Outbreak of Tuberculosis Among Homeless Persons ...

MAJOR ARTICLE

HIV/AIDS

Outbreak of Tuberculosis Among Homeless Persons Coinfected with Human Immunodeficiency Virus Peter D. McElroy,1,2 Karen L. Southwick,1,5,a Ellen R. Fortenberry,4 Elizabeth C. Levine,4 Lois A. Diem,3 Charles L. Woodley,3 Paula M. Williams,4 Kimberly D. McCarthy,6 Rene´e Ridzon,1,a and Peter A. Leone3,7

We investigated a cluster of patients with tuberculosis (TB) in North Carolina and determined the extent of transmission of 1 strain of Mycobacterium tuberculosis. A retrospective cohort study was conducted. Homeless shelter attendance and medical records for 1999 and 2000 were reviewed. The period of exposure to M. tuberculosis was determined, and shelter residents were offered TB screening. DNA fingerprinting was performed on 72 M. tuberculosis isolates. In addition to the initial index cluster of 9 patients, another 16 patients were identified. Isolates of M. tuberculosis from all 25 patients shared a matching DNA fingerprint pattern. All but 1 patient was male, 22 (88%) were African American, and 14 (56%) were human immunodeficiency virus–infected. An epidemiological link to a single shelter was identified for all but 1 patient. Earlier recognition of this shelter as a site of M. tuberculosis transmission could have been facilitated through innovative approaches to contact investigation and through genetic typing of isolates. The incidence of tuberculosis (TB) in the United States has declined to its lowest rate in recorded history. The 2001 rate of 5.6 TB cases per 100,000 population demonstrates that this is a rare disease among many strata of the US population [1]. Nevertheless, recently described TB outbreaks indicate ongoing transmission of Mycobacterium tuberculosis within certain demographic groups [2–5]. Homeless persons represent 1 high-risk group for whom TB control efforts must be strengthened to further reduce TB-associated morbidity [6]. Nationally, 6% of the TB cases reported in 2001 were

Received 6 August 2002; accepted 10 January 2003; electronically published 9 May 2003. a Present affiliations: New York State Department of Health, Albany, New York (K.L.S.) and Bill and Melinda Gates Foundation, Seattle, Washington (R.R.).

Reprints or correspondence: Dr. Peter D. McElroy, Surveillance and Epidemiology Branch, Div. of Tuberculosis Elimination, Mailstop E-10, Centers for Disease Control and Prevention, 1600 Clifton Rd., Atlanta, GA 30333 ([email protected]). Clinical Infectious Diseases 2003; 36:1305–12 2003 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2003/3610-0015$15.00

in people who were homeless at some point in the year preceding diagnosis [1]. There has been little change in this percentage since 1993, when the Centers for Disease Control and Prevention (CDC) began recording data on homelessness as part of TB surveillance [1, 7]. Yet, in some urban areas (population, 1200,000), the proportion of patients with TB who had a history of homelessness in the year preceding diagnosis is 125% [8] (CDC, unpublished data). Compared with the general population, homeless persons have a greater risk of latent M. tuberculosis infection (LTBI) [9–11]. The higher prevalence of substance abuse and HIV infection in this population increases the likelihood that such patients will progress from LTBI to active disease [12, 13]. Case management and contact investigation of homeless patients with TB present difficult challenges for local TB control programs and often require resource-intensive efforts [14]. We describe an outbreak investigation of a cluster of TB cases at a large homeless shelter in Raleigh, North Carolina, and how the frequency of stays at the shelter HIV/AIDS • CID 2003:36 (15 May) • 1305

Downloaded from http://cid.oxfordjournals.org/ by guest on January 3, 2016

1 Division of Tuberculosis Elimination, National Center for HIV, STD, and TB Prevention, 2Epidemic Intelligence Service, Epidemiology Program Office, and 3Division of AIDS, STD, and TB Laboratory Research, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia; and 4Wake County Human Services, 5North Carolina Department of Health and Human Services, and 6Mycobacteriology Section, North Carolina Laboratory of Public Health, Raleigh, and 7Department of Medicine, University of North Carolina at Chapel Hill, North Carolina

increased the risk of M. tuberculosis infection. Overcoming the challenge of finding additional undiagnosed cases and the many contacts in need of treatment for LTBI required novel approaches.

