Prevalence and determinants of unsuccessful tuberculosis treatment outcomes among children in Northern Ghana: a retrospective review

Abstract

Introduction

Tuberculosis (TB) remains a leading cause of childhood morbidity and mortality, yet evidence on treatment outcomes and associated determinants in Northern Ghana remains limited. This study examined the prevalence, trends, and determinants of unsuccessful treatment outcomes among children with TB at Tamale Teaching Hospital from 2016 to 2024.

Methods

A retrospective review of secondary data was conducted among children (< 15 years) diagnosed with TB at Tamale Teaching Hospital from 2016 to 2024. Data were extracted from the National Tuberculosis Programme registers. Treatment outcomes were defined according to WHO guidelines. Modified Poisson regression with robust standard errors was used to examine predictors of unsuccessful treatment outcomes, while logistic regression was performed as a sensitivity analysis. Analyses were conducted using Stata version 16 IC (StataCorp LLC, College Station, TX, USA).

Results

Of 1,842 TB cases, 245 were children, of whom 233 had documented treatment outcomes and were included in the analysis. The median age was 5 years (IQR: 2–10), and 58.4% were males. Pulmonary TB accounted for 63.5% of cases, while 13.7% had TB/HIV coinfection. The treatment success rate was 75.5% (176/233), while 24.5% (57/233) experienced unsuccessful outcomes, including 11.2% deaths and 13.3% loss to follow-up, with no cases of treatment failure. In multivariable analysis, children aged < 1 year (aRR: 3.04; 95% CI: 1.13–8.23), those aged 5–9 years (aRR: 3.51; 95% CI: 1.63–7.57), and rural residents (aRR: 1.70; 95% CI: 1.08–2.66) had a significantly higher risk of unsuccessful treatment outcomes. Sensitivity analysis using logistic regression confirmed these findings, and no significant associations were observed with sex, HIV status, type of patient, or disease classification.

Conclusion

Nearly one in four children experienced unsuccessful TB treatment outcomes, mainly due to death and loss to follow-up. Younger age and rural residence were key predictors of poor treatment outcomes. Strengthening early diagnosis, decentralised care, and targeted adherence support for vulnerable groups is essential to improve outcomes.

Introduction

Childhood tuberculosis (TB) remains a significant contributor to global morbidity and mortality. In 2023, an estimated 1.25 million children and young adolescents developed TB, resulting in nearly 200,000 deaths, most occurring in children under five years of age [1]. Despite advances in TB control, paediatric case detection remains suboptimal, with nearly half (50%) of cases estimated to be undiagnosed and untreated, and at risk of dying [2]. The Global Burden of Disease (GBD) 2021 analysis further revealed that children under five remain disproportionately affected [3]. Drug-resistant TB further complicates management, with limited access to diagnosis and treatment among affected children [4]. Globally, childhood TB is often under-recognised within national TB programmes, with many countries failing to disaggregate paediatric treatment outcomes [5, 6].

In sub-Saharan Africa, including Ghana, health system constraints contribute to gaps in TB case detection, reporting, and treatment [2]. Childhood TB outcomes remain suboptimal, with ongoing challenges related to mortality, loss to follow-up, and TB/HIV coinfection despite ongoing control efforts [7–14]. These gaps are further exacerbated by HIV, poverty, and weak health systems, all of which increase vulnerability and complicate disease management in children [14, 15].

In Ghana, childhood TB notifications remain low (4–6%), below the World Health Organisation expected range of 8–10%, reflecting persistent under-detection [10, 16, 17]. Treatment outcomes vary across regions, with success rates ranging from 68% to 90%, while mortality remains considerable [8, 10, 18]. Although TB services have improved in recent years, including recovery following COVID-19-related disruptions, a substantial proportion of cases remain undiagnosed. This continues to sustain community transmission [17, 19]. However, most TB research in Ghana has focused on adults, resulting in limited evidence on treatment outcomes among children, who often present with non-specific symptoms that delay diagnosis and management [18, 20, 21].

