- Split View
-
Views
-
Cite
Cite
Naile Malakmadze, Idalia M. González, Tanya Oemig, Ijeoma Isiadinso, Desiree Rembert, Mary M. McCauley, Philip Wand, Lois Diem, Lauren Cowan, Gabriel J. Palumbo, Michael Fraser, Kashef Ijaz, Unsuspected Recent Transmission of Tuberculosis among High-Risk Groups: Implications of Universal Tuberculosis Genotyping in Its Detection, Clinical Infectious Diseases, Volume 40, Issue 3, 1 February 2005, Pages 366–373, https://doi.org/10.1086/427112
- Share Icon Share
Abstract
Background. The initiation of universal genotyping revealed 3 clusters of 19 patients with tuberculosis (TB) in Wisconsin, with no apparent epidemiologic links among most of them. An epidemiologic investigation was conducted to determine whether genotype clustering resulted from recent transmission.
Methods. We conducted additional interviews with patients and reviewed medical records. Places frequented by the patients while they were infectious were visited to identify contacts.
Results. Our investigation revealed several previously unrecognized possible sites of TB transmission: a single-room occupancy hotel, 2 homeless shelters, 1 bar, and 2 crack houses. Seven patients with previously diagnosed TB were added to the clusters. Of 26 patients, we identified epidemiologic links for all but 1. Common risk factors among patients included alcohol abuse, crack cocaine use, homelessness, and unemployment. Additionally, 98 contacts missed during routine contact investigation were identified.
Conclusions. Transmission of TB, particularly among high-risk groups, may go undetected for years. Our investigation demonstrated the value of universal genotyping in revealing unsuspected recent TB transmission and previously unrecognized sites of transmission, which can be targeted for specific TB interventions.
Although tuberculosis (TB) rates are decreasing nationwide, pockets of high incidence remain within low-incidence areas [1]. Wisconsin reflects the national trend, experiencing almost a 2-fold decrease in the incidence of TB since 1993, from 2.1 cases to 1.4 cases per 100,000 population in 2002. During the 1990s, Milwaukee, Wisconsin's largest city, consistently reported a TB rate higher than the state's overall rate, with a peak in 1999 and a 25% increase in the number of TB cases in 2002 (35 cases), compared with 2001 (28 cases) [2]. However, an increase in the incidence of TB does not necessarily signify ongoing transmission; it might indicate an increase in the number of cases of reactivated disease.
Determining whether an increase reflects ongoing transmission is essential in developing appropriate TB-control interventions. Recent TB transmission contributes to a greater proportion of disease cases than was previously believed [3–5]. Furthermore, recent transmission appears to be more common among certain risk groups (alcohol and drug users, homeless persons, and HIV-infected persons) than among the general public [4, 6, 7].
Identification of high-risk groups and the early detection of TB transmission allow timely implementation of focused control measures [8, 9]. TB genotyping has been an effective tool that allows detection of unsuspected ongoing transmission and helps distinguish recent transmission from reactivated disease [8, 10]. Since 2000, the Wisconsin TB-control program has submitted isolates (including selected isolates from before 2000) that match by spoligotyping for further genotyping to the Centers for Disease Control and Prevention (CDC; Atlanta, GA). In 2002, three separate genotype clusters, which involved 19 patients, were identified in Milwaukee among patients with TB on the basis of spoligotyping, mycobacterial interspersed repetitive units (MIRU), and IS6110 restriction fragment–length polymorphism (RFLP) methods. Clustering of TB isolates often represents recent transmission [3, 4], but routine contact investigations of these patients revealed few epidemiologic links among them. Many patients shared certain risk factors, such as illicit drug use and alcohol abuse, thus increasing the likelihood that undiscovered epidemiologic links existed. In March 2003, an epidemiologic investigation was conducted to determine whether the clusters resulted from recent transmission and to identify possible factors facilitating the spread of TB.
Methods
Patient interviews and medical record reviews. A standardized data abstraction form and interview questionnaire included patients' demographic characteristics, history of current and previous residence and employment, history of previous TB and tuberculin skin test (TST) results, dates of symptom onset and diagnoses, hospitalizations, completion of treatment, laboratory results, chest radiograph findings, risk factors (e.g., substance and alcohol abuse and stays in long-term facilities, such as homeless shelters, detoxification centers, or correctional facilities), travel history, types and sites of social and leisure activities, underlying illness, and concurrent immunosuppressive conditions or treatment. Places frequented by the patients while they were infectious were visited to identify contacts. The list of patients with TB was cross-matched with the correctional facility and sexually transmitted diseases program databases to obtain TST screening results and to identify exposed contacts.
Definitions. A genotyping cluster was defined as ⩾2 isolates with identical genotypes by spoligotyping, MIRU, and IS6110 RFLP [11]. RFLP patterns that differed by only 1 band were considered to be matches if there were identical matches by the other 2 methods.
