Objective: To determine the sensitivity of bacteriologic studies in pediatric pulmonary tuberculosis.

Transcription

1 Microbiology of Pediatric Primary Pulmonary Tuberculosis* José M. Merino, MD; Teresa Alvarez, MD; Manuel Marrero, MD; Sara Ansó, MD; Ana Elvira, MD; Gemma Iglesias, MD; and José B. González, MD Objective: To determine the sensitivity of bacteriologic studies in pediatric pulmonary tuberculosis. Patients and methods: Between January 1988 and December 1996, 104 consecutive patients aged 0 to 18 years received a diagnosis of primary pulmonary tuberculosis at our institution. Demographic, clinical, laboratory, and bacteriologic data were collected. Clinical specimens were studied for acid-fast bacilli detection by Ziehl-Neelsen stain and cultured for Mycobacterium recovery by Lowenstein-Jensen culture medium. Statistical analysis was performed utilizing 2, t tests, and multivariate logistic regression analysis. Results: Bacteriologic results were available for 57 patients (54.8%). A positive smear or culture result for Mycobacterium tuberculosis was obtained in 9 of 54 patients (16.6%) and 25 of 50 patients (50%), respectively. Confirmation of M tuberculosis disease was achieved in 28 patients (49.1%). Ziehl-Neelsen stain and Lowenstein-Jensen culture recovery rates were 10.3% (14 of 135) and 52% (48 of 92) of specimens studied, respectively. Sputum, pleural fluid, and biopsy material cultures yielded M tuberculosis in 55%, 75%, and 63% of patients, respectively. Mean SD age ( years vs years) and number of samples submitted for culture ( vs ) were significantly higher in the confirmed tuberculosis disease group (p < 0.05). The presence of a pleural effusion was also more commonly found in the confirmed tuberculosis disease group (p < 0.05). Conclusion: The sensitivity of bacteriologic studies in pediatric pulmonary tuberculosis disease was 49.1%. Age is the main factor associated with the positivity of culture results. (CHEST 2001; 119: ) Key words: bacteriology; children; Lowenstein; tuberculosis; Ziehl-Neelsen Mycobacterium tuberculosis remains one of the pathogens causing the greatest amount of chronic disease and death throughout the world. 1 In the United States, pediatric rates increased by 15.9% in children 0 to 4 years old and 35.7% in children 5 to 14 years old between 1985 and Confirmation of the diagnosis in this group of patients is often cumbersome because of difficulties in obtaining adequate samples for bacteriology. 3 Even in secondary- or tertiary-care centers, the diagnosis is confirmed in no more than 30 to 40% of patients. 2,4 7 For this reason, in many cases, the diagnosis is based on a history of contact, a positive tuberculin test result, and characteristic abnormalities on chest radiography. 2,4 *From the Pediatric Department, Hospital General Yagüe, Burgos, Spain. Manuscript received October 1, 1999; revision accepted November 28, Correspondence to: José M. Merino, MD, Pediatric Department, Hospital General Yagüe, Avda. del Cid, 96, Burgos, Spain; jmmerino@hgy.es We have retrospectively reviewed our experience in pediatric primary pulmonary tuberculosis between 1988 and 1996 to determine the sensitivity of bacteriologic studies. Materials and Methods We selected from retrospective reviews all cases of primary pulmonary tuberculosis in children 18 years old admitted to our institution, a tertiary-care center with a reference population of 82,106 people under this age. Cases were identified from the hospital coding system and microbiology department records. The diagnosis of primary pulmonary tuberculosis was based on a chest radiograph that showed features suggestive of tuberculosis plus two or more of the following criteria: (1) suggestive symptoms and signs; (2) direct contact with a tuberculous adult, positive or negative smear result; (3) positive tuberculin test result; (4) positive detection of M tuberculosis by stain or culture; and (5) good response to antituberculous chemotherapy. After the clinical evaluation, the chest radiographs were reviewed independently by two groups of radiologists who were blind to the epidemiologic data. Chest radiographic interpretation was made on the basis of the following patterns: (1) 1434 Clinical Investigations

