Tuberculosis in Children
N Engl J Med 2012; 367:348-361July 26, 2012
- Widespread implementation of the strategy of directly observed treatment short course (DOTS) during the 1990s resulted in improved global control of tuberculosis.1 However, its effectiveness has been limited in areas where poverty and infection with the human immunodeficiency virus (HIV) or drug-resistant tuberculosis are prevalent, and the emphasis on a positive sputum smear as the diagnostic criterion actually excludes most children from care.2 Tuberculosis remains a major but often unrecognized cause of disease and death among children in areas where the disease is endemic3; service delivery in such areas is hampered by the absence of pragmatic strategies to guide diagnosis and management. This article provides a brief overview of basic principles, current controversies, and recent advances related to the care of children with tuberculosis, with an emphasis on intrathoracic disease.
Disease Burden and Recent Epidemiologic ShiftsPoor ascertainment and reporting of cases of tuberculosis prevent accurate estimation of the global burden of disease from tuberculosis in children.4 Among the 4,452,860 new cases reported in 2010 by the 22 countries with the highest burden of disease from tuberculosis, only 157,135, or 3.5% (range, 0.1 to 15.0), were in children. Best estimates suggest that children (defined as persons younger than 15 years of age) account for approximately 11% of the burden of disease from tuberculosis,5 suggesting that just over 332,000 cases of tuberculosis in children went undiagnosed or unreported in these countries. Although overdiagnosis does occur, underdiagnosis is the rule in most areas where there is a high burden of disease and children with tuberculosis can access services only through referral hospitals. The problem of underdiagnosis in children is illustrated by the low pediatric caseload reported in four countries with a high disease burden, where rates exceeding 10% of all reported cases would be expected: Russia, 0.8%; India, 1.1%; Nigeria, 1.4%; and Brazil, 3.5%.1 In areas such as North America and Western Europe, where there is minimal internal transmission and routine provision of postexposure prophylaxis, a smaller proportion of children is affected, and most cases of childhood tuberculosis occur in immigrant populations.6,7
Coinfection with HIV has had a major epidemiologic effect, especially in sub-Saharan Africa. Apart from leading to an increase in the absolute number of patients with tuberculosis, it has induced a pronounced shift in the age and sex of patients toward young women of childbearing age.8 The effect of this demographic shift can be seen in the high rates of exposure to tuberculosis among infants born to mothers infected with HIV9 and in the high rates of tuberculosis among infants infected with HIV.10 Early initiation of antiretroviral therapy is the single most important intervention for reducing overall mortality and the risk of tuberculosis among HIV-infected infants,11 with isoniazid preventive therapy providing additional benefit.12
The emergence of drug-resistant tuberculosis poses a major threat to global tuberculosis control.13 The initial complacency in addressing the problem was influenced by studies indicating that the acquisition of isoniazid resistance reduced the pathogenicity of the strain.14 However, the development of multidrug-resistant tuberculosis (characterized by resistance to isoniazid and rifampin) in children exposed to persons with infectious drug-resistant tuberculosis,15 as well as its clonal spread in New York City16 and the Russian prison system,17 has provided clinical evidence of the transmissibility of multidrug-resistant strains. Additional proof was provided by an explosive outbreak of extensively drug-resistant tuberculosis (multidrug-resistant bacteria with additional resistance to a fluoroquinolone and a second-line injectable agent) among patients with HIV infection in South Africa.18 Although the incidence of drug-resistant tuberculosis among children is unknown, pediatric cases provide a valuable epidemiologic perspective, since they reflect ongoing transmission within communities. In places where the rates of drug-resistant tuberculosis in children have been monitored, the rates among children were similar to those among adults from the same community.15 The World Health Organization (WHO) estimated that in 2008, 3.6% of incident tuberculosis cases globally were of the multidrug-resistant or extensively drug-resistant type, which suggests that there was a similar burden of this type of disease among children.13
Natural History of DiseaseAn understanding of the natural history of tuberculosis is required to appreciate both the variations in susceptibility to disease and the diverse spectrum of clinical manifestations observed in children. Meticulous descriptions of tuberculosis in the literature published before the introduction of chemotherapy provide valuable insight into the sequence of events that follows primary infection with Mycobacterium tuberculosis (Table 1Table 1Clinical Syndromes Associated with Tuberculosis in Children. and Figure 1Figure 1Clinical Syndromes of Intrathoracic Tuberculosis in Children.).19,20 An important observation documented in these earlier studies was the presence of transient hilar adenopathy, and even excretion of M. tuberculosis, in children who had never had progression to disease.21 This finding poses a major problem in case definition for studies, such as vaccine efficacy trials, that use active case-finding strategies in populations of asymptomatic children who have been exposed to persons with infectious disease.22 The recent formulation of an international consensus on reference standards and uniform research methodology should facilitate progress.23-25
The sequence of events that follows reinfection (which is common in areas where tuberculosis is endemic) remains poorly defined. In cases of recurrent tuberculosis, strain typing makes it possible to differentiate relapse from reinfection but cannot be used to quantify the risk of reinfection. Composite data analysis suggests that there is a 79% reduction in the risk of disease progression among previously infected immunocompetent adults as compared with previously uninfected adults after documented exposure26; however, the epidemic contribution made by reinfection depends on the frequency of its occurrence in a particular environment.
