• Research

Better detection of tuberculosis in children thanks to decentralised diagnosis

Updated on:

Two evaluations from a study exploring the decentralisation of childhood tuberculosis diagnosis in district hospitals and primary health centres have recently been published in the eClinicalMedicine journal. The study was conducted in six countries with a high incidence of tuberculosis as part of the TB-Speed project, coordinated by the University of Bordeaux.

Photo : TB-Speed team in Mozambique to test for tuberculosis © TB-Speed
TB-Speed team in Mozambique to test for tuberculosis © TB-Speed

Tuberculosis affects more than a million children each year, yet less than half of them are diagnosed and treated. This under-diagnosis is responsible for over 200,000 deaths every year. Diagnosing tuberculosis in children is complex, requiring specific procedures for taking samples for testing, clinical expertise and proper interpretation of chest X-rays.

In countries with limited resources, these diagnostic capabilities are lacking at the peripheral levels of health services where the majority of children seek treatment.
In 2022, the World Health Organisation recommended the decentralisation of childhood tuberculosis services to improve access to care in peripheral health centres.

Two evaluations from a study exploring the decentralisation of childhood tuberculosis diagnosis in district hospitals and primary health centres have recently been published in the eClinicalMedicine journal. The study was conducted in six countries with a high incidence of tuberculosis (Cambodia, Cameroon, Côte d'Ivoire, Mozambique, Sierra Leone and Uganda) as part of the TB-Speed project implemented by the Institut Pasteur (Cambodia), IRD (Cameroon), PACCI (Côte d'Ivoire), Instituto Nacional de Saude (Mozambique), Soltis (Sierra Leone) and MU-JHU (Uganda) and coordinated by the University of Bordeaux.

The first publication, led by the Ugandan research institute MU-JHU (a research collaboration in Uganda between Makerere University and John Hopkins University), the the French National Research Institute for Sustainable Development (IRD) and the University of Bordeaux, assesses the effect of decentralising the diagnosis of childhood tuberculosis in district hospitals and primary health centres. The second publication, led by the University of Sheffield (UK) and the University of Bordeaux, assesses the cost-effectiveness and budgetary impact of this decentralisation.

The low detection rate of tuberculosis in children is largely explained by their difficulty in spontaneously producing sputum and the poor diagnostic yield of current microbiological tests, due to the paucibacillary nature of tuberculosis in children. Another important factor explaining this low detection rate is linked to the organisation of healthcare systems. In the majority of countries with a high incidence of tuberculosis and limited resources, tuberculosis care services are often centralised at regional level, inaccessible to most children. Methods of collecting respiratory samples adapted to children and rapid molecular detection tests are often lacking at peripheral levels of health services. Chest X-rays, essential for diagnosing tuberculosis that has not been microbiologically confirmed, are only available in referral hospitals. Their use is limited by the poor quality of the X-rays and the lack of training for clinicians in interpreting them for the diagnosis of tuberculosis in children. In peripheral health centres, health staff are often unaware of and reluctant to diagnose tuberculosis in children without microbiological evidence. To date, there is little data on why, how and where to decentralise these services.

Recent advances in diagnostics could help to improve the diagnosis of childhood tuberculosis in peripheral health centres. Training and clinical mentoring could help clinicians to start treatment without microbiological confirmation.

Operational assessment

The first publication¹ reports the results of a cross-sectional operations research study using a pre-post intervention design that evaluated the effect of decentralising a diagnostic approach to childhood tuberculosis in 12 district hospitals and 47 primary health centres in 12 districts in Cambodia, Cameroon, Côte d'Ivoire, Mozambique, Sierra Leone and Uganda.

The intervention included :

  • a patient-level diagnostic approach with symptomatic tuberculosis screening of all sick children and young adolescents presenting to the health centre, clinical assessment, Xpert Ultra testing of nasopharyngeal aspirate and stool samples, and digital chest X-rays for those with symptoms suggestive of tuberculosis ;
  • two decentralisation models, one strengthening diagnostic capacity at district hospital level, and where children presenting with tuberculosis symptoms at primary health centre level were referred to district hospitals for diagnostic evaluation, and the other strengthening diagnostic capacity at primary health centre level. In each country, two districts were randomly assigned to one of the decentralisation models. New diagnostic equipment was installed in health facilities; health workers were trained and engaged in adopting new practices; supportive supervision and clinical mentoring were put in place to help health professionals practising in rural and remote settings.

