Annual risk of tuberculosis infection in West ... - Ingenta Connect

INT J TUBERC LUNG DIS 12(3):255–261 © 2008 The Union

Annual risk of tuberculosis infection in West Sumatra Province, Indonesia A. Bachtiar,* T. Y. Miko,* R. Machmud,* Besral,* Yudarini,* C. Basri,† F. Mehta,‡ V. K. Chadha,‡§ F. Loprang,‡ D. Manissero,‡¶ K. R. Palupi,‡ R. Jitendra§ * Centre for Health Research, University of Indonesia, Jakarta, † National TB Programme, Directorate General of Communicable Disease Control and Environmental Health, Jakarta, ‡ World Health Organization, Jakarta, Indonesia; § National Tuberculosis Institute, Bangalore, India; ¶ European Centre for Disease Prevention and Control, Stockholm, Sweden SUMMARY O B J E C T I V E : To estimate the annual risk of tuberculosis infection (ARTI) among children attending elementary school Classes I–III in West Sumatra Province, Indonesia. M E T H O D S : The survey was designed to estimate ARTI among children, irrespective of BCG scar status. Fieldwork was carried out in 72 schools selected by two-stage sampling. Children were administered 2 TU of purified protein derivative RT23 with Tween 80 intradermally on the mid-anterior aspect of the left forearm. The maximum transverse diameter of induration was measured at about 72 h. Analysis was carried out among 5653 satisfactorily test-read children aged 6–9 years.

R E S U L T S : The estimation by mirror-image method among all children (without and with BCG scar) revealed an ARTI of between 1% and 1.3%. The estimated ARTIs among children without and with BCG scar varied between 0.8–1.3% and 0.9–1.4%, respectively, using the mirror-image method, anti-mode method and mixture model. C O N C L U S I O N : Tuberculosis control efforts need to be intensified to reduce the risk of transmission of infection. K E Y W O R D S : tuberculosis; infection; tuberculin survey; annual risk; Indonesia

TUBERCULOSIS (TB) is a leading public health concern in Indonesia, which is ranked third among the high TB-burden countries.1 The DOTS strategy, introduced in the early 1990s, was extended in a phased manner to cover almost the entire population by 2000.1 A treatment success rate of 85% among new pulmonary smear-positive TB cases has consistently been attained, and case detection rates have improved in recent years to reach 66% in 2005.1 DOTS is expected to shorten the duration of infectiousness, leading to a reduction in the pool of infected individuals, which will eventually reduce the TB burden in the community. However, any increase in the present low-level prevalence of HIV infection could have an adverse influence on the TB epidemic.2 It is therefore important to periodically assess the epidemiological situation of TB in the community. The annual risk of TB infection (ARTI), defined as the average probability of a group of individuals acquiring new TB infection in the course of one year, is a sensitive indicator of the epidemiological situation in TB. It indicates the overall impact of incidence and the average duration of infectious cases in the community and the efficiency of TB control programmes

and socio-environmental factors influencing transmission of infection. ARTIs computed from the prevalence of infection estimated through cross-sectional tuberculin surveys among children reflect the current TB situation in the community. A tuberculin survey was therefore carried out in West Sumatra Province, Indonesia, to estimate the prevalence of TB infection among schoolchildren and to compute the ARTI from the estimated prevalence.

MATERIAL AND METHODS A survey was carried out among children attending Classes I–III in a sample of elementary schools. Elementary school enrolment is 90% in the various provinces of the country.3 As the majority of the children were vaccinated with bacille Calmette-Guérin (BCG) as per BCG scar surveys and Expanded Programme of Immunization (EPI) reports,4 all children, irrespective of their BCG scar status, were included in the study. Children in Indonesia are BCG-vaccinated at birth with a dose of 0.05 ml (Personal communication, EPI sub directorate, Ministry of Health, Indonesia).

Correspondence to: V K Chadha, National Tuberculosis Institute, 8 Bellary Road, Bangalore, India. Tel: (91) 080 23441192/ 3620. Fax: (91) 80 2344 0952. e-mail: [email protected] Article submitted 18 July 2007. Final version accepted 26 November 2007.

256

The International Journal of Tuberculosis and Lung Disease

Figure 1 Sampling design. PPS probability proportional to size; SRS simple random sampling. Kabupatens rural areas; Kotas urban areas excluding the provincial capital; Kecamatans urban blocks within the capital.