An epidemiologic link was defined as contact among ⩾2 patients who shared space while one of them was infectious, even if they did not name each other. Recovered isolates of additional cluster-related patients were genotyped.
A cluster-related patient was defined as (1) a patient with culture-confirmed TB whose isolate matched the cluster's genotyping pattern, (2) a patient with culture-confirmed TB with no isolates available for genotyping but with an epidemiologic link to a clustered patient, or (3) a patient with TB confirmed by a pathologic tissue specimen that tested smear-positive for acid-fast bacilli (AFB) who had an epidemiologic link to a clustered patient. Additional patients, who had previously received diagnoses and treatment from the health department, were added to a cluster on the basis of confirmed epidemiologic links or both epidemiologic links and genotype match to a cluster.
Determination of infectious period. Patients with culture-confirmed TB who had sputum smear results positive for AFB were considered to be potentially infectious for the 3 months preceding the date of collection of the first positive specimen. However, if the date of symptom onset occurred or the chest radiograph suggestive of TB was performed before this 3-month period, the infectious period was considered to have begun 6 months preceding the date of TB diagnosis. Patients with whose sputum smear results were negative for AFB were considered potentially infectious for the 1-month period preceding the date of TB diagnosis. Data were analyzed using EpiInfo, version 6.04b (CDC) [12].
Results
Cluster Ascertainment
At the beginning of the investigation, there were 6, 7, and 6 patients with TB in clusters A, B, and C, respectively. All patients in each cluster had genotype matches confirmed by 3 methods (spoligotyping, MIRU, and RFLP). The RFLP pattern for clusters A, B, and C consisted of 9, 9, and 15 bands, respectively. Investigation revealed 3 additional patients in cluster A and 4 additional patients in cluster B.
Patient Characteristics
The characteristics of the patients within each cluster are presented in table 1. Of the 26 patients, 23 (89%) completed their TB treatment with directly observed therapy (DOT). Two patients died before treatment completion, and 1 died before treatment initiation. Of the 3 deaths, only 1 was related to TB. Three patients interrupted TB treatment for 2–6 months before completion. Cross-matching with the correctional facilities database revealed that, before TB diagnosis, 2 patients had positive TST results in the city jail.
Identification of Epidemiologic Links
At the beginning of the investigation, epidemiologic links were known for 8 of the 26 patients. Our investigation revealed 29 previously unsuspected epidemiologic links among 17 other patients. Epidemiologic links were discovered for all but 1 patient. In the description of the epidemiologic links, patients are numbered by the date of diagnosis.
Cluster A. Cluster A comprised 9 patients, including the 3 additional patients found by our investigation (figure 1). Epidemiologic links were determined among all patients in the cluster. Patients A1, A3, A4, and A9 stayed in single-room occupancy (SRO) hotel A during 1997–1999. Patient A9 named patient A2 as the source of his TB through close social and work contact. Patients A5 and A7 stayed at shelter A, and patients A5 and A1 stayed at shelter B. Patients A6 and A8 were sex partners. Patient A8 frequented a local bar that all residents of SRO hotel A also frequented.
Cluster B. Cluster B comprised 11 patients, including the 4 additional patients found by our investigation (figure 1). Epidemiologic links were determined among all patients in the cluster. Patient B1 had no isolates available for genotyping but had a family history of TB and also smoked crack cocaine with patients B6 and B7. Patients B2, B3, B4, and B8 lived at SRO hotel B in an adjacent county. Patients B2 and B3 were sex partners, and patient B4 was the boyfriend of patient B2's daughter. Patient B3 worked at a car wash along with patients B1, B6, and B7. Patient B7 named his coworker and friend (patient B1) as the source of TB. Patient B7 was a sex partner of patient B9 but had no isolates available for genotyping. Patient B7 refused TB treatment initially and traveled to another state to visit relatives. His 2-year-old nephew died of TB soon thereafter and had an isolate that matched cluster B. Patients B5 and B6 were relatives and lived in the same household. Patient B6 also worked at a restaurant that patients B5 and B10 frequented. Patients B6 and B10 were friends and spent considerable time at a liquor store where patient B11 was a frequent visitor.
Cluster C. Epidemiologic links were known initially for 4 patients (patients C1, C2, C3, and C4) who shared the same household or were close friends (figure 1). No additional cluster-related patients were identified, but 1 additional epidemiologic link was established between patients C2 and C5, who were friends. These 5 patients retained close contacts for several years. The only patient with a matching genotype but without an identified epidemiologic link was patient C6, who worked as a nurse's aid in a hospital. She reported a history of a positive TST result, but latent TB infection (LTBI) treatment could not be completed because of adverse effects. Our investigation was unable to determine whether her case of TB represented recent transmission or the reactivation of a remote infection.
