Pulmonary Tuberculosis in Children in a Developing Country
Objective. We evaluated the clinical and epidemiologic characteristics of Peruvian children presenting with pulmonary tuberculosis (PTB) to determine whether features predictive of confirmed PTB could be identified.
Study Design. This was a cross-sectional study of 135 children (mean age: 6.8 years) presenting to the Hospital del Niño in Lima, Peru, with presumptive diagnosis of PTB. Clinical, epidemiologic, and laboratory findings were compared between 3 groups of pediatric patients with a presumptive diagnosis of PTB: those with positive Mycobacterium tuberculosis (MTB) cultures, those likely to have PTB based on clinical criteria but with negative cultures, and those who did not meet clinical diagnostic criteria or have positive cultures.
Results. A total of 50 (37%) patients were diagnosed with definitive PTB based on positive sputum culture. Another 55 (47%) patients were classified as having probable PTB based on meeting at least 2 of the following criteria: cough lasting for at least 2 weeks, typical chest radiograph changes, purified protein derivative (PPD) ≥10 mm, or history of tuberculosis family contact. Patients with definitive or probable PTB were significantly older than patients without clinical PTB, and those with symptomatic disease were significantly older than those with asymptomatic disease. Patients with PTB diagnosed by culture were significantly more likely than those diagnosed using clinical criteria to have cough lasting ≥2 weeks, fever, and a PPD ≥10 mm.
Conclusions. The typical presentation of PTB in Peruvian children includes symptoms of active pulmonary disease similar to those seen in adults. This presentation differs significantly from that reported in developed countries, where many children have minimal or no symptoms at the time of presentation. The diagnostic criteria for pediatric PTB must be modified in hyperendemic developing country environments where features may differ from those described in the United States. The triad of cough lasting ≥2 weeks, fever, and a PPD ≥10 mm was highly predictive for culture-positive PTB among children in this low-income Peruvian population.
Tuberculosis (TB) has reached epidemic proportions in many developing countries, but the burden of tuberculous disease in children often is underappreciated. In 1990, the World Health Organization estimated that there were approximately 1.3 million new cases of TB and 450 000 deaths worldwide from TB in children under age 15.1 In Peru the overall prevalence of pulmonary tuberculosis (PTB) is extremely high, ranging from 158.2 per 100 000 to >350 per 100 000 in hyperendemic shantytowns of Lima,2,,3 and our data suggest that there is also a high rate of Mycobacterium tuberculosis (MTB) infection among children in these communities. In a cross-sectional study of 1 such community, tuberculin reactions greater than 10 mm were found in 12% of children 0 to 1 year of age, 18% of children 2 to 4 years of age, and 24% of children 5 to 14 years of age.4
Diagnosing PTB in the pediatric population presents challenges. Symptoms of MTB infection are often nonspecific or absent in affected children.5 Adequate clinical diagnostic specimens often are difficult to obtain in children under age 8 because of a lack of sputum production.5 Furthermore, currently available diagnostic tests are costly, slow, and lacking in sensitivity, and even under the best circumstances MTB is isolated in fewer than 50% of pediatric cases.6
In developed countries, where the prevalence of MTB is low and resources for preventive therapy are available, the primary objective is to identify patients with asymptomatic infection. However, in regions where the prevalence of MTB is high and resources are limited, identifying all infected individuals would be an overwhelming and very expensive task, so only cases of active disease are sought. Worldwide, the majority of pediatric MTB cases are diagnosed using a combination of clinical and epidemiologic criteria, such as cough, lymphadenopathy, characteristic chest radiographs, and purified protein derivative (PPD) reactivity. Although cough lasting for longer than 2 weeks may be fairly sensitive for active PTB, it is also commonly seen in illnesses such as bronchitis and asthma.7,,8 Lymphadenopathy may be seen in a wide variety of infectious diseases or malignancies. Radiographic changes in children with PTB are quite variable5 and may mimic changes seen in other pulmonary diseases. Furthermore, in comparison to developed countries, where interpretation of PPD reactivity is based primarily on patients' risk factors, in developing countries such as Peru interpretation is complicated by the high prevalence of MTB, BCG vaccination, concurrent infections, and malnutrition.9–13 In developing countries such as Peru, rates of PPD skin test reactions greater than 10 mm among children often are more than 25%, in contrast to those in developed countries, where less than 5% of children are skin test positive.4 Diagnosis of PTB in children in developing countries thus relies heavily on clinical suspicion.
