Nuntaporn Klinjun, Apiradee Lim, and Kanitta Bundhamcharoen

pp. 711 - 718

Abstract

This study aimed to develop models based on Verbal Autopsy (VA) data and to estimate correct number of deaths from
external causes in Thailand from 1996 to 2009. Logistic regression was used to create models of the three external causes of
death classified by province, gender-age group and Vital registration (VR) cause-location group. Receiver operating characteristic (ROC) curves were used to validate the models by matching the number of reported deaths to the number of deaths
predicted by the models. The models provided accurate prediction results, with false positive error rates 1.6%, 2.0% and 0.6%
and sensitivities 73.8%, 46.3% and 62.0%, respectively. The results reveal that under-reporting of external causes of death
increased over the 14-year period. Our statistical method confirms that the Thai 2005 VA data can be used to estimate external
causes of death from VR report in Thailand to allow for the under-reporting rate.