Published ahead of print on December 4, 2003, doi:10.1164/rccm.200308-1101OC
© 2004 American Thoracic Society Cough-generated Aerosols of Mycobacterium tuberculosisA New Method to Study InfectiousnessDepartment of Medicine, Center for the Study of Emerging and Re-Emerging Pathogens, New Jersey Medical School, University of Medicine and Dentistry of New Jersey, Newark, New Jersey; Department of Medicine, National Jewish Medical and Research Center, Denver; Department of Microbiology, Colorado State University, Fort Collins, Colorado; and Department of Microbiology, University of Arkansas for Medical Sciences, Little Rock, Arkansas Correspondence and requests for reprints should be addressed to Kevin P. Fennelly, M.D., M.P.H., Center for the Study of Emerging and Re-Emerging Pathogens, New Jersey Medical School, University of Medicine and Dentistry of New Jersey, 185 South Orange Avenue, MSB A-901, Newark, NJ 07103. E-mail: fennelkp{at}umdnj.edu
The concentration and size distribution of infectious aerosols produced by patients with pulmonary tuberculosis (TB) has never been directly measured. We aimed to assess the feasibility of a method that we developed to collect and quantify culturable cough-generated aerosols of Mycobacterium tuberculosis. Subjects were recruited from a referral hospital and most had multidrug-resistant TB. They coughed into a chamber containing microbial air samplers while cough frequency was measured during two 5-minute sessions. Cough-generated aerosol cultures were positive in 4 of 16 subjects (25%) with smear-positive pulmonary TB. There was a rapid decrease in the cough-generated aerosol cultures within the first 3 weeks of effective treatment. Culture-positive cough aerosols were associated with lack of treatment during the previous week (p = 0.007), and there was a trend in the association with cough frequency (p = 0.08). The size distributions of these aerosols were variable, but most particle sizes were in the respirable range. Quantification of viable cough-generated aerosols is feasible and offers a new approach to study infectiousness and transmission of M. tuberculosis and other airborne pathogens.
Key Words: disease transmission • communicable disease control • air microbiology • infectivity • contact tracing Pulmonary tuberculosis (TB) continues to be a major source of morbidity and mortality worldwide (1), especially among patients infected with the human immunodeficiency virus. In response to nosocomial outbreaks of TB in the United States, the CDC has recommended the use of personal respiratory protection and engineering controls in healthcare facilities (2). To study the efficacies of these measures in decreasing exposure to infectious aerosols (3), we aimed to obtain data on the concentration and size distribution of the airborne particles containing Mycobacterium tuberculosis. Although Wells had inferred the size of infectious droplet nuclei on the basis of theory and experimental tuberculous aerosols generated in the laboratory (4–7), the size range of tuberculous aerosols produced by humans has never been directly measured. Riley and his colleagues estimated the concentration of infectious droplet nuclei of airborne M. tuberculosis by counting the number of pulmonary granulomatous lesions in infected guinea pigs exposed to the air from an experimental TB ward (8, 9). This approach provided valuable data and allowed sampling of air for longer periods of time than is possible by mechanical methods due to fungal overgrowth, desiccation of media and bacilli, and other problems. However, the use of animal models to study infectiousness has several limitations, including expense, variable susceptibility among species, and complex logistics, that restrict studies to specially constructed facilities. After several pilot studies of sampling the air from the isolation rooms of patients with TB failed to yield M. tuberculosis, we reasoned that collection of potentially infectious aerosols could be improved by having patients cough into a closed chamber to prevent dispersion. We hypothesized that viable aerosols of M. tuberculosis could be isolated and quantified from patients with active pulmonary TB using this method. Here, we report the feasibility of this approach and our initial data. Some of the results of these studies have been previously reported in the form of abstracts (10, 11).
