Tuberculosis surveillance indicators and chest radiograph phenotype clustering in southwestern Saudi Arabia,...

Subnational tuberculosis surveillance data in low-incidence settings are often reported descriptively, but can also support program monitoring when analyzed as performance indicators. We evaluated longitudinal trends in bacteriological confirmation and radiology-based indicators and characterized clustering of chest radiograph findings at diagnosis.

We performed a retrospective observational study using routinely collected tuberculosis surveillance data from Aseer Central Hospital (1 January 2010-31 December 2020). Patients were included if objective evidence of tuberculosis was documented by culture-confirmed Mycobacterium tuberculosis with or without sputum smear microscopy for acid-fast bacilli and radiological findings compatible with active disease. Categorical variables were summarized as frequencies and compared using Fisher's exact test; continuous variables as medians (interquartile ranges) compared using the Wilcoxon rank-sum test. Temporal trends in yearly proportions were modeled using log-linear Poisson regression with robust standard errors and a log-denominator offset; annual percentage change and 95% confidence intervals were derived using the delta method. Sensitivity to inclusion of the COVID-19 year was assessed by comparing 2010-2019 versus 2010-2020 models. Co-occurrence of radiological findings was quantified using phi coefficients with Chi-square testing.

Of 510 records reviewed, 422 patients met inclusion criteria; 386 (91.5%) had pulmonary tuberculosis and 36 (8.5%) had extrapulmonary tuberculosis. Among pulmonary cases, the proportion with abnormal chest radiographs increased from 63.6% (2010) to 93.8% (2020) (annual percentage change 2.2%; 95% confidence interval 0.7% to 3.8%). Smear positivity among patients with available smear results increased over time (annual percentage change 3.8%; 95% confidence interval 1.5% to 6.0%), while no meaningful trend was observed for advanced pulmonary tuberculosis (annual percentage change 1.5%; 95% confidence interval -3.5% to 6.4%). Trend estimates were stable when including 2020. Radiographic findings demonstrated clustering; strong co-occurrence was observed between nodules and cavitation (phi 0.46), consolidation and cavitation (phi 0.39), and nodules and consolidation (phi 0.34).

Smear positivity and abnormal chest radiographs increased over time, while advanced pulmonary tuberculosis remained stable, showing markers and severe disease may not align. Trends were unchanged when 2020 was included. Radiographic co-occurrence patterns support structured chest radiograph reporting, training, and routine program audit for surveillance.