OBJECTIVETo evaluate the completeness of the death registration system, so as to understand the death patterns in Xuanwei.

METHODSThe investigation on under-reported deaths was conducted in 30 villages selected with a multi-stage random sampling strategy. Participants were asked about changes of their family members (family members born or dead) during past 3 years with door to door visit. Then, death cases obtained in our investigation were matched with those from routine death registration system and under-reported rate of deaths during 2011-2013 was calculated employing capture-recapture method.

RESULTSTotal under-reported rate of deaths was 31.88%. For people aged between 0-14, 15-39, 40-69 and 60 above, under-reported rates of death were 33.35%, 34.93%, 29.10%, and 32.88%, respectively. And they were 31.72% and 32.02% for males and females, respectively. There was no significant difference shown in under-reported rates among deaths in different age groups (χ² = 7.24, P = 0.065) and genders (χ² = 0.06, P = 0.803). The under-reported rates in high-mortality, medium-mortality and low-mortality regions were 17.48%, 38.01%, and 36.22%, respectively with a significant difference (χ² = 213.25, P < 0.001). Death in local regions with mortality rate higher than 600.00/10(5), between 400.00/10⁵ and 600.00/10⁵ and lower than 400.00/10⁵ were adjusted with under-reported rates in three regions above respectively. The total adjusted morality rate in Xuanwei during 2011-2013 was 776.47/10⁵. For males and females, they were 918.73/10⁵ and 617.46/10⁵, respectively.

CONCLUSIONOverall under-reported rate of death was high in death registration system in Xuanwei. It was necessary to adjust mortality data reported with under-reported rate of death to estimate death patterns in this area.