Patients with pulmonary tuberculosis (TB) having

Transcription

1 Number of Negative Acid-Fast Smears Needed To Adequately Assess Infectivity of Patients With Pulmonary Tuberculosis* George Mixides, MD, FCCP; Vasanti Shende, MD; Larry D. Teeter, PhD; Robert Awe, MD ; James M. Musser, MD, PhD; and Edward A. Graviss, PhD, MPH Study objectives: To investigate the relationship between the number of negative acid-fast bacilli (AFB) smear results and infectivity of pulmonary tuberculosis (TB). Design: Retrospective analysis. Methods and subjects: We examined 122 index cases in Harris County, TX, reported in 1998 and All cases had only negative AFB smear results during the infectious period and were categorized in two groups: group A consisted of cases with only one or two sputum specimens collected and processed, and group B consisted of cases with at least three sputum specimens or at least one bronchoscopic specimen. Tuberculin skin test (TST) results of contacts were ascertained from the results of contact investigations performed by the City of Houston Department of Health and Human Services, Tuberculosis Control Division. Univariate and multivariate analyses were done to explore index case and contact attributes associated with tuberculosis (TB) transmission using positive TST results of contacts as a measure of recent transmission. Results: We found male gender and younger age of index cases along with Hispanic ethnicity of contacts to be independently associated with positive TST results, while younger contacts were less likely to be TST positive. Smear category of the index case (group A vs group B) was not independently associated with transmission. We also found that the first two sputum specimens in cases where three or more were performed yielded 90% of all positive culture results for Mycobacterium tuberculosis (MTB). Conclusions: We conclude that two sputum specimens negative for AFB stain are adequate for both assessing infectivity and for isolating MTB from patients with pulmonary TB. (CHEST 2005; 128: ) Key words: infectivity; sputum smear; transmission; tuberculin; tuberculosis Abbreviations: AFB acid-fast bacilli; CI confidence interval; HTI Houston Tuberculosis Initiative; IP infectious period; MTB Mycobacterium tuberculosis; OR odds ratio; OTC other than close; TB tuberculosis; TB Control Tuberculosis Control Division; TST tuberculin skin test Patients with pulmonary tuberculosis (TB) having consistently negative smears for acid-fast bacilli (AFB) are less infectious than patients with positive smear results as shown by various epidemiologic *From the Department of Internal Medicine (Drs. Mixides, Shende, and Awe), Section of Pulmonary and Critical Care Medicine, and Department of Pathology (Drs. Teeter and Graviss), Baylor College of Medicine, Houston, TX; and Laboratory of Human Bacterial Pathogenesis (Dr. Musser), Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Hamilton, MT. Deceased. This work was performed at Baylor College of Medicine. Funded in part by federal funds from the National Institute of Allergy and Infectious Disease, National Institutes of Health (N01-AO and DA09238). studies. 1 3 A positive AFB smear result is considered a major determinant as to whether a pulmonary TB case will transmit the bacilli to a contact. 4 However, according to a study 5 that utilized molecular characterization of Mycobacterium tuberculosis (MTB) isolates, smear-negative cases are still responsible for approximately 17% of recently transmitted TB. The Manuscript received September 17, 2004; revision accepted December 9, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( org/misc/reprints.shtml). Correspondence to: Edward A. Graviss, PhD, MPH, Department of Pathology, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030; egraviss@bcm.tmc.edu 108 Clinical Investigations

