Impact of Drug-Resistant Tuberculosis Prevention and Control Strategies in Southeast China: Evidence from a Longitudinal Study on Disease Epidemic Trends and Predictions Across 15 Years

Introduction

Tuberculosis (TB) is a major infectious disease worldwide.¹ According to the World Health Organization estimates (WHO), 10·7 million people will develop TB by 2024.² Recent global studies have shown that mycobacterial infections cause heavy health and economic burdens among working-age adults.³ The overall prevalence has declined since 1990; however, large regional disparities remain. Low- and middle-income countries bear most of the disease burden.⁴ Forecasts also indicate that the TB burden among older adults may increase again by 2045. These findings highlight the need for sustained and well-targeted TB control.⁵ Despite substantial global progress in TB control, the persistent prevalence of drug-resistant TB (DR-TB), particularly rifampicin-resistant TB and multidrug-resistant TB (MDR-TB), has emerged as one of the most severe challenges in global TB prevention and control efforts.⁶

The WHO predicts that by 2024, more than half of the global population will be affected by multidrug-resistant and rifampicin-resistant TB (MDR/RR-TB).² In 2023, approximately 400,000 global cases of MDR/RR-TB were observed; however, only approximately 175,650 people were diagnosed and initiated treatment, with a global treatment success rate for DR-TB standing at 63%.⁷ MDR/RR-TB accounts for only 5% to 10% of all TB cases;⁸ however, it requires longer-duration, more expensive treatment regimens with greater side effects.⁹ Compared to patients with drug-susceptible TB, those with DR-TB have a lower chance of cure. Furthermore, the treatment cost often accounts for over 50% of the TB budget of a country, imposing a severe burden on public health systems and emerging as a major barrier to achieving the “End TB Strategy” of the WHO.⁸

China remains a high-burden country for MDR/RR-TB.¹⁰ The number of MDR/RR-TB cases in China accounts for approximately 7·1% of the global total, ranking second in the world, only after India.² Notably, 81·07% of patients diagnosed with MDR-TB in China were detected and treated, and only approximately 67% of those receiving treatment completed the course or were cured.¹¹ Recently, with the popularization of molecular diagnostic methods, the detection rate of DR-TB has gradually increased.¹² Significant regional variations in the trends existed,¹³ particularly in economically developed regions with high population mobility.¹⁴ Discrepancies in laboratory capacity and screening coverage lead to a gap between patient detection and actual disease burden.¹⁵

Zhejiang Province is a densely populated and economically dynamic coastal region with a high proportion of floating populations, posing prominent challenges to the prevention and control of TB, particularly DR-TB.¹⁶ Supported by the China Global Fund TB project, Zhejiang Province initiated the prevention and control of drug-resistant TB in 2008. Five prefectures joined the Global Fund Project in 2009. Drug-resistance screening in patients at high-risk for TB, such as the retreated and treatment failure groups, was implemented, and standardized diagnosis and treatment were provided for the identified patients with MDR/RR-TB. In 2013, two other cities initiated the China Global Fund TB project. After the conclusion of the China Global Fund TB project in 2013, Zhejiang Province secured funds to continue and expand the prevention and control of MDR/RR-TB to all 11 prefecture-level cities across the province. After completion of the Global Fund project, the progress of MDR/RR TB control in Zhejiang Province was divided into three phases. Phase 1 (2014–2016): MDR/RR-TB control strategies were expanded to all 11 prefecture-level cities. Drug resistance screening was extended to high-risk patients for bacteria-positive TB province-wide, and diagnostic capacity was enhanced by combining conventional drug susceptibility testing with an increasing number of patients with TB who underwent molecular testing in prefectures equipped with GeneXpert to detect drug resistance faster. Financial subsidies for MDR/RR-TB treatment were sustained, maintaining out-of-pocket (OOP) expenditures of patients below 10%. In Phase 2 (2017–2020), the screening strategy was further broadened to include all bacteriologically confirmed patients with TB, and molecular diagnostic testing was fully implemented in all prefectures across the province. During this period, additional financial support was provided by the China-Gates Foundation, accompanied by an increasing reimbursement rate to 70% of medical insurance policies. Since 2021 (Phase 3), the MDR/RR-TB screening target population in Zhejiang Province has expanded to include all presumptive patients with TB. Molecular testing is routinely performed in all prefectures with continued financial support for patient treatment. Additionally, the government bought new second-line drugs to support treatment. Access to new second-line anti-TB drugs¹⁷ and short-course treatment regimens has been strengthened to improve the management and outcomes of MDR/RR-TB (Figure 1).

