How to Put Survival After Cardiothoracic Interventions in the General Population Context: A Case-Based Practical Guideline to Calculate Cumulative Matched-General-Population Survival

Abstract

BACKGROUND: Observed patient survival after cardiothoracic interventions should ideally be placed in the context of matched-general-population survival. This study outlines several methodologies of matching general population mortality to the study sample, subsequently calculating cumulative matched-general-population survival, highlighting their respective advantages, disadvantages, and limitations. METHODS: A multicenter data set containing survival data after the Ross procedure was used for methodological illustration. General population mortality was extracted from country-specific life tables in the Human Mortality Database. The matched-general-population mortality, also known as background mortality, was obtained by matching general population mortality to the study sample using different matching strategies, iteratively considering median/individual age and median/individual calendar year, besides country and sex. The corresponding cumulative matched-general-population survival was calculated subsequently. Sensitivity analyses were performed to assess the impact of varying patient ages on survival estimates by adding and subtracting 15 years from individual patients. A web-based Shiny Application (App) was developed to easily calculate cumulative matched-general-population survival. RESULTS: In total, 1431 hospital survivors from the Ross procedure from 5 countries (25.7% female; median age, 48.5 [interquartile range, 42.7-54.0] years) were included. Fifteen-year survival was 88.3% (95% CI, 85.3-90.6). Cumulative matched-general-population 15-year survival varied from 87.7% to 89.8% using the 3 methods of different complexities. For 15-year-older patients, the cumulative matched-general-population 15-year survival was 67.4%, 59.8%, and 63.2%, respectively, using the simplest to the most comprehensive matching methods; for 15-year-younger patients, it was 96.9%, 96.1%, and 96.7%, respectively. CONCLUSIONS: Different methodologies to match general population mortality to observed patient mortality yield variable estimates of cumulative matched-general-population survival, especially in older patients. The cumulative matched-general-population survival should ideally be calculated by considering country, sex, individual patient age, and calendar year (both updated annually). This method can be easily implemented using the web-based Shiny App enclosed in this article.