PATIENTS AND METHODS

1306 • CID 2003:36 (15 May) • HIV/AIDS

RESULTS Shelter and resident characteristics. The shelter (dimensions, 92 ⫻ 50 m), originally constructed as a seafood warehouse, was converted to a homeless facility in 1997. The occupancy often exceeded 350 persons per night, and, in 1999, it housed 13000 different individuals. The structure had forcedair heating and central air conditioning. The majority of residents were male (97%), African American (60%), and mostly young to middle-aged (median age, 37 years, with 95% aged !55 years). Meals were not served, and persons aged !18 years were not permitted. Until July 2000, when the shelter was converted to an all-male facility, a small number of women (!10) slept in an adjoining room separated by a wall and ceiling. Initial case cluster. The initially recognized cluster included 9 African American men aged 37–50 years. Six of these patients sought care at the county hospital emergency department, and 3 were identified through contact investigations. Six


Recognition of outbreak. In March 2000, Wake County Human Services (Raleigh) recognized a cluster of TB cases in 7 persons with a history of residence at the same homeless shelter in Raleigh. After 2 additional cases were diagnosed the next month (April 2000), an intensive investigation was initiated. Record review. A medical record review of the 9 clustered TB cases was conducted. Records for other homeless patients for whom TB was diagnosed in Wake County between 1999 and the start of the investigation were also reviewed. Shelter attendance records, collected at nightly check-in, were obtained for 1999 and 2000 and included name, age, ethnicity, and specific dates of overnight stays. Dates of previous incarceration in the Wake County jail and state prison system were obtained. Employment records from 3 temporary labor agencies were also reviewed. Infectious periods and the exposed cohort. Patients with culture-confirmed TB and sputum smears positive for acid-fast bacilli were considered potentially infectious for the 3-month period immediately preceding the date of TB diagnosis. Because patients were not permitted to stay in the shelter after receiving a diagnosis of TB, the infectious periods for shelter residents ended on the date of TB diagnosis. The shelter attendance records of the initial 9 patients were overlaid with the probable infectious periods of the patients with positive sputum smear results. This was performed to identify specific periods when other shelter residents were likely exposed to droplet nuclei that contained M. tuberculosis. Screening and LTBI treatment. During the course of the outbreak investigation, 4 tuberculin skin test (TST) screenings were conducted at the shelter (in December 1999 and in February, May, and October 2000), and screening results were made available to each shelter resident. TSTs were performed by the Mantoux method, and an induration of ⭓5 mm was considered a positive result for all contacts. On-site chest radiography was performed with a mobile unit. Radiographs were read and interpreted by the same radiologists who routinely participated in the TB program. Shelter residents with a positive TST result, a reported history of a positive TST result, symptoms suggestive of active TB, or HIV infection were eligible to undergo chest radiography. Persons with a positive TST result were encouraged to be tested for HIV infection. HIV counseling and testing were provided at the shelter for 5 consecutive nights in May 2000. Eligible candidates for the treatment of LTBI were offered a

2-month regimen of twice-weekly, directly observed therapy with rifampin and pyrazinamide [15]. All candidates underwent baseline laboratory testing and were encouraged to consent to an HIV test. A nursing contractor was hired to provide on-site therapy for and clinical monitoring of adverse reactions at each twice-weekly visit. Laboratory investigations. M. tuberculosis isolates from all patients in Wake County with culture-confirmed TB in 1999 and 2000 were forwarded to the CDC’s Tuberculosis/Mycobacteriology Branch. There, spoligotyping was used as a preliminary tool to determine whether isolates from the initial cluster of cases were genetically similar [16]. DNA fingerprinting was also performed at the CDC by IS6110-based restriction fragment–length polymorphism (RFLP) analysis of all isolates from Wake County [17]. A “unique” fingerprint pattern was defined as an IS6110 pattern that did not match the pattern of any other typed isolate. A “cluster” was defined as ⭓2 isolates with matching IS6110 and spoligotype patterns. Statistical analysis. The effect of shelter exposure duration on the risk of M. tuberculosis infection was explored with univariate and stratified analyses. The outcome variable was the TST result (negative or positive) determined at 1 of the 4 shelter screenings. The primary exposure variable was the number of overnight stays at the shelter during the 5-month exposure period (range, 0–153 days), categorized into 4 equally spaced groups (⭐38, 39–76, 77–114, and 115–153 nights). Age, sex, and race or ethnicity (white, African American, or Hispanic) were analyzed for the presence of confounding factors and interactions. Risk ratios with 95% CIs were estimated with SAS, version 6.12 (SAS Institute). Trends in the risk ratios, with respect to increased levels of shelter exposure, were assessed by a x2 test.