Sample size was determined to estimate the prevalence of infection within 15% of the true value at 5% level of significance. The expected prevalence was assumed to be at a minimum of 10% (the prevalence of infection estimated during most of the district-level surveys in the country during the 1970s and 1980s varied between 15% and 35%).5 The value of the design effect was taken as 3, as is generally recommended for tuberculin surveys.6 The estimated sample size of 4610 was increased by 20% to 5532 to take into account absenteeism, exclusions and drop-outs between registration and reading of tuberculin reactions, and then rounded to 6000. A stratified 2-stage cluster sampling was adopted in selecting schools, each denoting a cluster (Figure 1). The province was divided into three strata: Kabupaten, Kota and the provincial capital.* The first stage of sampling involved the selection of 30% Kabupatens, Kotas and Kecamatans in the respective strata, using the probability proportional to size method. The estimated sample size was allocated to three strata in proportion to their population size. Within the respective strata, sample size was distributed equally among the selected Kabupatens, Kotas and Kecamatans. The second stage of sampling involved the selection of schools within selected Kabupatens, Kotas and Kecamatans by simple random sampling. The number of schools selected from each Kabupaten, Kota and Kecamatan depended on the average number of children enrolled in Classes I–III. All children in Classes I–III of the selected schools who were present on the day of testing were included in the study. * Kabupatens are rural areas, Kotas denote urban areas excluding the provincial capital and Kecamatans denote urban blocks within the capital.

Fieldwork was carried out in 72 schools during August–December 2006 by two teams trained in all aspects of the survey by internationally experienced trainers. Readers who achieved the highest levels of agreement with the trainer-reader were employed as readers in the survey. Consent was obtained from the school authorities after acquainting them with the purpose of the survey and characteristics of the tuberculin test. Printed information about the purpose of the survey, the nature of the tuberculin test and dates of testing and reading was provided to parents 2–3 days before testing. Parents were given the option of refusing permission to have their children tested. All children enrolled in the attendance registers of Classes I–III were recruited for the study, and their age, sex and BCG scar status were recorded. The testing centre was set up in the shade at a suitable place in the school. Children with any of the following conditions were excluded from the test: high fever, severe malnutrition, known history of anti-tuberculosis treatment or a known immunocompromising condition, skin rash and refusal by the parent/guardian. The cotester inspected the upper third of both arms for the pea-sized hypo-pigmented shiny scar generally produced by BCG vaccination. The tester administered 0.1 ml of tuberculin containing 2 tuberculin units (TUs) of purified protein derivative (PPD) RT23 with Tween 80 intradermally on the mid-anterior aspect of the left forearm. Disposable tuberculin syringes with graduations of 0.1 ml and fitted with a 26 gauge needle 10 mm in length and 20 bevel were used. PPD vials were procured in a single batch from Statens Serum Institute, Copenhagen, Denmark, and the cold chain was maintained carefully. Each test was recorded as ‘satisfactory’ if it raised a flat pale wheal with clearly visible follicular

ARTI in West Sumatra

pits and well demarcated borders, or ‘unsatisfactory’ in the case of leakage or subcutaneous injection. Tuberculin reactions were read at approximately 72 h (3 days) after administration of the test. The reader measured the maximum transverse diameter of induration in mm using a transparent ruler, after carefully identifying its margins by palpation. The presence of bullae, vesicles, necrosis or lymphangitis, if any, at the test site was recorded. Care was taken to ensure that the BCG scar status of the child was not revealed to the reader. Children with reactions 10 mm were referred to the nearest health centre for further medical assistance if required. A total of 94 sputum smear-positive pulmonary TB cases were tuberculin-tested by the same field teams using the same batch of PPD. There were no known human immunodeficiency virus positive cases among those tested. Ethical clearance for the survey was obtained from the Ethical Review Committee of the Faculty of Public Health of the University of Indonesia. Statistical methods

