Discussion on data cleaning of vital registration system in Xuanwei, Yunnan, 2014-2016

Objective To investigate the problems existed in the vital registration system (VRS) and improve capacity building for primary public health staff by cleaning the database of VRS from 2014 to 2016 in Xuanwei,an area with high lung cancer mortality in Yunnan province of China. Methods The new VRS was used to report death cause surveillance data. The duplicate records,extreme values,abnormal values,missing values and logical consistency were checked,and proper deletion or modification were done. Frequency statistics and cause analysis were conducted for abnormal data. The annual constituent ratio of diseases (death causes) was calculated to evaluate the data quality after database cleaning. Results After data cleaning,there were 24 704 valid death records in the database;the deletion rate was 1.10%. The inconsistent rate of death case age obtained from identification card number,birth of date and the self-eported was less than 0.58%. Up to 68.82% of the death records were reported more than 1 month later,and 9% of the death records were reported 1 year later or even 3 years later,in which 454 were death records of lung cancers,and 78.63% of 357 lung cancer deaths were from Xuanwei. There were 131 duplicate death records,which were mainly reported during January-March of following year. The level of medical institutions making the second diagnosis was obviously lower than the level of those making the first diagnosis. There were around 2% of death records without identified causes. Conclusion The quality of death surveillance data has been greatly improved after the application of new VRS. However,the timeliness of reporting needs to be further strengthened. More strict review procedure is needed for the after-event death reporting in the following year. Specific functional upgrade of the system is needed to improve surveillance quality.