It is estimated that between 10 and 15 million people residing in the United States have asymptomatic tuberculous infection.1 It is not known how many children have tuberculous infection because only three states (Indiana, Kentucky, and Missouri) mandate reporting, and nationwide surveys of tuberculin skin testing in children were discontinued in the early 1970s. Several recent surveys in large cities indicate that the tuberculous infection rate among some groups of children is much higher than had been expected. Virtually all pediatricians in general or subspecialty practice will encounter situations where tuberculin skin testing is indicated, and decisions must be made about the evaluation and treatment of a child with possible tuberculous infection.
The terminology used for the various stages of tuberculosis can be confusing. Tuberculous infection without disease is the preclinical stage of infection with Mycobacterium tubercubsis. The tuberculin skin test is positive but the chest radiograph is normal, and the child is free of signs or symptoms. A child with tuberculous infection without disease is a candidate for so-called preventive therapy, usually with isoniazid as a single drug. Tuberculous disease occurs when clinical manifestations of pulmonary or extrapulmonary tuberculosis become apparent, either by chest radiograph or by clinical signs and symptoms. Ine word tuberculosis usually refers to disease, not infection. The time interval between initial tuberculous infection and the onset of disease may be several weeks or many years. In general, an infected child with any radiographic or clinical manifestations consistent with tuberculosis is considered to have disease, even if no symptoms are present.
WHAT IS TUBERCULIN?
The word tuberculin is used to describe one or several antigens used in tuberculosis skin testing. The two antigens currently used are purified protein derivative (PPD) and Old Tuberculin. Both of these preparations are derived from supernatant extracts of cultures of M tubercuhsis. Old Tuberculin is the older and more crude preparation and its use is restricted to several of the multiple puncture skin tests. Purified protein derivative is anything but pure. The exact protein and antigen components of this solution have not been defined. It is likely that the components vary greatly among batches, which may account for some variability in testing.
Because of their antigenic similarity, nontuberculous mycobacteria (NTM) and M tuberculosis can cause a great degree of cross-reaction in skin testing.2 One of the earliest identified NTM strains was known as Battey bacillus (now known as Mycobacterium avium), and the antigen prepared from this strain was designated PPD-B. Skin tests placed on patients with pulmonary disease due to the Battey bacillus showed strong reactions to PPD-B, but only a weak reaction to PPD standard. Similar dual skin testing, when applied to patients with pulmonary tuberculosis, showed a strong reaction to PPD-S but a weak reaction to PPD-B. Nontuberculous mycobacteria antigens were available in the United States until the early 1970s, when they were removed from the commercial market due to the lack of standardization in their preparation and testing. In the mid-1980s, the Centers for Disease Control and Prevention conducted a clinical trial with a new set of NTM antigens; the results for NTM infections in children were encouraging, but these antigens have not yet been made commercially available due to an unacceptably high rate of adverse reactions (mostly very mild vesicular reactions).3
MULTIPLE PUNCTURE SKIN TESTS
Multiple puncture skin tests (MPTs) have been used widely in pediatrics because of the ease of administration even for relatively unskilled personnel. The Aplitest (Parke-Davis, Morris Plains, New Jersey) and the Tine test (Lederle Biologicals, Wayne, New Jersey) use metal prongs coated with dried antigen, usually PPE^ although one form of the Tine test uses Old Tuberculin as the antigen. The Mono-Vacc Test (Connaught Laboratories, Swiftwater, Pennsylvania) uses plastic prongs and liquid Old Tuberculin as the antigen. Although MPTs have become popular in pediatric practice, they have never been endorsed by public health experts because of several serious problems.
The exact dose of antigen introduced into the skin cannot be controlled precisely, so interpretation of the reaction size cannot be standardized completely. As a result, MPTs can never be considered diagnostic unless vesiculation occurs, and children with a positive result must have a subsequent Mantoux test.4
The Booster Phenomenon
The need foT a subsequent Mantoux test can lead to a further problem, called the booster phenomenon. Skin testing, by itself, does not induce delayed hypersensitivity to tuberculin. Boosting is an increase in reaction size caused by serial or repetitive testing in a person already sensitized to mycobacterial antigens. This phenomenon is caused by stimulation of a waned immunologic response to M tuberculosis or to crossreaction stimulated by infection with NTM or previous BCG immunization.5 The booster phenomenon can occur in tuberculin tests performed less than 10 days to 12 months or longer. Using an MPT on a patient susceptible to boosting by previous BCG vaccination or exposure to NTM in the environment can increase the reaction size to a subsequent Mantoux test to greater than the cutoff size for a positive test (eg, 12 mm) whereas a Mantoux test without a prior MPT in the same patient might have yielded a reaction size less than the cutoff (eg, 7 mm). In this situation, the use of an MPT may create a falsepositive result on the Mantoux test.