Estimating the prevalence of infection by the mirror-image and anti-mode methods Tuberculin reaction sizes were plotted as frequency distribution graphs to identify the mode and anti-mode of reactions attributable to infection with tubercle bacilli and to set the criteria for estimating the prevalence of infection by the mirror-image and anti-mode methods. In the mirror-image method, the proportion of reactions larger than the mode was doubled and added to the frequency at the mode.7 In the anti-mode method, all reactions greater than the anti-mode were attributed to infection with tubercle bacilli.7 The weighted prevalence of infection was estimated by assigning weight to each Kabupaten/Kota/ Kecamatan as the ratio of the number of children originally allocated to those children actually investigated. The 95% confidence intervals (CI) of the estimates were obtained using the appropriate formula.8 The 2 test with continuity correction was used to test the significance of differences between proportions, and P 0.05 was considered statistically significant. Estimating the prevalence of infection by the mixture model An analysis using the basic mixture model was undertaken using the Bayesian method by the Markov Chain Monte Carlo approach using R software (version 2.4.1, http:/lib.stat.cmu.edu/R/CRAN) and scripts available at www.tbrieder.org.9 Two component distributions were assumed for children without BCG scars—one for infection with tubercle bacilli and one for crossreactions to infections with environmental mycobacteria. Three component distributions were assumed

257

for children with BCG scars, the third component attributable to cross-reactions due to BCG vaccination. Analysis was undertaken assuming different combinations of distributions for mixture components (Weibull, normal and log-normal). The Metropolis sampler was used to simulate from the posterior distribution of mixture model parameters.9 After a burnin period of 20 000 iterations, the results of which were discarded, a thinned sample of 50 000 values out of 500 000 was used to summarise the posterior distribution of model parameters.

Computing the ARTI from estimated prevalence The ARTI was computed from the estimated prevalence of infection using the following equation:5 Rba/2 1 (1 Pba)1/a where a and b are the mean age and the mid-point of the years of birth of test-read children, Pba is the prevalence of infection during the mid-point of the survey period and Rba/2 is the ARTI at mid-point between the year of birth of the cohort and the years of the survey. Sample for analysis After excluding 792 outliers in terms of age, a total of 6550 children aged 6–9 years were registered. Of these, 5948 were administered the tuberculin test, and reactions were read for 5655 children. The final analysis was conducted among 5653 children after exclusion of two children with an ‘unsatisfactory’ test, as explained in the methods. Thus, 86.3% of the registered children in the 6–9 years age group were satisfactorily test-read. Their mean age was 8.1 years, and 74% had BCG scars.

RESULTS Estimation by mirror-image and anti-mode methods

Frequency distribution of tuberculin reaction size among children Frequency distributions of reaction size among all children (without and with BCG scars) and by BCG scar status are presented in Figures 2 and 3, respectively. Approximately 70% of children without BCG scars and 60% with scars elicited no induration. Though clear bi-modality was not observed on the distribution of all children and that of children with BCG scars, a bi-modal distribution was seen in children without BCG scars, with the second mode at 15 mm probably representing reactions due to infection with tubercle bacilli. There was also the suggestion of an anti-mode among children without BCG scars, distinguishing the subgroup of children infected with tubercle bacilli from others at 11 mm. Visual inspection of the distributions revealed a terminal digit preference at 5, 10 and 15 mm. Expected frequencies at each mm of reaction size were therefore

258


Figure 2 Frequency distribution of tuberculin reaction size among children aged 6–9 years, irrespective of their BCG scar status (n 5653). BCG bacille Calmette-Guérin.

calculated by averaging the frequencies of five values (two before and two after the digit of preference). The mode on expected frequencies among children without BCG scars was observed at 16 mm. There were higher proportions of reactions in the 5–16 mm range among children with BCG scars than among children without scars.

Criteria for estimating the prevalence of infection with tubercle bacilli The prevalence of infection among all children and by BCG scar status was estimated by the mirror-image method using the mode observed among children without BCG scars. This was based on the assumption that the reactions due to infection with tubercle bacilli were distributed normally around this mode.5,7 The estimations were made by using the 15 mm mode and observed frequencies as well as by using the 16 mm mode and expected frequencies. Prevalence was also estimated by the anti-mode method among children without BCG scars, considering all reactions 11 mm as attributable to TB infection. This method was not applied to all children

or to children with BCG scars due to a lack of visible anti-modes on their frequency distribution.