2 parenchymal consolidation, (2) atelectasis, (3) mediastinal lymphadenopathy, (4) pleural effusion, (5) miliary disease, and (6) mixed patterns. Children with chest radiographic findings that were considered normal by radiologists were excluded from further analysis. Children administered isoniazid chemoprophylaxis for an isolated positive tuberculin test result were not considered in this study. We collected information regarding demographics, clinical course, laboratory, and bacteriologic results. Specimens of sputum, gastric washing, pleural fluid, pleural biopsy, cerebrospinal fluid, and others (synovial or ascitic fluid, BAL, adenopathy biopsy) were studied for acid-fast bacilli detection by Ziehl- Neelsen stain and microscopic evaluation. Mycobacterial cultures from the same specimens were performed utilizing Lowenstein- Jensen medium. A case was considered to be confirmed tuberculosis when M tuberculosis was detected by stain or culture in clinical specimens. Tuberculin test was performed by intradermal injection of 2 tuberculin units of purified protein derivative RT 23 (equivalent to 5 tuberculin units of purified protein derivative serum). Induration was measured within 48 to 72 h and recorded in millimeters. A positive test result was considered if the palpable induration was 5 mm. All of the children who received a diagnosis of primary pulmonary tuberculosis were given a 6- to 9-month chemotherapeutic regimen that included isoniazid and rifampicin in all cases. Patients with diagnosed tuberculous meningitis were given a 12- to 18-month chemotherapeutic regimen. Patients were grouped into three categories: those with no bacteriologic studies (the not-performed group), those with positive bacteriologic study results (the positive group), and those with negative bacteriologic study results (the negative group). Statistical analysis was performed utilizing software (Statistical Package for Social Sciences, version 8.0; SPSS; Chicago, IL). Absolute and relative frequencies of statistical variables are described. The association between categorized variables was studied by 2 test. The differences between continuous variables were measured by the analysis of variance and the t test if necessary conditions for test application were present. For establishing the concordance between the radiographic observers, we used the index. A multivariate logistic regression analysis was performed to determine the relations between dependent (confirmed tuberculosis) and independent (age, number of samples for Lowenstein culture, and number of samples for Ziehl-Neelsen stain) variables. These variables were selected by clinical and statistical criteria. Variables were included in the equation if the predicted probability for positive was 0.05 and excluded if predicted probability was 0.1. Statistical significance was assumed at p Results Between January 1988 and December 1996, 114 children aged 0 to 18 years received a diagnosis of primary pulmonary tuberculosis in our hospital. After chest radiographic evaluation, 10 patients (8.7%) had chest radiographic findings deemed normal by the two groups of radiologists blind to the clinical and epidemiologic data, but were deemed, at the time of diagnosis, to have mediastinal lymphadenopathy by the clinicians caring for the patients. None of these patients were found to have bacteriologic evidence of primary pulmonary tuberculosis and were excluded from the study. Of the remaining 104 confirmed cases of tuberculosis, 31 patients (29.8%) and 73 patients (70.2%) received a diagnosis from and were treated by adult