Evidence on childhood TB in Northern Ghana is particularly scarce, with most studies focusing on adults or mixed populations and few documenting treatment outcomes at Tamale Teaching Hospital [22–24]. This gap is concerning given the region’s socioeconomic and health system challenges, including poverty, malnutrition, and limited access to diagnostic and treatment services, which may adversely affect outcomes.

This study aimed to examine the prevalence, trends, and predictors of unsuccessful treatment outcomes among children with TB at Tamale Teaching Hospital in Northern Ghana.

Methods

Study design and setting

This retrospective cohort study examined the prevalence, trends, and determinants of unsuccessful treatment outcomes among children under 15 years receiving tuberculosis care at Tamale Teaching Hospital (TTH), Northern Ghana, from January 2016 to December 2024. TTH is the primary tertiary referral hospital for the five northern regions, serving over five million people. It plays a central role in TB diagnosis, treatment, and monitoring in Ghana [25]. Previous studies at TTH reported a 13% smear-positive TB prevalence between 2004 and 2010 and documented adverse drug reactions among TB patients [23, 26]. The hospital’s diagnostic capacity was enhanced with the reintroduction of GeneXpert testing in 2023, improving rapid case detection and identification of rifampicin resistance [27]. At the national level, TB case notification remains suboptimal, with only 19,081 of an estimated 44,000 cases notified in 2023, including 824 children [28]. TTH’s comprehensive TB services, high caseload, and research support make it a suitable setting for this study.

Study participants and sample size

Children aged 0–14 years with a confirmed tuberculosis (TB) diagnosis (pulmonary or extrapulmonary (EPTB)) registered for treatment at TTH during the study period were eligible. Children still undergoing treatment, those transferred out or those lacking documented treatment outcomes were excluded. A census approach was used to include all eligible cases, maximising statistical power and eliminating sampling error. Of 1,842 TB cases registered at TTH between 2016 and 2024, 245 were children under 15 years. Complete treatment outcome data were available for 233 children, which formed the final dataset for analysis (see Fig. 1).

Fig. 1.

Flowchart displaying childhood tuberculosis treatment outcomes at Tamale Teaching Hospital, Northern Region, Ghana, from 2016 to 2024

Study variables

The dependent variable was treatment outcomes, classified according to the National Tuberculosis Programme (NTP) and WHO guidelines as follows:

• Cured: cases initially positive for smear, culture, or GeneXpert, but negative at the end of treatment and on at least one prior occasion.

• Treatment completed: cases that completed treatment without documented negative smear or culture at the end of treatment.

• Died: death from any cause during treatment.

• Lost to follow-up (LTFU): cases that interrupted treatment for ≥ 2 consecutive months.

• Treatment failure: cases that have positive smear or culture at ≥ 5 months of treatment.

For regression analysis, outcomes were further grouped into two categories:

• Successful outcome: cases recorded as cured or treatment completed.

• Unsuccessful outcome: cases that resulted in loss to follow-up, death or treatment failure.

The primary outcome measure was unsuccessful treatment, while successful treatment served as the reference category.

The independent (explanatory) variables considered in this study were grouped into three domains: demographic, clinical, and programmatic characteristics. Demographic variables included the child’s age, sex, and place of residence. Clinical variables included the site of disease (pulmonary or extrapulmonary), the specific extrapulmonary site where applicable, and HIV status. Programmatic variables included the year of registration and the type of patient, categorised as new or retreatment.

Data sources and management

Data were abstracted from TB registers and treatment cards using a standardised form. Variables extracted included demographic, clinical, and diagnostic information. Each patient was assigned a unique identification number to maintain confidentiality. Data were entered into a password-protected Microsoft Excel file. Double data entry was performed independently by two researchers, and discrepancies were resolved through cross-checking against the source documents. Records with missing key outcome variables were excluded, while missing covariate data were retained as separate categories where applicable to minimise bias. A data manager reviewed the dataset for completeness, accuracy, and internal consistency, including checking for missing values and outliers. Selection bias was handled by including all eligible cases within the study period. Variable definitions were also derived according to the national TB control/WHO guidelines to reduce information bias.