patients were coinfected with HIV. The index patient (who received a diagnosis of TB in November 1999) was 37 years old and HIV uninfected, and he had a cavity (diameter, 4.5 cm) in the upper lobe of the left lung and a cavitary nodule (diameter, 1.0 cm) in the middle lobe of the right lung. In addition to a history of residence at the shelter, he was employed by a temporary labor agency through which he intermittently obtained work at outdoor construction sites. Shelter records indicated that 7 of the initial 9 patients stayed in the same shelter during the months preceding receipt of their TB diagnosis (figure 1). Patient 3 received a diagnosis of TB in a North Carolina state prison, where he had been incarcerated since December 1998. Records indicated that he stayed in the shelter during the fall of 1998 and that, before incarceration, he had worked for the same labor agency as did the index patient. Patient 8 spent no self-reported or documented nights in the shelter during the 6 months before diagnosis but reported staying there in the fall of 1998, 9 months before diagnosis. Of the initial 9 patients, 3 (the index patient and patients 2 and 5) had sputum smears positive for acid-fast bacilli. The estimated infectious periods of these 3 patients overlapped with the dates of many of their overnight stays in the shelter during the 5-month period from September 1999 through January 2000 (figure 1). Although the infectious period for patient 5 extended through the end of February 2000, attendance records showed he spent no nights in the shelter after late January 2000. M. tuberculosis isolates from the initial patients shared a common spoligotype (octal code, 777776777760771), suggest-

ing a unique outbreak strain, which was later confirmed by RFLP analysis [18]. Exposed cohort. The cohort of exposed shelter residents included persons who visited the shelter in the 5-month period from September 1999 through January 2000. A total of 1420 men spent ⭓1 night (median, 12 nights; range, 1–153 nights) in the shelter during this time. Only 138 men (10%) stayed 176 nights in the shelter; these men were significantly more likely to be African American (75% of the men) than white (13%; P ! .001) or Hispanic (12%; P ! .001). TB screening events. A total of 620 shelter residents and all employees (n p 26) underwent ⭓1 on-site TST screening. Among these persons, 455 had a TST placed and read at least once, 134 had a TST placed and not read, and 57 reported a history of a positive TST result. Table 1 presents TST results from the 4 screenings. The number of positive TST results was

Table 1. Results of tuberculin skin test (TST) screenings performed at a homeless shelter in Wake County, North Carolina, in 1999–2000. No. or no. (%) of patients, by month Variable

Dec 1999

Feb 2000

May 2000

Nov 2000

TST placed

229

280

120

167

TST read

175

205

105

138

Positive TST result

30 (17)

45 (22)

21 (20)

14 (10)

NOTE. On the basis of results of TST screenings conducted at other shelters in the Wake County area, the expected range of positive TST results in this population would be 8%–12% (Wake County TB Control Program, unpublished data).

HIV/AIDS • CID 2003:36 (15 May) • 1307


Figure 1. History of overnight stays at the Wake County, North Carolina, homeless shelter before receipt of tuberculosis (TB) diagnosis and the corresponding 3-month period of infectiousness for patients with sputum smear results positive for acid-fact bacilli. Index, index patient.

Table 2. Tuberculin skin test (TST) results among African American males according to number of nights spent in a homeless shelter in Wake County, North Carolina, during the 5-month highrisk exposure period (September 1999–January 2000).