and pediatric clinicians, respectively. Table 1 shows the principal epidemiologic and clinical data. There were 47 girls (45.2%) and 57 boys (54.8%). The mean age was significantly lower in the not-performed and negative bacteriologic study groups. No differences in sex ratio and temperature were observed among the three groups. Mean tuberculin skin test results were significantly lower in the positive group (Table 1). However, no differences in the number of patients with negative tuberculin skin test results were seen between the positive and negative groups (4 of 28 patients vs 1 of 29 patients, p 0.160). Erythrocyte sedimentation rate and C- Table 1 Epidemiologic, Clinical, and Skin Test Results of Bacteriologic Study* Variables Not-Performed Group Negative Group Positive Group Total Patients p Value Patients 47 (45.1) 29 (27.9) 28 (26.9) 104 (100) Age, yr Male/female gender, No. 26/21 17/12 14/14 57/47 NS Clinical findings Malaise 3 (6.3) 2 (6.8) 9 (32.1) 14 (13.4) 0.05 Anorexia 15 (31.9) 14 (48.2) 8 (28.5) 37 (35.5) NS Fatigue 4 (8.5) 10 (34.4) 11 (39.2) 25 (24) 0.05 Respiratory distress 4 (8.5) 2 (6.8) 5 (17.8) 11 (10.5) NS Thoracic pain 2 (4.2) 11 (37.9) 16 (57.1) 29 (27.8) 0.05 Weight loss 3 (6.3) 7 (24.1) 3 (10.7) 13 (12.5) NS Cough 29 (61.7) 13 (44.8) 18 (64.2) 60 (57.7) NS Temperature, C NS Tuberculin test, mm *Data are presented as No. (%) or mean SD. NS not significant. p 0.05 between negative group and/or positive group and not-performed group. p 0.05 between negative group and positive group. CHEST / 119 / 5/ MAY,

3 reactive protein levels were significantly higher in the positive group (Table 2). The presence of a pleural effusion was significantly more frequent in the positive group (p 0.05). However, mediastinal lymphadenopathy was more common in the other two groups. No other radiologic pattern differences among the three groups were observed (Table 2). There was a good agreement between the two groups of radiologists ( index of 0.62). Bacteriologic results were available for 57 patients (54.8%), 29 of 31 patients (93.5%) and 28 of 73 patients (38.3%) who were managed by adult and pediatric clinicians, respectively (p 0.05). Of the specimens obtained from these 57 patients, 9 of 54 specimens (16.6%) sent for study by Ziehl-Neelsen staining had positive results and 25 of 50 specimens (50%) sent for Lowenstein-Jensen culture grew M tuberculosis. Overall, confirmation of M tuberculosis disease was achieved in 28 patients (49.1%; 25 patients by the presence of positive culture results and 3 patients by the presence of positive Ziehl- Neelsen staining alone; Tables 3, 4). Patients cared for by adult clinicians showed a significantly higher culture-positive rate (73% vs 25%, p 0.05) and number of specimens (mean SD) submitted per patient for Lowenstein-culture ( vs , p 0.05). Bacteriologic results of clinical specimens are given in Table 4. Sputum, gastric washing, pleural fluid, and biopsy material cultures yielded M tuberculosis in 55%, 15%, 75%, and 63% of cases, respectively. The number of samples submitted for culture per patient was significantly higher in the positive group ( vs , p 0.05; Table 3). However, Ziehl-Neelsen stain recovery rates were very low in our study. Only 14 of 135 samples processed (10.3%) yielded a positive staining pattern. Discussion The diagnosis of tuberculosis in childhood continues to be surrounded by considerable uncertainty. 4 In previous reports, many of the children in whom sufficient criteria were present and were considered probable cases were subsequently thought not to have tuberculosis. 4 However, up to 10% of the children with confirmed tuberculosis had a normal chest radiographic finding. 4,8,9 Tuberculosis is particularly difficult to diagnose in children because of the poor yield of standard laboratory testing and the lack of characteristic symptoms. 10 Bacteriologic results were available for 45 to 83.6% of cases in previously reported series. 