Figure 1 Timeline of policy measures for MDR/RR-TB detection and care in Zhejiang Province (DST, drug susceptibility testing; OOP, the proportion of out-of-pocket. *High-risk populations included: 1. Chronic smear-positive patients or patients with treatment failure; 2. Smear-positive tuberculosis patients who have been in close contact with MDR-TB patients; 3. Smear-positive patients on retreatment; 4. Newly diagnosed patients with sputum smear-positive results at the end of the second or third month of treatment.).

Using data on MDR/RR-TB in Zhejiang Province from 2015 to 2023, this study aimed to describe the changes in disease burden and epidemic trends in recent years. We compared the estimated incidence rate with the number of cases reported in routine surveillance and identified key turning points. Additionally, we reviewed differences across regions and population groups to better understand the heterogeneity of MDR/RR-TB transmission and detection. Through these analyses, this study intended to evaluate the impact of existing prevention and control measures. We aimed to provide evidence that may help guide future planning and resource allocation for MDR/RR TB control in Zhejiang Province and other regions where the need to initiate similar work is required.

Methods

Data Sources

The case information of patients with MDR/RR-TB was extracted from the Tuberculosis Information Management System (TBIMS) between January 1, 2015, and December 31, 2023. All data were fully de-identified before the analysis to protect patient privacy. The first source was a patient with MDR/RR-TB records from TBIMS. We excluded duplicates, nontuberculous mycobacteria, and records with missing data or those not meeting the MDR/RR-TB criteria. Furthermore, 4,324 patients were included in this study. The second source was the drug resistance screening records in TBIMS. Records that were already present in their first sources were removed. Applying the same exclusion criteria, an additional 626 patients with MDR/RR-TB were included. This arm included patients diagnosed by screening but not treated because of death, moving out of the province, or refusal to undergo treatment. Finally, 4, 950 patients with MDR/RR-TB were included in the study (Figure 2).

Figure 2 Study Flow Diagram.

Key Indicators and Calculation

1. Notification rate of MDR/RR-TB: (number of notification cases/total population)×100,000. Notification data were obtained using TBIMS.
2. Estimated cases: The estimated cases of MDR/RR-TB were calculated using a WHO-standard estimation model:¹⁸

Parameters:

Rifampicin resistance rate among new TB cases (p_n): percentage of MDR/RR-TB among all new patients with TB. Derived from TBIMS.;

Rifampicin resistance rate among previously treated TB cases (p_r): percentage of all retreated patients with TB who are MDR/RR-TB. Derived from TBIMS.

Transition probability from new to retreatment (f): Proportion of drug-susceptible patients with TB in the previous year whose treatment outcomes were “failure, loss to follow-up, or not evaluated.” This parameter was calculated annually using patient-level surveillance data from the TBIMS in Zhejiang Province, China.

Relapse probability (r): The number of relapsed patients with TB in the current year divided by the sum of new and relapsed patients This parameter was derived annually from TBIMS surveillance records for each study year.

Relative risk of rifampicin resistance in relapsed patients compared to new patients (ρ): Standardized across countries, based on literature review and existing surveys (approximately 3.0). According to WHO methodological guidance, this parameter represents the relative risk of rifampicin resistance in relapsed TB patients compared with new TB patients. As this parameter was derived from the long-term monitoring results of drug resistance rates of initial and relapsed TB patients across the province by TBMIS, this parameter was used in model calculation as it reflected the local epidemiological context. Incident patients with TB (I): number of new and relapsed patients obtained from TBIMS.

3. Patient detection rate: number of notified MDR/RR-TB cases divided by estimated cases. This reflected the proportion of cases captured annually by the disease surveillance system.

Statistical Methods

Statistical analyses were performed using R version 4.3.1 software. Given that this study was based on continuous and traceable surveillance data, the WHO standard estimation model was applied to estimate the annual MDR/RR-TB cases.¹⁸ This model is preferred over dynamic models because the data are stable and are not subject to epidemic disruptions. The notification rate was calculated as the reported cases divided by the total population per 100,000, and the estimated cases were computed using the WHO Health Organization formula for the burden of DR-TB.