Subjects We recruited consecutive adult subjects with TB admitted to the inpatient service at National Jewish Medical and Research Center in Denver, Colorado. The Institutional Review Board approved the protocol, and informed consent was obtained from each subject. The major inclusion criterion was a sputum specimen positive for acid-fast bacilli (AFB) on microscopy before referral or on admission. Subjects were excluded if there were medical conditions proscribing against coughing, e.g., recent chest surgery. If subjects had positive AFB smears of the sputum collected during the first study, we repeated the studies once or twice weekly for 3 weeks. We obtained additional studies of two subjects after observing positive results on the initial tests.
Materials
After observing marked qualitative differences in cough strength and frequency between the first two subjects, we added an in-line pneumotach (MedGraphics, Inc., St. Paul, MN) connected to a pressure transducer (MP45–14–871; Validyne Engineering Corp., Northridge, CA). The signal was amplified and printed on a chart recorder. A stainless cylindrical chamber of the same volume was constructed to simplify disinfection and was used for the studies of the last two subjects.
Protocol We first sampled the ambient air in the isolation room for 5 minutes using both impactors simultaneously. The agar plates were removed and labeled, and fresh plates were replaced in the samplers at the bedside. Sputum induction was begun at this point if indicated. The subject was instructed to cough into the tubing for 5 minutes or for as long as was comfortable while we sampled the air from the chamber with both impactors and recorded the cough frequency. While the subject rested after the first session of coughing, we removed and labeled the plates and reloaded with fresh plates. We then repeated the sampling during 5 minutes of coughing if tolerated by the subject. Sputa were collected in sterile 50-ml polypropylene conical tubes and processed in the clinical mycobacteriology laboratory for smear and semiquantitative cultures according to standard methods (15). The outer surfaces of the plates from the aerosol collections and two negative control plates were wiped down with a 10% solution of household bleach (6% sodium hypochlorite), allowed to dry, and sealed in polyethylene bags and incubated at 37°C in an atmosphere of 5 to 10% carbon dioxide. The CASS components were disinfected in 2.6% activated glutaraldehyde (Metricide, Metrex Research Corp, MI). The agar plates from the aerosol collections were read weekly for 6 to 10 weeks, and the number of cfu consistent with M. tuberculosis on each plate were recorded. All colony types were noted, and convenience samples of bacterial contaminants were identified with the Gram stain or genus level in the clinical microbiology laboratory. Aerosol cfu confirmed by nucleic acid probes (Gen-Probe, San Diego, CA) as M. tuberculosis were subcultured and sent for restriction fragment length polymorphism analyses together with subcultures from the sputum to confirm the identity of the isolates. As a positive control for our air-sampling method, we placed two Andersen cascade impactors inside a Middlebrook inhalation exposure system (GlasCol, Terre Haute, IN). We nebulized Mycobacterium bovis bacillus Calmette–Guérin in water according to a standard protocol used for aerogenic infections of mice (16). We sampled the aerosol in the chamber at the midpoint of the nebulization cycle using two Andersen six-stage impactors for 5 minutes. For each subject, we collected the semiquantitative clinical mycobacteriology laboratory data on the sputum smears and cultures at the time of each study. Sputa smear (17) and culture data were converted to a grade scale ranging from 0 to 4. The culture grades are based on the reporting system of the National Jewish Center laboratory and so differ slightly from other scales (17, 18).