2,4 6,11,12 In our series, bacteriologic results were available for 54.8% of cases The mean age of patients without bacteriologic specimens was significantly lower ( years vs years, p 0.05) in our series (Table 1). This probably represents the difficulty in obtaining the appropriate specimens in younger patients. However, age has not been a problem for confirming the diagnosis in other series. 4 In our review, patients 10 years old (45 of 104 patients, 43.2%) showed a significant higher percentage of positive samples submitted for Ziehl-Neelsen stain (22.5% vs 0%, p 0.05) and Lowenstein-Jensen culture (63.8% vs 14%. p 0.05) than patients 10 years old (59 of 104 patients, 56.7%). Moreover, the number of samples submitted per patient for Lowenstein- Jensen culture ( vs , p 0.05) was significantly higher in patients 10 years old. In reported pediatric series, 22 to 42% of specimens tested were found to be culture positive for M tuberculosis. 2,4,6,10,11 In adult patients, recovery rates of Mycobacteria from Lowenstein-Jensen medium Table 2 Laboratory and Radiologic Findings of Bacteriologic Study* Findings Not-Performed Group (n 47) Negative Group (n 29) Positive Group (n 28) Total (n 104) p Value Laboratory Leukocyte count, 10 9 /L Neutrophils, % Lactate dehydrogenase, microkatal/l C-reactive protein, mg/l Erythrocyte sedimentation rate, mm/h Radiographic Atelectasis 3 (6.3) 4 (13.7) 5 (17.8) 12 (11.5) NS Parenchymal consolidation 23 (48.9) 14 (48.2) 15 (53.5) 52 (50) NS Mediastinal lymphadenopathy 32 (68) 18 (62) 11 (39.2) 61 (58.6) 0.05 Pleural effusion 1 (2.1) 8 (27.5) 14 (50) 23 (22.1) 0.05 Miliary tuberculosis 3 (6.3) 1 (3.4) 3 (10.7) 7 (6.7) NS *Data are presented as mean SD or No. (%). See Table 1 for expansion of abbreviation. p 0.05 between negative group and/or positive group and not-performed group. p 0.05 between negative group and positive group Clinical Investigations

4 Table 3 Bacteriologic Results* Variables Negative Group (n 29) Positive Group (n 28) Total Patients (n 57) p Value Ziehl-Neelsen stains NS Lowenstein-Jensen cultures Sputum NS Gastric washing NS Spinal fluid Pleural fluid Pleural biopsy Synovial and ascitic fluid NS Total *Data are presented as mean SD No. of specimens per patient. See Table 1 for expansion of abbreviation. ranged from 40 to 70% In our study, a total of 92 clinical specimens were processed. These 92 specimens yielded 48 isolates (52%; Table 4). Mean growth detection times ranged from 12.5 to days ( days in our study). 13 The sensitivity of Lowenstein-Jensen culture was similar to that seen with adult patients in other published reports (52%). Sputum, pleural fluid, and biopsy material culture findings showed a good yield (55%, 75%, and 63% of samples processed, respectively) in our study. However, only 15% of gastric washing specimens cultured by Lowenstein-Jensen media showed positive results. Age (r 0.58, p 0.01), number of samples submitted for culture (r 0.83, p 0.01), and caregiver (r 0.53, p 0.01) correlated with the positivity of culture results. However, after logistic regression analysis, only age predicted the positivity of culture results. Older children have a higher likelihood of having a positive smear result, probably because they are Table 4 Bacteriologic Results; Sensitivity of Test* Variables Patients Specimens Growth Detection Time, d Lowenstein-Jensen culture Sputum 18/7 (38) 27/15 (55) Gastric washing 15/4 (26) 19/3 (15) Spinal fluid 2/2 (100) 2/2 (100) Pleural fluid 19/14 (73) 20/15 (75) Pleural biopsy 11/7 (63) 11/7 (63) Synovial and ascitic fluid 7/3 (42) 13/6 (46) Total 50/25 (50) 92/48 (52) Ziehl-Nielsen stains Sputum 32/4 (12) 63/8 (12) Gastric washing 22/1 (4) 30/1 (3) Spinal fluid 2/0 (0) 2/0 (0) Pleural fluid 19/0 (0) 21/0 (0) Pleural biopsy 8/4 (50) 8/4 (50) Synovial and ascitic fluid 8/1 (12) 11/1 (9) Total 54/9 (16.6) 135/14 (10.3) *Data are presented as