Data analysis

Data analysis was carried out using Stata version 16 IC (StataCorp LLC, College Station, TX, USA). Descriptive statistics were used to summarise demographics and clinical characteristics of children. Categorical variables were presented as frequencies and percentages, while continuous variables were reported using the median and inter-quartile range (IQR). Bivariate associations between independent variables and treatment outcomes were assessed using Pearson’s chi-square or Fisher’s exact tests as appropriate.

Modified Poisson regression with robust standard errors was used to estimate risk ratios (RR) for unsuccessful treatment outcomes. It was selected for its suitability with binary outcomes of high prevalence (> 10%) and computational stability compared with log-binomial regression [29, 30]. Also, it is not subject to convergence issues and allows statistically valid inference, while accounting for overdispersion and variance misspecification [29, 31]. Predictor variables with p < 0.25 in univariate analysis or identified as clinically important confounders were included in the multivariable model. This was to ensure that most clinically relevant variables were included to minimise bias. HIV status was retained in all models regardless of statistical significance because of its established relevance [32]. Results were reported as adjusted risk ratios (aRR) with 95% confidence intervals (CIs). Sensitivity analysis using logistic regression was performed to estimate adjusted odds ratios (aORs) and compare effect estimates.

Ethical considerations

Ethical clearance for the study was obtained from the Committee on Human Research, Publications and Ethics (CHRPE) of Kwame Nkrumah University of Science and Technology (Reference: CHRPE/AP/549/25), with permission from the Department of Research and Development at the Tamale Teaching Hospital. The study was conducted in accordance with the Declaration of Helsinki. Given the retrospective nature of this study, which involved a review of de-identified medical records, the ethics committee waived the requirement for informed consent from individual participants. All data were anonymised before analysis and stored on password-protected systems to maintain confidentiality.

Results

Sociodemographic characteristics of respondents

A total of 233 children with documented treatment outcomes were included in the analysis, representing 12.6% of all TB cases registered during the study period. The median age was 5 years (IQR: 2–10), with 46.8% aged under five years. Males accounted for 58.4% of cases, and 67.8% resided in urban areas. The majority of cases were newly diagnosed (97.4%). Pulmonary TB was the most common form (63.5%), while 36.5% had extrapulmonary TB. HIV status was documented for most patients (91.8%), with a coinfection prevalence of 13.7%. Overall, the treatment success rate was 75.5%, while 24.5% experienced unsuccessful outcomes, including 11.2% deaths and 13.3% loss to follow-up, with no treatment failures reported (Table 1).

Table 1.

Sociodemographic characteristics of respondents

Characteristic	Number (n = 233)	Percentage (%)
Sex
Male	136	58.37
Female	97	41.63
Age group (in Years)
Median (Interquartile range)	5 (2–10)
< 1	28	12.02
1–4	81	34.76
5–9	57	24.46
10–14	67	28.76
Residence
Rural	75	32.19
Urban	158	67.81
Type of patient
New	227	97.42
Retreatment	6	2.58
Disease classification
Pulmonary positive TB	27	11.59
Clinically diagnosed TB	121	51.93
Extra-pulmonary TB	85	36.48
HIV status
Negative	182	78.11
Positive	32	13.73
Unknown	19	8.15
Treatment outcomes
Cured	15	6.44
Loss to follow-up	31	13.30
Died	26	11.16
Treatment completed	161	69.10
Summary outcomes
Successful treatment	176	75.54
Unsuccessful treatment	57	24.46

Figure 2 illustrates the trends in treatment outcomes of childhood tuberculosis over time. Treatment completion declined from 70.6% in 2016 to 54.5% in 2024, despite peaking at 88.9% in 2021. Mortality showed a general decline, although fluctuations were observed across the years. In contrast, loss to follow-up increased substantially over time, rising from 5.9% in 2016 to 21.2% in 2024, with a 28.6% peak in 2022. Cure rates rose from 9.1% in 2018 to 26.7% in 2019, declined in 2020, and then remained relatively stable through 2024.

