Estimating the incidence of heart failure: Insights from an illness-death model using statutory health insurance data from 70 million people in Germany

Introduction: Heart Failure (HF) is a considerable public health issue that affects more than 64 million people worldwide, with a prevalence of 1% to 3% in Western industrialized countries. However, reliable data on the incidence of HF are lacking, making it difficult to assess. The aim of our study was to estimate the age- and sex-specific incidence of HF in Germany using the illness-death model (IDM) and an associated partial differential equation (PDE), based on aggregated prevalence and mortality data. Methods: Prevalence data for 2009–2017 were used from the Central Institute for Statutory Health Insurance (Zi). Data on HF mortality were taken from a comparable Norwegian population from 2013. A PDE using a bootstrapping approach was used to calculate HF incidence. The estimated incidence was the median of 5,000 samples, complemented by 95% bootstrapping confidence intervals (CIs). We ran a ± 15% calibration sensitivity analysis by uniformly scaling Norwegian Mortality rate ratios (MRRs) to test transferability to the German population. The incidence was then re-estimated with the full 5,000-replicate bootstrap. Medians with 95% empirical intervals were reported for each scenario. Results: In all age groups, men had a higher incidence than women. The highest incidences were estimated in age groups over 90 years old, with approximately 88/1,000 person-years (py) (95% CI: 73–102) for men and 45/1,000 py (95% CI: 29–61) for women. A sharp increase in incidence rates could be seen in men from the age of 60 years and in women from the age of 65 years. The slope of the incidence was less pronounced among women aged above 60 years compared to men, without impacting the overall trend of increased incidence in higher age groups for women. Uniform ±15% scaling shifted incidence levels, but preserved the age-specific shape and male–female ordering. The effects were small to moderate in age groups from 50 to 90, and negligible in age groups ≤45, with the largest absolute shifts observed at very old ages. Conclusion: The IDM and an associated PDE were used to estimate a rising incidence of HF with age for both men and women. These findings underscore the importance of age- and sex-specific approaches in HF prevention and management strategies, particularly in older populations where the burden of HF is most pronounced. The method complements existing estimation techniques while circumventing the necessity of costly follow-up studies.