Published ahead of print on May 8, 2008, doi:10.1164/rccm.200803-381OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200803-381OC
Feasibility of Shortening Respiratory Isolation with a Single Sputum Nucleic Acid Amplification Test1 Division of Pulmonary and Critical Care Medicine, Department of Medicine Miller School of Medicine, University of Miami, Miami, Florida; and 2 A.G. Holley State Hospital, Lantana, Florida Correspondence and requests for reprints should be addressed to Michael A. Campos, M.D., Assistant Professor of Medicine, Division of Pulmonary and Critical Care Medicine, Miller School of Medicine, University of Miami, P.O. Box 016960 (R-47), Miami, FL 33101. E-mail: mcampos1{at}med.miami.edu
Rationale: Serial smear analysis to guide respiratory isolation (RI) of patients with suspected tuberculosis (TB), the majority of whom will be found not to have TB, leads to expensive and unnecessary isolation, and may potentially result in decreased vigilance of subjects with respiratory compromise. Objectives: To compare the performance of a single first-sputum, Mycobacterium tuberculosis–specific nucleic acid amplification (NAA) test with three sputum smears for assessing the need for RI. Methods: Prospective evaluation of 493 patients with suspected TB (74% HIV positive) admitted to RI in a major county hospital in the United States, who had at least three sputum smears and material available from the first sample for additional NAA testing. Measurements and Main Results: Accuracy of the first sputum NAA result and serial smears for identifying patients with potentially infectious TB who truly require RI was determined. Forty-six patients (9.3%) had TB confirmed by culture. First-sputum NAA test detected all patients with TB who had a positive smear (n = 35), even when the first of the three specimens was smear negative. In addition, when compared with serial smears, the first-sputum NAA had a higher sensitivity (0.87; 95% confidence interval [CI], 0.74–0.95) and specificity (1.0) in the detection of subjects with positive M. tuberculosis cultures (smear sensitivity, 0.76; 95% CI, 0.61–0.87; and specificity, 0.96; 95% CI, 0.94–0.98). Conclusions: A single first-sputum NAA testing can rapidly and accurately identify the subset of patients with suspected TB who require RI according to serial sputum smears. Its potential use to shorten RI time does not preclude the need to obtain subsequent specimens for culture.
Key Words: tuberculosis • nucleic acid amplification • diagnostic test • sensitivity • specificity
Tuberculosis (TB) remains among the major communicable causes of morbidity and mortality worldwide. In the hospital setting, the principal measure taken to prevent nosocomial spread is placement of patients with suspected TB in respiratory isolation (RI) until infectious risk is determined to be sufficiently low. Unfortunately, no real-time method exists for determining whether a patient is able to pass the disease to other individuals. Mycobacterium tuberculosis cultures, considered the most sensitive marker for infectiousness, are impractical for supporting decisions regarding RI because of their long turnaround time. As an alternative, microscopic detection of acid-fast bacilli on concentrated smears of expectorated sputum is used as a surrogate test for infectiousness. According to current guidelines, patients with suspected TB can be removed from RI if three consecutive smears obtained over at least a 2-day period are negative (1), in addition to a clinical assessment by the clinician that the risk of transmission is low. Unfortunately, patients with positive sputum cultures can be undetected by serial smear examinations due to low sputum organism loads (2–4), and may also represent a risk for transmission (5–7). Although an isolation strategy based on smear results is therefore imperfect, epidemiologic evidence suggests it is nonetheless highly effective because, in recent years, outbreaks in the health care setting, as well as episodes of transmission to health care workers and other patients, have declined markedly (1). The use of RI is not without cost. First, RI requires single-patient rooms with negative pressure and filtration, sterilization, or dispersion of waste air. Second, due to the nonspecific signs and symptoms of TB, only 1 of every 10–25 isolated subjects actually has the disease (8–10). Patients ultimately determined to be noninfectious typically remain in RI for 4 to 7 days while awaiting the collection and analysis of three sputum smears (11). Beyond monetary cost, RI leads to potential decreased patient contact with providers, which may negatively impact quality of care during the often-crucial first few days of hospitalization. Nucleic acid amplification (NAA) techniques are rapid, sensitive, and specific for the detection of M. tuberculosis complex DNA or RNA in respiratory specimens (12–18). NAA tests have higher sensitivity than acid-fast or fluorescent smears in detecting culture-positive samples because their detection threshold is as low as 1 organism per 100 ml of sample (19, 20). Given that patients with positive sputum smears have a bacillary load that likely exceeds this threshold, even if the initial sample is smear negative, we prospectively tested the hypothesis of whether first single-sputum NAA testing could efficiently detect patients identified as potentially infectious using the standard test of three sputum smears. Some of the results of these studies have been previously reported in the form of an abstract (21).