Joinpoint regression was applied to assess trends from 2015 to 2023, estimating the annual percentage change (APC) and average annual percentage change (AAPC).¹⁹ For long-term prediction, an Autoregressive Integrated Moving Average (ARIMA) model was fitted using WHO-estimated cases and population data from 2015 to 2023 to project incidence rates from 2024 to 2030. Because the 2023 estimated incidence rate was below one per 100,000 individuals, a logarithmic transformation was applied to stabilize the model and improve prediction accuracy.²⁰ The model parameters, including AIC and residual diagnostics, were selected based on the best-fit criteria. Statistical significance was set at p<0·050 significant.

Results

Baseline Characteristics

In total, 4,950 MDR/RR-TB cases reported in Zhejiang Province between 2015 and 2023 were included (Figure 2). Of these, 3,583 (72·38%) were males, and 1,367 (27·62%) were females. The age ranged from 10 to 97 years (mean:46·85±18·13 years). Urban residents (51·03%) and non-migrant populations (75·05%) were more commonly represented. Newly treated patients accounted for 59·62%; most received treatment (75·25%), and molecular testing was predominantly performed using GeneXpert MTB/RIF (70.63%). Occupational distribution varied, with farmers (40·14%) accounting for the largest proportion (Table 1).

Table 1 Analysis of Baseline Characteristics of MDR/RR-TB Cases, 2015–2023

From 2015 to 2023, the notification rate was significantly higher for males than for females (Figure 3a). Most patients are aged 21–60 years, which carries the main burden. Fewer cases were observed in the youngest (<20) and oldest (>80) age groups. The number of cases peaked in 2018; however, the proportion of the age groups remained stable over time. One exception was the 61–80 years age group, which showed a significant upward trend (p<0·050); the other age groups showed no significant trends (Figure 3b).

Figure 3 (a). Distribution of MDR/RR-TB Cases by Age Group and Gender, 2015–2023 (Orange: Male; Green: Female); (b). Trend in Proportion of MDR/RR-TB Cases by Age Group, 2015–2023; (c). Trend in Proportion of MDR/RR-TB Cases by Registered Residence, 2015–2023; (d). Heatmap of MDR/RR-TB Cases Counts by Region and Migrant Status, 2015–2023.

Regarding the household registration status, non-migrant populations consistently accounted for the majority, whereas the proportion of migrants was smaller. (Figure 3c) Analysis of the regional distribution heatmap revealed that non-migrant population cases were mainly concentrated in prefecture-level cities, such as Wenzhou, Ningbo, and Hangzhou. The overall number of migrant population cases was small; however, they showed a certain degree of concentration in cities. (Figure 3d).

Case Number Estimation and Trend Analysis

The MDR/RR TB in Zhejiang Province showed a downward trend from 2015 to 2023 (Table 2 and Figure 4). Notification cases increased in 2015, peaked in 2018, and declined thereafter. In contrast, the estimated number of cases decreased steadily from 2015 to 2023. The gap between estimated and notified cases narrowed year by year, with the patient detection rate improving from 24·55% in 2015 to 81·29% in 2023. (Table 2).

Table 2 Real-World Notification and Estimated Incident Cases of MDR/RR-TB, and Patient Detection Rate in Zhejiang Province, 2015–2023

Figure 4 Trend Analysis of Actually Notification and Estimated Incidence Rate of MDR/RR-TB (per 100,000 Population) in Zhejiang Province, 2015–2023. (APC, annual percent change. * indicates that the APC is statistically significant (P < 0.05). The shaded background indicates different phases of MDR/RR-TB control programs in Zhejiang Province.).

The three policy phases illustrated in Figure 4 closely correspond to these trends. The 2015–2016 Full Coverage Expansion extended routine screening to all smear-positive patients, the 2017–2020 China–Gates Foundation phase accelerated the province-wide rollout of molecular diagnostics, and the 2021–2023 Full Case-finding Expansion broadened testing to symptomatic TB suspects. These sequential expansions explain the initial increase in notifications, followed by convergence between the estimated and notified burdens as the case-finding capacity approached saturation. Estimated incidence decreased significantly (AAPC=−17·20, 95% CI: −22·50 to −11·50), while notification rates increased during 2015–2018 (APC=+10·51, 95% CI:+1·16 to +33·47, p=0·019) and declined thereafter (APC=−8·3995% CI: −19·27 to −4·67, p=0·0020).