Fig. 2.

Trends in treatment outcomes of paediatric tuberculosis at the Tamale Teaching Hospital, 2016–2024

Figure 3 shows the overall trends in treatment outcomes. Treatment success generally increased with slight fluctuations between 2016 and 2021, peaking in 2021, and then declining through 2024. Conversely, unsuccessful outcomes declined between 2016 and 2020, followed by an increase from 2021 to 2024.

Fig. 3.

Trend in successful and unsuccessful treatment of paediatric tuberculosis at Tamale Teaching Hospital, 2016–2024

Figure 4 shows the distribution of extra-pulmonary tuberculosis (EPTB) sites among the 85 reported cases. Lymphadenitis was the most frequently reported site, accounting for nearly one-third of cases, followed by spinal TB and cases with unspecified sites. Abdominal and disseminated TB were moderately represented, while bone TB, pericarditis, and pleural TB were relatively uncommon. However, no cases of TB meningitis were recorded during the study period.

Fig. 4.

Distribution of extra-pulmonary tuberculosis sites among children at the Tamale Teaching Hospital, 2016–2024

Childhood tuberculosis treatment outcomes by sociodemographic and clinical characteristics

Bivariate analysis revealed that treatment outcomes were significantly associated with age group, residence, and disease classification. Deaths were slightly high in children < 1 year (14.3%), while loss to follow-up was more pronounced among 5–9-year-olds (26.3%). Rural children experienced a higher loss to follow-up (24.0%) compared with urban children. Similarly, deaths were concentrated more among clinically diagnosed cases (15.7%), while extrapulmonary TB cases had the highest loss to follow-up (21.2%). No statistically significant associations were observed for sex, HIV status, or type of patient (Table 2).

Table 2.

Childhood tuberculosis treatment outcomes by sociodemographic and clinical characteristics at the Tamale Teaching Hospital, 2016–2024

Characteristic	Treatment outcomes				p-value
	Cure	Treatment completed	Died	LTFU
	n(r%)	n(r%)	n(r%)	n(r%)
Sex					0.347
Male	11(8.09)	88(64.71)	17(12.50)	20(14.71)
Female	4(4.12)	73(75.26)	9(9.28)	11(11.34)
Age group (years)					< 0.001
< 1	1(3.57)	19(67.86)	4(14.29)	4(14.29)
1–4	3(3.70)	59(72.84)	7(8.64)	12(14.81)
5–9	3(5.26)	31(54.39)	4(14.04)	15(26.32)
10–14	8(11.94)	52(77.61)	7(10.45)	0(0.00)
Residence					0.010
Rural	5(6.67)	44(58.67)	8(10.67)	18(24.00)
Urban	10(6.33)	117(74.05)	18(11.39)	13(8.23)
Type of patient					0.874
New	15(6.61)	156(68.72)	25(11.01)	31(13.66)
Retreatment	0(0.00)	5(83.33)	1(16.67)	0(0.00)
Disease classification					< 0.001
Pulmonary positive	14(51.85)	11(40.74)	2(7.41)	0(0.00)
Clinically diagnosed	1(0.83)	88(72.73)	19(15.70)	13(10.74)
Extra-pulmonary	0(0.00)	24(75.00)	5(5.88)	18(21.18)
HIV status					0.287
Negative	15(8.24)	123(67.58)	18(9.89)	26(14.29)
Positive	0(0.00)	14(73.68)	6(18.75)	2(6.25)
Unknown	0(0.00)	14(73.68)	2(10.53)	3(15.79)

%r: row percentage

Predictors of unsuccessful treatment outcome of childhood tuberculosis

In the multivariable modified Poisson regression model (Table 3), age and residence were significant predictors of unsuccessful treatment outcomes.

Table 3.