Protocol The study was conducted at Jackson Memorial Medical Center, a University of Miami–affiliated tertiary facility in Miami, Florida. This is a prospective study of all patients admitted to RI for possible TB between January 1, 2003, and December 31, 2003, except for those admitted directly to an intensive care unit and those with known active TB. Decisions regarding need for, length of stay in, and sample collection during RI were made on clinical grounds by patients' treating physicians. No consultation or guidance was provided by the study protocol or members of the study team, who were not involved in the care of these patients. When patients were admitted to RI more than once during the study period, only the first admission was counted. A waiver of need for informed consent was approved by the University of Miami Institutional Review Board. Demographic, clinical, radiologic, and microbiologic data were collected, and length of stay in RI was documented. Investigators visited all patients once during their stay in isolation to evaluate respiratory and mental status, oxygen saturation, and compliance with prescribed oxygen therapy.
Specimen Processing As part of the hospital's standard protocol, all smear-positive samples underwent in-house NAA testing using the fluorescence quencher polymerase chain reaction (FQ-PCR) (22). If sufficient sputum remained after standard hospital processing, aliquots from all first-sputum samples, regardless of smear results, were frozen at –20°C and sent in batches to the Florida State Laboratory (Jacksonville, FL) for additional NAA testing using Enhanced Amplified Mycobacterium tuberculosis Direct (EMTD) test kits provided by Gen-Probe. The EMTD tests were processed in batches at a later date so as not to influence patient care and RI usage.
Definitions
Statistical Analysis
Patient Demographics During the study period, there were 679 unique admissions to RI. Three or more sputum samples were collected from 550 patients during their initial admissions, and of these, 493 patients had sufficient material available from the first sputum for NAA testing and were included in the study (Figure 1). The demographic and clinical features of these subjects are summarized in Table 1. The majority (74%) had HIV infection. Ethnically, the isolated cohort was 64% African American, 19% Hispanic, and 17% white or other. There were 314 (46%) foreign-born subjects, predominantly from Haiti (29%) and Central America (34%).
Although concern about possible TB disease prompted physicians to admit patients to RI, clinicians caring for these patients considered it the most probable diagnosis in only 61 cases (12%), and started TB empiric therapy for only 13 cases. Among patients admitted to RI, those later confirmed to have TB were less likely to have presented with dyspnea upon admission than those without TB (31 vs. 54%, P < 0.001). Compared with the total isolated population, subjects with TB were more likely to be foreign born and less likely to have known HIV infection. Of the 364 isolated patients with known HIV infection, only 20 (5.4%) were subsequently found to have TB.
Microbiology Overall, TB was confirmed bacteriologically by sputum culture in 46 patients (9.3%). Only 14 of these patients had clinical suspicion of TB at the time of diagnosis (30.4%). Most (41 patients) had M. tuberculosis growth on the first specimen obtained, with only five growing M. tuberculosis on the second or third specimen only. Of the 35 patients with smear-positive TB, 3 had negative smears in their first sputum (Table 2). One other patient had five negative smears and a negative EMTD of the first sputum, but had M. tuberculosis grown in one of five sputum cultures. This prompted an investigation by the Florida Department of Health, which determined that this was a false-positive culture using clinical correlation and spoligotyping. Seventy patients (14%) had cultures that grew NTM.
Performance of NAA Testing for Detection of TB Infectiousness Among the first-sputum specimens, 27 were tested using the FQ-PCR assay only (under the hospital protocol for positive smears), 442 were tested using the EMTD only, and 24 specimens were tested using both methods. First-sputum NAA tests were positive for all the 35 patients with smear-positive TB, including the three who had negative smears of their first sputa. Therefore, first-sputum NAA testing had a sensitivity of 1.0, as well as a negative predictive value of 1.0, for the detection of smear-positive subjects (Table 3). The performance of NAA testing was similar in HIV-seropositive and HIV-seronegative patients. Among the 24 samples tested using both FQ-PCR and EMTD, there was a single discordant result, with a negative FQ-PCR and a positive EMTD in a patient confirmed to be positive by sputum culture.
In addition, the first-sputum NAA test proved significantly more sensitive than serial smears for identification of all cases with positive M. tuberculosis cultures (Table 3; P < 0.05 for difference between NAA and smears). The specificity and positive predictive value of the first sputum NAA were both 1.0, again markedly superior to smears (0.96 and 0.73, respectively). No positive NAA results were observed in smear-positive patients with cultures that grew only NTM. In contrast to NAA, serial smears had a lower positive predictive value in HIV-seropositive patients than those who were HIV seronegative (Table 3).