Of the 16 subjects invited to participate, all were recruited. There were eight men and eight women. Six subjects (38%) were non-Hispanic white, five (31%) were Hispanic, four (25%) were Asian, and one (6%) was non-Hispanic black. None were infected with human immunodeficiency virus. Nine subjects (56%) had multidrug-resistant TB, two (13%) had drug-resistant TB, and five (31%) had isolates susceptible to all drugs. At the time of the initial cough aerosol study, 12 (75%) of the subjects had sputa with AFB detected on smears and 12 (75%) of the subjects had M. tuberculosis isolated from sputa cultivated on Middlebrook 7H11 agar. Fourteen subjects (88%) had cavitary lung disease. We conducted 45 cough aerosol studies. Nine subjects (56%) had three or more studies, and seven subjects (44%) were studied only once. We isolated culturable M. tuberculosis from the aerosols generated by coughing in four (25%) of the subjects. All these subjects had multidrug-resistant TB, and their sputa were positive for AFB on smears and for M. tuberculosis on culture (Table 1) . These four subjects represented 33% of subjects who were either smear positive (n = 12) or culture positive (n = 12) on the sputa collected at the time of the initial cough aerosol study. They comprised 57% of the seven subjects with greater than 100 cfu from their sputum. All cough-generated aerosols decreased after initiation or modification of antituberculous therapy guided by in vitro susceptibility data (Figure 2) . Production of culturable aerosol was associated with lack of treatment in the preceding week(s) (Fisher's exact test; p = 0.007). There were no significant associations between culturable cough-generated aerosols and sputum smear or culture grades, age, sex, or presence of cavities. No cough-generated aerosols were cultured from subjects who had less than 100 cfu from the sputa or who were AFB smear negative, but all of those subjects had been on treatment. No subject with an initial negative study produced culture-positive aerosols on a subsequent study. Five subjects had three or more studies that were all negative.
We isolated 633 cfu from the initial aerosol generated by Subject 2 over 10 minutes during a sputum induction. Three cfu were isolated from the aerosol after 1 week of therapy, and thereafter, no cfu were cultivated from her cough aerosols despite persistently positive sputum smears and cultures (Figure 3 ; Table 2) . Laryngeal TB was suggested by hoarseness and multiple nodular lesions in the laryngeal mucosa observed with fiberoptic laryngoscopy. Directly observed therapy had been interrupted for less than 1 week for travel, and drug levels were undetectable on admission. Because of the persistence of positive sputa smears and the positive results on her initial aerosol cultures, we repeated a cough-generated aerosol study at 15 weeks, and there was no growth. In the interim she had received continuous inpatient therapy and a surgical resection of the right upper lobe to remove a large cavity.
We isolated 3 and 4 cfu from the initial aerosols generated by Subjects 12 and 13, respectively, during sputa inductions over 10 minutes. After 1 week of therapy no viable aerosols were isolated from either patient. The antituberculous therapy of Subject 12 had been discontinued because of adverse drug effects 3 weeks before the initial study. His sputum culture became negative at 3 weeks. Treatment of Subject 13 had been discontinued 5 months before the initial study due to the discovery of in vitro drug resistance to 11 antituberculous drugs. His sputum culture became negative at 7.5 weeks. All other patients were receiving antituberculous therapy at the time of their initial studies. Subject 14 had a cough productive of purulent sputa associated with extensive chronic cavitary and bronchiectatic disease, so sputum induction was not necessary. She generated 84 cfu on her first study, and after modification of her treatment regimen the aerosol cfu decreased more slowly than in Subject 2 (Figure 2). Her sputa became culture negative after 3 months of therapy and smear negative after 4 months. At this point, a cough aerosol study was repeated and yielded no growth. The restriction fragment length polymorphism (RFLP) analyses of colonies from the aerosol cultures and from sputa specimens from Subjects 2, 13, and 14 confirmed that the paired isolates were identical. Restriction fragment length polymorphism (RFLP) analyses could not be done on Subject 12 due to loss of viability of the subculture during transport. The particle size distributions of the viable aerosols are shown in Figure 4 . During the sputum-induction procedure on Subject 2, the mode was in Stage 5 (309 cfu; 49%) that collects particles of 1.1 to 2.2 µm s in aerodynamic diameter. A total of 572 cfu (90%) were isolated in Stages 4 to 6 that collect particles from 0.65 to 3.3 µm in aerodynamic diameter. We were unable to delimit the lower size range of the viable particles because there were cfu in the last stage. The 3 cfu from the aerosol of Subject 12 were all in Stage 6, and the 4 cfu from Subject 13 were in Stages 3, 4, and 5. These aerosols were also collected during sputa inductions. Subject 14 was the only subject able to produce sputa easily with voluntary coughing and without induction. The particle size distribution of her cough-generated aerosol was slightly larger, with a mode in stage 4 (2.1–3.3 µm in aerodynamic diameter). The lack of cfu in the last stage (6) of any of the 14 Andersen cascade impactors used in her five positive studies suggests that all the culturable aerosols produced were larger than 1.1 µm in aerodynamic diameter.