Predictors of unsuccessful treatment outcome among children at the chest clinic of the Tamale Teaching Hospital, 2016 to 2024

Characteristics	Unadjusted RR (95%CI)	Adjusted RR (95%CI)	Adjusted odds ratio (95%CI)
Sex
Male	1.311(0.818–2.129)
Female	1
Age group (years)
< 1	2.735(1.095–6.831)	3.044(1.126–8.234)	4.200(1.151–15.310)
1–4	2.245(1.003–5.025)	2.252(0.969–5.232)	2.768(0.969–7.905)
5–9	3.862(1.787–8.345)	3.509(1.626–7.574)	5.498(2.007–15.065)
10–14	1	1	1
Residence
Rural	1.767(1.134–2.754)	1.695(1.079–2.663)	2.186(1.103–4.331)
Urban	1	1	1
Type of patient
Retreatment	0.676(0.111- 4.118)
New	1
Disease classification
Pulmonary negative	0.977(0.617- 1.547)	1.015(0.645-1.600)	1.014(0.513–2.006)
Pulmonary positive	0.274(0.069–1.090)	0.359(0.090–1.436)	0.285(0.057–1.428)
Extra-pulmonary	1	1	1
HIV status
Positive	1.034(0.538- 1.990)	0.962(0.497–1.863)	0.947(0.365–2.460)
Unknown	1.089(0.491- 2.415)	1.000(0.538–1.862)	1.008(0.389–2.616)
Negative	1	1	1
Year	1.054(0.955–1.164)

RR: risk ratio; CI: confidence interval

Children younger than 1 year had a threefold higher risk of experiencing an unsuccessful outcome compared with those aged 10–14 years (aRR: 3.04; 95% CI: 1.13–8.23). Similarly, children aged 5–9 years were more than three times at risk of unsuccessful treatment outcomes compared with the 10–14-year age group (aRR: 3.51; 95% CI: 1.63–7.57). Residence also played an important role, with children from rural areas having 70% higher risk of experiencing an unsuccessful treatment outcome compared to urban residents (aRR: 1.70; 95% CI: 1.08–2.66).

These results were consistent with findings from the sensitivity analysis using logistic regression (Table 3). As expected, logistic regression overestimated effect sizes for this common outcome, but the direction of associations remained consistent. No significant associations were observed for sex, disease classification, type of patient, year of registration, or HIV status.

Discussion

This study found that nearly one in four children receiving tuberculosis treatment at Tamale Teaching Hospital experienced unsuccessful outcomes, primarily due to death and loss to follow-up. Younger age and rural residence were identified as key predictors of these outcomes, indicating important disparities in treatment outcomes among paediatric populations.

Children with documented treatment outcomes accounted for 12.6% of all TB patients registered at Tamale Teaching Hospital, similar to the 6–15% range reported in other settings [7, 33, 34].

Pulmonary tuberculosis (63.5%) was the predominant form of TB, aligning with evidence from Ghana, Brazil, Ethiopia, and Zambia [8, 35–37]. However, bacteriological confirmation was low, with only 11.6% of cases being smear-positive. Similarly, 16.5% of EPTB cases had undocumented disease sites, limiting the specificity of reporting. This reflects well-documented challenges in paediatric TB diagnosis, including difficulties in specimen collection and low bacillary load [38]. Similar diagnostic limitations have been reported in Ghana and other African settings [7, 8, 37]. Despite these limitations in our setting, other regions have reported higher confirmation rates, which may reflect variation in diagnostic capacity, access to molecular tools, and case mix [18, 35, 36]. The higher sputum smear positivity reported in those studies could also be linked to under-detection of smear-negative and clinically diagnosed cases, particularly in younger children [8].

Extrapulmonary TB accounted for 36.4% of cases, a finding that falls within ranges reported in other studies [8, 37, 39, 40] but is higher than the 17.7% previously reported in Ghana [18]. Lower proportions have also been observed in Zimbabwe, Zambia, and Pakistan, ranging from 9% to 23% [36, 41, 42]. This variation likely reflects differences in study settings and case profiles, and the relatively high proportion in this study may be explained by the larger number of older children, among whom EPTB is more frequently identified, alongside improved clinical recognition in this age group. Lymph node TB remained the most common extrapulmonary manifestation, consistent with findings from earlier studies [8, 11].

