[Quality of retrospective survey of deaths and its evaluation]

An imitative retrospective survey (RS) of deaths was carried out to evaluate the errors and quality of the retrospective survey by using the data on Qidong Death-Cause Registry. Results showed that the agreement rates for death-cause, death-age and death-date of RS were 79.69%, 67.97% and 58.98%, respectively. The total agreement rate for the three items was 31.95%, in which 52.71% in liver cancer, 25.00%, other cancers, 21.05%, other diseases, and 26.32%, accidents. Because some of the errors were cancelled out each other, the mis-report rates for the above diseases were 4.65%, 15.63%, 14.47%, and 10.53%, respectively, and the statistical breakdown of data, age-group and death-time distributions met basically with the data on registration (P greater than 0.05). This study indicates that the errors of the data from RS are rather higher compared with data on registration, yet the data on the whole can reflect the population distribution, time trends and epidemic intensity of diseases investigated, RS might be considered as a mean of getting historical data, but its application and evaluation should be cautious.