Introduction

Tuberculosis (TB) remains one of the world’s deadliest infectious diseases, despite being curable. In 2024, TB caused approximately 1.23 million deaths globally, highlighting its ongoing threat to public health [1]. The disease burden is particularly severe in Sub-Saharan Africa (SSA), which accounts for over 25% of global TB cases and reports the highest rates of HIV-associated Tanzania is classified among the 30 high-burden TB countries, with an estimated incidence of 183 cases per 100,000 population in 2023 [2]. Many countries, including Tanzania, continue to encounter significant obstacles in TB control. While global initiatives have reduced TB incidence in certain regions, persistent challenges, such as treatment failure and ongoing transmission, persist [3]. These factors impede national progress toward achieving the World Health Organization’s End TB Strategy, which seeks to substantially reduce TB incidence and mortality by 2030 [4].

TB primarily manifests as a pulmonary disease, affecting the lungs. However, if the infection is not contained, it may disseminate to other organs, resulting to, extrapulmonary TB. This form is often more severe and harder to diagnose and treat [5,6]. For instance, when TB infects the brain, it can cause life-threatening complications [7]. The clinical manifestation of TB is determined by a complex interplay of host and environmental factors [8]. Although HIV co-infection is a well-established biological factor that significantly complicates treatment and increases mortality risk [9,10], sociodemographic and structural determinants are increasingly recognized as crucial influences [11,12]. In Tanzania, marked regional disparities in treatment success rates, such as, 99% in Kagera compared to approximately 90% in Geita, underscore the significant impact of local contexts [13]. These disparities suggests underlying causes that extend beyond biology, likely rooted in environmental, occupational, and health system influences [8].

Geita is a region located in north-western Tanzania, is defined by its extensive gold mining industry, which shapes its demographic and health profile [14,15]. The region bears a high TB burden, with an estimated prevalence of 274 per 100,000, exceeding the national average [2,16]. Mining communities are recognized as high-risk environments where factors such as occupational dust exposure, overcrowding in work and living spaces, and a highly mobile workforce converge to create a perfect storm for TB transmission and progression [17]. Furthermore, the presence of distinct sub-communities within this context, such as prison staff and miners, each with unique exposure profiles, adds another layer of complexity to the local TB epidemiology.

Despite the recognition of these challenges, a critical knowledge gap persists. Existing studies in Tanzania have often focused narrowly on specific aspects, such as multidrug-resistant TB or prevalence within single communities [18–20]. There is a striking lack of comprehensive research that simultaneously analyzes how the distinct occupational groups, referral pathways, and clinical characteristics prevalent in a mining region interact to determine both the initial presentation of TB (pulmonary versus extrapulmonary) and the ultimate treatment outcome. Understanding these interconnected determinants is essential to moving beyond a one-size-fits-all approach to TB control

The primary contribution of this study lies in the integrated analysis of sociodemographic, clinical, and health-system factors; along with their interactions on both TB presentation and treatment outcome within the defined, high-burden context of Geita’s community. Therefore, we utilised routine data from Tanzania’s National Tuberculosis and Leprosy Programme (NTLP) to identify the key factors, including occupational group, referral pathway, HIV status, and demographic characteristics that influence TB presentation and treatment success in Geita. By offering a comprehensive perspective, these findings are intended to inform the development of targeted, effective, and context-specific interventions to improve TB control in Geita and other regions with high TB prevalence.

Materials and methods

Results

Study population and participant flow

A total of 14,999 tuberculosis cases were notified in Geita Region between January 1, 2017 and December 31, 2020. After applying the inclusion criteria (age ≥ 15 years, confirmed TB diagnosis per NTLP guidelines, and treatment initiation in Geita Region), 14,107 patients were eligible for further screening. Of these, patients were excluded for the following reasons: transferred in from other regions (n = 489(3.47%)), rifampicin-resistant TB at diagnosis (n = 296(2.09%), and missing data for primary exposure or outcome variables (HIV status: 98(0.69%); disease classification: 17(0.12%); referral source: 1(0.007%); occupation: 1(0.007%); treatment outcome: 141(1%)). Some patients met more than one exclusion criterion. The final analytical cohort comprised 13,107 patients. The flow of study participant selection is summarised in Fig 1.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Flowchart of study participant selection and cohort definition for TB cases in the Geita Region, 2017–2020.