The particle size distribution of the experimentally nebulized suspensions of M. bovis bacillus Calmette–Guérin was shifted to the smaller sizes, with most particles collected in the last two stages (0.65–2.1 µm in aerodynamic diameter). Again, the lower limit of the viable particles could not be defined.
No culturable aerosols of M. tuberculosis were isolated from the room air. Molds were isolated from the ambient air in 5 of 540 (0.9%) agar plates and from cough-generated aerosols in 6 of 936 (0.6%) agar plates. As these plates were removed from the incubator to avoid contaminating other plates, they were no longer available for cultivation and evaluation for M. tuberculosis. The cough aerosol of Subject 2 yielded a few cfu of Mycobacterium fortuitum and Mycobacterium flavescens in addition to M. tuberculosis on one occasion, although these were not isolated from her sputa, from other patients, or from our control plates. Colonies in nine ambient air plates were gram-positive rods suggestive of diptheroids and gram-positive cocci that were catalase positive suggestive of staphylococci and Cladosporium. Thirteen cough-generated aerosol plates grew a few colonies of diptheroids, staphylococci (not Staphylococcus aureus), Two 5-minute sessions of coughing were completed in 33 of the 45 studies (73%). Although there were more cfu isolated during the second cough session compared with the first in aggregated data, the difference was not statistically different (415 vs. 355; paired t test, p = 0.64). Three subjects (19%) completed only one coughing session, and four subjects (25%) were initially able to do two 5-minute sessions of coughing but then could only complete one session thereafter. All subjects were able to complete at least one 5-minute session of coughing, although Subject 7 had only one cough. The distribution of cough frequencies during the studies was skewed, with a median of 54 and a mean of 76.5 coughs (interquartile range, 32–113). There was a trend toward greater cough frequency in those who produced culturable aerosols compared with those did not (Mann–Whitney test; p = 0.08). There were no significant associations between cough frequency and age or sex.
To our knowledge, this is the first report of the direct isolation, quantification, and particle size determination of culturable airborne M. tuberculosis generated by patients with pulmonary TB. We detected cough-generated tuberculous aerosols in only 4 of 12 subjects (33%) with concurrent AFB-positive sputa, the current laboratory marker associated with infectiousness. Although these data are preliminary, they suggest that there may be considerable variability in the ability to produce potentially infectious aerosols among patients with TB. This is consistent with data from experimental infections of guinea pigs by exposures to air from a TB ward by Riley and colleagues. In one study, only 3 of 77 (4%) patients with TB produced 35 of 48 (77%) of the infections (19), and in another study only 8 of 61 patients (13%) produced all the infections (9). In Rotterdam from 1967 to 1969, only 28% of all patients who were sputum AFB positive transmitted infections (20). Brooks and colleagues also found variability in the number of latent infections among contacts of 21 patients with TB (21). Eight patients (38%) had no tuberculin skin test–positive contacts, another eight (38%) had only one tuberculin skin test–positive contact, but 11 of 13 contacts of one patient (5%) were tuberculin skin test positive. Although these household contact studies are hampered by the limitations of the tuberculin skin test, more recent molecular epidemiology studies also suggest variability in transmission (22–25). The size distributions suggest that most of the viable particles in these cough-generated aerosols are immediately respirable. These data support observations that M. tuberculosis can be transmitted by brief, close contact to an infectious case (26–30). Although Duguid (31) and Loudon (32) had provided data on the particle sizes generated by coughing, they measured total particle counts and not the particles containing pathogens as we describe here. A time-weighted average might suggest that these subjects could generate from 18 to 3,798 infectious particles per hour and that the data from Subject 2 represent the largest rate of tuberculous aerosol production reported in the literature. However, such an interpretation would probably be erroneous. This new method may represent a worst-case scenario or a potential to generate infectious aerosols. How it relates to the daily spontaneous production of cough-generated aerosols is unknown and will require additional research. The rate of decrease of cough-generated aerosol cultures appears to be considerably more rapid