Because of the lack of sensitive assays to confirm the diagnosis of pediatric TB, new diagnostic methods for TB in children are needed. Encouraging early results with a new polymerase chain reaction (PCR)-based sputum assay showed that MTB could be detected more rapidly and with equal sensitivity when compared with conventional sputum culture.14–17 Although this technique has not been fully evaluated in children, it may ultimately improve the diagnosis of PTB in this population.
We conducted a cross-sectional study of children with suspected PTB in a highly endemic population in Peru to provide a more thorough description of the epidemiology and clinical features of pediatric PTB in the developing world and to correlate results of a new heminested MTB PCR assay with clinical and microbiologic data used to categorize children with presumptive PTB in developing countries.
This study was conducted between November 1996 and September 1997 at the National Children's Hospital in Lima, Peru, the country's largest pediatric referral center. Study participants were recruited from inpatients admitted to the hospital's pulmonology ward or outpatients seen in the hospital's Tuberculosis Control Program who presented with a clinical features highly suspicious for MTB, including a positive PPD, history of contact, characteristic radiographic changes, or cough. Informed consent was obtained from a parent or legal guardian of all children who participated in the study. The study protocol and informed consent forms were approved by the institutional review boards of the Johns Hopkins University and the National Children's Hospital of Lima, Peru.
A physician investigator conducted a short questionnaire and brief physical examination of each patient at enrollment. A standardized study questionnaire was used to assess clinical symptoms, history of MTB contacts, previous MTB diagnosis, and demographic information. BCG vaccination was determined by the presence of characteristic scarring. A Mantoux test was placed and read after 72 hours. The majority of the patients received an anterior–posterior or posterior–anterior chest radiograph as part of the routine TB workup. The radiographs were interpreted by an independent pulmonologist blinded to the clinical diagnosis of the patient. Sputum samples were collected from the patients in the morning on study days 2 and 3. Nasogastric aspiration (NGA) was performed in young children unable to expectorate sputum. Samples were encoded and transported daily to the Department of Pathology at the Universidad Peruana Cayetano Heredia for analysis.
Criteria for Clinical Diagnosis of Pediatric Tuberculosis
The following criteria adapted from Migliori et al18 were used to diagnose PTB in our patients. Patients with a positive MTB culture of sputum or gastric aspirate were classified as having definitive PTB. Patients were classified as having probable PTB if they met 2 or more of the following criteria: history of tuberculous contact, cough lasting longer than 2 weeks, reactive tuberculin skin test ≥10 mm, or radiographic findings compatible with MTB infection such as miliary disease, cavitary lesions, hilar lymphadenopathy, or primary complex.
Sputum and Gastric Aspirate Cultures
All clinical samples were digested and concentrated using the standard N-acetyl-L-cysteine NaOH-Na citrate method for processing mycobacterial specimens.19 Acid-fast bacilli (AFB) smear microscopy using Ziehl–Nielsen and Auramine stains were performed using standard techniques.19
Mycobacterial Growth Indicator Tubes (Becton Dickinson, Sparks, MD) containing 10% OADC (oleic acid, albumin, dextrose, catalase; Becton Dickinson) and 100 μL PANTA Antimicrobic Supplement (polymyxin B, amphotericin B, nalidixic acid, trimethoprim, and azlocillin; Becton Dickinson) were inoculated with 500 μL decontaminated sample as per manufacturer's instructions. Lowenstein–Jensen slants (Difco, Detroit, MI) and Middlebrook's 7H11 medium plates (Difco) were inoculated with 250 μL decontaminated sample. Mycobacterial Growth Indicator Tubes were incubated at 37°C and examined for mycobacterial growth at least once a week for up to 6 weeks using a 365-nm ultraviolet transilluminator as described by the manufacturer. Lowenstein–Jensen slants and micro-agar 7H11 plates were incubated at 37°C ± 5% CO2 and examined by light microscopy for mycobacterial growth at least once a week from weeks 2 to 8 after inoculation.19 Criteria for positive mycobacterial growth were taken from the National Centers for Disease Control.19 The new and reliable Mycobacterium Alamar Blue Assay described by Franzblau et al20 was used to determine mycobacterial drug resistance.