No. of nightsa ⭐38 39–76

No. of subjects

Positive TST result, no. (%) of subjectsb

189

34 (18)

Referent

89

24 (27)

1.8 (0.9–3.4)

Risk ratio (95% CI)

77–114

47

20 (43)

2.7 (1.4–5.1)

115–153

38

18 (47)

3.5 (1.9–6.4)

a b

Categories are based on 4 equal groups (quartiles) of exposure duration. TST was placed and read.

Figure 2. Tuberculosis (TB) cases (n p 25) , by date of diagnosis, associated with a Wake County, North Carolina, homeless shelter. Data include the initially recognized case cluster, the subsequent cases that were diagnosed, and the additional outbreak-related cases found through record review of patients in the county who had previously received a diagnosis of TB. Index, index patient; +, HIV-infected patient. 1308 • CID 2003:36 (15 May) • HIV/AIDS


highest at the February 2000 screening (22%) and was lowest at the November 2000 screening (10%), which was after implementation of control activities. Chest radiography was performed on 85% of the eligible persons. Analysis of the TST data according to exposure duration revealed a significant trend (x2 test for trend, 21.2; df, 1; P ! .001) toward increased risk ratios as the number of nights spent in the shelter increased (table 2). After controlling for race or ethnicity, persons in the longest exposure category (i.e., 115–153 nights in the shelter) had a 3.5 times greater risk of having a positive TST result than did persons in the shortest exposure category (i.e., ⭐38 nights). Neither age nor sex confounded or modified these associations. Additional case findings. Seven additional infected pa-

tients, 6 of whom were African American and 1 of whom was Hispanic, received a diagnosis of TB between June 2000 and May 2002 (figure 2). Four patients received a diagnosis at the county hospital emergency department, 2 at the shelter following screening, and 1 at the county TB clinic. Five of these individuals resided in the shelter for extended periods (⭓77 nights) during the 5-month exposure window. The only female patient in this outbreak was the girlfriend of a patient in the outbreak group. Isolates from all 7 patients had spoligotype and RFLP patterns that matched those of the other outbreak isolates. Through May 2002, genetic typing continued on isolates from all patients with TB suspected to be outbreak associated. The patient who received a diagnosis of TB in May 2002 resided in the shelter for 89 days during the 5-month exposure window. Review of 13 patients in Wake County with TB and with a history of homelessness from 1998 through 2000 revealed 9 patients with potential links to the outbreak cluster. Our investigation then confirmed that 7 patients had a history at the Raleigh shelter and had isolates with RFLP patterns that matched those of the outbreak strain. Despite having isolates with RFLP patterns that matched those of the outbreak strain, the remaining 2 patients had no record of residing at this shelter. One was ultimately linked to the outbreak through a review of temporary labor agency records, but no link to the shelter or to any employment site was identified for the other patient. The outbreak cases constituted 48% of the culture-confirmed TB cases in Wake County from July 1999 though July 2000.

Table 3. Demographic and clinical characteristics of 25 patients with tuberculosis (TB) associated with a TB outbreak in a Wake County, North Carolina, homeless shelter from 1998 to 2002. Characteristic

Value

Demographic data Age, median years (range)

41 (27–50)

Male sex

24 (96)

Born in United States

22 (88)

Race or ethnicity African American Hispanic

22 (88) 3 (12)

Laboratory finding Positive sputum smear result

14 (56)

Positive culture result

25 (100) 25 (100)

History of STDa

15 (60)

HIV coinfection

14 (56)

Cavitary disease

4 (16)

Disease site Pulmonary only

20 (80)

Extrapulmonary only

1 (4)

Pulmonary and extrapulmonary

4 (16)

Incarceration history preceding TB diagnosis County jail State prison

18 (72) 6 (24)

NOTE. Data are no. (%) of patients, unless otherwise indicated. M. tuberculosis, Mycobacterium tuberculosis. a

Any history of diagnosis of a sexually transmitted disease (STD) at Wake County Human Services STD Clinic.