Variable Test Results
The MPTs have extremely variable and, in some studies and populations, very high rates of false-positive and false-negative results when compared with the results of Mantoux tests. Although some early studies demonstrated sensitivities of 95% to 99% and specificities of 98% to 99% for various MPTs compared with results of simultaneous Mantoux tests, other studies yielded false-positive rates of 10% to 20% and falsenegative rates greater than 10%.* In one of the few recent studies, conducted among Southeast Asian refugees in the United States at very high risk for tuberculosis, in which a 2-mm reaction size to a variety of MPTs was considered a positive result, the sensitivities of MPTs compared with the Mantoux test ranged from 68% to 96% and the specificities varied from 40% to 90%.6 The results of MPT testing will vary tremendously with the population studied.
The use of MPTs has led to the practice of asking parents to interpret the test results and report them by telephone or mail to the physician's office or clinic. No other screening test in pediatrics is routinely interpreted by nonprofessionals. This practice assumes parental adherence to a broad range of motivational and practical behaviors, an assumption that is unproven, unfounded, and certainly wrong in many cases. Many pediatricians do not adequately pursue the results of tests, assuming that nonreporting of results is equivalent to a negative result.
It is the opinion of virtually all tuberculosis experts in the United States that MPTs should no longer be used in pediatrics or any other subspecialty because of these major limitations. If a pediatrician continues to use these tests, it should be recognized that they are contraindicated for children who have had BCG vaccine, children who are known or suspected contacts of persons with infectious tuberculosis, or for testing children who are in groups at high risk for tuberculosis (especially in locales where tuberculosis is prevalent).
THE MANTOUX TEST
The Mantoux tuberculin skin test using five tuberculin units of PPD is the "gold standard" test. The tuberculin solution is injected intradermally on the volar surface of the forearm using a 27-ga needle and a tuberculin syringe. When done properly, a wheal 6 mm to 10 mm in diameter should result. If a smaller wheal or no wheal raises, the solution has gone in too deeply, where it quickly diffuses away; this may result in a false-negative test, and the test should be repeated immediately on the other arm.
Although some pediatricians maintain that the Mantoux test is too traumatic for the child or too difficult for the office staff to perform, experienced clinicians feel that with adequate training, the Mantoux technique is no more difficult or traumatic than placing a MPT A helpful technique is to anchor the side of one's hand that is holding the syringe against the side of the child's arm, and inject the solution in a transverse direction (Figure 1). With this technique, one maintains a firm grasp, can anticipate the child's tendency to move, and creates a fulcrum to guide proper entry of the needle. With practice, accurate application of the Mantoux test takes less than 60 seconds.
The tuberculin reaction is an expression of delayed hypersensitivity. In children infected with M tubercu' losis, the T-lymphocytes have been sensitized and macrophages activated. Some of these cells migrate toward the spot where tuberculin is deposited resulting in a firm and raised area of induration. This reaction usually reaches its peak in 48 to 72 hours when it is examined by inspection and palpation. In some individuals, a reaction occurs after 72 hours; this is considered a positive test. The indurated area is marked with a ball-point pen and measured in millimeters transverse to the axis of the administration of the tuberculin. In a more intense tuberculin reaction, the inflammatory process extends beyond the indurated area with redness and swelling (Figure 2). A reaction of this intensity almost always indicates infection with M tuberculosis.
Figure 1. The placement of the Mantoux skin test can be made easier if testers anchor their fingers as a fulcrum to guide the angle of the needle.
A negative Mantoux tuberculin skin test result never rules out tuberculous infection or disease in a child. A variety of factors can lower tuberculin reactivity (Table I).4 Approximately 10% of immunocompetent children with culture-documented tuberculosis do not react initially to five tuberculin units of PPD7; most become reactive after several months of treatment, suggesting that the disease, not factors intrinsic to the host, caused the anergy. Many adults co-infected with HIV and M tuberculosis have anergy to tuberculin, and it is likely that co-infected children frequently will be anergic as well.