In this study, 24.5% of children experienced unsuccessful treatment outcomes, which is substantially higher than the 2.2%–8.1% range (pooled estimate: 5.1%) reported in Zambia [36]. This higher proportion of unsuccessful outcomes may reflect the tertiary referral nature of the study setting, where more severe and complicated cases are managed, as well as delays in diagnosis and treatment initiation among referred patients.

The mortality rate (11.2%) was higher than facility-based estimates from some Ghanaian and Sub-Saharan African studies [18, 36, 37], but lower than national case fatality estimates [16]. It was also lower than the 17.4% mortality documented in an earlier Ghanaian study [8]. The elevated mortality observed may be attributable to late presentation and advanced disease at the time of diagnosis, which is common in referral facilities such as Tamale Teaching Hospital, as well as coexisting conditions such as malnutrition and HIV infection.

Loss to follow-up (13.3%) significantly exceeded both the national average (4.1%) and the global estimate of approximately 6% [4]. While it was higher than findings from Zambia [36], this rate was comparable to rates reported in other studies [37, 39]. This may reflect structural barriers to healthcare access in Northern Ghana, including long travel distances to treatment centres, transportation costs, and limited caregiver support, which can hinder consistent treatment attendance [39, 43].

The temporal trends in childhood TB treatment outcomes at Tamale Teaching Hospital showed an improvement in treatment success from 2016, peaking in 2021, followed by a decline through 2024, mirroring the increase in unsuccessful outcomes, particularly loss to follow-up, over the same period. This decline in treatment success and the observed rise in LTFU from 5.9% in 2016 to 21.2% in 2024, with a marked increase in 2022, aligns with regional and global reports on the severe disruption of routine TB services during the COVID-19 pandemic, which impacted patient access to care, treatment adherence, and follow-up activities [17, 44, 45].

Treatment failure remained uncommon, aligning with findings from other studies [36, 39]. This may be due to limitations in diagnostic follow-up, underreporting, or classification practices within routine program data.

The overall treatment success rate (75.5%) was below the global target of 85% and lower than many reported estimates from comparable settings [18, 35, 36, 40, 41, 46–49]. Nonetheless, it was comparable to some reports from Nigeria [7] and higher than others in West Africa [8, 9]. Variations in treatment success across settings may reflect differences in health system capacity, disease severity, HIV co-infection, and caregiver adherence to treatment [7].

The multivariable analysis identified younger age and rural residence as significant predictors of unsuccessful treatment outcomes. Children < 1 year and 5–9 years were more likely to experience poor treatment outcomes, consistent with findings from previous studies [7, 8, 46, 47, 50]. Younger children are biologically more vulnerable due to immature immunity, increased risk of severe disease, and diagnostic challenges that may delay treatment [46, 51, 52]. These risks may be compounded by comorbidities such as malnutrition and pneumonia, as well as caregiver-related factors affecting adherence [50].

The increased risk among children from rural areas may be explained by longer travel distances to health facilities, reduced availability of TB services at peripheral levels, and financial constraints that limit caregivers’ ability to sustain regular clinic visits [43, 53]. No significant associations were observed for sex, HIV status, type of patient, or disease classification, consistent with earlier studies [36, 37].

Strengths and limitations of the study

This study provides important programmatic evidence from an under-explored setting, drawing on nine years of routinely collected data to assess trends and determinants of childhood TB treatment outcomes. The use of longitudinal data and multivariable analysis strengthens the robustness of the findings. However, several limitations should be acknowledged: the retrospective design is subject to missing and incomplete data, particularly for key variables such as extra-pulmonary TB site and HIV status. The reliance on clinical diagnosis introduces the possibility of misclassification, especially among younger children. In addition, important predictors, including nutritional status, socioeconomic conditions, treatment adherence, and comorbidities, were not captured, limiting the depth of analysis. Exclusion of children without documented treatment outcomes and those transferred out may have introduced selection bias, potentially affecting the estimated proportion of unsuccessful outcomes. As a single-centre study conducted in a tertiary referral hospital, the findings may over-represent severe or complicated cases and may not be fully generalisable to primary care settings. Despite these limitations, the study provides valuable insights into real-world challenges in paediatric TB care and highlights priority areas for intervention.

