The impact of suicide news reporting (charcoal burning and non-charcoal burning respectively) in the previous day on suicide incidence in the following day in Taiwan, 1998–2002.

*Note: aPoisson generalized linear autoregressive model controlling for secular trend;bPoisson generalized linear autoregressive model controlling for secular trend, secular trend of non-charcoal burning suicide count and days of the week;cPoisson generalized linear autoregressive model controlling for secular trend, secular trend of charcoal burning suicide count and days of the week.*Note: The Poisson estimates can be transformed to proportions by exponentiation of the estimate and minus 1. For example, in the unadjusted models, the Poisson estimates 0.14 and −0.00 can be transformed to exp (0.14) −1 ≈15% and exp (−0.00) −1≈ 0% respectively. These estimates mean that before adjusting for potential confounders, each news article on charcoal burning suicide was associated with a 15% [95% CI = 0.08, 0.22] increase in charcoal burning suicide on the following day; whereas news reporting of non-charcoal burning suicide was not associated with an increase in non-charcoal burning suicide. After adjusting for potential confounders, for each reported news article on charcoal burning suicide was associated with a 16% (exp (0.15) −1 ≈ 16%) [95%CI = 0.07, 0.26] increase in charcoal burning suicide on the following day.