The researchers compared the proportion of children and young adolescents <15 diagnosed with tuberculosis between a period before the intervention (August 2018 to November 2019) and a period during the intervention (March 2020 to September 2021), overall and by decentralisation model.

Economic assessment

The second publication² focuses on assessing the cost-effectiveness and budgetary impact of decentralising a diagnostic approach to childhood tuberculosis in district hospitals and primary health centres compared with the standard of care in Cambodia, Cameroon, Côte d'Ivoire, Mozambique, Sierra Leone and Uganda.

A mathematical decision analysis model was developed to assess the cost-effectiveness and budgetary impact of the intervention from a health system perspective. Estimated outcomes included the number of children treated for tuberculosis, the number of deaths averted, the number of disability-adjusted life years averted, costs and cost-effectiveness ratios for the two interventions in each country. The budgetary impact of implementing these interventions at national level over a five-year period (2022-2026) was also calculated using cost functions and the results of decision models.

A nasopharynageal sample from a child in Mozambique to detect tuberculosis © TB-Speed
A nasopharynageal sample from a child in Mozambique to detect tuberculosis © TB-Speed

Main results

Decentralisation of the diagnostic approach to district hospitals and primary health centres almost tripled the detection of children diagnosed with tuberculosis compared with pre-intervention data. The decentralisation model based on district hospitals had a greater effect on tuberculosis detection than the model based on primary health centres. The effect varied from country to country, and was greater in countries where decentralisation of childhood tuberculosis services was non-existent or limited. The adoption of the various components of the diagnostic approach at district hospital and primary health centre level testifies to the efficacy of the intervention.

Relative to the standard of care, cost-effectiveness ratios ranged from 263 $ (Cambodia) to 342 $ (Côte d'Ivoire) per disability-adjusted life-year averted for the district hospital-based decentralisation model and from 477 $ (Cambodia) to 599 $ (Côte d'Ivoire) for the primary health centre-based decentralisation model. The additional costs of implementing the district hospital decentralisation model ranged from 12.8 M$ (10.8 to 16.4) (Cambodia) to 50.4 M$ (36.5 to 74.4) (Mozambique), and from 13.9 M$ (12.6 to 15.6) (Sierra Leone) to 134.6 M$ (127.1 to 143) (Uganda) for the primary health centre decentralisation model.

Decentralising the diagnosis of tuberculosis in children can improve the detection of tuberculosis cases and help to reduce infant mortality, but its implementation needs to take into account the low incidence of tuberculosis and the limited resources available at peripheral health centres. Sensitivity analyses suggest that decentralisation targeted at geographical areas where the prevalence of tuberculosis is very high is likely to be cost-effective. Decentralisation of diagnostic services for tuberculosis in children would require a substantial financial investment for implementation at national level, mainly due to the purchase of X-ray equipment and the Xpert MTB/RIF Ultra tests, particularly for decentralisation focused on primary health centres.

Bibliographic references

¹Effect of decentralizing childhood tuberculosis diagnosis to primary health center and district hospital level on disease detection in children from high tuberculosis incidence countries – An operational research with a pre-post study design

Link to the publication eclinm-D-23-02642  

²Cost-effectiveness and budget impact of decentralising childhood tuberculosis diagnosis in six high tuberculosis incidence countries: a mathematical modelling study

Link to the publication eclinm-D-23-02643  

Scientific contact

  • Olivier Marcy

    Epidemiologist and researcher at the Bordeaux Population Health Centre (BPH)
    Director of TB-Speed

    olivier.marcy%40u-bordeaux.fr

Scientific communication contact

  • Nicolas Koskas

    Communication manager

    nicolas.koskas%40u-bordeaux.fr

Find out more about the TB-Speed project

TB-Speed is a Unitaid-funded research programme that aims to reduce child mortality from tuberculosis by developing, testing and delivering an innovative, decentralised, cost-effective and feasible paediatric tuberculosis diagnostic strategy to increase case finding in children. The project is currently being carried out in seven countries in sub-Saharan Africa and South-East Asia. Several studies testing different diagnostic approaches in specific paediatric populations at risk of tuberculosis or in specific settings are being conducted within this framework. TB-Speed is coordinated by the University of Bordeaux.

Website