Conclusion

Childhood tuberculosis remains a significant public health concern in Northern Ghana, with nearly one in four children experiencing unsuccessful treatment outcomes. The high and increasing rates of loss to follow-up, alongside suboptimal treatment success, highlight critical gaps in treatment retention and continuity of care. Younger age and rural residence were key determinants of unsuccessful treatment outcomes, reflecting both biological vulnerability and structural barriers to accessing care. Addressing these gaps requires targeted, context-specific interventions to improve early diagnosis, strengthen adherence, and enhance access to care, particularly for vulnerable populations.

Implication for practice

The findings of this study highlight the need for strengthened health system responses to improve childhood TB outcomes in Northern Ghana. Given the high proportion of clinically diagnosed cases, expanding access to child-friendly diagnostic approaches, such as GeneXpert Ultra, stool-based assays, and radiographic tools, alongside continuous clinician training, is essential to improve early and accurate detection.

In response to the high and increasing rates of loss to follow-up observed, there is a critical need to strengthen community-based treatment support systems, including directly observed therapy, active follow-up, and structured caregiver counselling. Decentralisation of TB services to peripheral facilities would further reduce access barriers, particularly for rural populations identified in this study as being at higher risk of poor treatment outcomes.

Integration of TB services with maternal and child health, nutrition, and HIV programmes would support a more comprehensive continuum of care. Targeted resource allocation should prioritise younger children and rural communities, who are disproportionately affected.

Future research should include multicentre studies to improve generalisability and explore the social and behavioural determinants of treatment outcomes. Longitudinal studies assessing post-TB sequelae in children are also needed to inform long-term care and policy.

Abbreviations

HIV

Human immunodeficiency virus

DOT

Directly observed therapy

TB

Tuberculosis

EPTB

Extrapulmonary tuberculosis

WHO

World Health Organization

TTH

Tamale Teaching Hospital

cRR

Crude risk ratio

aRR

Adjusted risk ratio

aOR

Adjusted odds ratio

CI

Confidence interval

Author contributions

AS: Conceptualisation, Investigation, Formal Analysis, Visualisation, Writing - Original Draft. AIC: Conceptualisation, Formal Analysis, Writing - Original Draft. IH: Methodology, Writing - Review & Editing. AAB: Methodology, Writing - Review & Editing. AAK: Data Curation, Validation, Writing - Review & Editing. SOAJ: Data Curation, Validation, Writing - Review & Editing. ZI: Data Curation, Validation, Writing - Review & Editing. DN: Data Curation, Validation, Writing - Review & Editing. SN: Data Curation, Validation, Writing - Review & Editing. JSI: Data Curation, Validation, Writing - Review & Editing. AM: Methodology, Writing - Review & Editing, Supervision.

Funding

This study received no specific funding.

Data availability

The data supporting the conclusions of this article are not publicly available to protect patient confidentiality. Access to the data may be considered upon reasonable request to the corresponding author.

Declaration

Ethics approval and consent to participate

Ethical clearance for the study was obtained from the Committee on Human Research, Publications and Ethics (CHRPE) of Kwame Nkrumah University of Science and Technology (Reference: CHRPE/AP/549/25), with official permission from the Department of Research and Development at the Tamale Teaching Hospital. The study was conducted in accordance with the Declaration of Helsinki. Given the retrospective nature of this study involving review of de-identified medical records, the ethics committee waived the requirement for informed consent from individual participants. All data were anonymised before analysis and stored on password-protected systems to maintain confidentiality.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.