Joinpoint Analysis of Regional Distribution

The MDR/RR-TB Notification rate trends varied across the 11 prefectures in Zhejiang (Figure 5). Despite the differences in magnitude and timing, most trends followed the three policy phases. Most prefectures exhibited a rapid increase in notification rates during the 2015–2016 Full Coverage Expansion, reached peak levels during the 2017–2020 China-Gates Foundation phase, and showed stable or declining trends during the 2021–2023 Full Case-finding Expansion phase. For example, Ningbo and Zhoushan experienced pronounced rises during 2015–2018 (APC=+21·04 and +38·31, both p<0·010), followed by declines in 2018–2023 (APC=−8·25 and −28·59, both p<0·010). Taizhou and Wenzhou displayed similar rise-then-fall patterns; however, the trend in Wenzhou was not statistically significant. The other prefecture-level cities had no clear inflection points: Hangzhou trended downward; however, the change was not statistically significant (AAPC=−3·59%, 95% CI: −8·72 to +0·92, p=0·10). (Figure 5)

Figure 5 Joinpoint Analysis of MDR/RR-TB Notification Rates in Prefecture-Level Cities of Zhejiang Province, Aligned with Key Programmatic Phases, 2015–2023. (APC, annual percent change. * indicates that the APC is statistically significant (P < 0.05). The shaded background indicates different phases of MDR/RR-TB control programs in Zhejiang Province).

ARIMA Forecasting Analysis

The fitted ARIMA model showed a steady decline in the estimated incidence of MDR/RR-TB in Zhejiang Province. The incidence fell from 3·42 per 100,000 in 2015 to 0·91 per 100,000 in 2023. Forecasts indicate that this decline will continue until 2030. The point predictions (log scale) of the model range from 0·773 in 2024 to 0·283 in 2030. (Figure 6).

Figure 6 Incidence Rate of MDR/RR-TB and Forecast of Future Trends, 2015–2030 (The dashed vertical line indicates the year 2023, separating observed data (2015–2023) from forecasted trends (2024–2030). The shaded area represents the 95% confidence interval (CI) of the projected incidence rates.).

Discussion

To our knowledge, this study is the first to apply the WHO-recommended estimation model to quantify the provincial-level burden and long-term trends of MDR/RR-TB in China. From 2015 to 2023, the estimated incidence of MDR/RR-TB in Zhejiang Province showed a continuous decline. In contrast, the real-world notification rate first increased and then decreased, ultimately approaching the estimated incidence rate. This convergence suggests continuous improvements in case detection, treatment, and financial policies. Overall, the findings indicate the measurable impact of MDR/RR-TB control efforts in Zhejiang Province.

The estimated incidence in Zhejiang Province showed a sustained downward trend, indicating a continuous reduction in the underlying disease burden since 2015. This pattern is consistent with the national TB trends reported by the Global Burden of Disease Study and epidemiological surveys conducted in Zhejiang Province.^12,21 However, real-world notification trends vary across the policy phases.

The temporal patterns observed in this study are consistent with the progressive strengthening of MDR/RR-TB control strategies in Zhejiang Province. Before the study period, Zhejiang initiated MDR/RR-TB control through a global funding program between 2008 and 2013, during which diagnostic capacity, treatment access, and case management systems were gradually established. This early capacity-building provided an important contextual background for interpreting subsequent trends. During Phase 1 (2014–2016), when the MDR/RR-TB control framework was consolidated, and screening coverage continued to expand, the number of notifications increased after 2015. In Phase 2 (2017–2020), when the policy and management system reached broader implementation and government-supported activities were further scaled up, notification rates began to show a decreasing trend, accompanied by a gradual narrowing of the gap between notification rates and estimated incidence. Studies in the same province also support this view.²² Since 2021 (Phase 3), the surveillance and control system has operated in a relatively stable manner, and the notification rates have remained closer to the estimated burden. These policy phases were defined a priori, based on actual implementation processes rather than on study outcomes. The temporal alignment between the policy phases and the observed trends suggests that the evolving control system may be related to changes in the observed MDR/RR TB burden. However, this study did not assess the causal effects of specific interventions. It should also be noted that the decline in notifications observed in 2020 coincided with the COVID-19 pandemic. Disruptions in healthcare access and healthcare-seeking behavior during this period may have temporarily affected TB case detection.²³ At the same time, strict public health interventions implemented in China,^24,25 such as universal mask use, mobility restrictions, and strengthened respiratory infection control measures, may also have contributed to a reduction in TB transmission.^26,27 However, the downward trend in estimated MDR/RR-TB incidence had already begun before the pandemic (2015–2019), suggesting that the underlying disease burden was already declining.