A decontaminated sample (300–500 μL) was incubated in the presence of 500 μL absolute ethanol for at least 2 hours at room temperature. The material was pelleted by centrifugation at 10 000 rpm for 10 minutes. Cellular lysis was performed by resuspending the pellet in 500 μL of a solution of Tris-EDTA containing 1% Triton 100 and 10% chelex, vortexing vigorously, and then centrifuging for 5 minutes at 10 000 rpm. The supernatant was discarded, and the pellet was resuspended in 100 mL of the TE-Triton X solution and then placed in boiling water for 30 minutes. The remaining cellular debris was removed by centrifuging at 10 000 rpm for 10 minutes, and the supernatant was used directly in PCR reactions.
Nested N2 PCR for Sputum and Nasogastric Aspirates
The nested N2 PCR involves 2 sequential PCR reactions that recognize sequences in the insertion segment 1S6110 of the MTB genome.21 In the first, a larger fragment is amplified using primers PT8 and PT9. This amplification takes place in a reaction mixture containing 3 μL sputum, NGA, or fecal DNA, 0.2 mM dNTP, 0.25 μM primers (PT8 and PT9), 10 mM Tris-HCl, 50 mM KCl, 2 mM MgCl, and 0.325 U Taq polymerase under the following conditions: 2 minutes at 94°C and then 20 cycles of 20 seconds at 94°C, 45 seconds at 65°C, and 45 seconds at 72°C.
The second reaction amplifies an internal 360-bp fragment using primers TB290 and PT9. First, 1 μL of the first PCR reaction product was added to a mix containing 0.5 μM of primers TB290 and PT9 along with 0.2 mM dNTP, and 0.325 U Taq polymerase is added. The internal 360-bp fragment is amplified in a thermocycler (PTC-100; MJ Research, Watertown, MA) under the following conditions: 2 minutes at 94°C and then 35 cycles of 20 seconds at 94°C, 45 seconds at 65°C, and 45 seconds at 72°C. A negative control (water) and a positive control (genomic MTB DNA) were included in every amplification run as described for the single-step PCR assay. The amplified PCR products were electrophoresed on a 2% agarose gel containing 5 μg/mL ethidium bromide at 90 V for 1 hour, followed by inspection under ultraviolet light for a 245-bp PCR amplification product.
All data were coded and analyzed using the statistical software SPSS version 7.5 (SPSS Inc, Chicago, IL). Malnutrition was defined as weight for age below the 5th percentile according to National Center for Health Statistics (NCHS) standards. χ2 and, where necessary, Fisher's exact tests were used to measure strengths of association between categorical variables. A 2-tailed t test was used to compare continuous variables. Logistic regression was used to assess association between multiple variables.
A total of 135 children clinically suspected of having PTB were recruited into the study. MTB was isolated from sputum or NGA from 50 children, and these children were classified as definitive PTB diagnoses. AFB were identified in sputum smears of 35 (70%) of these culture-positive children; only 2 patients had positive AFB on smear despite negative sputum cultures.
Of the children with negative cultures, 55 were classified as having probable PTB after satisfying 2 or more Migliori clinical diagnostic criteria for PTB, including cough of at least 2 weeks' duration, history of TB family contact, Mantoux reaction ≥10 mm, or characteristic radiologic findings. The remaining 30 children were classified as negative for PTB. The demographic and clinical characteristics of these 3 groups of children are presented inTable 1; the mean age of the patients was 6.8 years (range 1 month–17 years). No significant differences in sex, BCG scar, presence of extrapulmonary TB, or socioeconomic status were observed between patients with a definitive or probable diagnosis of PTB and those who were classified as negative for PTB. The most common clinical symptoms reported were cough, fever, and weight loss.
Of the 105 patients diagnosed with definitive or probable PTB, 19% of infants, 52% of 2- to 4-year-olds, 37% of 5- to 12-year-olds, and 86% of 13- to 17-year-olds were diagnosed by positive sputum or NGA cultures. Children with definitive PTB were significantly older (9.1 years) than those with probable PTB (5.7 years) (P < .001). Furthermore, children with probable PTB were older than those classified as negative for PTB (4.3 years) (P = .1). Three or more Migliore clinical diagnostic criteria were met by 26/50 (52%) of children with definitive PTB and 28/55 (51%) of those with probable PTB (Table 2). However, 4/50 (8%) of the children with definitive PTB met less than 2 of the clinical criteria and therefore would not have been diagnosed with PTB according to these clinical criteria.