The demographic and clinical characteristics of the outbreak patients are summarized in table 3. The majority (56%) were HIV-infected at the time of TB diagnosis. RFLP analysis was performed on a total of 72 M. tuberculosis isolates (95%) from the 76 patients with culture-confirmed TB that was diagnosed in Wake County between January 1999 and April 2001. The RFLP patterns of these isolates ranged from 1 to 19 bands and included 41 unique patterns and 5 distinct clusters (figure 3). The outbreak cluster was the largest cluster during this period and included 19 isolates with a 3-band pattern and 4 isolates with a 4-band pattern (which had the same 3-band pattern as the other 19 isolates plus 1 additional band); all isolates were confirmed by results of spoligotyping to be a single strain. The remaining 4 clusters included only 2 isolates each (5-, 11-, 13-, and 14-band RFLP patterns). Treatment of LTBI. The investigation identified 119 candidates for the treatment of LTBI; 26 patients (22%) began the treatment course, and 16 (14%) successfully completed it. Among the 10 patients who did not complete the course, 1 developed a rash, 1 experienced symptoms of hepatitis, 2 refused to continue receiving therapy, and 6 patients were lost

DISCUSSION Our investigation revealed 12.5 times as many outbreakassociated cases of TB as were initially recognized. Transmission of a common strain of M. tuberculosis was responsible for at least one-half of the culture-confirmed TB cases from July 1999 through July 2000 and for three-quarters (23 of 31 isolates) of all genotype-clustered cases of TB in Wake County during 1999 and 2000 (23 of 31 clustered isolates). This represents a considerably higher than expected proportion of cases that were due to recent outbreak-associated transmission [19–21]. As in other locations, HIV coinfection continues to be an important factor contributing to the clustering of TB cases [5, 22, 23]. The transient nature of a homeless person’s life often interferes with the timely diagnosis and successful treatment of TB. Many strategies have been proposed for facilitating early detection of TB among residents of homeless shelters; they include symptom screening at entry to a shelter, receipt of proof that a TST has recently been performed, spot sputum screening, periodic chest radiography of all symptomatic persons, and education of shelter staff concerning the signs and symptoms of TB [8, 24–27]. Unfortunately, no consensus exists regarding the success of these approaches in homeless shelter settings. A recent analysis by Jones and Schaffner [28] found that screening for active TB with a miniature chest radiography unit can be a cost-effective approach in jail settings. It should be determined whether a similar approach is practical in homeless shelters. Given that at least one-half of the outbreak patients either were incarcerated in the county jail or had visited the Wake County Human Services Sexually Transmitted Diseases Clinic HIV/AIDS • CID 2003:36 (15 May) • 1309


Matching M. tuberculosis genotype

to follow-up. The 2-month course of twice-weekly, directly observed therapy with rifampin and pyrazinamide was well tolerated in the remaining patients. Approximately 100 persons agreed to undergo HIV testing and receive counseling; 2 were HIV-infected and encouraged to initiate treatment for LTBI. Control activities. Shelter staff members’ and residents’ awareness of TB signs and symptoms was increased through on-site visits and educational sessions conducted by the Wake County TB Program and the CDC. A symptom-screening checklist for regular shelter attendees was implemented in July 2000. To improve ventilation in the shelter, environmental measures were implemented in June 2000 and included the installation of 3 large fans (diameter, 1.5 m). Windows and doors, previously opened infrequently, were opened daily and nightly to help ventilate the facility. The rate of positive TST results from November 2000 indicated a return to background rates (table 1). This investigation was part of an emergency public health response and was determined not to be human subjects research. The shelter was demolished in early 2002.

