Prevalence and predictors of tuberculosis among children in Zambezi District, North-Western Province, Zambia

Abstract

Background

Tuberculosis (TB) remains a leading cause of morbidity and mortality among children in low- and middle-income countries, particularly in sub-Saharan Africa. Paediatric TB is often underdiagnosed due to nonspecific symptoms, limited access to diagnostic tools, and its tendency to mimic other common childhood illnesses. In Zambia, where TB and HIV are co-endemic, understanding the burden of childhood TB at the sub-national level is critical for tailoring effective interventions. This study aimed to determine the prevalence and identify key predictors of TB among children aged 0–14 years in Zambezi District, North-Western Province, Zambia.

Methods

A retrospective cross-sectional study reviewed routinely collected data from tuberculosis registers in 16 diagnostic health facilities in Zambezi District from January 2020 to December 2024. All records of children aged 0–14 years screened for TB were assessed. Data on sociodemographic and clinical variables (age, sex, HIV status, nutritional status, household TB contact, BCG vaccination status, household density, and caregiver education) were extracted using a standardised tool. Descriptive statistics summarised the data. Bivariate associations were examined with chi-square tests, and multivariable logistic regression identified independent predictors of TB diagnosis (p < 0.05).

Results

Of 412 children screened for TB, 57 (13.8%) were diagnosed with TB. The most affected age group was < 5 years. HIV-positive children were nearly seven times more likely to be diagnosed with TB than HIV-negative children (adjusted odds ratio [aOR] = 6.8, 95% CI: 3.4–13.6, p < 0.001). Malnutrition (WAZ < − 2; aOR = 3.7, 95% CI: 2.0–6.9, p < 0.001) and household TB contact (aOR = 5.2, 95% CI: 2.7–10.0, p < 0.001) were also significant independent predictors.

Conclusion

This study reveals a substantial burden of paediatric TB in rural Zambezi District. HIV infection, malnutrition, and household TB exposure are key drivers. Urgent integration of TB-HIV services, nutritional support, and enhanced household contact tracing is required to reduce the paediatric TB burden in rural Zambia.

Introduction

Tuberculosis (TB) remains a major global public health concern and is the leading cause of death from a single infectious agent [1]. While adult TB has traditionally received more attention, paediatric tuberculosis is increasingly recognised as a significant yet under-researched component of the global TB burden. According to the World Health Organization (WHO) Global Tuberculosis Report 2025, an estimated 1.2 million children and young adolescents under 15 years developed TB in 2024, representing approximately 11% of the 10.7 million global incident cases [1]. This figure includes 174,300 child deaths in 2024, accounting for 14.2% of all TB mortality, with the vast majority occurring in low- and middle-income countries (LMICs) [1, 2] Despite being preventable and curable, childhood TB continues to be overlooked in national control programmes due to diagnostic, clinical, and programmatic challenges [3, 4].

Diagnosis of TB in children is particularly difficult because symptoms are nonspecific and frequently mimic other common childhood illnesses such as pneumonia or malnutrition [4, 5] Children are also more likely to have paucibacillary disease, which limits bacteriological confirmation with standard tools, especially in resource-limited settings where access to advanced diagnostics such as GeneXpert or culture remains restricted [3]. These gaps contribute to substantial underreporting, delayed treatment initiation, and poorer outcomes, with modelling studies estimating that up to 43% of children with TB missed diagnosis and treatment in recent years [1, 6].

Zambia is one of the 30 high-TB-burden countries globally [6] The Zambia National Tuberculosis and Leprosy Programme (NTLP) has reported that children under 15 years accounted for approximately 11% of notified TB cases in recent years, with contributions from childhood notifications increasing from around 7% in earlier periods to 11% more recently [7, 8]. However, this national figure likely underestimates the true burden in rural and underserved districts, where diagnostic infrastructure, paediatric-specific services, and surveillance remain weak [9]. A 2025 hospital-based study in Zambia reported a 9.4% prevalence of active TB among hospitalised children under 15 years, with HIV infection and malnutrition identified as significant independent predictors [10].