This flowchart illustrates the transition from total notified tuberculosis (TB) cases to the final analytical cohort through the sequential application of inclusion and exclusion criteria. It tracks the removal of cases based on age, diagnostic verification, and data completeness. Abbreviations: HIV, Human Immunodeficiency Virus; NTLP, National Tuberculosis and Leprosy Programme; RR-TB, Rifampicin-resistant Tuberculosis; TB, Tuberculosis. Notes: The total exclusions (n = 1,000) represent unique patients, although some individuals met more than one exclusion condition. Missing data accounted for exclusions in HIV status (n = 98), disease classification (n = 17), referral source (n = 1), occupation (n = 1), and treatment outcome (n = 141).

https://doi.org/10.1371/journal.pone.0346249.g001

Participant characteristics

The majority of patients were male (63.0%, n = 8,252), and HIV co-infection was identified in 34.2% (n = 4,443) of the cohort. Regarding disease classification, 12,032 (91.9%) cases were pulmonary TB, 1,058 (8.1%) were extrapulmonary TB, and 17 (0.1%) had missing classification data. Most cases (97.9%, n = 12,835) were new diagnoses. The primary referral sources were self-referral (48.4%, n = 6,348) and community referral (28.1%, n = 3,685). Miners constituted 13.0% (n = 1,699) of the occupational cohort. Treatment outcome data were available for 12,966 patients (98.9% of the cohort). Among them, the overall treatment success rate was 95.5% (n = 12,385), while 4.5% (n = 581) had unsuccessful outcomes. The characteristics of the study population are summarized in Table 1.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Basic characteristics of tuberculosis patients in Geita Region, Tanzania, 2017-2020 (N = 13,107).

https://doi.org/10.1371/journal.pone.0346249.t001

Factors associated with tuberculosis presentation

In modelling TB presentation, between the two forms pulmonary and extra-pulmonary, the pulmonary form was modelled. Initially, a logistic regression model incorporating all exposure variables (age group, sex, occupation, referral source, HIV status, and TB history) in an additive structure was fitted. The model provided an adequate fit, with a deviance of 197.95 (P = 0.71) and a Pearson chi-square of 195.76 (P = 0.75). However, Type III analysis revealed that among these exposure variables, only sex, workplace category, referral source, and TB history were significant predictors (P < 0.05). When an extended model including interaction terms was fitted, significant interactions between age × sex (P = 0.0213) and sex × HIV status (P = 0.0045) were identified. Therefore, the final model was fitted using all significant terms from both the initial and extended models. The final model maintained a good fit (deviance = 181.17, P = 0.90; Pearson χ² = 179.74, P = 0.91; df = 207), with all included terms statistically significant. Table 2 displays the adjusted odds ratios (aORs). Notably, Prison staff had lower odds of pulmonary TB compared to miners (aOR = 0.26, 95% CI: 0.12–0.58; P = 0.001), suggesting fundamentally different exposure profiles. Community–referred cases showed higher pulmonary TB odds than self–referred patients (aOR=1.24, 95% CI: 1.06–1.45; P = 0.009), indicating community health workers effectively identify respiratory cases. New TB cases demonstrated 67% lower pulmonary TB odds than relapse cases (aOR=0.33, 95% CI: 0.15–0.75; p = 0.008). The inclusion of interaction effects indicated that these relationships were modified significantly by age and sex, and by sex and HIV status.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Adjusted odds ratios (aOR) for factors associated with pulmonary vs. extrapulmonary TB presentation.

https://doi.org/10.1371/journal.pone.0346249.t002

Interaction effects of sex with age and HIV status on pulmonary tuberculosis presentation

The significant interaction (P = 0.021) Sex × Age implies that the association between age and TB presentation differ by sex. Among males, odds of having pulmonary TB for those aged 40–59 is 1.26 times that of those who are ≥ 60 years (aOR=1.26, 95% CI: 1.01–1.58), while females showed no significant difference in odds between age groups (Table 3). This likely reflects occupational exposures during peak working years. The Sex × HIV status interaction (P = 0.0045) was of the form that HIV-negative males showed marginally higher pulmonary TB odds than HIV-positive males (aOR=1.21, 95% CI: 0.98–1.49), whereas HIV-negative females had marginally lower odds than their HIV-positive counterparts (aOR=0.80, 95% CI: 0.63–1.03) (Table 3).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Interaction effects of sex with age and HIV status on pulmonary tuberculosis presentation.