It is important to differentiate infection from disease, since infection is a common event and the approaches to managing the two conditions are very different. Disease progression is usually indicated by persistent, nonremitting symptoms, although the rate of progression is variable.21 In the vast majority of cases (>90%), disease occurs within 1 year after the primary infection, with the youngest children at greatest risk for progression. The risk profile is bimodal, with adolescents being at increased risk.21 Exploring the mechanisms underlying the increased risk and the sudden switch in phenotype toward adult-type cavitary disease that occurs with the onset of puberty should provide new insights into the immunopathogenesis of tuberculosis.27
Approaches to DiagnosisChildren are usually evaluated for tuberculosis after presenting with symptoms or signs suggestive of disease (passive case finding) or as a result of contact investigation or routine immigration screening (active case finding). The clinical presentation of children whose infection is detected through active case finding differs from that of children whose infection is detected through passive case finding, with the former group often having infection but not disease or having disease in a very early phase. Among children in whom M. tuberculosis infection is detected, young children and those with recent exposure are at increased risk for progression to disease. Knowledge of the child's status regarding the likelihood of exposure changes the pretest probability of disease and the positive predictive value of subsequent investigations.
Clinical EvaluationTaking a careful patient history is essential for exploring the nature of the exposure and accurately characterizing the symptoms.28 The diversity of the clinical presentation and the nonspecific nature of most symptoms complicate diagnosis. Constitutional symptoms often include failure to thrive (deviation from the expected growth-curve trajectory) and reduced playfulness; low-grade or intermittent fever is seen less frequently.28 With airway involvement, the usual presenting symptom is a persistent, nonremitting cough or wheeze that is unresponsive to the treatment for likely alternative causes. Clinical signs are often subtle, and no diagnostic scoring system has been adequately validated29; the sensitivity and specificity of the clinical diagnostic approaches for tuberculosis are particularly poor in children with HIV infection.30
Imaging StudiesIn clinical practice, chest radiography is one of the most useful diagnostic studies. Both frontal and lateral views should be obtained, since a lateral view assists in the assessment of the mediastinal and hilar areas. The radiographic findings vary, but pronounced hilar adenopathy, with or without airway compression, is highly suggestive of tuberculosis. The International Union against Tuberculosis and Lung Disease compiled an atlas of illustrative cases.31 Unfortunately, the technical quality of the radiographs obtained in areas where tuberculosis is endemic is often poor or radiographic facilities are not available.
Ultrasonography is useful in confirming the presence of pericardial or pleural effusions and abdominal lymphadenopathy. High-resolution computed tomography (CT) offers excellent anatomical visualization,32 but because of the high cost of CT and the high level of radiation to which the patient is exposed, as compared with other forms of imaging, it should be reserved for complicated cases. Both CT and magnetic resonance imaging (MRI) are particularly helpful in visualizing the intracranial effects of disease, although MRI is more sensitive to the detection of brain-stem lesions and early perfusion defects in patients with tuberculous meningitis, and it also allows superior evaluation of the spine and soft tissues.33
Table 2Table 2Diagnostic Studies for Tuberculosis in Children. provides an overview of the laboratory examinations used in the diagnosis of tuberculosis. (See the Supplementary Appendix, available with the full text of this article at NEJM.org, for a list of references that includes recent comprehensive studies that focus on children.) Microscopical examination of sputum smears is the cornerstone of diagnosis in most countries, but its usefulness is limited in young children with paucibacillary disease who are unable to expectorate. Both the tuberculin skin test and the interferon-γ release assay fail to differentiate M. tuberculosis infection from active disease. The WHO recommends that the assay not be used in place of the tuberculin skin test,34 although the two tests may be complementary, improving the sensitivity or specificity of the assessment in specific clinical circumstances.35
Collecting specimens of spontaneously produced sputum in young children is problematic; gastric aspiration and sputum induction (with or without laryngopharyngeal suction) are feasible alternative methods of collection.36 The “string test” (which involves the use of an esophagogastroduodenal nylon yarn that can absorb swallowed sputum) works well in adults with HIV infection who have little sputum,37 and preliminary test results in children seem promising.38 Fine-needle aspiration biopsy is very useful in children with a peripheral lymph-node mass.39 Although the Xpert-MTB/RIF assay (Cepheid) is less sensitive than liquid cultures for the detection of M. tuberculosis in both children and adults, it provides results quickly, is highly specific, and detects resistance to rifampin. When two sputum samples are used, the assay detects three times as many cases as when microscopy is used40 but only about 70% of the cases when liquid culture is used.41,42 Currently, access to the Xpert-MTB/RIF assay is limited, but the efforts of the Global Laboratory Initiative, a working group of the Stop TB Partnership, should increase its availability.