(Raleigh) within 2 years preceding receipt of TB diagnosis, some patients with LTBI potentially could have been identified if a TST had been offered at these facilities. Although prison systems routinely perform annual TB screening, jails are presumed to house inmates for a period of time insufficient to permit detection or treatment of LTBI or active disease. However, for many homeless persons, the frequency and duration of incarceration in local and county jails may necessitate a reexamination of policies regarding identification and treatment of persons with LTBI in these facilities [29]. The utility of routine DNA fingerprinting of M. tuberculosis isolates, outside of outbreak situations, remains unclear. In the absence of clear associations among outbreak-associated cases (i.e., among patients who did not report homeless status or identify a specific shelter visited), routine genotyping of isolates might have implicated a common strain that suggested recent transmission. We obtained and analyzed shelter attendance records in July 2000, one year after the first cases were diagnosed. Such data are often unavailable or too cumbersome for TB control staff to review. Had this county routinely performed DNA fingerprinting, the initial wave of outbreak cases may have been recognized in the early autumn of 1999. If the shelter had been identified as a likely transmission site at that time and if more aggressive TST screening and treatment for LTBI had been implemented, subsequent cases may have been prevented, particularly among HIV-infected persons. DNA fingerprinting of M. tuberculosis isolates should not be a substitute for sound epidemiological investigation of each patient who has recently received a diagnosis of TB. However, sometimes RFLP detects previously unrecognized links among cases and serves as an impetus for the TB program to further question a patient regarding the patient’s daily routine, contacts, and places frequented. This strategy may uncover social 1310 • CID 2003:36 (15 May) • HIV/AIDS

networks with ongoing transmission of M. tuberculosis. Recent TB outbreaks have been described for which DNA fingerprinting helped hone investigators’ use of ethnographic investigative techniques to uncover unique social networks and implement focused control strategies [5, 30, 31]. This outbreak illustrates the need for improved strategies for ensuring the initiation and successful completion of treatment for LTBI among the homeless. Regardless of HIV serostatus, a 9-month course of isoniazid therapy is the preferred treatment for LTBI. Among HIV-infected persons, a 2-month treatment regimen of rifampin and pyrazinamide provides an alternative to isoniazid therapy, but it is only recommended if the patient can be closely monitored (clinically and biochemically) [32]. Even with this short-course, directly administered therapy, the proportion of patients who initiated and successfully completed this regimen was no better than that of homeless persons in trials of isoniazid therapy [33, 34]. Recent reports [32, 35] that hepatotoxicity is associated with the 2-month regimen of rifampin and pyrazinamide makes transient homeless persons poor candidates for this alternative option because patient monitoring may be insufficient. Our investigation took place in an urban area with a homeless population that may be uncharacteristic of the homeless population in larger cities. The vast majority of persons who stayed at the Raleigh shelter did so for a matter of a few weeks, and they may have only been traveling through the Raleigh area. Thus, it is possible that other communities could have achieved greater success with initiating and completing treatment of homeless patients with LTBI. Likewise, although RFLP analysis may help reveal associations among cases, it is possible that the delay between the diagnosis of TB in a transient homeless person and the receipt of the corresponding RFLP pattern may reveal such associations too late. If the patient has moved away


Figure 3. Frequency distribution of IS6110 copies in restriction fragment–length polymorphism (RFLP) patterns of Mycobacterium tuberculosis isolates from all patients with culture-confirmed tuberculosis (TB) in Wake County, North Carolina, diagnosed from January 1999 through April 2001.

Acknowledgments

We thank Dee Foster (North Carolina Department of Health and Human Services), Gibbie Harris (Wake County Human Services), and Robert Pratt, Greg Andrews, Eric Williamson, and Paul Tribble (Division of Tuberculosis Elimination, Centers for Disease Control and Prevention) for their assistance with this investigation.

References 1. Centers for Disease Control and Prevention (CDC). Reported tuberculosis in the United States, 2001. Atlanta, GA: US Department of Health and Human Services, CDC, 2001. 2. Curtis AB, Ridzon R, Novick LF, et al. Analysis of Mycobacterium tuberculosis patterns in a homeless shelter outbreak. Int J Tuberc Lung Dis 2000; 4:308–13. 3. Jones TF, Craig AS, Valway SE, Woodley CL, Schaffner W. Transmission of tuberculosis in a jail. Ann Intern Med 1999; 131:557–63. 4. Kline SE, Hedemark LL, Davies SF. Outbreak of tuberculosis among regular patrons of a neighborhood bar. N Engl J Med 1995; 333:222–7. 5. Sterling TR, Thompson D, Stanley RL, et al. A multi-state outbreak of tuberculosis among members of a highly mobile social network: implications for tuberculosis elimination. Int J Tuberc Lung Dis 2000;4: 1066–73. 6. Institute of Medicine. Ending neglect: the elimination of tuberculosis in the United States. Washington, DC: National Academy Press, 2000. 7. Centers for Disease Control and Prevention (CDC). Reported tuberculosis in the United States, 1993. Atlanta, GA: US Department of Health and Human Services, CDC, 1994. 8. Kimerling ME, Shakes CF, Carlisle R, Lok KH, Benjamin WH, Dunlap NE. Spot sputum screening: evaluation of an intervention in two homeless shelters. Int J Tuberc Lung Dis 1999; 3:613–9. 9. Zolopa AR, Hahn JA, Gorter R, et al. HIV and tuberculosis infection in San Francisco’s homeless adults: prevalence and risk factors in a representative sample. JAMA 1994; 272:455–61.