total patients/no. positive (%) or mean SD. more likely to have a higher burden of organisms because of an increased chance of developing cavities, due to a more mature delayed-type hypersensitivity response. In our series, the presence of cavitations was more commonly found in children 10 years old as opposed to those 10 years old (4 of 45 children vs 0 of 59 children, p 0.05). Lowenstein- Jensen cultures yielded M tuberculosis in all patients who showed cavities on chest radiograph (9 of 9 specimens cultured, 100%), and 8 of 17 specimens from these patients (47%) showed a positive staining pattern. Ziehl-Neelsen recovery rates were very low in our study (10.3%, Table 4), similar to other reported series. 2,4,6 Only 12% (8 of 39) and 3% (1 of 30) of sputum and gastric washing samples, respectively, submitted for acid-fast bacilli detection showed positive findings. All patients 10 years old showed negative Ziehl-Neelsen stain results (0 of 14 patients) but 21.9% (9 of 41 patient) 10 years old had positive findings for acid-fast bacilli detection. The sensitivity of Ziehl-Neelsen stain was 16.6% (9 of 54) in our series. Our data support that the confirmation of M tuberculosis disease in pediatric primary pulmonary tuberculosis should be attempted, especially in patients 10 years old. ACKNOWLEDGMENT: We thank Dr. Francisco González Vilchez and Dr. José Cordero for their critical review of the article and helpful suggestions. References 1 Starke JR, Jacobs RF, Jereb J. Resurgence of tuberculosis in children. J Pediatr 1992; 120: Pineda PR, Leung A, Müller NL, et al. Intrathoracic paediatric tuberculosis: a report of 202 cases. Tuberc Lung Dis 1993; 74: Dutt AK, Moers D, Stead WW. Tuberculous pleural effusion: 6-month therapy with isoniazid and rifampin. Am Rev Respir Dis 1992; 15: Schaaf HS, Beyers N, Gie RP, et al. Respiratory tuberculosis in childhood: the diagnostic value of clinical features and CHEST / 119 / 5/ MAY,

5 special investigation. Pediatr Infect Dis J 1995; 14: Vidal ML, Del Cerro MJ, García MJ, et al. Tuberculosis pulmonar en la infancia: a propósito de 149 casos. An Esp Pediatr 1990; 32: Driver CR, Luallen JJ, Good WE, et al. Tuberculosis in children younger than five years old: New York City. Pediatr Infect Dis J 1995; 14: Agrons GA, Markowitz RI, Kramer SS. Pulmonary tuberculosis in children. Semin Roentgenol 1993; 2: Fox TG. Occult tuberculous infection in children. Tubercle 1977; 58: Toppet M, Malfroot A, Hofman B, et al. Tuberculosis in children: a 13 year follow-up of 1714 patients in a Belgian home care center. Eur J Pediatr 1991; 150: Smith KC, Starke JR, Eisenach K, et al. Detection of Mycobacterium tuberculosis in clinical specimens from children using a polymerase chain reaction. Pediatrics 1996; 97: Nemir RL, O Hare D. Tuberculosis in children 10 years of age and younger: three decades of experience during the chemotherapeutic era. Pediatrics 1991; 88: Medical Research Council Cardiothoracic Epidemiology Group. Tuberculosis in children: a national survey of notifications in England and Wales in Arch Dis Child 1994; 70: Palaci M, Mizuka SY, Sato DN, et al. Evaluation of Mycobacteria growth indicator tube for recovery and drug susceptibility testing of Mycobacterium tuberculosis isolates from respiratory specimens. J Clin Microbiol 1996; 34: Anargyros P, Astill DSJ, Lim ISL. Comparison of improved BACTEC and Lowenstein-Jensen media for culture of Mycobacteria from clinical specimens. J Clin Microbiol 1990; 28: Carpentier E, Drouillard B, Dailloux M, et al. Diagnosis of tuberculosis by Amplicor Mycobacterium tuberculosis test: a multicenter study. J Clin Microbiol 1995; 33: Wilson ML, Stone BL, Hildred MV, et al. Comparison of recovery rates for Mycobacteria from BACTEC 12B vials, Middlerbrook 7H11-selective 7H11 biplates, and Lowenstein-Jensen slants in a public health mycobacteriology laboratory. J Clin Microbiol 1995; 33: Clinical Investigations