Each of the diagnostic approaches described has limitations. However, when a combination of clinical, radiologic, laboratory, and histopathologic findings are consistent with a diagnosis of tuberculosis and there is epidemiologic evidence of exposure to tuberculosis or immunologic evidence of M. tuberculosis infection, an accurate diagnosis is possible in most cases.42
Principles of Disease ManagementAlthough every effort should be made to attain bacteriologic confirmation of disease, confirmation rates remain low, and treatment initiation should not be delayed in immunologically vulnerable children. Unfortunately, some tuberculosis-control programs will not initiate treatment without bacteriologic confirmation, citing the risk of adverse events from treatment and concerns about amplifying drug resistance. However, adverse events are rare in young children who are treated with first-line tuberculosis drugs, and they are at low risk for acquiring or transmitting drug-resistant tuberculosis. Despite differences between adult and pediatric tuberculosis (see Table S1 in the Supplementary Appendix), the principles of disease management are similar. The purpose of tuberculosis treatment is to cure the individual patient, whereas the intent of public health efforts is to terminate transmission and prevent the emergence of drug resistance. Rapidly metabolizing bacilli are quickly killed by bactericidal agents with high activity, thereby terminating transmission, ameliorating symptoms, and decreasing the risk of drug resistance (by reducing the population from which drug-resistant mutants emerge). The use of drugs with sterilizing activity is required to eradicate persistent subpopulations of intermittently metabolizing bacilli, thereby preventing relapse and effecting a long-term cure. Pragmatic disease classification should guide case management (Figure 2Figure 2Algorithm for the Diagnosis and Classification of Tuberculosis in Children.).The most important variables to consider in disease management are bacillary load and anatomical location. Drug resistance should be considered in children from areas with a high prevalence of drug-resistant tuberculosis and in those who have had documented contact with a person with drug-resistant disease, with someone who died during treatment for tuberculosis or who is not adhering to therapy, or with someone who is undergoing retreatment for tuberculosis. Young children with uncomplicated disease who are from areas with a low prevalence of isoniazid resistance can be treated with three drugs (isoniazid, rifampin, and pyrazinamide) during the 2-month intensive phase of treatment, followed by isoniazid and rifampin only during the 4-month continuation phase.43 However, children who have extensive or cavitary lung disease (either of which suggests a high bacillary load) or who are from areas with a high prevalence of isoniazid resistance should receive a fourth drug (ethambutol, which is safe in children of all ages) during the 2-month intensive phase of treatment.43 Table S2 in the Supplementary Appendix summarizes the mechanism of action, main adverse effects, and recommended pediatric dosages of drugs prescribed for the first-line treatment of tuberculosis.In the absence of drug resistance, the most frequent cause of a poor response to treatment is nonadherence to the regimen. Although empirical evidence of the value of DOT is limited, as a method of medication administration, it is preferable to unsupervised administration and to administration by a parent.44 In most instances, a recurrence of tuberculosis more than 12 months after treatment represents reinfection. Standard first-line treatment is appropriate in the absence of exposure to a person who is believed to have drug-resistant tuberculosis. Use of an escalated retreatment regimen that includes streptomycin is discouraged.43 When there is a poor clinical response in a patient with a history of adherence to treatment, a reevaluation of the diagnosis should be conducted, including consideration of the immune reconstitution inflammatory syndrome (IRIS) and drug resistance. Principles for the management of drug-resistant tuberculosis in children have been summarized elsewhere,45 and excellent outcomes have been reported.46Immune recovery after the initiation of antiretroviral treatment for HIV-coinfected individuals or nutritional rehabilitation may unmask subclinical disease or induce paradoxical deterioration, despite adequate treatment for tuberculosis. A finding of IRIS does not indicate treatment failure, and treatment should not be interrupted; patients with severe IRIS may require a course of glucocorticoids. Despite the risk of IRIS, data on adults indicate that antiretroviral therapy is most effective when initiated within 8 weeks after the start of tuberculosis treatment, or for patients with severely compromised immune systems, within 2 to 4 weeks after the start of treatment.47 The only exception would be patients with central nervous system tuberculosis, in whom IRIS can have devastating consequences.48 With HIV-associated tuberculosis, treatment should be given daily, and a prolonged course may be required, depending on the degree to which the patient's immune system has been compromised and the extent of disease.49,50