than that of sputa, at least in two of our subjects. The rapid decrease of culturable organisms in the cough-generated aerosols within the first few weeks of antituberculous therapy is consistent with epidemiologic data suggesting that patients become noninfectious within a few weeks of effective treatment (33). Our data suggesting an association between cough frequency and tuberculous aerosol production are consistent with those of Loudon and Spohn, who found an association between overnight cough frequency and increased transmission among household contacts (34). The major advantage to the CASS method is the ability to quantitatively assess the magnitude and size distribution of culturable aerosols generated by individual patients at multiple points in time. The CASS is relatively portable, making it amenable to the study of patients at the bedside. It is easily reproducible and could be used in multiple locations without the need for animal exposures, rendering it applicable to epidemiologic studies of infectiousness and transmission. Although polymerase chain reaction–based air-sampling techniques provide data more rapidly (35–37), they are unable to determine viability and they are not quantitative, in contrast to this method. This approach appears to be safe for both the research subjects and the staff. There were no adverse events associated with the CASS. The use of the chamber under negative pressure to collect the infectious aerosol probably renders these studies safer for personnel than if the patient were simply coughing into the open room. The three investigators (K.P.F., J.W.M., and K.E.F.) directly involved with collecting the cough-generated aerosols have all continued to have negative Mantoux tuberculin skin tests. Our use of selective agar combined with relatively short sampling times probably minimized the common problem of fungal contamination of culture plates. There are several limitations to this study. There is likely referral bias, as most of the cases had multidrug-resistant TB. We were unable to study patients before antituberculous therapy was first initiated, and these patients had been previously treated with various regimens and for various durations. Most subjects were studied during sputum-induction procedures. Therefore, these data may not be representative of most patients with pulmonary TB, especially of those in whom the disease caused by drug-susceptible bacilli has been newly diagnosed. The small number of subjects limited our statistical power to assess sources of variability in aerosol production. The study was not designed to study the relative infectiousness of sputum smear-positive versus smear-negative cases or of other factors such as human immunodeficiency virus infection or lung cavitation. The major disadvantage to this method is the delay in obtaining culture results. The reproducibility, sensitivity, specificity and the predictive value of the method will need to be assessed in future studies. The CASS can be used to study the mechanisms of aerosolization of bacilli from the human respiratory tract, an area of investigation that has received relatively little attention. This approach provides opportunities to simultaneously study contributions of host, pathogen, and environmental factors in transmission not only in TB but also in other airborne infections.
The authors thank the patients and staff of the Clinical Mycobacteriology Service at the National Jewish Medical and Research Center, especially Drs. Michael Iseman, James Cook, and Gwen Huitt. Dr. Fennelly thanks Drs. Jerrold Ellner and Richard Martin for their mentorship. The staff of the Clinical Mycobacteriology Laboratory provided valuable technical training and assistance, especially Ms. Pamela Lindholm-Levy. The authors also thank the staff of the Clinical Microbiology Laboratory for their technical assistance in identifying contaminants. Mr. Alan Roberts provided special technical assistance in the studies of the inhalation exposure system at Colorado State University.
Supported by the National Institute of Health Career Development Award 1K23 AI01676 and by the Cooperative Agreement U50/CCU810073 from the National Institute for Occupational Safety and Health, CDC. Conflict of Interest Statement: K.P.F. has no declared conflict of interest; J.W.M. has no declared conflict of interest; K.E.F. has no declared conflict of interest; I.M.O. has no declared conflict of interest; D.M.C. has no declared conflict of interest; L.B.H. has no declared conflict of interest. Received in original form August 7, 2003; accepted in final form November 25, 2003
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
-hemolytic streptococci, and rare gram-negative rods. No control plates showed growth of contaminants. 