Zambezi District in North-Western Province exemplifies these rural challenges; it is predominantly rural, with dispersed communities, limited health facilities, high poverty, chronic malnutrition, and food insecurity. High adult HIV prevalence further increases TB-HIV co-infection risk in children through household exposure and vertical transmission [11]. Few studies have examined paediatric TB at the sub-national level in rural Zambia, creating a critical evidence gap. Recent research in comparable rural African settings has highlighted the compounded effects of household contact, HIV, malnutrition, and structural factors such as overcrowding [12, 13]. This study therefore aimed to determine the prevalence and predictors of tuberculosis among children aged 0–14 years in Zambezi District between January 2020 and December 2024. By generating local evidence from routinely collected registers, the study seeks to inform context-specific TB control strategies for paediatric populations in resource-constrained rural environments.

Materials and methods

Study design and setting

This retrospective cross-sectional study used routinely collected data from tuberculosis (TB) registers in Zambezi District, a predominantly rural area in Zambia’s North-Western Province. Sixteen public health facilities provide TB services under the Zambia National Tuberculosis and Leprosy Programme (NTLP), employing symptom screening, chest radiography, sputum smear microscopy, and GeneXpert testing (with variable capacity across sites).

Study population and inclusion criteria

The population included all children aged 0–14 years screened for TB at the 16 facilities between 1 January 2020 and 31 December 2024. Records were included only if TB screening and diagnostic evaluation were completed and key variables (HIV status and TB diagnosis outcome) were documented. Records with missing key information or where the child was transferred out before diagnosis confirmation were excluded. Complete-case analysis was applied for secondary variables with occasional missing values.

Sampling procedure and sample size

All eligible paediatric records meeting the inclusion criteria during the study period were included (census approach). No formal sample-size calculation was performed because the study utilised every available eligible record to maximise power and provide a comprehensive district-level assessment. The final analytic sample comprised 412 children.

TB screening and diagnostic procedures

Screening followed Zambia’s national paediatric TB algorithm using a standardised symptom checklist (chronic cough ≥ 2 weeks, fever, night sweats, weight loss, or poor weight gain). Positive screens prompted physical examination, chest radiography, and bacteriological testing (sputum smear microscopy or GeneXpert). Gastric aspirates or induced sputum were used for young children unable to produce spontaneous sputum. TB diagnosis was clinical or bacteriological per national guidelines.

Data collection

Data were abstracted from standardised TB facility registers using a structured tool. Key variables included age, sex, HIV status, nutritional status (classified using weight-for-age Z-scores [WAZ] computed from recorded weight/age or documented by health workers), household TB contact history, BCG vaccination status (from linked immunisation records or presence of BCG scar), household density (persons per room), and caregiver education level (from patient records). Records were screened for completeness before inclusion. Nutritional status was consistently classified as malnourished if WAZ < − 2. To minimise information bias, abstraction was performed by trained personnel using a standardised form, with double-checking of 10% of records.

Data analysis

Data were analysed in STATA version 17. Descriptive statistics (frequencies, proportions, means) summarised characteristics. Chi-square tests assessed bivariate associations between variables and TB diagnosis. Variables with p < 0.20 in bivariate analysis entered a multivariable logistic regression model to adjust for confounding. Adjusted odds ratios (aOR) with 95% confidence intervals (CI) identified independent predictors; p < 0.05 was considered statistically significant. The study followed STROBE guidelines.

Results

Of the 412 children aged 0–14 years screened for tuberculosis (TB) in Zambezi District health facilities between January 2020 and December 2024, 57 (13.8%) were diagnosed with TB. The mean age of participants was 6.9 ± 4.2 years, with 31.6% (n = 130) under 5 years old. Males comprised 53.4% (n = 220) of the sample.

Table 1 summarizes the associations between demographic, clinical, and exposure characteristics and TB diagnosis. Younger age groups showed a significant association with TB diagnosis (p = 0.03), with the highest prevalence among children < 5 years (18.5%) compared to 9.9% in 5–9 years and 8.6% in 10–14 years. HIV status was strongly associated with TB diagnosis (p < 0.001), with a prevalence of 41.3% among HIV-positive children (n = 46) versus 8.2% among HIV-negative children (n = 366). Nutritional status, assessed using weight-for-age Z-scores (WAZ), was significantly associated with TB (p < 0.001); malnourished children (WAZ < − 2; n = 93) had a TB prevalence of 27.9% compared to 7.8% among those with normal WAZ (n = 319). History of household TB contact was strongly linked to TB diagnosis (p < 0.001), with 29.9% prevalence among those with contact (n = 97) versus 6.3% without (n = 315). Incomplete BCG vaccination (n = 54) was associated with higher TB prevalence (29.6%) than complete vaccination (9.8%; n = 358; p = 0.01). Household overcrowding (≥ 5 persons per room; n = 186) showed a significant association (16.7% prevalence) compared to lower density (9.3%; n = 226; p = 0.04). Gender was not significantly associated with TB diagnosis (p = 0.42). No data on continuous variables such as exact age medians by TB status or treatment duration were available in the registers; thus, associations are reported categorically. Other variables extracted (type of TB, diagnostic modality, treatment outcomes) were not included in the primary analysis due to incomplete or non-routine recording.

