Cause of death statistical information is important for health planning. The information enables policy planners to understand the medical nature of fatal health problems of the country. Priority can be set based on the information, which can also be baseline for evaluation of various health programs. However, in Thailand, the standard of death registry is still poor. My deaths took place outside the hospital and registered by non-medical personnel. The information is therefore not accurate enough for planning. In 2019, in order to solve this problem, a research team from the International Health Policy Program (IHPP) did a survey on causes of deaths using so-called ‘verbal autopsy’ method. After the data were analyzed, IHPP invited relevant experts to review the report. Tuberculosis (TB) was reported to be the 10th cause of death in male and 17th cause of death in female population. These figures indicated that TB is still a very important health problem. However, the above findings are in contrast with reports from the National TB registry, which show that TB notification rate is rapidly and steadily declining. To solve this problem of inconsistency, our research team carried out the current study with the following objectives: 1. The primary objective is to review the records of the verbal autopsy of the so-called TB deaths in order to check the accuracy of the reports 2. The subsidiary objectives include to assess clinical cause of these deaths Our research team were kindly supported by the IHPP team who provided the verbal autopsy data Our team consist of the following experts: (1) Senior Medical Epidemiologist who conceptualized and designed the study. (2) Senior Infectious Disease Expert (3) National Senior Expert on Tuberculosis (4) Statistician Experts No (2) and No (3) initially independently reviewed the data, write their comments and finally scored how much they agreed with the verbal autopsy reports. The score could be one of these 5 levels 1 Not agree 2 Somewhat disagree 3 Indecisive 4 Somewhat agree
5 Agree Out of 198 records provided by IHPP, the 63 duplicated records were removed, leaving 135 unique records. Frequency of the 1, 2, 3, 4 and 5 scores given by the National TB expert were 10, 12, 94, 11 and 8 Frequency of the 1, 2, 3, 4 and 5 scores given by the Infectious Disease expert were 14, 7, 21, 21 and 72. Weighted kappa statistic = 0.226, indicating fair agreement between the two experts. There were 3 types of weighting scheme were applied. The unweighted, linear weighted and quadratic weighted kappa statistic were 0.86, 0.141 and 0.226 respectively, which can be concluded as relatively low. In conclusion, the revision of history of cause died from TB in the verbal autopsy result got inconsistent rate of likelihood from two independent experts. The findings suggest the need to improve such data in the future.