False-positive reactions to tuberculin also can occur. Recent exposure to environmental NTM can result in cross sensitization and a false-positive reaction to a Mantoux test. This problem appears to be particularly important in the southeastern region of the United States, where historical studies with NTM skin test antigens showed high background rates of asymptomatic infection with NTM in the population.
Figure 2. A positive reaction to a Mantoux tuberculin skin test in a child with pulmonary tuberculosis. The central area of induration often is surrounded by a zone of redness and swelling.
The effect of the various BCG vaccines on subsequent tuberculin skin test reactions has been studied extensively. Unfortunately, the effect is variable, and there is no reliable method of distinguishing tuberculin reactions caused by BCG vaccination from those caused by infection with M tuberculosis. In various studies with different populations and characteristics, the proportion of previously BCG-vaccinated individuals with significant tuberculin skin test reactions has ranged from 0% to 90%.' In a large number of studies of children who received BCG vaccine, the mean reaction from a tuberculin skin test ranged from 0 mm to 19 mm, although most experts feel that reactions greater than 10 mm to 15 mm following vaccination are unusual.
Tuberculin reactivity then wanes rapidly over the next few years. Several studies have shown that less than 50% of infants given BCG vaccine shortly after birth will have a positive tuberculin skin test at 6 to 12 months of age, and virtually all children will have nonreactive skin tests by 5 years of age.8 A recent study from Canada investigated children who had received BCG vaccine either in infancy or as older children.9 Among children with a mean age of 11 years at skin testing, 4.9% vaccinated in infancy had a positive tuberculin reaction compared with 12.5% of those vaccinated as an older child. A similar study found that only 15% of US naval recruits who had received BCG vaccine 8 to 15 years previously had a positive tuberculin skin test at ≥10 mm.10
Severe reactions to a tuberculin skin test are extremely rare in individuals who previously have received BCG vaccine and are not infected with M tuberculosis. Prior BCG vaccination is never a contra' indication for tuberculin testing. In general, tuberculin skin test reactions should be interpreted in the same manner for persons who previously have received BCG vaccine as they are for persons who have not been vaccinated.4
Factors That Cause Decreased Response to Tuberculin
During the last decade, the approach to interpretation of the Mantoux tuberculin skin test reaction has changed significantly. The interpretation of the reaction should be influenced by the purpose for which the test was given and by the consequences of false classification.4 The appropriate size of induration indicating a positive reaction varies with the person being tested and related epidemiologic factors. The critical information is whether the child is in a population group likely to have been exposed to an adult with infectious pulmonary tuberculosis.
As a result of the contribution of epidemiology to skin testing, the Centers for Disease Control and Prevention (CDC) and the American Academy of Pediatrics (AAP) recommend varying cutoff points for a positive reaction in different groups (Table 2).4,11 For children at the highest risk for tuberculous infection - those who are contacts of adults with infectious tuberculosis, who have abnormalities on a chest radiograph or clinical evidence of tuberculosis, or who are HIVseropositive - a reactive area ≥5 mm is classified as a positive result. For other high-risk groups, including all infants, and for children living with adults in high-risk groups, a reactive area ≥10 mm is a positive result. For all other persons who are at low risk for tuberculosis, a reactive area ≥15 mm is considered a positive result.
Cutoff Size of Reactive Area for Positive Mantoux Tuberculin Reaction
Although this scheme is scientifically valid, it presents some problems for clinicians who care for children. First, classifying children by risk group requires the willingness and ability of the physician to obtain and the family to provide a thorough history, not only about the child, but also about the adults caring for the child, since it is really the risk factors of the adults that determine the risk for the child.
Second, within an office, clinic, or school, children with identical reactions may be evaluated differently depending on the risk factors for tuberculosis of the individual children; this can be difficult to adequately explain to medical staff, school administrators, and parents in a manner that precludes the perception of cultural, economic, or even racial bias.
Many public health officials have dealt with this issue by using induration classifications of only ≥ 5 mm and ≥10 mm as cutoff points, especially in communities with high tuberculosis case rates where most children are at higher-than-average risk. In communities where the tuberculosis prevalence is low and tuberculin testing of children is continued, increasing the cutoff size of the reactive area to ≥ 15 mm for truly low-risk children is a scientifically sound practice. The clinician must have a clear understanding of the tuberculosis case rates and characteristics within the community in order to make this decision. The geographically focal nature of tuberculosis in the United States means that each local public health authority should determine appropriate cutoff sizes for its community.
