Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
March 2006

Risk of Active Tuberculosis Among Schoolchildren in Hong Kong

Author Affiliations

Author Affiliations: Tuberculosis and Chest Service, Department of Health (Drs Leung, Chang, Tam, Chan, Law, Wong, and Lee, and Ms Leung), and Tuberculosis and Chest Unit, Grantham Hospital (Dr Yew), Hong Kong.


Copyright 2006 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2006

Arch Pediatr Adolesc Med. 2006;160(3):247-251. doi:10.1001/archpedi.160.3.247

Objective  To estimate the risk of active tuberculosis (TB) and its implication on preventive treatment among BCG-vaccinated schoolchildren.

Design  Cohort and case-control designs.

Setting  Community settings in a high-prevalence area.

Participants  Children in primary school. The main exposure was their tuberculin response.

Main Outcome Measures  Of 94 928 primary schoolchildren tuberculin tested during a routine school revaccination program in 1999, 656 with a tuberculin response at 20 mm or more were followed up prospectively through the territory-wide TB registry up to December 31, 2003, for the development of TB. In a separate case-control analysis, the tuberculin responses of children who subsequently had active TB (at the age of 10-15 years) were compared with those of their sex- and age-matched classmates to ascertain the relative risks of TB for different tuberculin reaction categories. The absolute and relative risks were applied to the 1999 cohort for estimating the incidence of TB among different tuberculin reactors.

Results  The annual incidence (95% confidence interval) of active TB was estimated to be 13.4 (5.6-40.6) per 100 000 for the entire cohort and 7.5 (2.4-24.5), 7.5 (1.7-32.0), 16.0 (4.4-57.2), 92.6 (26.6-320.2), and 340.6 (163.3-626.4) per 100 000 for children with a tuberculin reaction at 0 to 4, 5 to 9, 10 to 14, 15 to 19, and 20 mm or more, respectively. By using 10 mm as the cutoff, 482 (95% confidence interval, 163-1391) children have to be treated to prevent a single case of active TB within 5 years. Treatment will cover 17.5% of the cohort, but prevent only 54.1% of all active TB cases.

Conclusion  It is desirable to reexamine the existing screening method for BCG-vaccinated children from high-prevalence countries.

Tuberculin skin testing has been used for a long time as a diagnostic test for latent tuberculosis (TB) infection. Tuberculin-positive children from high-prevalence countries are regularly given treatment for latent TB infection in the United States,1 while such treatment is seldom recommended in the absence of other risk factors like recent contact or human immunodeficiency virus infection in their home countries. It would be desirable to quantify the actual risk of TB among these children, so that the appropriateness of the recommendation can be examined in light of better scientific evidence. While a prospective cohort might be the ideal approach to tackle this issue, the relatively low incidence of disease2 among children poses a significant problem in relation to sample size and follow-up.

In many countries with a high prevalence of TB, neonatal BCG vaccination is regularly performed.3 For the past 30 years, BCG (Glaxo strain) has been given to all newborns in Hong Kong, with a coverage rate of about 99%.4 BCG revaccination was also practiced among primary schoolchildren (aged mainly from 6-12 years) until its cessation in 2000. All primary schools in Hong Kong were covered. The BCG team (a group of well-experienced inoculators) of the Tuberculosis and Chest Service used to visit each primary school once every 4 years in rotation to conduct tuberculin skin testing with 1 U of purified protein derivative (PPD) RT-23. Schoolchildren with a tuberculin reaction of 9 mm or less were offered revaccination. In general, treatment for latent TB infection was not offered. The revaccination program was stopped in 2000 on the advice of the local Advisory Committee on Immunization. A postcessation surveillance program was launched from January 1, 2001, to June 30, 2003, to record the BCG revaccination history of all TB patients aged 10 to 15 years on notification. With data available from these regular services, we engaged in a 2-part study, combining the benefits of cohort and case-control design, to quantify the risk of active TB among schoolchildren in Hong Kong.


