A retrospective analysis of tuberculosis notification trends and spatial distribution in South Africa, 2005-2012.

BACKGROUND: Tuberculosis (TB) remains a major public health challenge in South Africa. Despite substantial policy and programme changes since 2012, including the GeneXpert rollout, antiretroviral therapy expansion, Universal Test and Treat, and tuberculosis preventive therapy, historical surveillance data remain essential for establishing baselines against which subsequent progress can be measured. This retrospective study analysed TB notification data from 2005-2012 to characterise demographic disparities and district- level spatial patterns preceding these major interventions. METHODS: A retrospective analysis of 3,474,320 TB cases recorded in the Electronic TB Register (2005-2012) was conducted. The analysis was limited to the period for which complete national notification data were available. Annual notification rates were calculated by age, sex, province, and district. Trends were assessed using chi-square tests and linear regression. Poisson regression was used to assess the variation across provinces. Age-and sex-standardised rates were derived using the 2011 census population. Spatial autocorrelation and clustering were evaluated using Global and Local Moran's I and Getis-Ord Gi* statistics. A spatial lag regression model assessed associations with district- level HIV prevalence and poverty. RESULTS: The mean annual TB notification rate was 885. 2 per 100,000 population at 95% CI [772.0, 945. 3], with no statistically significant temporal trend over the eight years. Males had higher notification rates than females (974.6 vs 800.2 per 100,000), with the highest burden among adults aged 25-44 years. KwaZulu-Natal (1,309.3 per 100,000) and North West (1,113.2 per 100,000) provinces recorded the highest rates. There was a statistically significant variation in TB notification rates across provinces (likelihood-ratio (omnibus) test (LR χ² = 442,769.6; df = 8; p < 0.001). Significant spatial autocorrelation was observed (Moran's I = 0.556, p < 0.001), with high-burden clusters concentrated in eastern districts, predominantly in KwaZulu-Natal, Eastern Cape, and Free State. Spatial regression demonstrated strong spatial dependence (ρ = 0.53-0.64, p < 0.001) but no statistically significant associations with HIV prevalence or poverty at the district level. CONCLUSIONS: This retrospective analysis reveals persistent demographic and geographic disparities in TB burden across South Africa during 2005-2012. The identification of high- burden districts and population subgroups provides a critical historical baseline for evaluating the impact of subsequent TB and HIV interventions. Districts identified as hotspots in this era warrant renewed attention to assess whether they have improved or whether entrenched vulnerabilities persist. By establishing a detailed baseline from the pre-GeneXpert, pre-Universal Test and Treat, and pre tuberculosis preventive therapy era, this study enables policymakers and programme managers to evaluate where progress has been made and where geographic and demographic disparities have proven resistant to intervention. The findings support evidence-based targeting of resources to populations and districts with a historically entrenched burden. PUBLIC HEALTH IMPACT: Although the data presented here are from 2005-2012, they provide a vital historical baseline for interpreting subsequent TB trends in South Africa. By enabling an assessment of progress and persistent gaps, this study offers actionable evidence to guide more efficient, equity-focused TB control strategies, particularly in high-burden and underserved settings. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12982-026-02685-5.