Objective To evaluate the quality of cause-of-death data in Hubei province from 2015 to 2019 by using three international assessment systems, compares their similarities and distinctions and explore their supplementary role to the national quality control framework.

Methods Between 2015 and 2019, the cause-of-death data from 22 national disease surveillance areas in Hubei was analyzed to analyze the main types and proportions of unsuitable underlying causes of death (UCODs) and garbage codes. This analysis was based on the unsuitable UCOD list, the Analysis of Causes of National Deaths for Action (ANACONDA) tool, and the Global Burden of Disease (GBD) garbage code list. Cochran-Armitage test and Joinpoint regression analysis were used to evaluate the temporal trends and calculate the annual percent change (APC). Additionally, Spearman correlation analysis and Venn diagrams were used to evaluate the consistency and complementarity of the three assessment systems.

Results From 2015 to 2019, there were significant increases in the proportions of unsuitable UCODs (14.30%), ANACONDA garbage codes (9.22%), and GBD garbage codes (12.40%) (Z=24.074, 10.154, 13.034, all P < 0.001). Non-specific underlying causes of death accounted for a relatively large proportion and the main low-quality codes identified by the three systems all exhibited high similar annual increases of 0.23%, 0.23%, and 0.24%, respectively. The proportions of unsuitable UCOD code and garbage code were high in children aged < 5 years and adults aged ≥85 years. All three systems demonstrated significant consistency in identifying major low-quality codes (r: 0.82-1.00). Furthermore, Venn diagram analysis revealed that all the three systems collectively identified 247 problematic codes (e.g., I64, I69.4, I46.1, R96.1), with each system also identifying a specific number of unique codes.

Conclusion The quality of cause-of-death data needs to be improved in Hubei, the problem of non-specific underlying causes of death is obvious. The three assessment systems demonstrate considerable consistency in identifying critical low-quality garbage codes, despite variations in the overall coverage. Therefore, it is suggested to strengthen the training on high-frequency non-specific garbage codes, include priority populations and home deaths in routine verification, establish a prioritized intervention list of major problematic codes based on the consensus of multi-system assessment results. Additionally, exploring a multidimensional quality control evaluation framework suitable for China is essential to improve the quality of cause-of-death data.