Tuberculosis (TB) is a chronic infectious disease caused by Mycobacterium tuberculosis (MTB), which is mainly transmitted through the respiratory tract (1, 2) and is the leading cause of infectious disease deaths worldwide (3). The Global Tuberculosis Report 2025 indicates that 10.7 million new TB cases occurred globally in 2024 (incidence rate of 131 per 100,000), of which 15.89% were retreatment TB (4). Patients with retreatment TB included those who failed initial treatment and others who had recurrent TB after previous successful treatment (5). For the recurrent TB cases, evidence has found two sources of recurrence involved endogenous reactivated TB and exogenous reinfected TB (6). In low-incidence settings, the risk of reinfection is usually considered to be minimal, and most of the recurrent TB cases are expected to be due to reactivation (7). In contrast, in high-incidence areas, the proportion of recurrent TB cases due to reinfection is higher due to increased risk of exposure (8). Available literature demonstrated that patients with recurrent TB commonly require an additional round of treatment and require longer treatment duration, leading to reduced probability of treatment success and further contributing to additional transmission of MTB strains and increasing the burden of TB (9). This may present a challenge to END TB planning and the attainment of treatment objectives (10).

Several factors such as poor medication efficacy, poor adherence to anti-tuberculosis drugs, and the patient’s immune status (co-infection with HIV or diabetes mellitus), were related to the development and treatment outcome of recurrent TB (11–13). In addition, patients with recurrent TB have poorer treatment outcomes with high mortality, low treatment completion rates, and high rates of drug resistance compared to new-onset TB cases (14). Despite significant progress in TB control in China, however, limited efforts have been made to identify the main causes of recurrent TB in the country, including in Zhejiang Province. However, Quzhou City in the west of Zhejiang Province had accounted for the top case findings of all identifications, which had its potential value to identify the influencing factors of recurrent development (15). To help address these issues, we analyzed surveillance data to characterize the epidemiology of recurrent TB and identify specific demographic and clinical risk factors for both recurrence and adverse retreatment outcomes, which would contribute to the further development of targeted interventions for TB control.

Although TB is recognized as a preventable and treatable disease, it still contributes to a considerable global burden of disease (16, 18, 19). Patients with recurrent PTB are usually more serious, face greater difficulties in medication adherence, poor health, and need to receive a second or even multiple treatment regimen, which may increase the risk of drug resistance of Mycobacterium tuberculosis (20, 21). In order to solve the problems of the high rate of recurrent PTB and unsatisfactory treatment results. Our study was the first to explore the epidemiological characteristics, changing trends, and risk factors of recurrent PTB in western Zhejiang. The results of the study showed that recurrence remains a challenge, although the majority of recurrent cases achieved treatment success. Among them, males were more likely to be recurrent than females, and older TB patients were more likely to be recurrent. Although the rate of recurrent PTB in Quzhou exhibited fluctuations, it has shown a downward trend since 2017. The results of this study can provide a reference for future policy formulation and research on the prevention and treatment of recurrent PTB in China.

Recurrent PTB is defined as a new TB episode in patients who have been considered cured after completing treatment (22). Its epidemiological characteristics vary significantly across different regions and countries, generally mirroring the local TB burden. In low-incidence settings—such as Israel, Finland, the United Kingdom, and the United States—recurrence rates are typically low, ranging from 0.6 to 1.8% (23–27). Conversely, recurrence rates are substantially elevated in high-burden regions or high-risk populations. For instance, a pooled analysis of Indian cohorts reported a recurrence rate as high as 8% (28), and a systematic review indicated that recurrence estimates could range broadly from 4.9 to 47% depending on regional epidemiology and study definitions (22, 29). In our study, the recurrent PTB rate was 3.15%. This figure places our region in an intermediate position globally. It is higher than the average recurrence rates observed in European countries (30, 31) and the 2.06% recurrence rate reported by Shao et al. in Jiangsu, China (32). However, it remains significantly lower than the rates reported in high-burden countries such as India (10.9%) (33) and South Africa (11%) (34). The variation in recurrence rates likely reflects differences in the baseline TB incidence, as well as variations in risk factors that drive both infection rates and the progression to active disease (35).

Regarding the temporal dynamics of recurrence, existing literature consistently identifies the initial post-treatment phase as the critical risk window, with the majority of recurrences occurring within 1–2 years (11, 26, 36, 37). For instance, data from South Korea indicated that 50.3% of recurrences occurred within the first year (37), while a Dutch study quantified the reactivation rate at 228/100,000 person-years in the first 2 years, dropping to 57 thereafter (37). In our study, we conducted a granular analysis of these intervals and observed a novel “acceleration phenomenon” in repeated episodes. Specifically, the median recurrence-free interval significantly shortened for the first recurrence compared to the second recurrence. This rapid progression in subsequent episodes suggests a cumulative compromise in host immunity or a persistent endogenous reservoir. This aligns with the biological distinction between endogenous reactivation and exogenous reinfection. In low-incidence settings, reactivation is the predominant driver, accounting for approximately 80% of recurrent cases (24, 25, 38). The acceleration we observed might indicate a cumulative compromise in host defense mechanisms, highlighting that patients with a history of recurrence are progressively more vulnerable and warrant intensified, prolonged surveillance beyond the standard regimen.