2 latter observation lends support to the current Centers for Disease Control and Prevention recommendations stating that in order to discontinue airborne precautions, patients with pulmonary TB in addition to having three negative, consecutive, sputum smear results must be receiving efficacious therapy and show adequate clinical response. 6 However, there is a paucity of data supporting a specific number of negative smear results as adequate in assessing the infectious potential of such patients. In addition, some authors have suggested that airborne precautions be discontinued after two negative smear results have been collected before 7,8 or during adequate treatment, 9 although direct evidence that this would not result in higher disease transmission is lacking. In order to investigate the impact of the number of negative smear results on transmission of pulmonary TB, we conducted a retrospective study of patients (index cases) and their corresponding contact investigations. We included cases with only negative AFB smear results collected during a specified infectious period. We then grouped and compared the index cases according to the number of specimens collected from each. The rationale for such a study design is as follows: When the initial two sputum specimens collected from an active pulmonary TB case are smear negative, according to current guidelines 6 at least one more specimen is needed in order to fully assess whether a patient is infectious. In theory, this third specimen can be smear positive and can thus signify a case with a higher infectious potential, information that would be missed if this third specimen is not collected. However, if disease transmission from cases in which a third specimen is not collected is similar to that of cases with three or more negative smears results, one may conclude that in practice a third or later specimen adds little information on evaluating the infectious potential of the patient. Index Case Selection Methods and Materials All index cases meeting the following criteria were initially included in this study: (1) culture-proven pulmonary TB; (2) no positive smear results for AFB from pulmonary specimens collected during the infectious period (as defined below); and (3) reported to the City of Houston Tuberculosis Control Division (TB Control) during 1998 and The infectious period (IP) was defined as the time from the start of cough and/or hemoptysis until 14 days after the start of effective therapy (at least two drugs with in vitro activity) or, if therapy was not started, until the time the patient died or moved out of Harris County. When neither cough nor hemoptysis were reported, the date of the first specimen collection was considered the start of the IP. The above information was extracted from the Houston Tuberculosis Initiative (HTI) database. The HTI is an ongoing population-based project involving TB patient interview, medical record review, and molecular characterization of MTB isolates. The HTI has been enrolling Harris County (Houston), TX, TB cases since October Cases were cross-referenced with the City of Houston TB Control database to confirm that no positive AFB smear results were ever reported (during the period noted above) and that all specimens collected and processed were captured. Patient demographics and additional factors potentially related to disease transmission and severity were extracted from the HTI database. Culture results together with the timing and type of respiratory specimens were also recorded. The HTI project was approved by the Baylor College of Medicine and Affiliated Hospitals Institutional Review Board, and all enrolled cases signed informed consent, allowing access to contact investigation data for each index case. Index Case Groups For the purpose of statistical analysis, index cases were classified in two groups: group A included those with one or two sputum smears for AFB collected and processed during the IP, and group B included those with three or more sputum AFB smears or at least one smear from a bronchoscopic specimen collected during the IP. In cases in which more than one specimen was collected the same day, the specimens were considered as one for the purpose of the above classification, but all had to be smear negative in order to be included in the study. Contact Investigation Trained workers from City of Houston TB Control performed contact investigation on index cases according to the American Thoracic Society recommendations. 4 Briefly, the investigation protocol was as follows: all index cases were interviewed to identify all possible contacts. Then, all household contacts were investigated as well as anyone who shared the same closed space with an index case for at least 4 h/wk during a period of at least 3 months prior to TB diagnosis. Household contact was defined as anyone staying in the same house as his/her permanent residence at the time of TB diagnosis. The above contacts were defined as close contacts for our analysis. The remaining contacts were defined as other than close (OTC). Contact investigation was expanded if 30% of the close contacts were positive tuberculin skin test (TST) reactors. The first TST is administered within 2 weeks of identification of a contact using five tuberculin units of purified protein derivative of standard strength. The TST was repeated in 3 months if the first test result was negative. Contacts with previously documented positive TST results or contacts treated for TB in the past were not tested with TST, but underwent chest radiography to assess for active disease. These individuals were classified for our analysis as nonsusceptible. Contact Classification Contacts with at least a 5-mm induration on the first or second TST were considered positive (infected by the index case). 