Predictors of unsuccessful treatment outcomes among adults with pulmonary tuberculosis in a resource-limited setting

Abstract

Introduction: Tuberculosis (TB) remains a major health burden in low- and middle-income countries, where successful treatment is central to control efforts. This study aimed to determine the predictors of unsuccessful TB treatment in a resource-limited setting.

Method: A retrospective, record-based study was conducted at a district general hospital in Banten Province, Indonesia, and included 231 adults with pulmonary TB who were registered between January 2020 and December 2023. Outcomes were classified as successful (cured or treatment completed) or unsuccessful (treatment failure, loss to follow-up, or death) according to WHO definitions. Factors associated with unsuccessful outcome were assessed using chi-square tests, and variables with p < 0.25 were entered into a multivariable logistic regression model.

Results: Of 231 patients, 34 (14.7%) had an unsuccessful treatment outcome, comprising 18 deaths, 11 lost to follow-up, and 5 treatment failures. Independent predictors of unsuccessful outcome were age ≥60 years (adjusted odds ratio [aOR] 2.81, 95% CI 1.14–6.92), underweight nutritional status (aOR 2.47, 95% CI 1.08–5.65), previously treated TB (aOR 3.56, 95% CI 1.39–9.11), and HIV-positive status (aOR 5.74, 95% CI 1.71–19.24). HIV-positive status demonstrated the strongest association with unsuccessful treatment outcome.

Conclusion: Older age, undernutrition, retreatment, and HIV co-infection independently predicted unsuccessful TB treatment outcomes in this district setting. These factors help frontline clinicians to identify high-risk patients early and target nutritional support, integrated TB-HIV care, and closer monitoring.

Keywords: pulmonary tuberculosis; treatment outcome; unsuccessful outcome; predictor; HIV

Introduction

Tuberculosis (TB) remains one of the most pressing global health challenges. TB returned to being the world’s leading cause of death from a single infectious agent.^[1] The burden is concentrated in low- and middle-income countries (LMICs): just five nations accounted for 56% of global cases, with Indonesia ranking second and contributing an estimated 10% of the world’s TB burden.^[2,3] This concentration reflects the close link between TB and poverty, undernutrition, and constrained health systems, making the disease a defining condition of socioeconomic deprivation.^[4] 

Successful completion of treatment is the cornerstone of TB control and a key indicator tracked under the World Health Organization (WHO) End TB Strategy.^[5] Unsuccessful outcomes comprising treatment failure, loss to follow-up, and death prolong infectiousness, facilitate the emergence of drug-resistant strains, and impose substantial costs on patients and health systems. Identifying patients at risk of unsuccessful outcomes early in their treatment course is therefore essential for targeting supportive interventions.^[6,7] Although Indonesia reports a high national treatment success rate, this aggregate figure conceals considerable variation across regions and patient subgroups, and outcomes remain poorer among vulnerable populations.^[8]

Most data on the determinants of TB treatment outcomes in Indonesia derive from national datasets or tertiary referral centres.^[8,9] Far less is known about district-level hospitals, where the majority of patients are actually diagnosed and treated but where data are seldom reported and published. This represents an important gap, because the local determinants of unsuccessful treatment in a resource-limited district setting may differ from those identified in large, aggregated datasets and may be more directly actionable by frontline clinicians. The present study addresses this gap by examining treatment outcomes at a district hospital in Indonesia. 

The aim of this study was to determine the predictors of unsuccessful TB treatment outcomes for pulmonary TB in a resource-limited district setting.

Method

This was a retrospective record-based study conducted at a district general hospital in Banten Province, Indonesia. The hospital is a referral facility for the surrounding district and provides TB diagnosis and treatment under the national TB programme. Banten is a densely populated province adjacent to greater Jakarta, with a mixed urban–rural population, representing a typical resource-limited district context.

