Revealing Gaps in Our Understanding of Finding Children With TB and Our Ability to Inform Policy

Tuberculosis (TB) has only recently been officially recognized as a top 10 cause of age under-5 mortality globally.¹ The vast majority of the 216 000 children who die each year from TB are never diagnosed with TB.^2,3 TB often presents differently in children than adults, and microbiological tests have low sensitivity in children,⁴ making case detection a major problem. Therefore, identifying efficient screening algorithms to find children with TB disease is both important and challenging.

In this issue of Pediatrics, Robsky et al⁵ present the findings from a systematic review of active case finding (ACF) efforts other than contact tracing to identify children with TB disease. They focus on the measure of “number needed to screen” (NNS) to identify a single case of TB disease with different ACF strategies.

Their most striking, but unfortunately not surprising, takeaway is the lack of high-quality studies to inform these NNS estimates. Despite a thorough search of >27 000 publications and inclusion of 31 studies, it was not possible to stratify by several critical factors when estimating NNS. However, to inform policy regarding the settings and characteristics of children that should be the focus of screening programs, we need large, high-quality, setting-specific studies, using varying screening tools and algorithms, that can sufficiently scrutinize patient-level characteristics. Therefore, this review is most helpful in identifying opportunities for future studies to fill some important data gaps.

Robsky et al found that ACF among all children aged <5 years and children living with HIV decreased the estimated NNS substantially, which is consistent with the fact that these 2 groups are known to have high rates of TB in high-burden settings. However, these are also precisely the groups of children who are less likely to have microbiological confirmation (required in only 8 studies), and because of the fact that they are at the highest risk of poor outcomes and mortality, they will likely be overdiagnosed as a precaution to avoid these poor outcomes.⁶ This raises the specter of the burden of false positives, because it is unlikely that all of the children diagnosed with TB actually had TB. The nature of this systematic review precluded using a standardized case definition of childhood TB when microbiologic confirmation was negative or not attempted, so it is not reported how the children in the included studies were diagnosed. A focus on these children must therefore acknowledge the risks of overdiagnosis in the forms of stigma, effect on the child’s education, disruption to children and families, adverse drug effects, and expense. Although some degree of overtreatment has been deemed acceptable,⁷ the importance of more studies that are sufficiently powered to identify and stratify by presence and absence of specific clinical findings and microbiological diagnosis would help us minimize both over- and underdiagnosis.

The local setting can have a substantial effect on the yield from a screening algorithm. Clearly, children in high-burden settings are more likely to have TB than those in low-burden settings. The majority of included studies were from medium- and high-burden settings. However, there can be order-of-magnitude differences between the per-capita incidence in medium- and high-burden settings,² not just at the national level, but also subnationally, given the substantial heterogeneity that we know exists within 1 country.^8,9 It would be highly beneficial if estimated NNS’s could reflect this more precisely. In addition, the extent and effectiveness of contact tracing in the local area would affect yield from additional ACF strategies, another aspect that would vary by setting and 1 that could not be analyzed in this systematic review. Finally, the role of the family, household, and social structure could affect yield from a screening strategy. Although these types of data are hard to collect, they could be important, especially for younger children when we consider child care versus home care, and household size for the children infected at home, even if the majority are infected in the community.¹⁰

An important finding of this systematic review is that lower NNS’s were identified when ACF was conducted in both inpatient and outpatient health care settings compared with community settings. Increased screening for TB in health care settings in high-TB-burden areas could substantially increase pediatric TB diagnoses and reduce deaths.¹¹ Unsurprisingly, symptom screening was the primary modality used in the identified studies and there were too few studies using other screening methods (eg, chest radiography; the rapid, sputum-based TB PCR test called GeneXpert; being underweight) to draw specific conclusions about their effectiveness. We need more studies that assess ACF using expanded screening algorithms, given the wide range of extrapulmonary forms and presentations among children with TB.

Current TB laboratory diagnostics have known poor sensitivity in children⁴ and radiography has lower-than-desired specificity.¹² There are several promising diagnostic techniques in development that might increase our ability to correctly diagnose TB in children.¹³ However, even with improved diagnostics, the other piece in the puzzle of finding children with TB is identifying children who truly need to be evaluated and tested in the first place. Although the ostensibly highest yield is contact tracing in a household and community where an adult is known to have TB,¹⁴ this activity assumes 2 things: first, that we know about all the adults that have TB, and second, that the resources exist to conduct these investigations in a timely fashion, neither of which might be feasible. To wit, case detection globally among adults was estimated to be only 60% in 2021² and was lower in many high-burden settings. Identifying additional and alternative case-finding strategies and appropriate settings to use them would inform policy to identify more children with TB and ultimately save lives. However, without sufficient data from high-quality studies to inform these policies, children with TB will continue to suffer and be neglected.

Glossary

ACF

active case finding

NNS

number needed to screen

TB

tuberculosis