11 Those with both TSTs 0 to 4 mm were considered negative (not infected). If only one TST was performed, was 0 to 4 mm, and was performed at least 8 weeks after the end of the IP, a contact was also considered negative. 12 If only one TST was performed with 0- to 4-mm induration 8 weeks from the end of the IP, or no TSTs were done, the contact was classified as incompletely investigated. CHEST / 128 / 1/ JULY,

3 Specimens Respiratory specimens consisting of spontaneous or induced sputum and bronchoscopic washings were collected, decontaminated, concentrated, stained, and cultured according to Centers for Disease Control and Prevention guidelines 13 by various hospital public and private laboratories. DNA probes were used for identification of MTB. 14 All microscopic examinations of smears for AFB were performed using the auramine-rhodamine staining method. 15 Statistical Analysis Index cases were included in the analysis if they had at least one susceptible contact adequately investigated. Susceptible was defined as a contact with no previously documented positive TST or documented history of active TB. A univariate analysis was performed to explore associations between index case characteristics and positive TST result. Infection rate was defined as the ratio TST-positive susceptible contacts to the total number of completely investigated susceptible contacts. 16 This was used as a measure of infectivity of the index case. Contacts who were incompletely investigated or nonsusceptible were not included in the analysis. Variables with univariate p values 0.25 were added to a multiple logistic regression model to explore any independent associations with positive TST results. 17 Variables with p values 0.05 after adjustment for other covariates were removed from the final model. The multivariate model initially included all susceptible completely investigated contacts and was then repeated using only the close contacts. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using logistic regression. A p value 0.05 was considered significant. Statistical software (STATA 7.0SE; StataCorp; College Station, TX) was used to analyze the data. Index Cases Results A total of 601 TB cases with MTB culture-positive respiratory specimens were reported in Harris County in 1998 and 1999, including 152 cases (21%) that met study criteria. Of these specific cases two met criteria for cross contamination, 18 one had specimens collected only after being on therapy for 14 days, and one had an unknown treatment regimen before entering the United States. These four cases were not used in subsequent analyses. Median age of the remaining 148 cases was 40 years (range, 14 to 85 years). Sixteen cases (10.8%) had no contacts identified; in another 10 cases (6.8%), none of the contacts were completely investigated. The remaining 122 cases had a total of 420 sputum and 12 bronchoscopic specimens during the IP: 1 sputum AFB smear was performed on 11 cases (9%), 2 were performed on 29 cases (24%), 3 were performed on 70 cases (57%) [median, 4; range, 3 to 8], and at least 1 bronchoscopic specimen in 12 cases (10%). In 3 of the 122 index cases, the end of the IP was defined by the date of death and, in 1 case, by the date the patient moved out of the area. In 40 cases (32.8%), neither cough nor hemoptysis were reported, and thus the date of first specimen was used as start date. (These cases either reported symptoms other than cough or hemoptysis, ie, fever and weight loss, and/or diagnosis was made through chest radiographic screening, ie, on entry to jail or on immigration.) No case had a diagnosis of laryngeal TB. Two cases had an isolate resistant to a drug in addition to isoniazid (one resistant also to streptomycin and ethambutol, and the other also resistant to streptomycin). There were no significant differences between included and excluded index cases