From the paper records of the School Revaccination Program, all schoolchildren with a tuberculin response at 20 mm or more in 1999 were retrieved, and cross matched with the computerized statutory TB notification registry in Hong Kong every 6 months until December 31, 2003, using name (in English), sex, and date of birth as identifiers. The medical and public health records of all identified TB cases were retrieved from relevant chest clinics or hospitals for verification of identity and confirmation of the diagnosis. An active case of TB was defined as disease proved by isolation of Mycobacterium tuberculosis or, in the absence of bacteriological confirmation, disease diagnosed on clinical, radiological, and/or histological grounds together with an appropriate response to treatment. The incidence of active TB was quantified, with the 95% confidence interval estimated by assuming a Poisson distribution in the occurrence of the cases.


The school BCG records of active TB patients aged 10 to 15 years, from January 1, 2001, to June 30, 2003, were retrieved together with those of their classmates. Each case was matched to 8 controls by class, sex, and age to control for the potential confounding effects of these variables, and to maximize the statistical power to detect any significant differences among multiple tuberculin reaction categories. All students who had not been tested with a tuberculin skin test during the school BCG revaccination were excluded. Univariate conditional logistic regression analysis was used because of the matched design to estimate the odds ratios for active TB among different tuberculin reaction categories.5 A commercially available software program (SPSS version 10; SPSS Inc, Chicago, Ill) was used for analysis. P<.05 was considered statistically significant.

The odds ratios obtained in part 2 were multiplied to the absolute incidence derived for the strongest reactors in part 1 to obtain the annual risks of active TB for other tuberculin reaction categories. These were then applied to the 1999 cohort to estimate the numbers needed to treat to prevent a single case of active TB. For simplicity, only those active TB cases occurring within the first 5 years were considered, and the protective efficacy of treatment was assumed to be 100%.

The study has been approved by the Ethics Committee of the Department of Health of Hong Kong.


In 1999, a total of 94 928 primary school students were successfully tested for tuberculin response in the School Revaccination Program. The distribution of tuberculin skin test results is summarized in the Figure. Of those with a tuberculin test reaction of less than 10 mm, 99.7% were given a BCG revaccination. Of 662 students (0.7%) with a tuberculin reaction at 20 mm or more, 6 with missing data on date of birth were excluded, leaving 656 students for analysis. All except 6 (0.9%) were Chinese. Approximately 54% were females. The mean (SD) age was 9.80 (2.13) years. The mean (SD) duration of follow-up was 4.48 (0.26) years.

Image not available

Distribution of tuberculin response among 94 928 children in primary schoolchildren in Hong Kong.

Up to December 31, 2003, 10 active TB patients (7 pulmonary, 1 extrapulmonary, and 2 combined) were identified. Four cases were confirmed by culture, including one with a positive sputum smear result. The diagnosis of the rest was verified by their clinicoradiological picture and subsequent response to treatment. There were 7 female and 3 male patients (Fisher exact test, 2-tailed; P = .36). The median age at disease onset was 12.5 (range, 9.5-13.6) years. The median interval from tuberculin skin testing to the development of disease was 286.5 (range, 20-1649) days. All of them had received a BCG vaccination at birth. Three reported a history of TB contact (2 household and 1 nonhousehold) within 5 years. All enjoyed good health in the past, except for one with Down syndrome. The incidence rate of active TB was 340.6 (95% confidence interval, 163.3-626.4) per 100 000 person-years among schoolchildren with a tuberculin reaction at 20 mm or more.