Recurrence is the result of a complex interplay involving host susceptibility, treatment-related factors, and environmental contexts. In our multivariable Cox regression analysis, male sex was identified as an independent risk factor for recurrence. This finding aligns with a broad consensus in the literature identifying male gender as a high-risk determinant (23, 28, 36), although it contrasts with a previous systematic review from India which found no such association (39). The male predisposition to recurrence may be attributable to behavioral factors, such as higher rates of smoking (11, 37), alcohol consumption (11, 22, 29), and generally poorer treatment adherence compared to females. Additionally, we observed that advanced age is a robust predictor of recurrence. Compared to patients aged 0–44 years, the risk was significantly elevated for those aged 45–64 years and ≥65 years. This is consistent with earlier studies highlighting the vulnerability of older populations (40, 41). We speculate that this association is driven by immunosenescence—the gradual decline in immune function with age—and the higher prevalence of comorbidities. Conditions such as diabetes mellitus (22, 36), chronic lung disease (25), and malnutrition (low BMI) (28), which are more common in the older adults, jeopardize the host’s ability to maintain containment of Mtuberculosis, thereby facilitating both the rekindling of primary occult foci and susceptibility to exogenous reinfection.

Socioeconomic status and accessibility to healthcare systems also play a crucial role. In China, a significant proportion of the TB burden is concentrated in rural areas (42), complicating disease control efforts (43). Our study confirms that farmer status is an independent risk factor for tuberculosis recurrence. These associations should be interpreted as surrogate indicators of broader socioeconomic and healthcare system determinants rather than independent biological risk factors. Rural populations often face multiple barriers to accessing healthcare, such as long distances to designated TB hospitals and lower health literacy, which may increase the risk of delayed diagnosis and more severe baseline disease. Furthermore, despite the widespread implementation of DOTS strategies, treatment supervision and follow-up intensity in remote rural areas may be compromised compared to urban centers due to resource constraints. Additionally, even with health insurance coverage, the financial burden of tuberculosis treatment disproportionately impacts low-income farming households. This can compromise nutritional support and treatment adherence, thereby increasing the risk of relapse. Clinically, patients with a bacteriologically positive diagnosis at the index episode faced a higher risk of recurrence. A positive baseline pathogen test indicates a higher initial bacterial load; if not sterilized completely, residual strains may serve as a source of recurrence (44). This aligns with findings that cultural positivity, particularly delayed conversion, predicts recurrence (45). Our study also identified protective factors related to healthcare access. Patients detected through active screening had a significantly lower recurrence risk compared to symptomatic visits, and those visiting municipal-level hospitals fared better than those at county-level institutions. These findings suggest that active case finding allows for diagnosis at earlier disease stages (46), and higher-level medical institutions may offer superior diagnostic and treatment quality, thereby reducing the likelihood of recurrence.

Although 83.1% of patients with recurrent PTB in our study achieved treatment success, this figure remains below the 85% target set by the Revised National Tuberculosis Control Programme (47), underscoring the importance of optimizing therapeutic strategies. Multivariable analysis identified advanced age (≥65 years) as an independent predictor of adverse outcomes, likely attributable to immunosenescence and comorbidities. In contrast, other studies (46), alongside HIV coinfection (48), consistently identify it as a critical driver of treatment failure. Beyond biological factors, treatment adherence is a pivotal determinant; poor compliance (23) and treatment interruption (26) significantly escalate reactivation risks. Additionally, adverse drug reactions (24) and the use of fixed-dose combinations may complicate therapeutic courses. Suboptimal outcomes in recurrence may also stem from a higher prevalence of acquired drug resistance (49) or pathological lung changes that impede drug penetration (50, 51). Consequently, we suggest that for high-risk patients, particularly the older adults, clinical management could be optimized by considering regimen intensification or extending the consolidation phase, coupled with enhanced post-treatment surveillance.

The demographic profile of tuberculosis in Quzhou has shifted significantly over the 18-year study period. We observed a consistent rise in the proportion of recurrent PTB cases among individuals aged ≥65 years, a trend that reflects the region’s aging population. The older population was identified as the main risk group for recurrence and poor re-treatment results in our model (Table 3). Apart from the overall risk level, our subgroup analysis also highlights the significant interaction between age and occupational risk. Among younger adults, farming is a strong predictor of recurrence; however, this association was not significant in the older adults, and there are significant interactions (Supplementary Figure 5). This observation shows that the influence of biological aging effectively outweighs the influence of socio-economic factors in later life. While the temporal pattern of early recurrence may suggest a potential role for endogenous reactivation due to waning immunity, we should interpret this with caution. Given the high TB burden in rural Western Zhejiang and the identified risk among farmers, exogenous reinfection still maybe a plausible mechanism driven by ongoing transmission and environmental exposure. The lack of molecular genotyping data in our study prevents us from definitively distinguishing between relapse (reactivation) and reinfection. Therefore, the observed recurrence patterns in our study likely reflect a complex interplay of biological vulnerability in the older adults and environmental exposure risks in rural settings. Future studies integrating genomic epidemiology are essential to quantify the specific contribution of each mechanism.