All patients with pulmonary TB who were registered for anti-TB treatment at the hospital between 1 January 2020 and 31 December 2023 were included. Data extraction was performed only after treatment outcomes had been recorded in the TB register. Consequently, even patients registered at the end of 2023 were followed through routine programme records until a final treatment outcome was available, typically after completion of the standard treatment course. Inclusion criteria were: (1) a diagnosis of pulmonary TB (bacteriologically confirmed or clinically diagnosed) according to national guidelines; (2) age ≥ 18 years; and (3) a documented final treatment outcome in the medical records. Exclusion criteria were: (1) extrapulmonary TB only; (2) patients transferred out to another facility whose final outcome could not be ascertained; and (3) records with incomplete data on key variables.

Data were extracted from hospital medical records and/or the TB register using a standardised form. Information collected included demographic characteristics, clinical variables, and treatment outcomes. To ensure data accuracy, a proportion of entries were re-checked against the original medical records, and discrepancies were resolved through source-document verification.

The primary outcome was TB treatment outcome, classified according to WHO definitions. Outcomes were divided into successful (cured or treatment completed) and unsuccessful (treatment failure, lost to follow-up, or died).

The dependent variable was unsuccessful treatment outcome. Independent variables comprised age, sex, residence/address, body weight or nutritional status, TB classification (new or previously treated), bacteriological status (smear- or rapid molecular test-positive vs clinically diagnosed), HIV status, diabetes mellitus status, and treatment regimen category. 

Statistical analysis

Data were analysed using SPSS version 30. Categorical variables were summarised as frequencies and percentages, and continuous variables as mean ± standard deviation. The distribution of continuous variables was assessed using the Shapiro–Wilk test and visual inspection of histograms. Variables with approximately normal distributions were presented as mean ± standard deviation. The association between each independent variable and unsuccessful treatment outcome was first assessed using the chi-square test for categorical variables. Variables with a p-value < 0.25 in bivariate analysis were entered into a multivariable logistic regression model to identify independent predictors of unsuccessful outcome. Adjusted odds ratios (aOR) with 95% confidence intervals (CI) were calculated. A two-sided p-value ≤ 0.05 was considered statistically significant. Model performance was evaluated using the Hosmer–Lemeshow goodness-of-fit test and Nagelkerke’s R². The overall classification accuracy of the model was also assessed.

Ethical Approval

Ethics approval was obtained from the Health Research Ethics Committee of Banten Regional General Hospital, protocol number 98/KEPK-RSUD/8/2024. Because this study used retrospective and anonymized data, the requirement for informed consent was waived. This study was conducted in accordance with the Declaration of Helsinki.

Results

A total of 254 patients with pulmonary TB were registered during the study period: 231 met the inclusion criteria  for analysis. Twenty-three patients were excluded (8 extrapulmonary TB only; 6 transferred out with unknown outcome; 9 incomplete data) (Figure 1). 

Figure 1. Flow diagram of patient selection. (LTFU = Lost to follow-up)

Of the 231 patients included, 197 (85.3%) had a successful treatment outcome (cured or treatment completed), 34 (14.7%) had an unsuccessful outcome, comprising 5 treatment failures, 11 lost to follow-up, and 18 deaths. The baseline characteristics of the study population are presented in Table 1.

Table 1. Baseline characteristics of pulmonary TB patients and treatment outcomes (N = 231)

The mean age was 46.8 ± 15.2 years, and 142 (61.5%) were male. Most patients were newly diagnosed cases (198; 85.7%), 33 (14.3%) had previously received TB treatment. The majority were bacteriologically confirmed (171; 74.0%), whereas 60 (26.0%) were clinically diagnosed. HIV positivity was identified in 12 patients (5.2%), and diabetes mellitus was present in 49 (21.2%). In bivariate analysis, age group (p = 0.018), nutritional status (p = 0.006), TB classification (p = 0.002), HIV status (p < 0.001), and diabetes mellitus (p = 0.097) were associated with unsuccessful treatment outcome (Table 2). Older age, underweight nutritional status, previously treated TB, and HIV positivity were associated with a greater proportion of unsuccessful outcomes.

Table 2. Bivariate analysis of factors associated with unsuccessful TB treatment outcome

Chi-square test was used for categorical variables. Variables with p < 0.25 were entered into the multivariable logistic regression model.