Estimated prevalence of infection and ARTI The prevalence of infection in the group of all children estimated by the mirror-image method using the 15 mm mode was 10.4% (95%CI 8.1–12.7), with the ARTI computed as 1.3%. It was 8.0% (95%CI 6.2– 9.8) using the 16 mm mode, with ARTI computed as 1.0% (Table 1). Using the criteria described above, ARTI estimates among children without and with BCG scars varied between 0.8–1.1% and 1.0–1.4%, respectively (Table 2). Estimation by mixture model The best fit to the observed frequency distribution was obtained using the Weibull distribution for crossreactions to infection with environmental mycobacteria, log-normal distribution for reactions due to infection with tubercle bacilli and log-normal for cross-reactions to BCG vaccination. Mixture component distributions among children without and with BCG scars are summarised in Figure 4. The median of distribution of infection with tubercle bacilli was estimated at 15 mm among children without and those with BCG scars. Estimates of prevalence of infection and ARTI by BCG scar status are presented in Table 2. There was no predictive failure for any reaction size among children without BCG scars. However, a predictive failure of 23% was observed among children with BCG scars. Table 1 Estimated prevalence of infection among all children (N 5653) by mirror-image method and computed ARTI Mode

Figure 3 Comparisons of frequency distributions of tuberculin reaction size among children aged 6–9 years, by BCG scar status. BCG bacille Calmette-Guérin.

15 mm (observed frequencies) 16 mm (expected frequencies)

Prevalence n (95%CI)

ARTI %

10.4 (8.1–12.7) 8.0 (6.2–9.8)

1.3 1.0

ARTI annual risk of tuberculosis infection; CI confidence interval.


Table 2

259

Estimated prevalence of infection among children without and with BCG scars, by method of estimation Children without BCG scar (n 1462)

Estimation method Mirror-image method Mode 15 mm (observed frequencies) Mode 16 mm (expected frequencies) Cut-off point method Anti-mode 11 mm (observed frequencies) Anti-mode 11 mm (expected frequencies) Mixture model

Children with BCG scar (n 4185)


ARTI %


ARTI %

8.0 (4.9–11.0)* 6.2 (4.8–7.6)*

1.0 0.8

11.0 (8.2–13.7)* 7.9 (6.4–9.4)*

1.4 1.0

1.0 1.1 1.3

NA NA 7.2 (3.5–11.7)

NA NA 0.9

7.7 (5.3–10.2) 8.4 (5.9–10.9) 10.0 (7.5–12.9)

* Denotes significant difference between children with and those without BCG scars. BCG bacille Calmette-Guérin; ARTI annual risk of tuberculosis infection; CI confidence interval; NA not applicable.

Figure 5 Frequency distribution of reaction size among sputum smear-positive cases of pulmonary tuberculosis (n 94).

DISCUSSION

Figure 4 Distribution of mixture components. A. Tuberculin reaction sizes among children without BCG scar. B. Tuberculin reaction sizes among children with BCG scar. Histograms display observed reaction sizes; a denotes mixture distribution; b, c, d denote estimated underlying distributions due to infection with environmental mycobacteria, cross-reactions due to BCG vaccination and infection with tubercle bacilli respectively; n denotes the number of reactors. BCG bacille Calmette-Guérin.

Frequency distribution of tuberculin reaction size among sputum smear-positive TB cases All cases of sputum smear-positive TB elicited reactions of 13 mm following a unimodal distribution with the mode at 18 mm (Figure 5).

The present survey to obtain information on the prevailing epidemiological situation of TB in terms of ARTI was the first provincial-level survey to be conducted in West Sumatra. It was designed to estimate the ARTI among schoolchildren, irrespective of their BCG scar status. The estimation by the mirror-image method among children aged 6–9 years revealed that the ARTI may be between 1% and 1.3%. This indicates a fairly high rate of transmission of TB infection, considering that 1000–1300 individuals per 100 000 population acquire new infection with tubercle bacilli every year, thereby adding to the pool of infected people potentially at risk of developing TB disease in future. The survey results therefore suggest the need for further intensification of TB control efforts. The estimated ARTIs among children without and with BCG scars varied between 0.8–1.3% and 0.9– 1.4%, respectively, using various methods of estimation. However, the survey was not designed to estimate the difference in ARTI between the two subgroups. The frequency distributions of tuberculin reaction sizes revealed considerable proportions of intermediate reactions even among children without BCG scars, which could be attributable to non-specific tuberculin sensitivity due to infection with environmental