Children with definitive PTB were significantly more likely than PTB-negative children to report cough lasting more than 2 weeks (P < .001) or fever (P = .02) or to have a Mantoux reaction ≥10 mm (P < .001). In contrast, history of family MTB contact was significantly more likely in both the probable and negative PTB groups (P < .001). When these variables were examined for their association with culture results in a logistic regression analysis, fever (P = .05), cough for greater than 2 weeks (P = .01), and Mantoux reaction ≥10 mm (P = .001) were found to be significantly associated with positive culture results. When we examined the sensitivity of PTB case detection using the latter 3 criteria alone (fever, cough >2 weeks, and positive PPD ≥10 mm), 40 (80%) of the definitive PTB patients, 36 (65%) of the probable PTB patients, and 7 (23%) of the negative PTB patients presented with at least 2 out of the 3 criteria (Tables 2 and 3). We call this set of clinical features the Peru criteria. The sensitivity of this approach as a screening strategy to detect culture-positive pediatric patients was similar to that of the Migliore criteria, with the added advantage of not requiring a chest radiograph (sensitivity of detection of definitive PTB: 92% based on 2 of 4 Migliore criteria, 80% based on 2 of 3 Peru criteria). The combination of all 3 Peru criteria had a positive predictive value of 73% for definitive PTB, so in our study group this triad was useful in identifying patients at exceptionally high risk for culture-positive PTB. However, when we compared screening based on 2 versus 3 Peru criteria, the clinical triad was of more limited use as a screening tool because sensitivity of detection of definitive PTB was only 44%. Because MTB culture is not a highly sensitive way to identify PTB in children, calculations of specificity and negative predictive value are difficult to interpret.
When symptoms not found to be as useful as case-defining criteria were examined as a function of age, patients with definitive PTB who had hemoptysis (12.3 years vs 8.6 years, P < .01), dyspnea (11.5 years vs 8.2 years, P = .04), or weight loss (10.7 years vs 6.8 years, P < .01) were significantly older than patients without these symptoms. Children with probable PTB who reported hemoptysis were also significantly older than those not reporting this symptom (9.6 years vs 5.3 years,P < .01).
A Mantoux test was performed in 132 of the 135 patients enrolled in the study. Overall, 67/132 (51%) of patients had a positive PPD result, defined as induration ≥10 mm (Table 4). There was no significant difference in response to PPD according to previous BCG vaccination or presence of extrapulmonary MTB infection. A negative Mantoux test was more common in malnourished children than in nonmalnourished children, 5/8 (63%) versus 9/36 (25%) in the definitive PTB group (P = .05) and 8/11 (73%) versus 17/39 (44%) in the probable PTB group (NS), respectively. There was no significant association between age and malnutrition or a negative Mantoux test in our study. A chest radiograph was performed in 130 of the study participants, and characteristic findings of PTB such as lymphadenopathy, miliary disease, cavitation, or primary complex were observed in 32/53 (60%) of the probable PTB patients and 31/48 (65%) of the definitive PTB patients. Nonspecific radiologic findings included pneumonia, atelectasis, interstitial disease, and pleural effusion; only 1 patient (1%) with a diagnosis of probable PTB had a normal chest radiograph. Extrapulmonary TB was noted in 21/135 (16%) patients in the study and included lymphadenopathy,5intestinal–intraperitoneal TB,7 intra-abdominal lymphadenopathy,2 miliary disease,5meningitis,1 and optic involvement.1 One of the patients with miliary disease and 2 of the patients with lymphadenopathy did not meet our clinical diagnostic criteria for PTB. Children with extrapulmonary TB were no more likely to be symptomatic or malnourished than those without extrapulmonary MTB. Children with definitive TB who were found to have extrapulmonary TB were younger than those without signs or symptoms of extrapulmonary TB (P = .02).
MTB PCR was performed on 104 of the sputum or NGA samples. A positive PCR was obtained in 29 (76%) of the definitive, 9 (18%) of the probable, and 4 (27%) of the negative PTB samples.
Sensitivity to anti-TB drugs was performed on all positive sputum or NGA cultures. Of the 50 MTB culture-positive patients, 9 (18%) were resistant to isoniazid (INH), 2 (4%) were resistant to rifampicin (RMP), 7 (14%) were resistant to streptomycin (SMP), and 1 (2%) was resistant to ethambutol. In total, 5 (10%) patients were resistant to 2 or more drugs: 4 to INH and SMP and 1 to INH, RMP, SMP, and ethambutol. Thus, multidrug-resistant TB as defined by resistance to both INH and RMP was found in only 1 patient (2%).