Table 1.

Associations between demographic, clinical, and exposure characteristics and TB diagnosis among children screened

Variable	Category	n (%)	TB Diagnosis n (%)	TB Prevalence (%)	p-value
Gender	Male	220 (53.4)	30	13.6	0.42
	Female	192 (46.6)	21	10.9
Age group	< 5 years	130 (31.6)	24	18.5	0.03*
	5–9 years	142 (34.5)	14	9.9
	10–14 years	140 (34.0)	12	8.6
HIV status	Negative	366 (88.8)	30	8.2	< 0.001*
	Positive	46 (11.2)	19	41.3
Nutritional status (WAZ)	Normal (≥ − 2)	319 (77.4)	25	7.8	< 0.001*
	Malnourished (< − 2)	93 (22.6)	26	27.9
Household TB contact	No	315 (76.5)	20	6.3	< 0.001*
	Yes	97 (23.5)	29	29.9
BCG vaccination	Complete	358 (86.9)	35	9.8	0.01*
	Incomplete	54 (13.1)	16	29.6
Household density	< 5 persons/room	226 (54.9)	21	9.3	0.04*
	≥ 5 persons/room	186 (45.1)	31	16.7

*Statistically significant (p < 0.05). TB prevalence (%) = proportion diagnosed with TB within each category. Nutritional status classified using weight-for-age Z-scores (WAZ). Household density = persons per room. BCG status from immunisation records or scar presence

Univariable analysis

In univariable logistic regression, several factors showed significant crude associations with TB diagnosis. Younger age (< 5 years vs. 10–14 years reference) was associated with increased odds (crude OR [COR] = 2.4, 95% CI: 1.3–4.6, p = 0.03). HIV positivity showed markedly higher odds (COR = 7.9, 95% CI: 4.1–15.2, p < 0.001). Malnutrition (WAZ < − 2) was linked to higher odds (COR = 4.5, 95% CI: 2.5–8.1, p < 0.001). Household TB contact had the strongest crude association (COR = 6.3, 95% CI: 3.4–11.6, p < 0.001). Incomplete BCG vaccination (COR = 3.8, 95% CI: 2.0–7.3, p < 0.001) and household overcrowding (COR = 1.9, 95% CI: 1.1–3.5, p = 0.004) were also significant. Gender was not associated (p = 0.42). Caregiver education showed a significant crude association in initial bivariate tests (p = 0.03) and was retained for multivariable modelling.

Multivariable analysis

After adjustment for potential confounders in the multivariable logistic regression model, several factors remained independently associated with tuberculosis (TB) diagnosis among children. HIV positivity was the strongest predictor, with HIV-positive children having significantly higher odds of TB compared to HIV-negative children (adjusted OR [aOR] = 6.8, 95% CI: 3.4–13.6, p < 0.001). Malnutrition, defined as weight-for-age Z-score (WAZ) < − 2, was also a significant predictor (aOR = 3.7, 95% CI: 2.0–6.9, p < 0.001), indicating increased vulnerability among undernourished children. A history of household TB contact was strongly associated with TB diagnosis (aOR = 5.2, 95% CI: 2.7–10.0, p < 0.001), highlighting the importance of close exposure in disease transmission. Incomplete Bacillus Calmette–Guérin (BCG) vaccination was linked to higher odds of TB (aOR = 3.1, 95% CI: 1.5–6.4, p = 0.003). Younger age (< 5 years) remained a significant predictor compared to children aged 10–14 years (aOR = 2.0, 95% CI: 1.0–3.9, p = 0.04), while household overcrowding (≥ 5 persons per room) was also independently associated with increased odds of TB (aOR = 1.8, 95% CI: 1.1–3.1, p = 0.02).

Caregiver education (primary or less versus secondary or higher) showed a non-significant trend toward increased odds of TB (aOR = 1.7, 95% CI: 0.9–3.4, p = 0.08), suggesting a possible influence that did not reach statistical significance after adjustment.

