Abstract:

Objective: To analyze the long-term trends of pulmonary tuberculosis (PTB) incidence in Liangping District, Chongqing from 2011—2025, and to compare the predictive performance of different time series models, thereby providing a scientific basis for precision prevention and control in the context of establishing a “tuberculosis-free district”. Methods: Annual and monthly reported PTB incidence data and demographic data in Liangping District from 2011 to 2025 were collected from the “Infectious Disease Surveillance” subsystem of the “China Disease Prevention and Control Information System”. Joinpoint regression was applied to analyze the long-term trend of annual reported incidence, and the annual percent change (APC), monthly percent change (MPC), and their 95% confidence intervals (CI) were calculated. Using monthly reported PTB incidence data from 2011 to 2023 as the training set, and R 4.5.2 software to automatically select optimal models, three models were constructed: seasonal autoregressive integrated moving average (SARIMA), error-trend-seasonal (ETS), and log-transformed SARIMA. The data from 2024 to 2025 were used as the test set to evaluate the predictive performance of the models. The optimal model was selected based on the root mean squared error (RMSE) and mean absolute percentage error (MAPE), and was then used to forecast the monthly reported PTB incidence for 2026—2027. Results: From 2011 to 2025, a total of 5151 PTB cases were reported in Liangping District, with an annual average reported incidence of 51.21/100000 (5151/10059411) and a monthly average reported incidence of 4.26/100000 (5151/120848683). The annual reported incidence decreased from 80.44/100000 (553/687498) in 2011 to 34.26/100000 (214/624635) in 2025. Joinpoint regression showed an overall downward trend in the annual reported incidence of PTB in Liangping District (APC=-6.14%, 95%CI: -6.85% to -5.42%, t=-17.998, P<0.001). Using R 4.5.2 software, the three optimal models were automatically established and selected from the training set: SARIMA(5,1,1)(2,0,0)_[12], ETS(A,Ad,A), and log-transformed SARIMA(0,1,1)(2,0,0)_[12]. Based on the RMSE and MAPE calculated on the test set, the SARIMA(5,1,1)(2,0,0)_[12] mode exhibited the best predictive performance, with RMSE=0.798 and MAPE=27.016%. The testing and forecasting results of this model indicated that the monthly reported PTB incidence in Liangping District from 2024 to 2027 showed a slowly decreasing trend (MPC=-1.11%, 95%CI: -1.57% to -0.66%, t=-4.916, P<0.001). Conclusion: From 2011 to 2025, the reported incidence of PTB in Liangping District showed a downward trend, indicating remarkable achievements in prevention and control. The SARIMA(5,1,1)(2,0,0)_[12] model demonstrates good short-term predictive capability for monthly PTB incidence in this area, and suggests that more proactive intervention strategies are needed in subsequent prevention and control efforts in Liangping District. This provides solid methodological support for dynamic surveillance and early warning in the creation of a “tuberculosis-free district”.

Key words: Tuberculosis, pulmonary, Models, statistical, Statistical distributions, Regression analysis, Incidence, Prediction