in regards to age, sex, or presence of cough (data not shown). However, included cases were less likely to be white (OR, 0.21; p 0.001). Contact Investigation The 122 index cases had 1,577 contacts identified through contact investigation. This investigation was incomplete in 233 cases (15%), from whom 178 cases did not have TSTs completed and 55 individuals had only one TST result that was 5 mm and were not repeated at least 8 weeks later. Another 163 individuals (10%) had documented previously positive TST results or were treated for active TB prior to contact with the current index case. A total of susceptible 1,181 contacts were completely investigated regarding TST outcomes, and used for analysis. Median age was 26 years (range, 1 month to 93 years). Among these contacts, 343 persons were close contacts (29%) and 838 persons (71%) were OTC contacts. The first (and only) TST result was positive in 277 contacts, the second TST result was positive (conversion) in 44 contacts, and in the remaining 860 contacts the TST results were negative. TST results were positive in 134 of the close contacts (39%) and in 187 of OTC contacts (22%). Active TB developed in 9 of the 1,181 contacts within 2 years of exposure to the index case. Culture Yield In patients who had at least three sputum specimens, the first or second specimen grew MTB in 90% of cases (data not shown). Variables Associated With Positive TST Results Univariate analysis of characteristics for index cases and contacts in relation to TST results of contacts are shown in Tables 1, 2, respectively. Variables with univariate p values 0.25 were then introduced in a multiple logistic regression model first using all completely investigated susceptible contacts and then using only close contacts. Younger age and male gender of index cases were indepen- 110 Clinical Investigations

4 Table 1 Univariate Analysis of Index Case Characteristics Associated With Positive TST Results Characteristics of Index Cases Index Cases, No.* Positive Infection Rate, TST Results/Total TSTs (%) OR (95% CI) p Value Gender Male /795 (30) 1.49 ( ) 0.01 Female 37 85/386 (22) 1 Age, yr 40 (median) /750 (30) 1.59 ( ) /431 (22) 1 Race/ethnicity Black 36 82/369 (22) 1 White 14 18/93 (19) 0.84 ( ) 0.55 Hispanic /574 (34) 1.81 ( ) Asian 24 25/145 (17) 0.73 ( ) 0.21 Country of birth Foreign country /741 (33) 2.37 ( ) United States/Canada 56 76/440 (17) 1 Ever incarcerated Yes 50 (114) 80/356 (22) 0.78 ( ) 0.11 No /673 (27) 1 Ever homeless Yes 20 (114) 21/137 (15) 0.49 ( ) 0.01 No /892 (27) 1 Household income, US$/yr 10, (108) 87/335 (26) 1.04 ( ) , /656 (25) 1 Persons in household 4 41 (114) 139/455 (31) 1.61 ( ) /574 (21) 1 Illicit IV drug use Yes 12 (114) 14/77 (18) 0.63 ( ) 0.13 No /952 (26) 1 Infectious period, d Below median /611 (26) 0.92 ( ) 0.51 Above median /570 (28) 1 HIV status Seropositive 23 30/174 (17) 0.51 ( ) 0.05 Seronegative /1,007 (29) 1 Dead within 6 mo Yes 6 (121) 10/74 (14) 0.40 ( ) 0.01 No /1,105 (28) 1 Drug failure Present 5 20/51 (39) 1.78 ( ) 0.05 Absent /1,130 (27) 1 Isolate in vitro susceptibility Isoniazid resistant 8 23/75 (39) 1.20 ( ) 0.48 Isoniazid susceptible /1,106 (27) 1 Cavity on chest radiograph Present 35 (113) 103/395 (26) 1.05 ( ) 0.76 Absent /658 (25) Cultures after 14 d of treatment Positive for MTB 29 65/184 (35) 1.58 ( ) 0.01 Negative for MTB /997 (26) 1 MTB-positive cultures during IP /361 (31) 1.34 ( ) /820 (25) 1 First specimen to treatment, d 17 (median) 58 (118) 146/515 (28) 1.05 ( ) /511 (31) 1 Smears during IP Group A 40 82/356 (23) 0.73 ( ) 0.05 Group B /825 (29) 1 *n 122 unless indicated in parentheses (due to missing data). Within 6 months of first positive culture result for MTB. Defined as positive culture result for MTB after 3 months of therapy. Three index cases did not receive therapy. CHEST / 128 / 1/ JULY,