https://doi.org/10.1371/journal.pone.0346249.t003

Predictors of treatment outcome

In modelling treatment outcome, between the two forms successful and unsuccessful, the successful form was modelled. The fit of the initial logistic regression model with all 5 variables (age group, sex, referral source, workplace, HIV status, and TB history) was satisfactory, with a deviance of 180.58 (P = 0.92) and a Pearson chi–square of 195.10 (P = 0.75) with 209 degrees of freedom. However, significant predictors included age group, referral source, HIV status, and TB history. Extended model with interaction effects detected no significant interaction between factors on treatment outcome. Thus, the final model was fitted with significant factors identified in the initial model. The model’s adequacy remained acceptable in the final fitting, with a deviance of 53.95 (P = 0.59) and a Pearson chi–square of 51.85 (P = 0.67) with 57 degrees of freedom. Table 4 presents the adjusted odds ratios (aORs). Younger patients had significantly better outcomes: Patients aged <40 (aOR=1.30, 95% CI: 1.04–1.62; P = 0.021) and 40–59 (aOR=1.31, 95% CI: 1.05–1.63; P = 0.019) had higher success odds than those with age ≥ 60. HIV–negative status increased success odds compared to HIV–positive (aOR=1.90, 95% CI: 1.53–2.37; P < 0.001). New cases had 2.45-fold higher success than relapses (95% CI: 1.53–3.93). Both Inpatient referrals (aOR = 0.55, 95% CI: 0.31–0.98), and other referral sources (combined non-self/non–IPD) showed lower success odds (aOR = 0.46, 95% CI: 0.30–0.69) compared to that of self-referral (Table 4).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Odds ratios estimates and confidence intervals for associated factors after fitting the logistic regression model on the proportion of TB treatment outcome.

https://doi.org/10.1371/journal.pone.0346249.t004

Discussion

This study investigated the determinants of tuberculosis (TB) successful treatment outcomes and factors associated with TB presentation in the region of Geita, Tanzania, from 2017 to 2020. Key factors influencing these outcomes include age, occupation, referral pathways, and HIV co-infection. Our findings highlight the complex interplay of individual characteristics and systemic factors affecting TB presentation and treatment success in this high-burden setting. Understanding these determinants can inform targeted interventions to improve TB management in mining communities.

Workplace exposure is a key determinant of tuberculosis (TB) presentation. Prison workers, for instance, exhibit a distinct profile: they have 74% lower odds of developing pulmonary tuberculosis (PTB) but a higher likelihood of extrapulmonary tuberculosis (EPTB) compared to miners (aOR for PTB = 0.26; P = 0.001). This pronounced difference demonstrates how occupation-specific risks influence disease manifestation. Miners are exposed to environments that facilitate respiratory transmission, including silica dust, inadequate ventilation, and crowded underground spaces, which create optimal conditions for pulmonary infections. This exposure likely accounts for their representation in 13% of all PTB cases. These findings are consistent with research from South Africa, where TB rates among miners exceed 7,000 per 100,000, more than ten times the national average [28]. Comparable trends are observed in Zambia and other sub-Saharan African countries, where mining significantly contributes to TB incidence, particularly PTB [29,30]. Although prison settings are recognized as high-risk environments for TB transmission [31] due to overcrowding and limited ventilation, which could theoretically increase PTB risk, the observed lower PTB odds among prison staff compared to miners indicate that additional factors may be involved. These may include differences in health-seeking behavior, access to regular health checks in prison employment, or distinct immunological responses. Further investigation is warranted to elucidate the mechanisms underlying this occupational difference in TB presentation. The substantial proportion of miners among PTB cases reinforces the association between occupational hazards and pulmonary TB. Collectively, these results emphasize the necessity for occupation-specific interventions. In mining communities, strategies such as improved ventilation, health education, and regular screening are essential to reduce PTB transmission. The observed occupational variation in TB presentation confirms that work environments significantly influence both TB risk and disease manifestation, highlighting critical health disparities that require targeted interventions.

Referral pathways significantly influence the presentation of tuberculosis (TB). Community referrals are associated with a 24% higher likelihood of pulmonary TB (PTB) than self-referrals (aOR 1.24; P = 0.009), suggesting that self-referred patients more frequently present with extrapulmonary TB (EPTB). This trend suggests that community health workers (CHWs) are effective at identifying classic respiratory symptoms, such as chronic cough, during active case-finding, thereby facilitating timely PTB diagnosis. In contrast, self-referred cases often involve the non-specific manifestations of EPTB, such as lymphadenopathy or spinal pain, resulting in delayed recognition. These findings align with evidence from Ethiopia, where health extension workers, a type of CHW, significantly increased TB case detection, doubling notifications in supported districts from 64 to 127 per 100,000 population and improving treatment success rates from 76% to 96% over 4.5 years [32]. Similarly, in Mozambique, CHW316 facility-based screening and contact tracing increased TB notifications by 14.6% in intervention areas, while control districts experienced declines [33]. Despite these advances, diagnosing EPTB remains challenging. Reports from Uganda indicate that many EPTB cases self-present after prolonged symptoms [34,35], underscoring the need for targeted community education regarding EPTB signs, such as persistent lymph node swelling.

