Township-level spatial heterogeneity and associated factors of diagnostic delay among pulmonary tuberculosis patients in Central China

Pulmonary tuberculosis (PTB) remains a major public health burden, and delayed diagnosis may prolong transmission before treatment. Township-level evidence is limited, particularly on whether patient delay and health system delay show different spatial patterns. This study analyzed PTB diagnostic delays in Yichang City, identified high-delay townships, and examined individual-level correlates of long delays. We used routinely reported PTB surveillance records from the Tuberculosis Information Management System. Notified PTB cases registered in Yichang City from January 1, 2015 to December 31, 2019 were included, representing a pre-COVID-19 baseline period. Long patient delay (LPD) and long health system delay (LHD) were defined as delays exceeding 14 days, and long diagnostic delay (LDD) as total delay exceeding 28 days. Modified Poisson regression using generalized estimating equations with township-level clustering was used to estimate adjusted prevalence ratios for individual-level correlates, while township-level spatial analyses included Global Moran’s I, Getis-Ord Gi* hot spot analysis, Local Indicators of Spatial Association, and global bivariate Moran’s I. A total of 16,099 PTB patients were included. The median total diagnostic delay was 31 days (IQR: 15–47), with patient delay accounting for the larger recorded component; the median patient delay was 24 days (IQR: 9–38) and the median health system delay was 2 days (IQR: 0–7). Patient delay exceeded health system delay in 80.3% of patients. Older age groups, minority nationality, and household/unemployed status were associated with a higher prevalence of LPD, whereas rural residence was associated with a lower prevalence of LHD. Significant positive spatial autocorrelation was observed for LPD, LHD, and LDD at the township level ( Global Moran’s I = 0.318–0.495, all P < 0.001). LPD and LHD showed a significant negative cross-spatial association ( bivariate Moran’s I = − 0.192, P = 0.020). After false discovery rate correction, LPD and LDD hot spots were mainly concentrated in southern Yichang, particularly in the Yidu–Changyang area, whereas LHD hot spots were mainly located in Dangyang and the central urban districts. Within the diagnostic pathway recorded in TBIMS, patient delay accounted for the larger recorded component of diagnostic delay, and patient delay and health system delay showed distinct township-level spatial patterns.