These findings are presented in Table 2.

Table 2.

Multivariable logistic regression of factors associated with tuberculosis diagnosis among children

Variable	Category	COR (95% CI)	p-value	aOR (95% CI)	p-value
Age group	10–14 years	1 (Ref)	0.03	1 (Ref)	0.04
	5–9 years	1.2 (0.6–2.3)		1.3 (0.7–2.5)
	< 5 years	2.4 (1.3–4.6)		2.0 (1.0–3.9)
HIV status	Negative	1 (Ref)	< 0.001	1 (Ref)	< 0.001
	Positive	7.9 (4.1–15.2)		6.8 (3.4–13.6)
Nutritional status (WAZ)	Normal (≥ − 2)	1 (Ref)	< 0.001	1 (Ref)	< 0.001
	Malnourished (< − 2)	4.5 (2.5–8.1)		3.7 (2.0–6.9)
Household TB contact	No	1 (Ref)	< 0.001	1 (Ref)	< 0.001
	Yes	6.3 (3.4–11.6)		5.2 (2.7–10.0)
BCG vaccination	Complete	1 (Ref)	< 0.001	1 (Ref)	0.003
	Incomplete	3.8 (2.0–7.3)		3.1 (1.5–6.4)
Household density	< 5 persons/room	1 (Ref)	0.004	1 (Ref)	0.02
	≥ 5 persons/room	1.9 (1.1–3.5)		1.8 (1.1–3.1)
Caregiver education	Secondary or higher	1 (Ref)	0.03	1 (Ref)	0.08
	Primary or less	2.2 (1.1–4.3)		1.7 (0.9–3.4)

Abbreviations: COR, crude odds ratio; aOR, adjusted odds ratio; CI, confidence interval; Ref, reference category; WAZ, weight-for-age Z-score. Model adjusted for all listed variables (gender included but non-significant and omitted from final display for brevity)

Discussion

This study documented a prevalence of paediatric TB of 13.8% (57/412) among children screened in rural Zambezi District health facilities and higher than the national notified proportion of approximately 11% for children under 15 years and the 9.4% reported in a recent Zambian hospital-based study [9]. The elevated burden in this remote, high-poverty setting underscores how geographic isolation, limited diagnostic capacity, and socio-economic vulnerabilities amplify paediatric TB risk in rural high-burden areas [10].

Younger children (< 5 years) had significantly higher odds of TB diagnosis (aOR 2.0), consistent with biological vulnerability to rapid progression from infection to disease [14]. Recent African cohort and modelling studies confirm that children under five carry the highest risk of severe forms (e.g., miliary or meningeal TB), with progression risks up to 19–30% within two years following exposure in high-burden households [12, 14].

HIV infection emerged as a strong independent predictor (aOR 6.8), closely aligning with the aOR of 6.30 reported in the 2025 Zambian paediatric study [10] and global estimates that HIV increases TB risk 5–16-fold through impaired immunity [11]. In HIV-endemic rural settings like Zambezi, vertical and household transmission drive this association, with HIV primarily accelerating progression to active disease, particularly when antiretroviral therapy coverage or immune reconstitution is suboptimal [11, 14].

Malnutrition (WAZ < − 2) conferred nearly fourfold increased odds (aOR 3.7), consistent with strong associations observed in urban Zambia (aOR up to 10.38) [10] and broader evidence from LMICs showing bidirectional causality undernutrition impairs immunity while TB exacerbates nutritional deficits through appetite loss and metabolic demands [10, 12]. In rural Zambia, where food insecurity is widespread, integrated nutritional support within TB programmes is essential to interrupt this cycle [12].

Household TB contact was the strongest predictor (aOR 5.2), reinforcing the dominant role of close exposure in paediatric transmission, with meta-analyses and recent rural African studies (e.g., in pastoralist Kenya) reporting odds ratios exceeding 20 in high-exposure settings [12] Household overcrowding (≥ 5 persons/room; aOR 1.8) further facilitates airborne transmission, highlighting overcrowding as a modifiable structural driver in hyper-endemic rural districts [12, 15].

