Table 6. Annual influenza-associated deaths in Singapore, Hong Kong, and United States.

Author	Country	Statistical method	Influenza-associated mortality rate/100,000 person-years
			All-cause	Underlying pneumonia and influenza deaths	Underlying circulatory and respiratory deaths
Chow et al.	Singapore	Negative binomial regression model was used to estimate mortality outcomes. The model was developed by using monthly number of deaths and monthly proportion of positive influenza test results. Linear and nonlinear time trends, 3–4 pairs of seasonality variables, monthly mean temperature and relative humidity, and monthly proportion of positive respiratory syncytial virus (RSV) test results were included as covariates in the model.	All ages: 14.8	All ages: 2.9	All ages: 11.9
			>65 y: 167.8	>65 y: 46.9	>65 y: 155.4
Wong et al. (6)	Hong Kong	Poisson regression model was used to estimate mortality outcomes. The model was developed by using weekly number of deaths and weekly proportion of positive influenza test results. Dummy variables for each year, 2 pairs of seasonality variables, weekly mean temperature and relative humidity, and weekly proportion of positive RSV test results were included as covariates in the model.	All ages: 16.4	All ages: 4.1	All ages: 12.4
			>65 y: 136.1	>65 y: 39.3	>65 y: 102.0
Thompson et al. (7)	United States	Age-specific Poisson regression models were used to estimate mortality outcomes. Each model was developed by using weekly number of deaths for the specific age group and weekly proportion of positive influenza test results. Age-specific population size, linear and nonlinear time trends, 1 pair of seasonality variables, and weekly proportion of positive RSV test results were included as covariates in each model.	All ages: 19.6	All ages: 3.1	All ages: 13.8
			>65 y: 132.5	>65 y: 22.1	>65 y: 98.3