Although the majority of TB cases occur in the developing world, much of our understanding of the clinical presentation of pediatric TB comes from the developed world and differs markedly from our experience in Lima, Peru. Indeed, Peruvian children diagnosed with PTB were more likely to be symptomatic at the time of diagnosis, less likely to have a positive Mantoux, less likely to have specific chest radiograph findings, and, if infants, less likely to have a positive culture than patients with PTB from developed countries. To improve the diagnosis of pediatric PTB in the developing world, we identified characteristics predictive of positive MTB culture in Peruvian children: fever, cough for more than 2 weeks, and a Mantoux reaction ≥10 mm.
The majority of our patients with PTB presented with signs of active disease, such as cough for more than 2 weeks and fever. In contrast, the majority of children with PTB in the United States are asymptomatic at the time of diagnosis,5 reflecting the fact that many cases are identified by contact screening and population-based skin testing instead of by evaluation of children with symptomatic disease. Because the burden of MTB infection among children in the developing world is very high, as reflected by the high prevalence of positive PPD skin tests, pediatric PTB cases are both more numerous and present at a more advanced stage of disease. In this environment, screening of active disease is a higher priority for TB control programs than contact investigation and population-based screening. Interestingly, a family contact could be identified in 57% of our patients with definitive PTB and 92% of our probable PTB patients, rates similar to those in reports from developed countries5,,22 but much higher than those seen in other developing countries.23–25 This may reflect the successful case identification component of the Peruvian TB control program.
Infants and children with extrapulmonary TB have been reported to be more symptomatic than older children or children with PTB.5,,22 However, we found no difference in the frequency of presenting symptoms in children with extrapulmonary TB compared with those with PTB. Children in the definitive PTB group reporting weight loss or dyspnea and those in the probable PTB group reporting hemoptysis were significantly older than asymptomatic children. Presentation with other symptoms was not found to be significantly associated with age.
The yield of sputum and NGA cultures and smears in children often is reported to be very low and varies greatly depending on frequency of sampling, techniques used, and stringency of criteria used to define true positive cases. In our patient population we found that culture-positive patients were older than culture-negative patients. This is in contrast to reports from developed countries, where the highest rates of MTB recovery in pediatric populations (up to 70%) are reported from infants.22 Nutritional status, socioeconomic level, and specific symptoms were not different in culture-positive and culture-negative patients.
Radiographic evidence of pulmonary disease often is observed in pediatric patients with PTB,5,,22,26 but such findings are neither sensitive nor specific. Indeed, although the majority of our patients' chest radiographs were abnormal, only 62% of definitive PTB and 65% of probable PTB cases demonstrated patterns highly characteristic of PTB (lymphadenopathy, miliary disease, upper lobe cavities, or primary complex disease). Other pediatric TB studies have reported much higher rates of positive radiographic changes in their patients, perhaps because nonspecific findings such as consolidation and atelectasis were considered.18,,25
In our study, a negative Mantoux test correlated strongly with malnutrition (defined as weight for age <5th percentile NCHS standards), in contrast to a community study in Peru, where little effect was noted.4 Indeed, the levels of antigen-specific anergy noted from other developing countries are greater than those reported from children in developed countries and may reflect concurrent infections as well as malnutrition.5,,11,12,27No relationship between age and anergy was found in our study, although other studies have shown high levels of anergy in children under age 2.5 Furthermore, we did not find a correlation between previous BCG immunization and reaction to PPD.
The significance and value of PCR for diagnosis of pediatric TB remains controversial, especially because confirmation of false-positive results in a hyperendemic setting is almost impossible because culture lacks sufficient sensitivity. Although some recent studies suggest that MTB-specific PCR is no more sensitive than culture in pediatric patients,15–17,28 recent improvement in sensitivity and specificity with heminested PCR assays may ultimately improve their diagnostic importance. Our N2 PCR data must be considered speculative, but they do suggest several observations when analyzed in conjunction with the clinical and microbiologic data described. Our current PCR-based detection in children may be limited by the some of the same factors that limit recovery of MTB by culture from children because children with probable PTB were no more likely than children without clinical PTB to have a positive PCR result. In early disease or subclinical infections, MTB may be present in culture-negative patients with mild or no symptoms, given that one-fifth to one-fourth of both the probable PTB group and the group without clinical PTB had positive N2 PCR results. Because the specificity of the N2 PCR has been shown to be ≥98% in studies done in the United States using the same techniques and controls used in our lab,21 a large number of the PCR-positive culture-negative cases in this high-risk population probably represent early or subclinical MTB disease. Additional studies are needed to evaluate these hypotheses critically.