Age and HIV status are associated with TB presentation, with sex acting as a modifier of these relationships. Males aged 40–59 had 1.26 times higher adjusted odds of PTB than older males (aOR 1.26; 95% CI: 1.01–1.58), suggesting a lower prevalence of EPTB in this group. This pattern likely reflects increased occupational exposures, such as silica dust in mining, during peak working years, consistent with Marçôa et al.’s (2018) findings of higher PTB rates among Portuguese men aged 50–59 [36]. HIV co-infection further complicates these associations: HIV-negative males exhibited higher PTB odds than HIV-positive males, whereas HIV-positive females had higher PTB odds than HIV-negative females. Although estimates within specific groups did not reach statistical significance, the significant sex and HIV interaction (P = 0.0045) demonstrates that biological sex modifies the effect of HIV on TB presentation. A 2018 Ethiopian meta-analysis confirms that HIV-positive individuals have increased TB susceptibility due to immune system weakening [37,38]. These findings suggest that HIV-positive women experience a disproportionate burden of both PTB and EPTB, likely attributable to more advanced immunosuppression. These results underscore the need for occupational lung health programs targeting mining men aged 40–59, enhanced TB screening for HIV-positive women that includes EPTB symptoms such as lymphadenopathy, and clinician training on sex-specific HIV-TB interactions.

The analysis revealed that tuberculosis (TB) history significantly influenced the type of clinical presentation. New TB cases (aOR = 0.33; P = 0.008) demonstrated substantially lower odds of pulmonary TB (PTB) relative to extrapulmonary TB (EPTB) when compared to relapse cases. These results indicate that patients experiencing a TB relapse are significantly more likely to present with pulmonary disease than those with new or undefined prior histories. This finding partially aligns with studies suggesting that relapse may be driven by reactivation of persistent pulmonary foci or undiagnosed drug resistance, which favors PTB [39,40]. In contrast, other research has reported higher EPTB rates in relapse, often associated with immunosuppression or disseminated disease [41]. This discrepancy underscores the complex interplay among local epidemiology, drug resistance patterns, and host factors in relapse presentation. Based on these findings, targeted interventions should prioritize enhanced respiratory symptom screening and sputum testing for all relapse patients, and provide specific education on recognizing pulmonary TB symptoms to encourage prompt care-seeking.

Referrals from hospital inpatient departments (IPD) were associated with a 45% lower likelihood of successful tuberculosis (TB) treatment compared to self-referred patients (aOR 0.55; P = 0.044). This association likely reflects the advanced disease severity, including disseminated extrapulmonary TB (EPTB), HIV co-infection, and malnutrition, which are prevalent among hospitalized individuals and complicate treatment while increasing mortality risk. These findings are consistent with Nidoi et al. (2021) in Uganda, where hospital-referred TB patients experienced higher mortality due to late presentation [42]. Similarly, Mpagama et al. (2020) identified diagnostic delays among hospitalized TB patients with comorbidities in Tanzanian mines [43]. In contrast, these results differ from Ethiopian studies advocating hospital-based screening, underscoring context-specific health system barriers [44]. It is therefore recommended to implement continuous systems to expedite diagnosis and treatment for hospital referrals, conduct regular screening of hospitalized patients for EPTB and comorbidities, and strengthen community-based treatment support following discharge.

HIV-negative patients demonstrated a 90% greater likelihood of successful treatment compared to HIV-positive individuals (aOR 1.90; P < 0.001), highlighting the essential role of immune competence in tuberculosis (TB) cure. HIV-associated immunosuppression impairs treatment response, increases the risk of disseminated extrapulmonary TB (EPTB) such as meningeal or miliary disease, and raises relapse risk due to inadequate drug penetration at sterile sites. These findings are consistent with Gupta-Wright et al. (2022) in Southern Africa, where HIV co-infection was associated with a twofold increase in TB mortality (aHR 2.1), particularly in cases of disseminated EPTB [45]. The increased relapse risk among HIV-positive patients, attributed to persistent subclinical EPTB reservoirs, was further supported by a Ugandan cohort study that reported relapse rates 3.3 times higher than those observed in HIV-negative patients [46].

