Improving the monitoring of suicide incidence by estimating the probability of news reporting.

A timely estimate of suicide incidence is important for surveillance and monitoring but always difficult if not possible. The delay in reporting suicide cases between the time of occurrence of the deaths and them being registered is unavoidable. There is at least one year if not more of the delay time in the latest WHO website reporting the suicide statistics of most countries. Based on the daily newspaper reporting on suicide incidence, this study proposes a method to estimate the unknown incidence in a timely manner. It is shown that demographic characteristics such as age, suicide methods, and the districts of the deceased were significantly associated with the probability of the newspapers reporting the suicides. By incorporating this information on the daily suicide news reports into estimating the probability of the newspapers reporting the suicides, the daily number of suicide cases can be estimated. The proposed method is applied to estimate the number of suicides in Hong Kong where there is the Coroner's Court to investigate into suicide deaths, but it takes at least six months to deliver a verdict. The present method can generate timely and accurate estimations on the daily count of suicide deaths with only a one day lag. In a threefold nested cross-validation, the proposed approach has achieved an average RMSE of 1.38, MAE of 1.10, and R of 0.24. It can also serve as a surveillance system in providing estimations of temporal clusters of suicides with certain characteristics timelessly and accurately.