The logistic regression model demonstrated acceptable fit to the data (Hosmer–Lemeshow χ² = 6.84, df = 8, p = 0.553). The model explained 28.7% of the variation in unsuccessful treatment outcomes according to Nagelkerke’s R² (Cox & Snell R² = 16.8%) and correctly classified 86.1% of cases, indicating satisfactory predictive performance. In the multivariable logistic regression model, independent predictors of unsuccessful treatment outcome were age ≥60 years (aOR 2.81, 95% CI 1.14–6.92, p = 0.025), underweight nutritional status (aOR 2.47, 95% CI 1.08–5.65, p = 0.032), previously treated TB (aOR 3.56, 95% CI 1.39–9.11, p = 0.008), and HIV-positive status (aOR 5.74, 95% CI 1.71–19.24, p = 0.005) (Table 3). HIV-positive status demonstrated the strongest association with unsuccessful treatment outcome.

Table 3. Multivariable logistic regression of independent predictors of unsuccessful TB treatment outcome

aOR = adjusted odds ratio; CI = confidence interval. Variables with p < 0.25 in bivariate analysis were entered into the multivariable logistic regression model.

Discussion

In this retrospective record-based study of 231 patients with pulmonary TB treated at a district hospital in Indonesia, 14.7% experienced an unsuccessful treatment outcome. Four factors were independently associated with unsuccessful outcomes: age ≥60 years, underweight nutritional status, previously treated TB, and HIV-positive status, with HIV infection showing the strongest association. These findings identify readily measurable, routinely available characteristics that frontline clinicians can use to recognise patients at higher risk early in the treatment course.

The treatment success rate of 85.3% in this cohort fell just below the WHO End TB Strategy target of ≥90% and was broadly consistent with the national Indonesian figure of approximately 86%.^[9] This suggests that, despite operating in a resource-limited district setting, the hospital achieved outcomes comparable to national performance, while still leaving a meaningful gap relative to the global target. Notably, deaths accounted for the largest share of unsuccessful outcomes (18 (53%) of 34), followed by loss to follow-up, a pattern that points to the clinical severity of cases rather than programmatic dropout alone.

HIV-positive status was the strongest independent predictor of unsuccessful outcome (aOR 5.74). This is consistent with a large body of evidence across high-burden settings. A six-year Ethiopian cohort reported substantially lower treatment success among TB-HIV co-infected patients than among HIV-negative patients,[10] and a Cameroonian study found HIV positivity to be the strongest predictor of death during TB treatment.^[11] The biological basis is well established: HIV-related immunosuppression increases the risk of disseminated disease, treatment complications, and mortality. The relatively high proportion of unsuccessful outcomes among the small number of HIV-positive patients in our cohort reinforces the importance of integrated TB-HIV care, including early antiretroviral therapy, in district settings.

Previously treated TB was also an independent predictor (aOR 3.56), in keeping with findings from both Indonesian and other LMIC studies, where retreatment cases consistently show poorer outcomes than new cases.^[12,13] Patients requiring retreatment may harbour drug-resistant strains, have a history of poor adherence, or carry a higher burden of comorbidity all  impacting cure. This underscores the value of drug-susceptibility testing and closer monitoring for retreatment patients, particularly where rapid molecular diagnostics may not be uniformly available.

Underweight nutritional status independently predicted an unsuccessful outcome (aOR 2.47). This aligns with a multisite Indonesian cohort that reported being underweight as an independent risk factor for unfavourable outcomes (aOR 1.93),^[8,14] and a similar Ethiopian study.^[10] The relationship between undernutrition and TB is bidirectional: malnutrition impairs cell-mediated immunity and worsens treatment response, while active TB accelerates wasting. This supports the integration of nutritional assessment into routine TB care, an intervention that is low-cost and feasible in constrained settings.

Older age (≥60 years) was likewise an independent predictor (aOR 2.81), consistent with the same multisite Indonesian study (age ≥50 years, aOR 2.54)^[8,9] and with broader evidence that elderly patients experience higher mortality and poorer outcomes owing to comorbidity, frailty, atypical presentation, and reduced treatment tolerance. As populations age in many LMICs, tailored care for older adults with TB is likely to become increasingly important.