Prevention and ControlTransmission of tuberculosis within health care facilities is a particular concern in settings where immunologically vulnerable children may be exposed. In hospitals and clinics, careful consideration should be given to areas where patients are treated and to air-exchange patterns. It is also important to recognize that symptomatic parents or caregivers may pose transmission risks.51 Vaccination with bacille Calmette–Guérin (BCG) reduces the risk of disseminated (miliary) disease and tuberculosis meningitis in young children but offers no consistent protection against adult-type tuberculosis.52 No benefit of BCG vaccination has been established in HIV-infected children, and it is contraindicated in such children because of the risk of disseminated BCG disease.53 The development of a safe and effective vaccine remains a top priority among global health researchers.With good adherence, a 6-month course of isoniazid preventive therapy provides excellent protection against tuberculosis disease.54 Despite universal recommendations regarding the provision of preventive therapy and strong evidence of the greatly increased risk of tuberculosis and the increased mortality among children in close contact with persons who have tuberculosis,55 the implementation of preventive strategies remains poor. Pragmatic solutions are required to close the pronounced gap between policy and practice.56 Parents are often reluctant to provide preventive treatment for an otherwise well child, and the long duration of preventive therapy is a source of further discouragement. One study showed that a 3-month course of preventive therapy with isoniazid and rifampin was similar in efficacy to a 9-month course of isoniazid alone.57 A regimen of 12 doses of weekly rifapentine and isoniazid has been shown to be efficacious in adults,58 but this regimen is not yet recommended for children younger than 12 years of age because specific data on safety and efficacy in this age group are required. The efficacy of abbreviated regimens has not been well studied in children with HIV infection. A disadvantage of the regimens that include rifampin or rifapentine is the interactivity of these drugs with the protease inhibitors included in the antiretroviral therapy provided for infection with HIV49,50; rifabutin is less reactive, but its use in preventive therapy regimens has not been evaluated.Although the value of postexposure prophylaxis is universally acknowledged, the value of preexposure prophylaxis remains in question. Successive randomized, controlled trials of preexposure prophylaxis in children with HIV infection have had contradictory findings. The first of these trials, involving children with minimal access to antiretroviral therapy, was discontinued because of increased mortality in the placebo group.59 The reduction in mortality among those receiving isoniazid preventive therapy was confined to the first 2 to 3 months of treatment, raising the possibility that subclinical tuberculosis was present at trial entry. The second trial enrolled young infants (3 to 4 months of age) who had been exposed to HIV but had no known exposure to tuberculosis.60 The infants were randomly assigned to receive open-label isoniazid or placebo; those who were infected with HIV also received early antiretroviral therapy. All infants were closely monitored for subsequent exposure to tuberculosis. The investigators found no significant difference in the incidence of tuberculosis or mortality between the treatment and placebo groups, suggesting that preexposure prophylaxis against tuberculosis has little value if HIV-infected infants are enrolled in management programs early, with meticulous monitoring for tuberculosis exposure and provision of postexposure prophylaxis. However, the value of preexposure prophylaxis in areas where monitoring for tuberculosis exposure is likely to be poor remains unresolved.49,54
With the use of isoniazid preventive therapy after the completion of tuberculosis treatment in HIV-infected adults, it has been estimated that 83 recurrences can be prevented for every 1000 cases treated.12 The WHO recommends isoniazid preventive therapy for 6 to 36 months after the completion of tuberculosis treatment in all patients with HIV infection, including children who live in areas with a high prevalence of tuberculosis. However, the added value of preventive therapy as compared with ongoing screening for tuberculosis exposure and meticulous postexposure prophylaxis has not been evaluated.
It is possible to drastically reduce the morbidity and mortality associated with pediatric tuberculosis if case detection is improved and preventive therapy and curative treatment are made more accessible globally. Many challenges and research priorities remain (Table S3 in the Supplementary Appendix), but while we await the development of new vaccines, better diagnostics, and shorter treatment regimens, much can be achieved with pragmatic approaches and sensible application of existing tools.
No potential conflict of interest relevant to this article was reported.
We thank Mario Perez, M.D., for his valuable assistance with an earlier version of Figure 1.
Source InformationFrom Grupo Tuberculosis Valle-Colorado and Clínica León XIII, IPS Universidad de Antioquia, Medellín, Colombia (C.M.P.-V.); and the Sydney Emerging Infections and Biosecurity Institute and Children's Hospital at Westmead, University of Sydney, Sydney (B.J.M.).
Address reprint requests to Dr. Marais at the Clinical School, Children's Hospital at Westmead, Locked Bag 4001, Westmead, Sydney, NSW 2145, Australia, or at firstname.lastname@example.org.