The gap between cause-of-death statistics and Household Registration reports in Shandong, China during 2011-2013: Evaluation and adjustment for underreporting in the mortality data for 262 subcounty level populations.

Underreporting is a quality concern in mortality statistics. The purpose of this study was to assess and adjust underreporting in the population-based cause-of-death statistics. The total population (96 million) in Shandong, China was divided into 262 subcounty level populations geographically and by residence type (urban/rural). For each subpopulation, the total number of deaths during the years 2011-2013 was determined using data from the Household Registration System (HRS), and was used as a reference to assess the underreporting rate (UR) in the cause-of-death data from the Shandong Death Registration System (SDRS). It was estimated that 454,615 deaths, or 21.5% (95% CI: 21.4-21.5%) were unreported. Underreporting was more pronounced in rural (22.1%) versus urban communities (20.0%), in economically underdeveloped regions (32% versus 16% in least disadvantaged areas), and in newly included sites with no prior experience in cause-of-death reporting (24% versus 17%). Geographic variation was large with a UR at the prefectural level ranging from 11.2% to 43.7%. A stratified analysis showed that UR was higher in rural populations in high-income regions, but in middle- and low-income regions, was higher in urban communities. An adjustment factor (AF) was calculated for each of the 262 subpopulations (ranging from 0.9 to 2.5 with an average of 1.27). The total morality rate was adjusted from 6.03 to 7.67 deaths per 1000 persons. Underreporting in the SDRS varies greatly between areas and populations and is related to residence type, prior experience and local economy. Correcting underreporting at a local level is needed especially for comparative analyses across geographical areas or populations.