From January 1, 2001, to June 30, 2003, a total of 103 active TB patients aged 10 to 15 years were reported to the Department of Health. Fourteen had not met the school BCG team because of recent immigration or cessation of the School Revaccination Program. Five students missed the tuberculin skin testing because of parental refusal or absence from school, and the school BCG records of 1 student could not be retrieved, leaving 83 patients with complete tuberculin skin testing records (from December 8, 1992, to March 27, 2000) for analysis. The mean (SD) age of these 83 patients at notification of disease was 14.18 (1.54) years, and the mean (SD) interval from tuberculin skin testing to disease was 5.21 (2.00) years. All 83 patients (59 pulmonary, 20 extrapulmonary, and 4 both) were new cases. Only 16 (19.3%) were positive for TB by sputum smear result, and 30 (36.1%) were confirmed to have TB by culture result. The rest met the definition for active TB by their clinicoradiological picture and subsequent response to treatment. The neonatal BCG vaccination history was available for only 75 patients. Seventy-four reported BCG vaccination and 1 did not. BCG scars were checked in 76 patients, with 16 showing 0 scars, 45 showing 1 scar, and 15 showing 2 scars. Matching the 83 patients by class identified 2505 classmates in 79 different schools. Eighty-four students did not participate in the regular revaccination program, leaving 2421 with tuberculin skin testing results.

The proportions of males were 39.8% and 49.1% among patients and their classmates, respectively (P = .10). The mean (SD) age at tuberculin skin testing was 8.97 (1.88) years for patients and 8.90 (1.79) years for their classmates (unpaired t test, P = .72). The mean (SD) tuberculin test response was 9.64 (7.95) mm for patients and 4.09 (5.13) mm for their classmates (unpaired t test, P<.001). All students with a tuberculin response at 9 mm or less had been revaccinated.

A total of 649 sex-, age-, and class-matched controls were successfully matched to the 83 patients in a ratio of 8:1, except for 4 patients in special classes (for whom 2, 4, 5, and 6 controls were matched, respectively). The selected controls differed significantly from those disease-free classmates not chosen as controls by sex (39.4% vs 52.6% males; χ2 test, < .001) and age (mean ± SD, 9.05 ± 1.95 vs 8.85 ± 1.74 years; unpaired t test, P = .02), but there was no significant difference in their distribution of tuberculin response (χ2 test, P = .17).

The cases and matched controls were comparable by sex (39.8% vs 39.4% males; χ2 test, P = .96) and age (mean ± SD, 8.97 ± 1.88 vs 9.05 ± 1.95 years; paired t test, P = .75). The distribution of tuberculin responses between cases and controls is summarized in Table 1. Table 2 shows the results of univariate conditional logistic regression analysis of the tuberculin responses for all identified active TB cases and their controls. Similar results were obtained when analysis was restricted to the 30 culture-confirmed cases and their matched controls.

Table 1. 
Image not available
Comparison of Tuberculin Responses Between Active TB Patients and Their Sex-, Age-, and Class-Matched Controls During the School BCG Revaccination Program*
Table 2. 
Image not available
Effect of Tuberculin Responses on Relative Odds of Active TB in Univariate Conditional Logistic Regression Analysis*

On applying the incidence rate derived in part 1 and the odds ratios of Table 2 to the 94 928 primary schoolchildren tuberculin tested in 1999, the respective incidences of active TB among different tuberculin reactors are summarized in Table 3. Table 4 lists the proportions of the student population to be covered and the number of active TB cases preventable within 5 years by using different cutoffs, based on the estimations shown in Table 3.

Table 3. 
Image not available
Estimated Incidence of Active TB Among Children in Primary School Tested With Tuberculin Skin Test in 1999
Table 4. 
Image not available
Effect of Using Different Cutoffs in Tuberculin Skin Testing on Proportions of Schoolchildren Eligible for Treatment of Latent TB Infection and Potential TB Cases Preventable Within 5 Years

In this study, tuberculin reaction size was a strong predictor of active TB. The overall pattern was similar to the J-shaped pattern reported in northern Malawi, apart from the absence of a definite trough among those with low-grade sensitivity.6 The risk started to increase for the 10- to 14-mm group, and increased sharply for a reaction size of 15 mm or more, supporting previous studies7 that the latter is more indicative of infection among subjects with a previous BCG vaccination. By using 10 mm as the cutoff, 482 children have to be treated to prevent a single case of active TB within 5 years (Table 4). Treatment will cover 17.5% of the cohort, but prevent only 54.1% of all active TB cases. If the cutoff is increased to 15 mm, only 141 children need treatment to prevent a case of active TB. Only 3.6% of children will be subjected to treatment, but an additional 16.6% of cases will be missed.