260


mycobacteria. The distributions also revealed a lower mode of reactions attributable to infection with tubercle bacilli among children than among confirmed TB cases. However, the mode used for estimating the prevalence of infection by the mirror-image method was supported by the mixture model. The mixture model has been proposed as a possible solution to overcome problems in the interpretation of tuberculin surveys due to difficulties commonly encountered in the identification of modes and anti-modes of reactions due to infection with tubercle bacilli on the frequency distributions of reactions.9,10–16 In the present survey, the mixture model provided a good fit for children without BCG scars, but high predictive failures were observed among children with BCG scars. This, as observed in Figure 4, could be due to the considerable overlap of reactions due to BCG vaccination and those due to infection with tubercle bacilli.9 Further experience is needed before accepting mixture analysis as the standard method for analysing tuberculin surveys. It is not feasible to comment on the past trends of ARTI in West Sumatra Province due to a lack of provincial-level data from previous years. ARTI estimates in one of the districts in the province (Padang Pariaman) varied between 1.9% and 3.1% during three rounds of tuberculin surveys carried out in the 1980s.5 Finally, given the mean age of the children, the present survey results would most closely correspond to the year 2002 and may be used as baseline information for monitoring ARTI trends in future. As trends in ARTI have been observed to reflect trends in the prevalence of TB, this would help to measure progress towards the Millennium Development Goals.1,10–21 Acknowledgements The authors are indebted to Ms E Adams and Ms P Suganthi for training the field teams, the provincial authorities and heads of schools for their support and the field staff for their commitment. The contributions of H L Rieder for assistance in mixture analysis and of M van der Werf, C Dye, C Titaley, F X Budhyono and E Sasongko for valuable input are duly acknowledged.

References 1 World Health Organization. Global tuberculosis control: surveillance, planning and financing. WHO report 2007. WHO/ HTM/TB/2007.376. Geneva, Switzerland: WHO, 2007. 2 Director General Communicable Diseases & Environmental Health. Statistik Kasus HIV/AIDS di Indonesia Dilapor s/d December 2006. Jakarta, Indonesia: Department of Health, 2006. http://www.spiritia.or.id Accessed February 2007. 3 Statistics Indonesia (Badan Pusat Statistik). Provincial profiles. Jakarta, Indonesia: BPS. www.bps.go.id Accessed 24 January 2007.

4 National Institute of Health Research and Development. Indonesia, tuberculosis prevalence survey in Indonesia 2004. Jakarta, Indonesia: Ministry of Health, 2005. 5 Cauthen G M, Pio A, ten Dam H G. Annual risk of TB infection. WHO/TB/88.154. Geneva, Switzerland: WHO, 1988. 6 Nagelkerke N J D, Borgdorff M W, Kalisvaart N A, Broekmans J F. The design of multi-stage tuberculin surveys: some suggestions for sampling. Int J Tuberc Lung Dis 2000; 4: 314–320. 7 Bleiker M A, Sutherland I, Styblo K, ten Dam H G, Misljenovic O. Guidelines for estimating the risk of TB infection from tuberculin test results in a representative sample of children. Bull Int Union Tuberc Lung Dis 1989; 64: 7–12. 8 ten Dam H G. Surveillance of tuberculosis by means of tuberculin surveys; WHO/TB/85.145. Geneva, Switzerland: WHO, 1985. 9 Neuenschwander B E. Bayesian mixture analysis for tuberculin induration data. Paris, France: International Union Against Tuberculosis and Lung Disease. http://www.tbrieder.org Accessed 27 October 2007. 10 Chakraborty A K, Chaudhuri K, Sreenivas T R, Krishnamurthy M S, Shashidhara A N, Channabasavaiah R. TB infection in a rural population of South India: 23-year trend. Tubercle Lung Dis 1992; 73: 213–218. 11 Tupasi T E, Radhakrishna S, Rivera A B, et al. The 1997 nationwide tuberculosis prevalence survey in the Philippines. Int J Tuberc Lung Dis 1999; 3: 471–477. 12 Bosman M C J, Swai O B, Kwamanga D O, Agwanda R, Idukitta G, Misljenovic O. National Tuberculin Survey of Kenya, 1986–1990. Int J Tuberc Lung Dis 1998; 2: 272–280. 13 Tanzania Tuberculin Survey Collaboration. Tuberculosis control in the era of the HIV epidemic: risk of tuberculosis infection in Tanzania, 1983–1998. Int J Tuberc Lung Dis 2000; 5: 103–112. 14 Tuberculosis Research Centre, Chennai. Trends in the prevalence and incidence of tuberculosis in South India. Int J Tuberc Lung Dis 2001; 5: 142–157. 15 Gopi P G, Subramani R, Narayanan P R. Trend in the prevalence of TB infection and ARTI after implementation of a DOTS programme in South India. Int J Tuberc Lung Dis 2006; 10: 346–348. 16 Egwaga S M, Cobelens F G, Muwige H, et al. The impact of HIV epidemic on tuberculosis transmission in Tanzania. AIDS 2006; 20: 915–921. 17 Chakraborty A K, Singh H, Srikantan K, Rangaswamy K R, Krishnamurthy M S, Stephen J A. Tuberculosis in a rural population of South India: report on five surveys. Indian J Tuberc 1982; 29: 153–167. 18 Fourie P B, Gatner E M, Glatthaar E, et al. Follow-up tuberculosis prevalence survey of Transkei. Tubercle 1980; 61: 71–79. 19 Hong Y P, Kim S J, Lew W J, Lee E K, Han Y C. The seventh nationwide tuberculosis prevalence survey in Korea, 1995. Int J Tuberc Lung Dis 1998; 2: 27–36. 20 Neuenschwander B E, Zwahlen M, Kim S J, Engel R R, Rieder H L. Trends in the prevalence of infection with Mycobacterium tuberculosis in Korea from 1965 to 1995: an analysis of seven surveys by mixture models. Int J Tuberc Lung Dis 2000; 4: 719–729. 21 Gopi P G, Subramani R, Santha T, et al. Relationship of ARTI to incidence and prevalence of tuberculosis in a district of South India. Int J Tuberc Lung Dis 2006; 10: 115–117.