Common risk factors among patients in the 3 clusters included alcohol abuse, crack-cocaine use, homelessness, and unemployment (table 1).
Transmission at homeless shelters and SRO hotels. While infectious, patient A5 from cluster A intermittently stayed in shelter A, where patient A7 stayed for several years (figure 2). Although patient A5 did not name patient A7, epidemiologic links between these 2 patients were established by overlapping days of stay at shelter A. In 1997, before the patient stayed at shelter A, patient A5 (from cluster A) stayed at shelter B for 2 months at the same time as patient A1 (from cluster A). We identified exposed contacts for TB screening at shelter A using a computerized log of residents, but shelter B did not keep records from previous years.
Four patients from cluster A stayed at SRO hotel A. Patient A1, whose genotype pattern matched cluster A, was not named by anybody, but when it was presumed that this person was infectious, the patient stayed at SRO hotel A at the same time as patients A3, A4, and A9. Before staying at this hotel, patient A1 stayed at shelter B, where patient A5 was also staying. Patients A3 and A9 were close friends for several years. In 1998, patient A3 died of cavitary TB within 1 month after treatment initiation. Patient A9 refused to undergo a TST test in 1998 when his coworker, patient A2, had pulmonary TB diagnosed. Four years later, TB was diagnosed in patient A9, and his genotype pattern was determined to have matched that associated with cluster A by the 3 methods, except for 1 extra band noted in the RFLP.
Four patients from cluster B stayed at SRO hotel B in an adjacent county during a TB outbreak there in 1998. A subsequent investigation revealed contacts with LTBI. Most of them received treatment at that time.
Illicit drug use. We identified 2 houses used for crack-cocaine smoking. Persons who gathered there were commonly unacquainted with each other or acquainted only by nicknames. Sixteen patients from the 3 clusters reported crack-cocaine use, and some of these patients reported that they shared pipes during the 3–6 months preceding their TB diagnosis. However, they were reluctant to name the location of the “crack houses” unless a patient was the owner of the house.
Delay in Diagnosis
At least 15 (58%) of the 26 patients received diagnoses 2–18 months after the onset of symptoms, with 7 patients having had delays in diagnosis of ⩾5 months. Ten of the 15 patients had sputum smears test positive for AFB, and 6 had cavitary disease. TB was not initially suspected for 6 of these patients, despite respiratory complaints or abnormal chest radiograph findings. In addition, at least 6 other patients did not seek timely medical care, despite having respiratory symptoms.
Contact Identification and Management
Of the total 292 contacts, 194 (66%) were previously identified by the health department, 185 (95%) of whom were tested. Twenty-nine (16%) of the 185 contacts had positive TST results. Twenty-six patients (90%) completed LTBI treatment, and 3 refused treatment. Ninety-eight additional contacts (34%) were found during our investigation; 30 (31%) of these contacts were tested, 5 (17%) of whom had positive TST results. Two began receiving LTBI treatment, and 3 refused treatment. The TST results for contacts and their relationships to patients are shown in tables 2 and 3. Sites of exposure for TST-positive contacts are shown in table 4.
Discussion
Our investigation demonstrated the value of universal genotyping in revealing unapparent multiyear community TB transmission among high-risk groups in a low-incidence state. Without any genotyping results, recent TB transmission would not have been suspected for most of the patients described here. Furthermore, if the Wisconsin TB program had submitted some but not all Mycobacterium tuberculosis isolates for genotyping, the true scope of these clusters would probably not have been recognized, because the suspicion of recent transmission was not great enough to have led to the genotyping of most of the isolates. These clusters were identified only because the Wisconsin TB program had a policy of genotyping isolates from all patients with culture-positive TB.
The combination of genetic clustering and shared risk factors provided critical clues that unrecognized recent transmission might be occurring, and our focused and patient-tailored investigation revealed unsuspected epidemiological links and additional sites of transmission missed during the routine contact investigation. The necessity of persistent and meticulous contact tracing among high-risk groups was indicated in previous studies [6, 14]. Our experience shows that, when genotyping results suggested potential transmission links between patients, specific questions based on information from previous interviews of patients from the same cluster and extension of a 2-year exposure period contributed to establishing these epidemiologic links. Our identification of epidemiologic links among clustered patients suggests that the combination of spoligotyping, MIRU analysis, and IS6110-based RFLP genotyping provides an accurate approach to the identification of patients with TB who might be involved in previously unsuspected recent transmission. These 3 tests are now available to genotype isolates from all patients with TB in the United States [13].
