Validity and completeness of death reporting and registration in a rural district of Vietnam

Aims: Assessment was made of the validity of mortality estimates based on data collected during 1999 - 2000 by quarterly follow-up visits and compared with other methods (re-census, communal death registration, and neighbourhood survey). Methods: This study was carried out within a longitudinal epidemiological laboratory in Bavi District, Vietnam (called FilaBavi), covering a sample of 11,089 households with 51,024 inhabitants. Deaths within FilaBavi during 1999 - 2000 were collected by four methods and compared: quarterly household follow-ups, the re-census carried out in 2001, the Commune Population Registration System (CPRS), and a neighbourhood survey. Results: Within these four methods, a total of 471 deaths were detected in the FilaBavi sample. Quarterly household follow-ups detected 470 deaths (99.8%). The re-census missed 19 deaths, of which eight were infants, and two-thirds of the missed deaths fell in 1999. The CPRS missed 89 cases (19%), the majority being infant and elderly deaths. The neighbourhood survey over-reported deaths. Conclusions: Quarterly follow-ups were the best method for death registration. The re-census approach was less complete, with problems of recall bias. The completeness and quality of death registration by CPRS was low, especially for infant and elderly mortality.