5 Table 2 Univariate Analysis of Contact Characteristics Associated With Positive TST Results Characteristics of Contacts Positive Infection Rate, TST Results/ Total TSTs (%) OR (95% CI) p Value Gender Male 164/597 (27) 1.04 ( ) 0.75 Female 155/582 (27) 1 Age, yr 26 (median) 123/587 (21) 0.53 ( ) /578 (34) 1 Race/ethnicity Black 49/303 (16) 1 White 23/207 (11) 0.65 ( ) 0.11 Hispanic 185/501 (37) 3.03 ( ) Asian 62/165 (38) 3.12 ( ) Type of contact Close 134/343 (39) 2.23 ( ) OTC 187/838 (22) 1 dently associated with positive TST, as well as Hispanic ethnicity of the contacts. Younger contacts were less likely to be TST positive. In a model that included only close contacts, only Hispanic ethnicity and younger age of contacts remained significantly associated with positive TST (Table 3). Smear group of the index cases (group A vs group B) was not independently associated with positive TST results in either of the two models (Table 3). Discussion Our findings suggest that index cases assessed with only two smear-negative sputum specimens can be considered no more infectious than cases with at least three smear-negative specimens. Previous studies have concentrated on the likelihood of a positive smear or culture in relation to the order of specimen collection. Craft and coworkers 7 found no positive smears results after two AFB-negative sputum samples. Other investigators 8,19,20 also found a low likelihood of the third smear being positive after a two negative results, ranging from 0.2 to 3.2%. Nelson et al, 21 however, found that the probability of a smear being positive for AFB following two negative results was 13%. Telzak et al, 9 using a different approach, investigated smear-positive cases who converted to consecutively negative smears after receiving effective therapy; only 4 of 100 cases had a repeatedly positive smear results after two negative specimens. None of the above or other investigations to our knowledge have examined the infection rates resulting from index cases in which less than three smears were collected. In our study, we also found a high positive culture yield of the first two sputum specimens (90%), which is in agreement with other studies (93 to 99%). 7,21 23 Although not the focus of this study, a third sputum specimen may still be useful if a clinician wants to maximize the diagnostic yield of sputum specimens before embarking on more invasive testing such as bronchoscopy and lung biopsy. Our study is unique in that we retrospectively evaluated transmission of TB to contacts exposed to index cases with less than three negative smear results. We were able to assess the infectivity of cases with only one or two smear-negative sputum specimens collected and compare it with cases in which more than two negative smears were available. We chose to classify cases in groups of one or two sputum samples (group A) vs at least three sputum samples or at least one bronchoscopic specimen (group B), because three specimens are recommended in the current guidelines. 6 Due to the small number of cases with only one sputum sample collected (n 11), insufficient statistical power was available to compare these cases to those with more than one specimen. Because the chance of an AFB smear becoming positive for the first time decreases with an increasing number of negative smears preceding it, 7,19 21 then potentially group A should be more infectious than group B. This is true because patients who only have one sputum smear could in theory have been identified as smear positive if more sputum smears were to be done. It has also been shown that bronchoscopic specimens can be AFB smear positive even if the patient already had three negative sputum smears. 24,25 Therefore, patients with AFB-negative smears collected by bronchoscopy can be considered even less infectious than Table 3 Results of the Multivariate Logistic Regression Analysis Characteristic* OR 95% CI p Value All contacts Index case attributes Male gender Age 40 yr Group A vs group B Contact attributes Hispanic ethnicity Close contact Age 26 yr Close contacts Index case attributes Group A vs group B Contact attributes Hispanic ethnicity Age 26 yr *Variables with p 0.05 (nonsignificant) are not shown except for smear groups A vs B. 112 Clinical Investigations

6 cases with three negative sputum smears. Including patients with bronchoscopic specimens in group B would also favor higher infectivity in group A. Thus, groups A and B were so defined as to increase the chance of finding a higher infectivity rate for group A patients if indeed there was one. By excluding specimens collected after the end of a defined IP, we were able to stratify the effect of treatment on infectivity of all index cases. There is considerable literature supporting that a patient is no longer infectious after receiving effective therapy for at least 14 days Although this is not universally accepted, 29 it seems more likely to be true for patients with negative smear results prior to initiating treatment than those with positive smear results. 