Ongoing transmission was sustained by a constellation of factors, including crack-cocaine use; alcohol abuse; unemployment; frequent social gatherings in bars, restaurants, or liquor stores; a transient lifestyle; residence in homeless shelters or SRO hotels; and delays in diagnosis. Previous studies have described similar patient characteristics as being associated with belonging to a TB cluster, suggesting a profile of patients more likely to be involved in recent transmission [3–5, 14–18]. These were difficult-to-reach populations, which resulted in incomplete contact tracing and contributed to ongoing transmission, as revealed by our investigation and as described elsewhere [19]. Patients in these high-risk groups are likely to delay seeking timely medical care and to not adhere to TB treatment, causing a prolonged period of infectivity [7, 20]. In addition, delays in diagnosis propagated this ongoing transmission. These delays have been described elsewhere as a major reason for community-based TB outbreaks [20–22].
Shelters increasingly contribute to recent TB transmission because of crowding and poor ventilation, as well as risk factors (e.g., illicit drug use and alcohol abuse) among the homeless population [5, 23, 24–27]. Moreover, intermittently homeless persons (those who work outside but take their meals and sleep at the shelters or use only daytime services) are more likely to acquire and introduce TB not only into the shelter but also into the community during their social contacts. Frequent moves of infectious homeless patients may contribute to a greater spread of TB among shelters. A study in Los Angeles showed that recent TB infection was less frequent among long-term homeless persons because they were more socially isolated than those who stayed in shelters intermittently [27]. Furthermore, identification of contacts of homeless persons appears to be exceedingly difficult, particularly if the investigation has been reopened after several years. As shown previously, only one-half of the patients who resided in homeless shelters had contacts identified by traditional approaches [28]. Therefore, detection of transmission among homeless persons might be easier if location-based screening is implemented [29].
The association between crack-cocaine use and recent TB transmission has been described elsewhere [7, 30, 31]. Coughing induced by crack cocaine and pipe sharing during crack-cocaine smoking commonly occurs in poorly ventilated settings, thus facilitating the airborne transmission of M. tuberculosis. Lack of cooperation of drug users with health department personnel contributed to the delay in recognizing the transmission and initiating timely TB-control measures.
When TB transmission occurs among illicit drug–users, the inability or reluctance of patients to share names and sites, limited recall of full names and activities, and behavioral patterns (i.e., illicit drug use, high occurrence of casual contacts, and frequent changes of residence) appeared to be barriers for relevant contact tracing and have resulted in limited effectiveness of the traditional contact investigation [6, 16, 19]. We could not interview 11 patients: 5 were deceased, 3 had moved out of state, and 3 were not found because of their transient lifestyle. Isolates from 3 patients were not available for genotyping to confirm the discovered epidemiologic links with other clustered patients. Besides, the other difficulties of reopening the contact investigation after considerable time has elapsed, such as the absence of a database of clients from previous years in the shelter and the closing of the bar identified as a potential site of transmission, likely resulted in our missing unnamed casual contacts during the investigation. Unnamed nonhousehold contacts accounted for a large proportion of the TB cases from failed contact identifications, because casual transmission appears to be hard to detect by conventional methods of contact tracing [3, 19, 20, 32].
Despite these difficulties, which are common for an outbreak investigation among high-risk groups, genotyping allowed identification of the unsuspected epidemiologic links among all patients but 1, supporting the assumption that clustering was a result of recent transmission. Moreover, the child who had TB disease diagnosed and matched by genotype with cluster B validates recent interstate transmission.
Although thorough traditional contact investigation should be the cornerstone of detection of TB transmission, genotyping has been shown to aid in detection of unapparent transmission in the community even before an increase in incidence [8, 33]. As we found during this investigation, the first patients in each cluster had epidemiologic links with later patients, suggesting that TB transmission had started several years earlier. If genotyping had been available to the state TB program in 1998, it might have helped detect TB clusters earlier, allowing timely implementation of control measures and potentially averting at least 14 cases.
In summary, transmission of TB among high-risk, hard-to-reach groups may go undetected for several years [19, 20, 29]. Universal genotyping, when combined with intensive epidemiological methods, can help overcome the difficulties inherent in conducting TB investigations among these groups and can assist TB-control officials in detecting unsuspected recent transmission.
Acknowledgments
We are indebted to the staff of the Tuberculosis Control Clinic, City of Milwaukee Health Department (Wisconsin) and Tuberculosis Program, Bureau of Communicable Diseases, Division of Public Health, State of Wisconsin Department of Health and Family Services (Madison), for their assistance and participation in this investigation. We would also like to thank the Wisconsin State Laboratory of Hygiene (Madison) for assistance in laboratory analysis. We extend special thanks to Dawn Tuckey, for her contribution to the programmatic aspects of the investigation; Douglas Gieryn, for his overall coordination and the support of his staff; Carol A. Johnsen, for her valuable assistance during the investigation; Drs. Seth Foldy and Jack T. Crawford, for critical review of the manuscript; and Dr. Thomas R. Navin, for essential and valuable input during the review of the manuscript.
Potential conflicts of interest. All authors: no conflicts.
Comments