Assessing trends and predictors of tuberculosis in Taiwan

BMC Public Health

Table 3 Summary of Poisson regression models used in this study

Poisson regression model^a	RMSE
Hwalien County
$Y (t) = exp β 0 + β 1 t + β 2 t 2 + β 3 ∑ n = 1 5 sin (2 n π t ∕ 12) + β 4 ∑ n = 1 5 cos (2 n π t ∕ 12) + β 5 T mean, t - 3$	2.66	(T1)
$Y (t) = exp β 0 + β 1 t + β 2 t 2 + β 3 ∑ n = 1 5 sin (2 n π t ∕ 12) + β 4 ∑ n = 1 5 cos (2 n π t ∕ 12) + β 5 T mean, t - 3 + β 6 a g e t$	2.48	(T2)
$Y (t) = exp β 0 + β 1 t + β 2 t 2 + β 3 ∑ n = 1 5 sin (2 n π t ∕ 12) + β 4 ∑ n = 1 5 cos (2 n π t ∕ 12) + β 5 T mean, t - 3 + β 6 a g e t + β 7 m a l e t + β 8 f e m a l e t$	2.34	(T3)
Taitung County
$Y (t) = exp β 0 + β 1 t + β 2 t 2 + β 3 ∑ n = 1 5 sin (2 n π t ∕ 12) + β 4 ∑ n = 1 5 cos (2 n π t ∕ 12) + β 5 T max, t - 2$	2.64	(T4)
$Y (t) = exp β 0 + β 1 t + β 2 t 2 + β 3 ∑ n = 1 5 sin (2 n π t ∕ 12) + β 4 ∑ n = 1 5 cos (2 n π t ∕ 12) + β 5 T max, t - 2 + β 6 a g e t$	2.63	(T5)
$Y (t) = exp β 0 + β 1 t + β 2 t 2 + β 3 ∑ n = 1 5 sin (2 n π t ∕ 12) + β 4 ∑ n = 1 5 cos (2 n π t ∕ 12) + β 5 T max, t - 2 + β 6 a g e t + β 7 m a l e t + β 8 f e m a l e t$	2.63	(T6)

^a Y(t) represents expected incidence rate of tuberculosis, β ₀ stands for intercept, β ₁ stands for linear time trend, β ₂ stands for quadratic time trend, β ₃ and β ₄ stands for the seasonality, β ₅ stands for monthly maximum and mean temperature (°C), t - 3 in the subscript represents the 3-month lag time and t - 2 in the subscript represents the 2-month lag time, β ₆ stands for the effect for age, β ₇ stands for the effect for male, and β ₈ stands for the effect for female.
RMSE = Root mean squared error.

ISSN: 1471-2458