30 To further explore this issue, we introduced a variable accounting for patients who had positive culture findings after 14 days of therapy in our logistic regression models. We did not find this variable to be independently associated with transmission, thus supporting the above statement. 30 While defining the end of the IP was well documented and accurate, the start of IP (and thus duration) was less so, because it was based only on patient recall. However, since contact investigation covered at least 3 months prior to the diagnosis of TB regardless of reported duration of symptoms, we do not think that this lack of accuracy influenced assessment of infectivity. The higher infectivity noted in Hispanics might be due to a higher incidence of background positive TST (including foreign-born contacts) or more crowding in the household. In Harris County, Hispanic TB cases have larger households than blacks or whites, but not Asians (median, 4 persons vs 3 persons vs 2 persons vs 4 persons, respectively) [HTI database 1995 to 2001; unpublished data]. Also, when analysis was restricted to contacts of ages 1 to 15 years, Hispanic ethnicity was still significantly associated with TST positivity (OR, 2.13; p 0.01). Household crowding may be responsible for this association. However, when the analysis was restricted to OTC contacts only, Hispanic ethnicity was still significantly associated with a positive TST result (OR, 2.02; p 0.001), which favors increased incidence of background positive TST in this ethnic group. Similar findings have been reported elsewhere. 31 A higher background of TST positivity among Hispanics may be due to a higher proportion of foreign birth in this group and consequently a higher background level of previously undocumented latent TB infection and perhaps bacille Calmette-guérin vaccination. Regrettably, neither country of birth nor bacille Calmette-guérin vaccination history were available for most contacts; however, ethnicity was obtained. When we studied the ethnicity of contacts, we found that the contacts of patients in group A were less likely to be Hispanic than contacts of patients in group B (p 0.001), although contacts in group A were also more likely to be household contacts (p 0.05). The higher infectivity of male and younger index cases may be related to their ability to generate a more forceful cough. Lower infectivity associated with younger age of contacts has been noted by others 32,33 and seems to be mostly observed in early childhood. Of interest is the association we found between white race and the inability to adequately investigate or identify any contacts. We did not explore this finding further as this was not within the scope of this study and specific information on possible reasons was not readily available. Our findings of no difference in infectivity of HIV-seropositive vs HIV-seronegative patients are different than similar reports, but are in agreement with a metaanalysis. 35 However, if one considers that the reason HIV patients might be less infectious is because of a lower bacterial load than HIV-seronegative patients, 32 then our findings are not surprising. In a report by Elliott and coworkers, 36 the association of HIV with infectivity was weakened when an adjustment was made for the bacterial load according to sputum smear examination. Similarly, in our study bacterial load, at least as evidenced by smear status, was adjusted for by design since all smears of all included patients were negative. Low mycobacterial load may also explain the lack of association between infection rates and presence of cavities on radiography. There are several limitations inherent to this study design. The deficiencies of traditional contact investigation have received increased attention recently. 31,37 One such deficiency is distinguishing a recent tuberculin conversion from a remote one when the first TST result is positive during a contact investigation and a TST was not done before the exposure in question. Molecular epidemiology based on molecular characterization techniques is thought to be more accurate in detecting recent transmission. 38 However, the current molecular approach ignores persons with only latent TB infection who constitute the majority of secondary infections after exposure to an active TB case. 39 Although a number of contacts were not adequately investigated, the distribution of incompletely investigated contacts was similar between groups A and B (OR, 1.27; p 0.12). Similarly, the distribution of nonsusceptible contacts was similar between the two groups (OR, 1.02; p 0.93). We attempted to minimize bias by including many parameters in our univariate analysis potentially associated with infectivity (Table 1). Although accurate information on sputum volume was not available, we did adjust for closely related indexes of infectivity, CHEST / 128 / 1/ JULY,