Patients under 60 years of age demonstrated significantly higher treatment success than those aged 60 years and older. The adjusted odds ratio (aOR) for treatment success among younger patients was 1.30–1.31, representing a 30–31% increase (P < 0.05). Several biological factors may contribute to this disparity, including stronger immune responses that reduce the risk of extrapulmonary tuberculosis (EPTB) dissemination, fewer comorbidities that limit drug interactions, and improved medication tolerance, which decreases the risk of hepatotoxicity—a common cause of treatment interruption in older adults. These findings are consistent with studies from South Korea, where patients under 60 years exhibited higher rates of treatment completion and cure [47]. However, the present results contrast with studies from high-resource settings, such as the Netherlands, which reported smaller age-related effects [48]. This contrast underscores the heightened vulnerability of older adults in resource-limited settings such as Geita.

Newly diagnosed tuberculosis patients demonstrated 2.45 times higher odds of treatment success compared to relapse cases (aOR 2.45; P < 0.001), highlighting retreatment as a critical vulnerability. This outcome is attributed to factors such as undetected drug resistance, extrapulmonary tuberculosis (EPTB) involving sanctuary sites with limited drug penetration, and previous treatment interruptions leading to immunological non-response [49]. These results are consistent with the WHO Global TB Report (2024), which indicates that retreatment cases contribute disproportionately to poor outcomes due to a higher prevalence of drug-resistant TB (16% versus 3.2% in new cases), resulting in lower treatment success for MDR/RR-TB (68% versus 88% for drug-susceptible TB) (2). Recommended interventions include universal drug susceptibility testing, extended EPTB retreatment regimens, decentralized access to second-line drugs, and pharmacist-led adherence support.

This study provides important insights into the determinants of tuberculosis in Geita’s mining communities; however, several limitations should be considered. The retrospective design, which relies on routine program data, limited our ability to include key variables such as socioeconomic status and detailed clinical metrics. The patient cohort was heterogeneous in terms of occupation and referral sources, which, although representative of the region, may have introduced unmeasured confounding. Additionally, the routine national program database aggregated all occupations outside the specified categories of miner, healthcare worker, and prison staff into a single “Other” field, restricting our ability to analyse other occupational groups in detail. While our findings are highly relevant to mining settings in Tanzania, their applicability may differ in regions with distinct health systems or mining operations. Although we did not have complete drug resistance test results for all relapse patients, this limitation does not diminish the significance of our core findings regarding HIV status, referral pathways, age, and occupation-related risks. These factors remain essential for enhancing tuberculosis care in this high-risk population.

Conclusions

The findings of this study indicate that in the high tuberculosis (TB) burden mining region of Geita, the clinical profile and treatment outcomes of TB are systematically influenced by occupational, clinical, and demographic factors. The elevated risk of pulmonary TB among miners, the demonstrated effectiveness of community referrals in identifying pulmonary cases, and the complex interactions between sex and HIV status highlight the necessity for precisely targeted public health interventions. To reduce the TB burden in this and similar contexts, a shift towards differentiated care strategies is warranted. This approach should include implementing mandatory, periodic occupational health programs in mines, with a focus on silica dust control and active case-finding to protect miners. Additionally, community-based screening should be strengthened by training health workers to identify both classic pulmonary symptoms and subtle extrapulmonary manifestations of TB, thereby minimizing diagnostic delays. Integration of TB-HIV care should also incorporate a gender-specific perspective, ensuring that HIV-positive women receive intensified screening for both pulmonary and extrapulmonary TB. These evidence-based recommendations, directly informed by the study’s results, offer a clear and actionable framework for policymakers and health planners to enhance TB control in Tanzania’s mining communities.

Acknowledgments

We gratefully acknowledge the National Tuberculosis and Leprosy Programme (NTLP) of the United Republic of Tanzania, for granting access to routine program data and for their valuable technical support throughout the study. We also express our appreciation to the healthcare workers in the Geita Region whose meticulous record-keeping enabled this analysis. Special thanks are extended to GlaxoSmithKline (GSK) for supporting training and mentorship in data cleaning, validation, analysis, and reporting through the Department of Biostatistics at Kwame Nkrumah University. We further thank the Tanzania National Health Research Ethics Committee (NatHREC) for their ethical oversight and for ensuring adherence to national ethical standards.