Characteristics of the Stay in RI
This investigation has shown that for patients with suspected TB, NAA testing of the first sputum sample outperforms the current standard of serial smears as a screening test for the need to continue RI. NAA testing of the first sample detected all patients considered as high risk of TB transmission by using serial smears, and proved more sensitive and specific than smears for identifying patients who were ultimately proven to have TB by culture. These results offer the possibility of improving current "rule out TB" protocols, both by enhancing detection rates before releasing patients from isolation and by liberating noninfectious patients from isolation earlier. Our study also confirms that the practice of obtaining serial smears takes much longer than initially envisioned (5 instead of 3 d in our study). RI length may currently be shorter following the publication of the 2005 Centers for Disease Control and Prevention (CDC) guidelines, which now recommend the collection of three respiratory samples within 2 days (1). Replacement of serial smears with a single NAA test could permit the vast majority of patients to be discharged from RI within a single day, reducing the demand for RI facilities by perhaps 75%. Even if three smears could be obtained and interpreted within a 48-hour period, using an NAA test as a screening tool would still reduce the RI time by half. NAA tests thus offer the possibility of significant economic savings. In addition, shortening RI time allows more contact between patients and health care providers, which potentially offers the opportunity to improve the quality of patient care. By design, the study did not interfere with the regular RI process in our institution, nor were the techniques for sputum collection enhanced in any way for study purposes. We would therefore expect our results to be applicable to any setting that collects sputum samples in accord with CDC-recommended techniques. Whether our findings can be replicated in other settings (i.e., in outpatients, in emergency room settings, or in facilities where routine sputum induction is not routinely performed) should be a topic of future trials. Our study shows that routine NAA testing can be used as a rapid tool to assess the need of RI, as one single test can detect all subjects with positive smears. It is important to emphasize that the NAA test does not replace the need of obtaining multiple specimens to enhance the yield of M. tuberculosis cultures, required to diagnose smear-negative, NAA-negative cases (13% in our study) and to obtain drug sensitivity. For NAA test–negative subjects, these additional sputum samples can be obtained out of RI, and in many cases as outpatients. In addition, infection control efforts will potentially be improved by the detection and continued isolation of about half of patients with TB with negative smears, who present a lower but nonnegligible risk of TB transmission (5–7). Collection of serial sputum samples from NAA-positive, smear-positive subjects should be done to determine discontinuation of RI. Some limitations of our study deserve comment. Our data represent the experience of a single institution with a moderate annual prevalence of TB cases. Previous cost-effectiveness analyses have shown that routine use of NAA tests is more cost-effective in settings with high prevalence of TB and smear-positive cases (23, 24), but performance in lower prevalence areas should still benefit from reducing hospitalization costs. The use of two different NAA tests in our study can be considered as another limiting factor of our study. So as to not interfere with the participant's clinical care, respiratory samples had to be appropriately processed using the hospital's protocol, including FQ-PCR testing of positive smear specimens. This technique, as well as other PCR protocols, has been shown to perform as well as the two currently approved NAA tests in the United States (22, 25, 26), including the EMTD. We anticipate our experience can be replicated using this and other NAA tests in other institutions with different TB prevalence. We chose to exclude from analysis the 129 isolated patients from whom fewer than three sputum specimens were collected. Generally, this occurred when treating physicians became highly confident that patients had alternative diagnoses rather than TB, and is reflective of actual clinical practice, which, rightly or wrongly, is not always congruent with published guidelines. Among these excluded patients, five had positive smears due to NTM (all with negative in-house PCR and EMTD) and two had smear-positive TB (both with positive first-sputum in-house PCR and EMTD). These results are consistent with the findings from our principal analysis. HIV-positive patients with suspected TB may be particularly well served by substituting an NAA test for serial smears because the performance of serial smears is inferior in this population (3). This is because TB often presents with negative smears in patients with HIV infection, and false-positive smears due to NTM infection are relatively common. In our study, we did not observe that performance of NAA tests was affected by HIV status. Despite the efficacy of NAA tests, the CDC has not yet formally recommended the use of these tests as replacements for serial smears, or as alterations of existing isolation policies (1). We believe that, with the growing body of evidence supporting NAA tests, such changes should be considered. Substituting the late 20th century technology of NAA testing for the 19th century technique of smears offers the opportunities to improve infection control by earlier recognition of TB, and to liberate patients without the disease from unnecessary isolation.
The authors thank the following individuals for their contribution in this project: Gladys Roudel (Jackson Memorial Hospital, Microbiology Laboratory) for sample processing, Susan Dean and Candice Stevens (Florida State Laboratory) for EMTD analysis. None of these individuals received compensation for their contribution.
Originally Published in Press as DOI: 10.1164/rccm.200803-381OC on May 8, 2008 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form March 6, 2008; accepted in final form May 7, 2008
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
2 test for categorical variables, and the t test or Wilcoxon test for continuous variables, depending on the data distribution. Statistical analysis was performed using NCSS 2004 (Kaysville, UT). 
