Background: Assessing the impact of influenza epidemics provides useful information to assess both population and healthcare system burden and can inform prevention and control measures for seasonal epidemics, such as vaccination and antivirals. Furthermore, it is an important component of pandemic preparedness.

Methods: We assessed and compared three influenza impact parameters: influenza-associated excess respiratory mortality, hospitalizations and ICU admissions, under the World Health Organization Pandemic Influenza Severity Assessment framework. We used a generalized additive model to estimate these parameters from 1998 through 2019 in Hong Kong based on historical mortality, hospitalization, ICU admission and influenza surveillance data. Intensity thresholds by influenza type were estimated using quantiles from the distribution of peak values of the parameters from 1998 through 2017 and were compared to the real-time estimates of excess parameters in 2018-2019. Influenza death and hospitalization data were used for validation.

Findings: There was good agreement between the different impact parameters after comparing the 2018-2019 data to the thresholds. The 2019 influenza A epidemic was characterized as having moderate impact overall and in all age groups, except 0-64 years for whom the excess ICU impact was high; whereas the 2018 influenza B epidemic was characterized as having very high impact overall and in all age groups.

Interpretation: The impact of influenza epidemics can vary from year to year. The PISA framework facilitates the impact assessment of seasonal influenza epidemics using different data sources and can be implemented in both real-time or at the end of seasons as policy makers and public health officials prepare for the next seasonal epidemic.