Resistance to antituberculous drugs in our pediatric patient population was low; only 1 child, who had been previously treated for TB, was noted to have multidrug-resistant MTB. The prevalence of drug resistance in our pediatric population is similar to that which has been reported from Peru in a recent World Health Organization report on global antituberculous drug resistance.29 Because children with MTB are infected in the community and progress from infection to disease more rapidly than adults,30 we suggest that the patterns of drug resistance in children may serve as an early marker of patterns present in the wider community.
Clearly, improved diagnostic tools are needed to identify pediatric patients with PTB because the sensitivity of the current gold standard, sputum or gastric aspirate culture, is approximately 50%. In countries with minimal resources, a pediatric PTB screening protocol, based solely on history and clinical findings, would provide an inexpensive and simple means of identifying patients at highest risk for active pulmonary disease. Most pediatric PTB diagnostic systems that have been proposed for resource-poor countries have included retrospective data such as response to TB treatment; initial laboratory or radiographic examinations such as culture or chest radiograph, which are time-consuming and costly; or historical and clinical data presumed but not proven to be relevant to the diagnosis of pediatric PTB.16,,23,31,32 We have identified a triad of cough lasting more than 2 weeks, fever, and a positive Mantoux test that has a positive predictive value of 73% and sensitivity of 44% for culture-positive disease. Screening for children meeting 2 of these 3 characteristics identifies only slightly fewer of the culture-positive patients in our study (80% vs 92%) than screening for 2 of the 4 Migliore adapted clinical criteria, with the added benefit that our criteria do not require an initial chest radiograph. In areas of the developing world with a similar epidemiology of PTB, we propose that children in the developing world be screened initially for possible PTB using fever, cough for more than 2 weeks, and a Mantoux reaction ≥10 mm. For children meeting 2 out of the 3 criteria, a chest radiograph and sputum or NGA smear and culture would be needed to identify those with active PTB.
This study was supported in part by the National Institutes of Health (NIH Grant AI-135894) and by a grant from the Fogarty International Center and NIH (“Peru Emerging Infections Training and Research,” Grant TW 00910-05).
- Received May 26, 1999.
- Accepted February 27, 2001.
Reprint requests to (R.H.G.) Department of International Health, Johns Hopkins University School of Hygiene and Public Health, 615 N Wolfe St, Baltimore, MD 21205. E-mail:
- TB =
- tuberculosis •
- PTB =
- pulmonary tuberculosis •
- MTB =
- Mycobacterium tuberculosis •
- PPD =
- purified protein derivative •
- PCR =
- polymerase chain reaction •
- NGA =
- nasogastric aspiration •
- AFB =
- acid-fast bacilli •
- NCHS =
- National Center for Health Statistics •
- INH =
- isoniazid •
- RMP =
- rifampicin •
- SMP =
- ↵Suarez PG, Canales R, Alarcon E, Zavala D, Portoarrero J, eds.Tuberculosis in Peru. New Paradigms Facing the New Millennium. Lima, Peru: Ministry of Health, National Tuberculosis Control Program; 1999:67
- Sanghavi D,
- Gilman RH,
- Lescano-Guevara A,
- Checkley W,
- Cabrera L
- Getchell WS,
- Davis CE,
- Gilman J,
- Urueta G,
- Ruiz-Huidobro E,
- Gilman RH
- Starke JR,
- Taylor-Watts KT
- Osborne CM
- Committee on Infectious Diseases
- American Thoracic Society
- Kiehn TE,
- Cammarata R
- Dalovisio JR,
- Montenegro-James S,
- Kemmerly SA,
- et al.
- Connelly Smith K,
- Starke JR,
- Eisenach K,
- Ong LT,
- Denby M
- ↵Kent PT, Kubia GP. Public Health Mycobacteriology: A Guide for the Level III Laboratory. Atlanta, GA: Centers for Disease Control and Prevention; 1985. US DHHS Publ. No. (CDC) 86-216546
- Franzblau SG,
- Witzig RS,
- McLaughlin JC,
- et al.
- Caviedes L,
- Lee TS,
- Gilman RH,
- et al.
- Vallejo JG,
- Ong LT,
- Starke JR
- Houwert KAF,
- Borggreven PA,
- Schaaf HS,
- Nel E,
- Donald PR,
- Stolk J
- ↵Brailey ME. The epidemiologic aspects of Harriet Lane study. In:Tuberculosis in White and Negro Children. Cambridge, MA: Harvard University Press; 1958
- Fourie PB,
- Becker PJ,
- Festenstein F,
- et al.
- Copyright © 2001 American Academy of Pediatrics