The findings in this study are consistent with much larger national and multisite studies, despite originating from a single district hospital. This strengthens confidence that the determinants of unsuccessful treatment identified here are not artefacts of the local setting but reflect generalisable risk factors that operate even at the district level where most patients are actually managed. The practical implication is that a simple risk profile based on age, nutritional status, treatment history, and HIV status could help district clinicians prioritise intensified support without requiring sophisticated tools.

Study limitations

This study has several limitations. First, its retrospective single-centre design limits generalisability and is subject to the completeness and accuracy of routinely recorded data; patients with incomplete records were excluded, which may have introduced selection bias. Second, the number of unsatisfactory outcomes was relatively small (n = 34), limiting the statistical power and stability of the multivariable model, resulting in wider confidence intervals, particularly for HIV status. HIV status was unavailable for 18 patients (7.8%), which may have introduced some degree of misclassification. Therefore, the magnitude of associations should be interpreted cautiously. Third, several potentially important variables were not available, including drug-susceptibility results, diabetes control (HbA1c), smoking, treatment adherence, and socioeconomic status, raising the possibility of residual confounding. Fourth, the study period (2020–2023) overlapped with the COVID-19 pandemic, which disrupted TB services in Indonesia and may have influenced follow-up and outcomes. Finally, the division of outcomes, while standard, may obscure differences between distinct unsuccessful outcomes such as death and loss to follow-up.

Recommendations and future research

The findings support several practical actions. District TB programmes should incorporate routine nutritional assessment and support, ensure HIV testing and timely antiretroviral therapy for all co-infected patients, and apply closer monitoring to retreatment cases and older adults. A simple, locally validated risk score incorporating these readily available predictors could help target limited resources to the patients most likely to benefit. Future research should include larger, multicentre, and prospective studies to confirm these associations with greater precision and to incorporate variables unavailable here, such as drug-susceptibility patterns, glycaemic control, and adherence. Studies evaluating whether targeted interventions, including nutritional support, enhanced adherence counselling, and integrated TB-HIV care, improve outcomes in district settings would be especially valuable for informing policy in resource-limited contexts.

Conclusion

In this resource-limited district hospital in Indonesia, nearly one in seven patients with pulmonary TB experienced an unsuccessful treatment outcome. Age ≥60 years, underweight nutritional status, previously treated TB, and HIV-positive status were independent predictors, with HIV co-infection carrying the greatest risk. Because these factors are routinely recorded and require no advanced technology to identify, they offer a practical basis for early risk stratification at the district level. Strengthening nutritional support, integrated TB-HIV care, and targeted monitoring of older and retreatment patients may help improve treatment success and advance progress towards the End TB Strategy targets.

Declaration of Generative AI and AI-assisted Technologies in the Writing Process

During the development of this manuscript, AI-assisted tools were utilized solely to support language refinement and clarity. All generated content was carefully reviewed, revised, and validated by the authors, who assume full responsibility for the accuracy and integrity of the final manuscript.

Conflicts of Interest:  None.

Funding: None.