7 namely duration of IP and, perhaps more importantly, the number of positive culture results per patient during the IP. 40 Finally, our study did not include index cases with multidrug-resistant TB, and it could be argued that our conclusions cannot be generalized to this setting. However, evidence indicates that these cases are equally or less infectious than drug-sensitive TB. 41,42 In conclusion, we found that collecting two sputum specimens from patients with pulmonary TB is adequate for assessing the infectious potential of MTB. We believe that the reasons for finding similar infectivity of cases belonging to group A compared with those of group B is twofold. First, the likelihood of a third sputum specimen being smear positive is low 7,19,20 ; second, even if the result is positive, the number of bacilli in a third specimen is much smaller compared to positive smears of the first specimen, 23 signifying a lower bacterial load in such patients. Thus, our findings indirectly support the recommendation that patients with two negative smears and receiving efficacious therapy can be safely removed from respiratory isolation. 9 This shortening of the isolation period would promote decreased hospital costs 7 and perhaps adverse events. 43 However, since efficacy of therapy (before culture results are known) is presumably based on clinical response, clinician judgment still plays a major role. If a hospitalized patient with an unspecified respiratory disease is placed in an isolation room and is not receiving effective TB therapy, discontinuing isolation is justified only if the pretest probability for TB is quite low. 8 This decision must be weighed with the knowledge that smear-negative patients can still transmit TB disease. 5 If a clinician chooses to use negative smears as one of the parameters to assess the probability of TB transmission, two negative smears seem to be as informative as three or more negative smears. References 1 Van Zwanenberg D. The influence for the number of bacilli on the development of tuberculous disease in children. Am Rev Respir Dis 1960; 82: Shaw JB, Wynn-Williams N. Infectivity of pulmonary tuberculosis in relation to sputum status. Am Rev Tuberc 1954; 69: del Castillo Otero D, Penafiel Colas M, Alvarez Gutierrez F, et al. Investigation of tuberculosis contacts in a nonhospital pneumology practice. Eur J Clin Microbiol Infect Dis 1999; 18: Control of tuberculosis in the United States. American Thoracic Society. Am Rev Respir Dis 1992; 146: Behr MA, Warren SA, Salamon H, et al. Transmission of Mycobacterium tuberculosis from patients smear-negative for acid-fast bacilli. Lancet 1999; 353: Guidelines for preventing the transmission of Mycobacterium tuberculosis in health-care facilities, 1994: Centers for Disease Control and Prevention. MMWR Recomm Rep 1994; 43: Craft DW, Jones MC, Blanchet CN, et al. Value of examining three acid-fast bacillus sputum smears for removal of patients suspected of having tuberculosis from the airborne precautions category. J Clin Microbiol 2000; 38: Mathew P, Kuo YH, Vazirani B, et al. Are three sputum acid-fast bacillus smears necessary for discontinuing tuberculosis isolation? J Clin Microbiol 2002; 40: Telzak EE, Fazal BA, Pollard CL, et al. Factors influencing time to sputum conversion among patients with smearpositive pulmonary tuberculosis. Clin Infect Dis 1997; 25: El Sahly HM, Adams GJ, Soini H, et al. Epidemiologic differences between United States- and foreign-born tuberculosis patients in Houston, Texas. J Infect Dis 2001; 183: Diagnostic standards and classification of tuberculosis in adults and children: This official statement of the American Thoracic Society and the Centers for Disease Control and Prevention was adopted by the ATS Board of Directors, July 1999; this statement was endorsed by the Council of the Infectious Disease Society of America, September Am J Respir Crit Care Med 2000; 161: Menzies D. Interpretation of repeated tuberculin tests: boosting, conversion, and reversion. Am J Respir Crit Care Med 1999; 159: Kent PT, Kubica GP. Public health mycobacteriology: a guide for the level III laboratory. Atlanta, GA: US Department of Health and Human Services, Center for Disease Control, Gonzalez R, Hanna BA. Evaluation of Gen-Probe DNA hybridization systems for the identification of Mycobacterium tuberculosis and Mycobacterium avium-intracellulare. Diagn Microbiol Infect Dis 1987; 8: Strumpf IJ, Tsang AY, Sayre JW. Re-evaluation of sputum staining for the diagnosis of pulmonary tuberculosis. Am Rev Respir Dis 1979; 119: Self-study modules on tuberculosis: contact investigations for tuberculosis. Atlanta, GA: Centers for Disease Control and Prevention, 1999, Hosmer DW Jr, Lemeshow S. Applied logistic regression. 