Incomplete tuberculosis reporting and registration to the surveillance system in southwestern China of Yunnan Province: an inventory survey

The real-world tuberculosis (TB) surveillance data was generally incomplete due to underreporting and underdiagnosis. The inventory study aimed to assess and quantify the incompletion of surveillance systems in southwestern China. The inventory study was conducted at randomly selected health facilities (HF) by multi-stage stratified cluster sampling. The participants were included in the period between August of 2020 in province-level and prefecture-level HF, and in the period between June to December of 2020 in other categories of HF respectively. The clinical committee confirmed medical records were matched to the National Notifiable Disease Reporting System (NNDRS) and the Tuberculosis Information Management System (TBIMS) to define the report and register status. The underreporting and under-register rates were evaluated based on the matched data, and factors associated with underreport and under-register were assessed by the 2-level logistic multilevel model (MLM). We enrolled 7,749 confirmed TB cases in the analysis. The province representative overall underreport rate to NNDRS was 1.6% (95% confidence interval, 95% CI, 1.3 - 1.9), and the overall under-register rate to TBIMS was 9.6% (95% CI, 8.9-10.3). The various underreport and under-register rates were displayed in different stratifications of background TB disease burden, HF level, HF category, and data source of the medical record in HF among prefectures of the province. The intraclass correlation coefficient (ICC) was 0.57 for the underreporting null MLM, indicating the facility-level cluster effect contributes a great share of variation in total variance. The two-level logistic MLM showed the data source of medical records in HF, diagnostic category of TB, and type of TB were associated with underreporting by adjusting other factors (p