Incomplete BCG vaccination tripled the odds of TB diagnosis (aOR 3.1). Neonatal BCG provides robust protection against severe disseminated forms in infants, with recent meta-analyses estimating 60–80% efficacy against life-threatening TB (e.g., meningitis) in young children, though protection against pulmonary TB varies geographically (higher in some settings, lower in tropical Africa) [16, 17]. Recent systematic reviews support maintaining high neonatal coverage and exploring adjunct strategies in high-burden contexts [15]. Lower caregiver education showed a non-significant trend toward increased odds (aOR 1.7), likely mediated by delayed care-seeking and poorer adherence, as documented in sub-Saharan African health-literacy studies [18–20].

Strengths

This study has several important strengths. First, it used a complete census of all eligible children screened for TB across all 16 public health facilities in Zambezi District over five full years (2020–2024), which avoided sampling bias and gave a realistic picture of paediatric TB in a rural setting. Second, the multivariable logistic regression adjusted for multiple factors at once, helping to identify truly independent predictors while reducing the chance that one variable was masking the effect of another. Third, the key variables (HIV status, nutritional status using WAZ, household TB contact, BCG vaccination, age group, and household density) are routinely collected in Zambia’s national TB programme registers, making the results directly comparable to national data and other studies in similar settings. Fourth, by focusing on a remote, high-poverty rural district in North-Western Province, the work fills an important gap, most Zambian paediatric TB research comes from urban hospitals or Lusaka, so these findings add valuable context-specific evidence for decentralised TB control in underserved areas. Finally, the study followed STROBE guidelines for reporting observational research, which improves transparency and makes the methods and results easier to understand and build on.

Limitations

Several limitations should be kept in mind when reading these results. The cross-sectional design means we can show associations but cannot prove cause and effect or determine which came first, for instance, whether malnutrition increased the risk of TB or whether TB led to weight loss. The study relied entirely on routine health facility records, so errors or incomplete recording of nutritional status (WAZ), BCG scar presence, or household details could have introduced misclassification. Important information was often missing from the registers, such as the exact timing and intensity of TB exposure, antiretroviral therapy adherence and viral load in HIV-positive children, detailed socioeconomic measures, or use of preventive therapy, which may have left some confounding unaccounted for. The analysis only included children who actually came to public facilities for TB screening; many children with TB in the community who never reached a clinic are not represented, which could mean the true district burden is higher and the risk profile among unscreened cases might look different. Although the overall sample of 412 children was reasonable for detecting strong associations (such as those with HIV and household contact), it was still modest and may not have had enough power to pick up weaker effects or to explore subgroups in detail.

Conclusion

This study shows a clear and substantial burden of paediatric tuberculosis, 13.8% among screened children in rural Zambezi District, driven mainly by HIV infection, severe malnutrition, close household TB contact, incomplete BCG vaccination, young age under five years, and living in overcrowded conditions. These risk factors line up closely with what has been seen in other Zambian and African studies, but the rural focus highlights how poverty, distance from services, and structural issues make the problem worse in places like Zambezi.

The findings point to practical steps that could make a real difference; fully integrate TB and HIV care so every HIV-positive child is screened and offered preventive treatment, include nutritional support and ready-to-use foods as part of household contact tracing, actively find and screen all close contacts of adult TB patients, expand access to child-friendly diagnostics like GeneXpert in rural clinics, improve housing to reduce overcrowding where possible, and run community campaigns to help caregivers recognise symptoms early and seek care quickly. Implementing these actions in a coordinated way would help lower preventable cases and deaths from childhood TB in rural Zambia and support progress toward national and global TB elimination goals. Future work should use prospective designs with richer data on treatment, exposure, and social factors to better understand what works best and how to scale up effective interventions in similar resource-limited settings.

Author contributions

MCM conceived and designed the study, supervised data collection, performed the statistical analysis, and drafted the initial manuscript. MC provided technical guidance on study design and statistical analysis, contributed to the interpretation of findings, and critically revised the manuscript for important intellectual content. CH participated in data collection and management and assisted in drafting sections of the manuscript. KC contributed to the literature review, data entry, and editing of the manuscript, including final formatting. All authors read and approved the final manuscript.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due to restrictions from the Ministry of Health but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Ethics approval was obtained from the University of Lusaka Research Ethics Committee (Reference No: SMHS REC IORG0010092). Permission to access and use the data was also granted by the Zambezi District Health Office. This study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. The study utilized de-identified secondary data extracted from routine tuberculosis registers; therefore, individual informed consent was not required. All data were handled with strict confidentiality, and no personal identifiers were included during data collection, analysis, or reporting.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.