Similar temporal patterns were previously reported in China. National data showed that the detection rate of DR-TB increased from 14·3% in 2015 to 28·7% in 2019.²⁸ Some studies, including those from Yunnan,²⁹ Jiangsu,³⁰ and Shenzhen,³¹ have described a pattern characterized by a rapid increase in detection associated with expanding diagnostic capacity, followed by a subsequent decrease. In parallel, surveillance data indicated a long-term decline in rifampicin resistance among newly diagnosed patients with TB in China,³² with comparable trends reported in neighboring provinces such as Fujian.³³ International evidence also suggests that sustained screening coverage, stable financial support, and standardized treatment systems are associated with a reduction in MDR-TB burden. Collectively, these findings indicate that the temporal trends observed in Zhejiang align with broader national and international experiences.

Considerable heterogeneity was observed across the prefectures. These differences reflected the timing of entry of each city into the global funding program from 2008 to 2014. Prefectures that joined the MDR/RR-TB control programs earlier (2008–2009) generally showed stable notification trends from 2015 to 2023. Hangzhou has higher population mobility and urban factors;³⁴ however, the overall trend still shows a decline, which is similar to previous studies.³⁵ Later-joining prefectures (2013–2014) showed an initial increase in notifications, followed by a decline. These trends resemble the overall provincial patterns. These later-joining prefectures likely contributed substantially to provincial trends. This variation is related to the differences in program start times and system readiness. This pattern suggests that earlier implementation of MDR/RR-TB control strategies could help stabilize and reduce the provincial disease burden sooner. Other prefecture-level studies report similar results.^36,37

The ARIMA projections suggest that the estimated MDR/RR-TB burden in Zhejiang Province will continue to decline in the near future. However, the rate of decline was expected to be slow. This may indicate that current interventions are approaching a plateau. Additional efforts are required to achieve further reduction. These include strengthened screening and treatment of migrant populations,³⁸ optimization of second-line treatment regimens, application of new therapeutic strategies, and improved follow-up of close contacts.³⁹ Challenges remain, particularly the high financial burden placed on patients. Despite insurance reimbursements and subsidy programs, OOP costs remain substantial. In 2023, the median direct medical expenditure for patients with MDR/XDR-TB reached 10,491 US dollars, with medications accounting for more than 70%.⁴⁰ Sustained financial investments, improved subsidy schemes, and reduced drug prices are essential for equitable access.

Limitations

This study had some limitations. First, it relied on routine surveillance data, which may have been affected by underreporting and healthcare-seeking behaviors. Second, the uncertainty in the estimation model was not fully quantified. Third, Joinpoint regression and ARIMA modeling assume linearity and stationarity. These assumptions may not capture abrupt structural changes. Fourth, the alignment observed between policy phases and trends in the MDR/RR-TB burden was based on temporal associations only. Our study design did not allow for causal inference, and unmeasured factors may have influenced the observed patterns. Therefore, long-term projections and policy-related interpretations should be considered with caution.

Conclusion

His study provided a comprehensive assessment of the MDR/RR-TB burden in Zhejiang Province using a WHO-recommended estimation model and multisource data. The results showed that a series of MDR/RR-TB control policies implemented over the past decade have effectively reduced the provincial disease burden. The narrowing gap between the estimated incidence and notification rates further indicates an improved case detection and surveillance capacity. Current strategies have made clear progress; however, further strengthening of targeted interventions is required to achieve additional reductions in MDR/RR-TB burden.