FINDING INFECTED CHILDREN
The most efficient method for finding children infected with tuberculosis is through contact investigations of adults with infectious tuberculosis.12 On average, 30% to 50% of household contacts of adults with infectious disease will have positive Mantoux skin test results. An associate investigation is the examination of all adults and children in contact with a child infected with M tuberculosis in order to find the source case and other infected contacts. Many health departments routinely perform associate investigations for children 6 years of age or younger who have a positive Mantoux skin test reaction. In general, children older than 6 years of age have had many more potential sources of infection, making associate investigations prohibitively expensive and timeconsuming. However, an individual clinician whose pediatric patient of any age has a positive Mantoux reaction should examine all household members with a tuberculin skin test. Only the Mantoux test should be used for contact or associate investigations.
All children with tuberculous disease or asymptomatic infection acquired the organisms from an infectious person. If health departments were able to do perfect contact and associate investigations, and if every immigrant to the United States were screened and appropriately treated for tuberculous infection, tuberculin testing programs for other children would not be necessary. The true purpose of testing programs for children is to find persons with asymptomatic tuberculous infection who were not found by these other means and who can be given isoniazid treatment to prevent further disease.
One limiting factor in designing tuberculin testing programs for children is the accuracy of the Mantoux tuberculin skin test, which varies with the risk of tuberculosis in the population being tested.13 The Mantoux skin test has both a sensitivity and specificity of approximately 90% at a 10-mm reaction size (Table 3). If the test is applied to a large population that has a 90% prevalence of tuberculous infection, the positive predictive value of the Mantoux skin test is 0.99, an excellent result. However, if the Mantoux test with the same characteristics is given to a group with an infection prevalence of only 1%, the positive predictive value drops to 0.08. The low prevalence group will have fewer positive results, but the vast majority of positive results will be falsely positive, most likely due to cross sensitization with NTM. Although false-positive test results usually are more acceptable in screening programs than false-negative results, the former create considerable expense and anxiety for both the family and the physician. In general, the limitations of the Mantoux tuberculin skin test preclude routine frequent testing of children at low risk for tuberculosis. Unless testing is performed strictly for epidemiologic purposes, it should be undertaken only when there are adequate resources to ensure that children with positive test results can be evaluated and treated.
Effect of Tuberculous Infection Rate on the Accuracy of Mantoux Tuberculin Skin Testing (1000 Subjects)*
Several recent studies suggest that tuberculous infection rates are high among certain groups of children in the United States. A study performed in several Boston public schools found that 5.1% of seventh-grade students and 8.9% of 10th graders had positive tuberculin skin tests. About 85% of the students with positive test results were foreign-born.14 Rom 1985 to 1989 in Los Angeles, the rate of positive tuberculin skin tests in US-born children was 2.8% for grades 1 through 12, compared with 20.5% for foreign-born students.15 A 1987 Mantoux skin test survey of kindergarten and first-grade students in Houston public schools showed that 5 . 1 % of children - 3.8% of US-born and 20.3% of foreign-born students - had a reactive area ≥10 mm.16 It is clear that in some cities, tuberculous infection rates among children are much higher than the 0.2% rate found among 5 and 6 year olds in the United States in 1970.1 These children are part of the pool from which future cases of tuberculosis will arise.
It should be emphasized that large-scale tuberculin testing programs for school-aged children will have little immediate impact on the incidence of tuberculous disease in children in the United States. The majority of children with reported tuberculous disease are below school age, and school -based programs will have virtually no impact on disease in these children. The vast majority of children who are infected between ages 5 and 14 years and do not get treated will not develop immediate primary tuberculous disease, although some will go on to develop pulmonary tuberculosis during adolescence or adulthood. The true purpose of testing school-aged children is to prevent future cases of tuberculosis among adolescents and adults, a noble goal, but not one that will lower the incidence of childhood tuberculosis. The most effective way to lower the incidence of childhood tuberculosis is to control the disease in adults.