1. References
2. Goletti D, Meintjes G, Andrade BB, Zumla A, Lee SS. Insights from the 2024 WHO Global Tuberculosis Report – More Comprehensive Action, Innovation, and Investments required for achieving WHO End TB goals. Int J Infect Dis. 2025 Jan 1;150. doi: https://doi.org/10.1016/j.ijid.2024.107325 PubMed PMID: 39631498.
3. Xue Y, Zhou J, Wang P, Lan J hong, Lian W qin, Fan YY, et al. Burden of tuberculosis and its association with socio-economic development status in 204 countries and territories, 1990–2019. Front Med. 2022 Jul 22;9. doi: https://doi.org/10.3389/fmed.2022.905245 
4. Nidoi J, Muttamba W, Walusimbi S, Imoko JF, Lochoro P, Ictho J, et al. Impact of socio-economic factors on Tuberculosis treatment outcomes in north-eastern Uganda: a mixed methods study. BMC Public Health. 2021 Nov 26;21(1):2167. doi: https://doi.org/10.1186/s12889-021-12056-1 
5. Harries AD, Kumar AMV, Satyanarayana S, Takarinda KC, Timire C, Dlodlo RA. Treatment for latent tuberculosis infection in low- and middle-income countries: progress and challenges with implementation and scale-up. Expert Rev Respir Med. 2020 Feb 1;14(2):195–208. doi: https://doi.org/10.1080/17476348.2020.1694907 PubMed PMID: 31760848.  
6. Chakaya J, Petersen E, Nantanda R, Mungai BN, Migliori GB, Amanullah F, et al. The WHO Global Tuberculosis 2021 Report – not so good news and turning the tide back to End TB. Int J Infect Dis. 2022 Nov 1; Commemorating World Tuberculosis Day, March 24th, 2022:“Invest to End TB. Save Lives”124:S26–9. doi: https://doi.org/10.1016/j.ijid.2022.03.011 
7. Boutilier JJ, Jónasson JO, Yoeli E. Improving Tuberculosis Treatment Adherence Support: The Case for Targeted Behavioral Interventions. Manuf Serv Oper Manag. 2022 Nov;24(6):2925–43. doi: https://doi.org/10.1287/msom.2021.1046 
8. Marley G, Zou X, Nie J, Cheng W, Xie Y, Liao H, et al. Improving cascade outcomes for active TB: A global systematic review and meta-analysis of TB interventions. PLOS Med. 2023 Jan 3;20(1):e1004091. doi: https://doi.org/10.1371/journal.pmed.1004091 
9. Lolong DB, Aryastami NK, Kusrini I, Tobing KL, Tarigan I, Isfandari S, et al. Nonadherence to anti-tuberculosis treatment, reasons and associated factors among pulmonary tuberculosis patients in the communities in Indonesia. PLOS ONE. 2023 Agu;18(8):e0287628. doi: https://doi.org/10.1371/journal.pone.0287628 
10. Yani DI, Juniarti N, Lukman M. Factors Related to Complying with Anti-TB Medications Among Drug-Resistant Tuberculosis Patients in Indonesia. Patient Prefer Adherence. 2022 Dec 31;16:3319–27. doi: https://doi.org/10.2147/PPA.S388989  PubMed PMID: 36568917.
11. Fekadu G, Turi E, Kasu T, Bekele F, Chelkeba L, Tolossa T, et al. Impact of HIV status and predictors of successful treatment outcomes among tuberculosis patients: A six-year retrospective cohort study. Ann Med Surg. 2020 Dec 1;60:531–41. doi: https://doi.org/10.1016/j.amsu.2020.11.032 
12. Tanue EA, Nsagha DS, Njamen TN, Assob NJC. Tuberculosis treatment outcome and its associated factors among people living with HIV and AIDS in Fako Division of Cameroon. PLOS ONE. 2019 Jul 30;14(7):e0218800. doi: https://doi.org/10.1371/journal.pone.0218800. 
13. Zhang L, Han X, Ge Q, Shu W, Sun Y, Gao J, et al. Treatment outcomes of retreated patients with isoniazid/rifampicin resistant pulmonary tuberculosis. BMC Infect Dis. 2024 Jan 2;24(1):7. doi: https://doi.org/10.1186/s12879-023-08909-2   
14. Opolot G, Olupot-Olupot P, Okware S, Izudi J. Retreatment TB is a risk factor for multidrug-resistant TB among people with HIV in rural eastern Uganda: A nested case-control study. J Clin Tuberc Mycobact Dis. 2025 Dec 1;41:100562. doi: https://doi.org/10.1016/j.jctube.2025.100562. 
15. Soeroto AY, Pratiwi C, Santoso P, Lestari BW. Factors affecting outcome of longer regimen multidrug-resistant tuberculosis treatment in West Java Indonesia: A retrospective cohort study. PLOS ONE. 2021 Feb 8;16(2):e0246284. doi: https://doi.org/10.1371/journal.pone.0246284 

.

More from the August 2026 issue