RÉSUMÉ O B J E C T I F : Estimer le risque annuel d’infection tuberculeuse (ARTI) parmi les enfants fréquentant les classes I à III de l’école élémentaire dans la province de Sumatra Ouest, Indonésie.

L’enquête a été conçue pour estimer l’ARTI chez les enfants quel que soit le statut de leur cicatrice BCG. Le travail de terrain a été mené dans 72 écoles sélectionnées par un échantillonnage à deux étapes. On a

MÉTHODES :


administré aux enfants par voie intradermique 2 UT de PPD RT 23 avec Tween 80 sur la partie moyenne de la face antérieure de l’avant-bras gauche. Le diamètre transversal maximal de l’induration a été mesuré après environ 72 h. L’analyse a été menée sur 5653 enfants de 6 à 9 ans dont le test avait été lu de manière satisfaisante. R É S U L T A T S : L’estimation par la méthode des images en miroir parmi tous les enfants (qu’ils aient ou non une cicatrice de BCG) a révélé un ARTI situé entre 1% et

261

1,3%. Les ARTI estimés chez les enfants sans et avec cicatrice vaccinale de BCG ont varié respectivement entre 0,8% et 1,3% et 0,9% et 1,4% selon les diverses méthodes d’estimation (la méthode de l’image en miroir, la méthode anti-modale et le modèle mixte). C O N C L U S I O N : Il est nécessaire d’intensifier davantage les efforts de lutte contre la tuberculose pour réduire le risque de transmission de l’infection.

RESUMEN

Estimar el riesgo anual de infección tuberculosa (ARTI) en niños que acuden a la escuela primaria del primero al tercer grado en la provincia de Sumatra occidental en Indonesia. M É T O D O S : El estudio se diseñó con el fin de estimar el ARTI en los niños, sin tener en cuenta la presencia o ausencia de cicatriz BCG. El trabajo en el terreno se llevó a cabo en 72 escuelas escogidas mediante un muestreo en dos tiempos. Los niños recibieron por vía intradérmica 2 unidades de tuberculina (derivado proteínico purificado RT23) con polisorbato 80, en la parte media de la cara anterior del antebrazo izquierdo. A las 72 h se midió el máximo diámetro transverso de la induración. Se anaOBJETIVO :

lizaron los datos de 5653 aplicaciones satisfactorias de tuberculina en niños de 6 a 9 años. R E S U L T A D O S : El cálculo por el método de la imagen en espejo en todos los niños (incluidos aquellos sin cicatriz BCG) puso en evidencia un ARTI entre 1 y 1,3%. El ARTI en niños sin cicatriz BCG osciló entre 0,8% y 1,3%) y en aquellos con cicatriz BCG entre 0,9% y 1,4%, utilizando diversos métodos de cálculo : el método de la imagen en espejo, el método del valor discriminatorio y el análisis de modelos mixtos. C O N C L U S I Ó N : Es necesario intensificar aún más los esfuerzos en la lucha contra la tuberculosis a fin de disminuir el riesgo de transmisión de la infección.