2nd ed. Hoboken, NJ: John Wiley & Sons, Braden CR, Templeton GL, Stead WW, et al. Retrospective detection of laboratory cross-contamination of Mycobacterium tuberculosis cultures with use of DNA fingerprint analysis. Clin Infect Dis 1997; 24: Van Deun A, Salim AH, Cooreman E, et al. Optimal tuberculosis case detection by direct sputum smear microscopy: how much better is more? Int J Tuberc Lung Dis 2002; 6: Wu ZL, Wang AQ. Diagnostic yield of repeated smear microscopy examinations among patients suspected of pulmonary TB in Shandong province of China. Int J Tuberc Lung Dis 2000; 4: Nelson SM, Deike MA, Cartwright CP. Value of examining multiple sputum specimens in the diagnosis of pulmonary tuberculosis. J Clin Microbiol 1998; 36: Cascina A, Fietta A, Casali L. Is a large number of sputum specimens necessary for the bacteriological diagnosis of tuberculosis? J Clin Microbiol 2000; 38: Finch D, Beaty CD. The utility of a single sputum specimen in the diagnosis of tuberculosis. comparison between HIVinfected and non-hiv-infected patients. Chest 1997; 111: So SY, Lam WK, Yu DY. Rapid diagnosis of suspected 114 Clinical Investigations

8 pulmonary tuberculosis by fiberoptic bronchoscopy. Tubercle 1982; 63: Chawla R, Pant K, Jaggi OP, et al. Fibreoptic bronchoscopy in smear-negative pulmonary tuberculosis. Eur Respir J 1988; 1: Abeles H. Early hospital discharge of tuberculous patients with sputum containing acid-fast bacilli on microscopic examination. Am Rev Respir Dis 1973; 108: Brooks SM, Lassiter NL, Young EC. A pilot study concerning the infection risk of sputum positive tuberculosis patients on chemotherapy. Am Rev Respir Dis 1973; 108: Riley RL, Moodie AS. Infectivity of patients with pulmonary tuberculosis in inner city homes. Am Rev Respir Dis 1974; 110: Sepkowitz KA. How contagious is tuberculosis? Clin Infect Dis 1996; 23: Menzies D. Effect of treatment on contagiousness of patients with active pulmonary tuberculosis. Infect Control Hosp Epidemiol 1997; 18: Marks SM, Taylor Z, Qualls NL, et al. Outcomes of contact investigations of infectious tuberculosis patients. Am J Respir Crit Care Med 2000; 162: Carvalho AC, DeRiemer K, Nunes ZB, et al. Transmission of Mycobacterium tuberculosis to contacts of HIV-infected tuberculosis patients. Am J Respir Crit Care Med 2001; 164: Espinal MA, Perez EN, Baez J, et al. Infectiousness of Mycobacterium tuberculosis in HIV-1-infected patients with tuberculosis: a prospective study. Lancet 2000; 355: Cauthen GM, Dooley SW, Onorato IM, et al. Transmission of Mycobacterium tuberculosis from tuberculosis patients with HIV infection or AIDS. Am J Epidemiol 1996; 144: Cruciani M, Malena M, Bosco O, et al. The impact of human immunodeficiency virus type 1 on infectiousness of tuberculosis: a meta-analysis. Clin Infect Dis 2001; 33: Elliott AM, Hayes RJ, Halwiindi B, et al. The impact of HIV on infectiousness of pulmonary tuberculosis: a community study in Zambia. Aids 1993; 7: Reichler MR, Reves R, Bur S, et al. Evaluation of investigations conducted to detect and prevent transmission of tuberculosis. JAMA 2002; 287: Behr MA, Hopewell PC, Paz EA, et al. Predictive value of contact investigation for identifying recent transmission of Mycobacterium tuberculosis. Am J Respir Crit Care Med 1998; 158: Ferebee SH, Mount FW, Murray FJ, et al. A controlled trial of isoniazid prophylaxis in mental institutions. Am Rev Respir Dis 1963; 88: Mitchison DA. Infectivity of patients with pulmonary tuberculosis during chemotherapy. Eur Respir J 1990; 3: Teixeira L, Perkins MD, Johnson JL, et al. Infection and disease among household contacts of patients with multidrugresistant tuberculosis. Int J Tuberc Lung Dis 2001; 5: Nitta AT, Knowles LS, Kim J, et al. Limited transmission of multidrug-resistant tuberculosis despite a high proportion of infectious cases in Los Angeles County, California. Am J Respir Crit Care Med 2002; 165: Stelfox HT, Bates DW, Redelmeier DA. Safety of patients isolated for infection control. JAMA 2003; 290: CHEST / 128 / 1/ JULY,