Years of life lost associated with deaths of patients with laboratory-confirmed 2009 Pandemic (H1N1) virus infection in Hong Kong in 2009

Abstract

BACKGROUND: While the overall mortality impact of the first wave of the 2009 pandemic appeared to be lower than in most annual seasonal epidemics, mortality among children was thought to be higher. Years of life lost (YLL) has therefore been proposed as an alternative measure of the impact of influenza epidemics and pandemics. However, reported estimates of YLL have typically ignored the presence of underlying chronic conditions or health risk factors in most individuals who died. METHODS: Data were available on patients in Hong Kong with laboratory-confirmed 2009 influenza A (H1N1) virus infection who died between May 2009 and March 2010. The YLL for each confirmed death was estimated based on the remaining life expectancy at the age at death, which was based on the 2009 Hong Kong life tables for males and females. For each confirmed death with documented underlying conditions or a health-risk behavior, such as smoking, the YLL was corrected with hazard-based modifications to the life tables to account for reduced life expectancy in those individuals. RESULTS: There were 72 deaths among patients with laboratory-confirmed 2009 influenza A(H1N1) virus infection. Patient ages ranged from 0-95 years of age, and 56% had underlying health conditions or risk behavior that reduced their life expectancy. We estimated that the 2009 pandemic was associated with 1560 YLL (95% confidence interval, 1410-1700 YLL), adjusted for chronic conditions and health risk behaviors. This estimate of YLL was 25% less than the naive estimate of YLL ignoring chronic conditions and health risk behavior and using only the sex-specific life expectancy at age of death. On average, the life expectancy was reduced by 50% for patients with chronic conditions or health risk behaviors. CONCLUSION: Our estimates of the YLL corrected for chronic conditions and health-risk behaviors provide a framework for similar calculations elsewhere. Our analysis demonstrates the potential scale of bias in YLL estimation if chronic conditions are ignored.