A critical finding of our analysis is the temporal robustness of the identified risk determinants. Despite the evolution of TB control policies over 18 years, the ‘study period’ was not an independent driver of recurrence or outcomes in our multivariable models, suggesting that biological determinants (e.g., immunosenescence) and clinical severity (e.g., bacterial load) override programmatic changes. However, our stratified analysis offered nuanced insights: the rising risk among farmers in recent years mirrors the concentration of the disease burden in rural areas, while the attenuated prognostic risk in the older adults during the second period may reflect improvements in geriatric care and supportive management.

In the multivariate model, study period (2007–2016 vs. 2017–2024) did not emerge as an independent predictor of recurrence or adverse outcomes. This observation coincided with a decline in overall reporting rates. Given that standard DOTS strategies effectively curbed transmission in the general population, the residual disease burden likely concentrated among clinically refractory core populations characterized by higher immune senescence and multiple comorbidities, thereby offsetting overall improvements in healthcare delivery. However, our stratified analysis revealed critical epidemiological shifts. Notably, the association between agricultural occupation and recurrence risk strengthened in recent years compared to earlier periods. This trend suggests disease burden is shifting toward rural areas, likely driven by widening disparities in healthcare access and socioeconomic resilience between urban and rural settings. Conversely, regarding prognosis, the risk of adverse outcomes associated with advanced age (≥65 years) has significantly decreased in recent years compared to earlier period. This narrowing of the prognostic gap suggests improvements in the healthcare environment for older patients and enhanced post-treatment monitoring. In summary, these data underscore that while intrinsic biological susceptibility (e.g., age, bacteriological characteristics) remains constant, extrinsic socioeconomic drivers are evolving. This necessitates dynamically adjusting public health resources to better serve rural older populations.

Our study had several limitations. First, this study was based mainly on a retrospective analysis of TB Information Management System case data rather than a prospective cohort study, which may underestimate the proportion of patients with recurrent PTB. Second, as this study was based on routine surveillance records, detailed information on several potential risk factors was not available. Specifically, variables such as smoking history, alcohol use, body mass index, and household economic status were not systematically collected in the database. Consequently, we could not adjust for these unmeasured factors in our multivariable models, and the observed associations may be subject to residual confounding. Future prospective studies with more granular clinical and socioeconomic data are needed to validate our findings. Third, a significant limitation of this study is the lack of systematic data on drug susceptibility testing (DST) and rapid molecular assays, particularly for the earlier years of the study period (2007–2015). Due to the high proportion of missing data, we could not determine the true prevalence of baseline drug resistance for the entire cohort or evaluate the specific impact of multidrug-resistant tuberculosis (MDR-TB) on recurrence risks (52–54). Therefore, the risk factors identified in our study should be interpreted with caution, as they may partially reflect unmeasured confounding by drug resistance status. Fourth, our findings were limited by geographic location, which makes them less appropriate for individuals who have recently migrated or who have migrated to different regions. Finally, it is unclear whether recurrence was endogenous rekindling or exogenous reinfection, as genotyping data were not available, and differentiating both mechanisms is important for optimizing post-treatment strategies.

Despite these limitations, our study has several strengths. First, few studies have conducted epidemiological studies of recurrent PTB, and we collected 18 years of surveillance data in areas with a high prevalence of TB. Second, we comprehensively identified independent risk factors for recurrence and evaluated adverse treatment outcomes. Crucially, we documented a distinct “acceleration phenomenon” in repeated episodes, characterized by a significant shortening of the recurrence-free interval. Although existing policies cannot completely block the occurrence of recurrent PTB, there is a need for more intensive and targeted clinical management of patients requiring retreatment.

This study indicates that recurrent tuberculosis remains a persistent public health challenge in Quzhou City. Our analysis identified advanced age (≥65 years) as a dominant predictor associated with both recurrence susceptibility and adverse retreatment outcomes, while male sex and farming occupation were also significant risk factors. Notably, nearly half of the recurrent cases occurred within the first 2 years after treatment completion, and the recurrence-free interval shortened significantly with repeated episodes. These findings suggest that current TB control strategies could be optimized by prioritizing age-specific management and extended post-treatment surveillance for high-risk groups. We suggest that integrated care models focusing on geriatric support warrant further consideration to address the dual burden of recurrence and poor prognosis in this vulnerable population.