The AAP currently recommends tuberculin testing with a Mantoux skin test every 1 to 2 years for children known to be at high risk for tuberculous infection.11 Periodic testing of high-risk children should be a priority for any practice, clinic, or community, especially those with high rates of tuberculosis in the general population or in identifiable groups. For children not in any of the recognized high-risk groups, the AAP suggests that no tuberculin testing is necessary. However, if the clinician feels uncomfortable with a total lack of testing, the AAP recommends testing at one or several stages of childhood - age 12 to 15 months, age 4 to 5 years, and during adolescence. One historical rationale for testing at 12 to 15 months was the fear that a measles vaccination could activate a dormant tuberculous infection. This concept is no longer considered valid, and the yield from tuberculin testing at this age - even in high-risk populations - is quite small. In low-risk populations, the vast majority of "positive" tuberculin skin tests in these children will be falsepositives.
Although the exact magnitude of the problem is not known, it is likely that there is a vast number of children currently residing in the United States with undiagnosed asymptomatic tuberculous infection. Most of these children come from easily recognizable groups known to be at high risk for tuberculosis, and our public health efforts should be aimed at finding, evaluating, and treating these children. The correct interpretation of the tuberculin skin test cannot be divorced from the epidemiology of the disease. It is apparent that tuberculin skin testing is most accurate among children at high risk for tuberculosis, and unfortunately, it is often inaccurate in children at low risk for the infection. The best strategy for the United States is to identify children truly at high risk for tuberculous infection and to test them with the best available test, which is the Mantoux tuberculin skin test.
1. Starke JR, Jacobs RF, Jereb J. Resurgence of tuberculosis in children. J Pediatr. 1992;12:839-855.
2. Hsu KHK, Jenkins DE1 Soriano LR. Trie tuberculin reaction associated with tuberculous infection. Am Rev Respir Dis. 1963;87:493-503.
3. Huebner RE, Schein MF, Cauthen GM, Geiter LJ, O'Brien RJ. Usefulness of skin testing with mycobacterial antigens in children with cervical lymphadenopathy, Pediatr ifa De}. 1992; 1 1 :450-456.
4. American Thoracic Society. Diagnostic standards and classification of tuberculosis. Am Rev Respir Dis. 1990;142:725-735.
5. Sepulveda RL, Burr C, Ferrer X, Sorensen RU. Booster effect of tuberculin testing in healthy 6-year-old school children vaccinated with Bacille Calmette-Guérin at birth in Santiago. Chile. Pediatr infect Dis ). 1988;7:578-581.
6. Cantanzaro A, Multiple-puncture skin test and Mantoux [est in Southeast Asian refugees. Chest. 1985;87.-34f5-?50.
7. Steiner P, Rao M, Victoria MS, Jabbar H. Steiner M. Persistently negative tuberculin reactions: their presence among children culture positive for Mycobacterium tuberculosis. Am J Dis Child. 1980;134:747-750.
8. Karalliede S, Katugha LP, Uragoda CG. The tuberculin response of Sri Lankan children after BCG vaccination at birth. Tubercle. 1987;68:33-38.
9. Menzies R, Vissandjee B. Effect of Bacille Calmette-Guérin vaccination on tuberculin reactivity. Am Rev Respir Dis. 1992;141:621-625.
10. Comstock GW1 Edwards LB, Nabangxang H. Tuberculin sensitivity 8 to 15 years after BCG vaccination. Am Rev Respir Dis. 1971:103:572-575.
11. American Academy of Pediatrics. Report of the Committee on Infectious Diseases. Elk Grove Village, 111: American Academy of Pediatrics; 1991.
12. Hsu KHK- Contact investigation: a practical approach to tuberculosis eradication. Am J Public Healm. 1963;53:1761-1769.
13. Snider DEJr. The tuberculin skin test. Am Rev Respfr Dis. 1982;125(suppl):108-118.
14. Barry MA, Shirley L, Grady MT, et al. Tuberculosis infection in urban adolescents: results of a school based testing program. Am} Public Health. 1990;80:439-441.
15. Davidson PT, Ashkar B, Salem N. Tuberculosis testing of children entering school in Los Angeles County, California. Am Rev Resprr Du. 1990;141(suppl):A336. Abstract.
16. Starke JR, Taylor KT, Martindill CA, PyIe NO1 Herrin CM. Extremely high rates of tuberculin reactivity among young school children in Houston. Am Rev Resptr Dis. 1988;137(suppl):22. Abstract.
Factors That Cause Decreased Response to Tuberculin
Cutoff Size of Reactive Area for Positive Mantoux Tuberculin Reaction
Effect of Tuberculous Infection Rate on the Accuracy of Mantoux Tuberculin Skin Testing (1000 Subjects)*