Among schoolchildren incidentally found to have a tuberculin reaction at 20 mm or more, the annual incidence of active TB was as high as 341 per 100 000. However, their risk of disease falls far short of the figure of 12.9% within the first year,8 and 1.6% per year from the second year to the seventh year1 for unvaccinated tuberculin-positive TB contacts. Among those children with a tuberculin reaction between 15 and 19 mm, the risk was 92.6 per 100 000, which was comparable with the figure of 90.4 per 100 000 reported among similar vaccinated children in Singapore.9 Such risk is just a quarter of those with a tuberculin reaction at 20 mm or more. Because most of both groups were likely to have been infected, more recent infection might have contributed to the greater risk among those with the strongest tuberculin reaction.

Children from birth to the age of 14 years have a lower risk of active TB compared with those at other ages.2 In 2001, the TB notification rates were 5.4, 3.5, 13.5, and 52.5 per 100 000 for those aged 0 to 4, 5 to 9, 10 to 14, and 15 to 19 years, respectively. A question naturally arises whether tuberculin-positive children at the mean age of 9 years may have a higher risk of developing active TB in the late teens. However, the risk of developing disease has been reported to decrease substantially with the passage of time after infection.8 In high-prevalence areas, substantial ongoing transmission is expected, and the protective effect of isoniazid treatment has been well reported to decrease with time, at least among human immunodeficiency virus–infected subjects.10,11 Among a cohort of tuberculin reactors (aged 1-18 years) detected during a BCG trial from 1949 to 1951, the annual risk of developing active TB (averaged over 20 years) was also lowest among those aged 7 to 12 years at recruitment.12 Even in low-prevalence areas, 55% of immigrants with TB had the condition diagnosed within the first 5 years of arrival.13 Considering the less than 100% protective efficacy of the recommended treatment, compounded by often poor adherence and sometimes isoniazid resistance, in field conditions,14,15 the numbers needed to treat to prevent an active TB case are unlikely to deviate excessively from those previously estimated.

One unit of PPD RT23 was used in this study, not the 5 U of PPD standard commonly used in the United States. Stronger reactions might be expected if 2 U of PPD RT23 (the equivalent of 5 U of PPD standard16,17) were used. This would increase the percentage of children to be covered with each cutoff, and the number of children to be treated to prevent a single case of active TB. Because children with a tuberculin reaction at 10 mm or more were not revaccinated, BCG revaccination would not confound the results for any cutoffs more than 9 mm, even if there was any protective effect.

Undernotification is always a point of concern for all notifiable diseases.18 In a local audit of TB notifications by reviewing discharge codes, microbiological reports, or hospital reports in 1995, the undernotification rates were 3.0%, 27.6%, and 31.4% for chest clinics, chest hospitals, and general hospitals, respectively.19 After a series of improvement measures, notifications from sources other than chest clinics increased drastically,20 in line with what was previously observed in 2 London, England, hospitals.21,22 With incomplete notification, some degree of underestimation is likely, but it would be expected to affect all tuberculin reactor categories. Because tuberculin testing was done during primary school age, the bulk of the 1999 cohort should be in the age range of 10 to 14 years from the time of tuberculin testing to the end of 2003. The estimated incidence of active TB for the study cohort was 13.4 per 100 000 person-years (Table 3), while the reported incidences from the territory-wide TB notification registry were 13.5, 8.5, and 10.3 per 100 000 person-years for children aged 10 to 14 years in 2001,2 2002,23 and 2003,24 respectively. Even if we apply the relative risks derived from part 2 directly to the notification figures for the same years, there will not be any substantial changes in the various estimates in Table 3 and 4.

