Epidemiological trajectories and predictive modeling of latent tuberculosis prevalence in South Asia: an in-depth global burden of disease analysis (1990-2021) with advanced forecasting to 2050.

Background: Latent tuberculosis infection (LTBI) remains a critical reservoir for the global tuberculosis (TB) epidemic, particularly in South Asia, where high population density and socioeconomic disparities amplify its burden. However, LTBI epidemiological trajectories and long-term forecasts have received limited attention. This study characterizes LTBI prevalence trends in South Asia from 1990 to 2021 and forecasts prevalence to 2050 to inform targeted interventions. Methods: We analysed age-standardised LTBI prevalence rates (ASPR per 100 000 population) from the Global Burden of Disease Study 2021 for five high-burden South Asian countries (Bangladesh, Bhutan, India, Nepal, Pakistan) from 1990 to 2021, stratified by sex. Joinpoint regression identified temporal inflection points and calculated average annual percent change (AAPC) and segment-specific annual percent changes (APC) with 95% confidence intervals. Auto-ARIMA models, selected by minimised Akaike Information Criterion with first differencing for stationarity, generated projections up to 2050 with 95% prediction intervals (PI) based on 1000 bootstrap simulations. Negative point forecasts were interpreted as signals of potential near-elimination, with lower bounds truncated at zero. Results: From 1990 to 2021, Bangladesh (AAPC -3.55%, 95% CI -3.61 to -3.49), Bhutan (-3.38%, -3.45 to -3.32), Nepal (-2.17%, -2.31 to -2.06), and Pakistan (-2.53%, -2.64 to -2.42) achieved steep declines. India showed only modest reduction (AAPC -0.76%, -0.93 to -0.59) with a significant resurgence during 2015-2019 (APC + 1.20%, +1.05 to +1.39). Sex-specific patterns revealed steeper declines among females in Nepal, Bangladesh, and Bhutan, but a more pronounced resurgence among Indian females (+1.32%) than males (+1.09%). Auto-ARIMA projections to 2050 indicate continued rapid declines in Bangladesh, Bhutan, Nepal, and Pakistan, with point estimates falling below 5000 per 100 000 (many strata reaching zero) by mid-century. India's ASPR remains essentially stable at approximately 26,300-26,400 per 100 000 through 2050. Regional aggregation masks this heterogeneity, projecting only marginal decline (ASPR 23,283 in 2021 to 23,085 in 2050; 95% PI 12,554-33,616). Wide prediction intervals beyond 2035 highlight substantial long-horizon uncertainty. Conclusions: Four of five high-burden South Asian countries are on trajectories compatible with very low LTBI prevalence or pre-elimination by 2050, driven by strong primary-healthcare integration and community engagement. India's persistently high burden, explained by massive scale, urban transmission hotspots, internal migration, and a large multidrug-resistant reservoir, will continue to dominate regional and global metrics unless transformative interventions are implemented. The extraordinarily wide 2050 prediction intervals underscore that favourable outcomes are plausible but not inevitable; achieving them will require sustained political commitment, equitable financing, and accelerated deployment of new tools (shorter preventive regimens, TB vaccines). Targeted, gender-responsive, and adaptive strategies are essential for South Asia to meet, and in several countries exceed, WHO End TB Strategy milestones.