. Relative survival was higher in older patients. CONCLUSIONS: Changes in patient demographics account for some but not all of the crude survival changes seen during the study period. Relative survival is an under-used methodology in interventional settings like PCI and should be considered wherever survival is compared between populations with different demographic characteristics, such as between countries or time periods.