Considering the sharp contrast in disease risk between those with a tuberculin reaction of less than 15 mm and those with a reaction of 15 mm or more (Table 2), it might be worthwhile to reexamine the existing tuberculin test screening criterion in children with clear documentation of BCG vaccination, at least outside the contact settings. A tuberculin reaction is known to wane rapidly after neonatal BCG vaccination.25,26 However, in a study27 on 2588 randomly selected Saudi children, while the tuberculin-positive rate was not significantly different between vaccinated and unvaccinated children at the age of 5 years, the tuberculin sensitivity increased more steeply with age in the BCG-vaccinated than in the unvaccinated children, so that the difference between the 2 groups became statistically significant in those aged 12 and 13 years. Because booster phenomenon is also well reported in BCG-vaccinated children,28 the possibility of interaction with ongoing environmental antigenic stimulation cannot be completely excluded. An interferon γ assay measuring the production of interferon γ in T lymphocytes on stimulation with PPD is not expected to differentiate false-positive cross-reactors from those with genuine infection.29 However, a modified version of that test using ESAT-6 and CFP-10, antigens not present in BCG, has recently been reported to show a high degree of sensitivity and specificity for culture-confirmed TB, which is unaffected by previous BCG vaccination.30,31 The modified test also showed a better correlation with the degree of exposure than the tuberculin test in a mass contact screening exercise.32 It is conceivable that an adapted version of the test may help to sort out the truly infected children among BCG-vaccinated immigrants from high-prevalence areas. More systematic studies are, therefore, warranted in this direction.

Back to top
Article Information

Correspondence: Chi Chiu Leung, MB, Wanchai Chest Clinic, 99 Kennedy Rd, Wanchai, Hong Kong (

Accepted for Publication: September 14, 2005.

Author Contributions: Dr Leung had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Financial Disclosure: None.

