Assessments of mortality at oldest-old ages by province in China's 2000 and 2010 censuses

This study examined the possible underestimation and age-trajectories of mortality at oldest-old ages in China’s 2000 and 2010 censuses. By linking logit-transformed conditional probabilities of dying from 13 countries with the highest data quality in the world, this study found that many Chinese provinces had underestimations of mortality at oldest-old ages when a relatively lenient criterion was applied. When a relatively strict criterion was applied, most provinces had a 30% or more underestimation in the probability of dying. We also investigated age trajectories of death rates after age 80 in these two censuses by applying the Kannisto model. Results showed that the age trajectories were distorted in most provinces after age 95. Overall, eastern-coastal provinces had higher data quality — in terms of low underestimation rates and less distorted age trajectories — whereas western China had provinces with problematic data. Females had greater rates of underestimation yet less distorted age-trajectories than males; and the 2010 census had greater rates of underestimation yet less distorted age-trajectories than the 2000 census. We conclude that appropriate adjustments with simultaneous applications of the Kannisto model are needed for direct estimates of mortality at oldest-old ages in the 2000 and 2010 censuses for China and for its provinces.