American Thoracic Society, Targeted tuberculin testing and treatment of latent tuberculosis infection MMWR Recomm Rep 2000;49(RR-6)1- 51
Hong Kong Tuberculosis and Chest Service, Annual Report of Tuberculosis and Chest Service for 2001  Hong Kong Dept of Health2002;
WHO, Expanded Program on Immunization: immunization schedules in the WHO Western Pacific region, 1995 WHO Wkly Epidemic Rec 1996;71133- 138
Leung  CCTam  CMChan  SL  et al.  Efficacy of the BCG revaccination program in a cohort given BCG vaccination at birth in Hong Kong Int J Tuberc Lung Dis 2001;5717- 723
Campbell  MJ Statistics at Square Two: Understanding Modern Statistical Applications in Medicine  London, England BMJ Books2001;51- 55
Fine  PESterne  JAPonnighaus  JMRees  RJ Delayed-type hypersensitivity, mycobacterial vaccines and protective immunity Lancet 1994;3441245- 1249
American Thoracic Society, Diagnostic standards and classification of tuberculosis in adults and children Am J Respir Crit Care Med 2000;1611376- 1395
Ferebee  SH Controlled chemoprophylaxis trials in tuberculosis: a general review Bibl Tuberc 1970;2628- 106
Chee  CBSoh  CHBoudville  IC  et al.  Interpretation of the tuberculin skin test in Mycobacterium bovis BCG-vaccinated Singaporean schoolchildren Am J Respir Crit Care Med 2001;164958- 961
Quigley  MAMwinga  AHosp  M  et al.  Long-term effect of preventive therapy for tuberculosis in a cohort of HIV-infected Zambian adults AIDS 2001;15215- 222
Woldehanna  SVolmink  J Treatment of latent tuberculosis infection in HIV infected persons Cochrane Database Syst Rev 2004; ((1)) CD000171
Comstock  GWLivesay  VTWoolpert  SF The prognosis of a positive tuberculin reaction in childhood and adolescence Am J Epidemiol 1974;99131- 138
McKenna  MTMcCray  EOnorato  I The epidemiology of tuberculosis among foreign-born persons in the United States, 1986-1993 N Engl J Med 1995;3321071- 1076
International Union Against Tuberculosis, Committee on Prophylaxis Efficacy of various durations of isoniazid preventive therapy for tuberculosis Bull World Health Organ 1982;60555- 564
Ferebee  SHMount  FW Tuberculosis morbidity in a controlled trial of the prophylactic use of isoniazid among household contacts Am Rev Respir Dis 1962;85490- 510
Teixeira  LMaciel  EDutra  MEPerkins  MDJohnson  JLdo Valle Dettoni  V Simultaneous comparison of reactivity to purified protein derivative RT-23 and Tubersol in health care workers in Vitoria, Brazil Int J Tuberc Lung Dis 2000;41074- 1077
Munoz-Barret  JMMacias-Hernandez  AEHernandez-Ramos  I  et al.  Comparative tuberculin reactivity to two protein derivatives Rev Invest Clin 1996;48377- 381
Jenkinson  D Whooping cough: what proportion of cases is notified in an epidemic? BMJ 1983;287185- 186
Health Services Research Group, Department of Community Medicine, University of Hong Kong, Services for the Treatment of Patients With Tuberculosis in Hong Kong: Final Report to the Project Steering Group  Hong Kong University of Hong Kong2000;
Hong Kong Tuberculosis and Chest Service, Annual Report of the Tuberculosis and Chest Service for 1999  Hong Kong Dept of Health2000;
Sheldon  CDKing  KCock  HWilkinson  PBarnes  NC Notification of tuberculosis: how many cases are never reported? Thorax 1992;471015- 1018
Brown  JSWells  FDuckworth  GPaul  EABarnes  NC Improving notification rates for tuberculosis BMJ 1995;310974
Hong Kong Tuberculosis and Chest Service, Annual Report of Tuberculosis and Chest Service for 2002  Hong Kong Dept of Health2003;
Hong Kong Tuberculosis and Chest Service, Annual Report of Tuberculosis and Chest Service for 2003  Hong Kong Dept of Health2004;
Bozaykut  AIpek  IOOzkars  MY  et al.  Effect of BCG vaccine on tuberculin skin tests in 1-6-year-old children Acta Paediatr 2002;91235- 238
Karalliedde  SKatugaha  LPUragoda  CG Tuberculin response of Sri Lankan children after BCG vaccination at birth Tubercle 1987;6833- 38
Al-Kassimi  FAAbdullah  AKal-Orainey  IO  et al.  The significance of positive Mantoux reactions in BCG-vaccinated children Tubercle 1991;72101- 104
Sepulveda  RLBurr  CFerrer  X  et al.  Booster effect of tuberculin testing in healthy 6-year-old school children vaccinated with Bacillus Calmette-Guerin at birth in Santiago, Chile Pediatr Infect Dis J 1988;7578- 581
Mazurek  GHLoBue  PADaley  CL  et al.  Comparison of a whole-blood interferon γ assay with tuberculin skin testing for detecting latent Mycobacterium tuberculosis infection JAMA 2001;2861740- 1747
Lalvani  APathan  AADurkan  H  et al.  Enhanced contact tracing and spatial tracking of Mycobacterium tuberculosis infection by enumeration of antigen-specific T cells Lancet 2001;3572017- 2021
Mori  TSakatani  MYamagishi  F  et al.  Specific detection of tuberculosis infection: an interferon-γ–based assay using new antigens Am J Respir Crit Care Med 2004;17059- 64
Ewer  KDeeks  JAlvarez  L  et al.  Comparison of T-cell–based assay with tuberculin skin test for diagnosis of Mycobacterium tuberculosis infection in a school tuberculosis outbreak Lancet 2003;3611168- 1173