10. Gelberg L, Panarites CJ, Morgenstern H, Leake B, Andersen RM, Koegel P. Tuberculosis skin testing among homeless adults. J Gen Intern Med 1997; 12:25–33. 11. Paul EA, Lebowitz SM, Moore RE, Hoven CW, Bennett BA, Chen A. Nemesis revisited: tuberculosis infection in a New York City men’s shelter. Am J Public Health 1993; 83:1743–5. 12. Linn LS, Gelberg L, Leake B. Substance abuse and mental health status among homeless and domiciled low-income users of a medical clinic. Hosp Community Psychiatry 1990; 41:306–10. 13. Culhane DP, Gollub E, Kuhn R, Shpaner M. The co-occurrence of AIDS and homelessness: results from the integration of administrative databases for AIDS surveillance and public shelter utilisation in Philadelphia. J Epidemiol Community Health 2001; 55:515–20. 14. Nolan CM, Elarth AM, Barr H, Saeed AM, Risser DR. An outbreak of tuberculosis in a shelter for homeless men: a description of its evolution and control. Am Rev Respir Dis 1991; 143:257–61. 15. Centers for Disease Control and Prevention. Targeted tuberculin testing and treatment of latent tuberculosis infection. MMWR Morb Mortal Wkly Rep 2000; 49(RR-6):26–39. 16. Kamerbeek J, Schouls L, Kolk A, et al. Simultaneous detection and strain differentiation of Mycobacterium tuberculosis for diagnosis and epidemiology. J Clin Microbiol 1997; 35:907–14. 17. van Embden JDA, Cave MD, Crawford JT, et al. Strain identification of Mycobacterium tuberculosis by DNA fingerprinting: recommendations for a standardized methodology. J Clin Microbiol 1993; 31:406–9. 18. Dale JW, Brittain D, Cataldi AA, et al. Spacer oligonucleotide typing of bacteria of the Mycobacterium tuberculosis complex: recommendations for standardised nomenclature. Int J Tuberc Lung Dis 2001; 5: 216–9. 19. Bishai WR, Graham NMH, Harrrington S, et al. Molecular and geographic patterns of tuberculosis transmission after 15 years of directly observed therapy. JAMA 1998; 280:1679–84. 20. Barnes PF, Yang Z, Preston-Martin S, et al. Patterns of tuberculosis transmission in central Los Angeles. JAMA 1997; 278:1159–63. 21. Small PM, Hopewell PC, Singh SP, et al. The epidemiology of tuberculosis in San Francisco. N Engl J Med 1994; 330:1703–9. 22. Alland D, Kalkut GE, Moss AR. Transmission of tuberculosis in New York City. N Engl J Med 1994; 330:1710–6. 23. Moss AR, Hahn JA, Tulsky JP, Daley CL, Small PM, Hopewell PC. Tuberculosis in the homeless: a prospective study. Am J Respir Crit Care Med 2000; 162:460–4. 24. Rendleman NJ. Mandated tuberculosis screening in a community of homeless people. Am J Prev Med 1999; 17:108–13. 25. Southern A, Premaratne N, English M, Balazs J, O’Sullivan D. Tuberculosis among homeless people in London: an effective model of screening and treatment. Int J Tuberc Lung Dis 1999; 3:1001–8. 26. Layton MC, Cantwell MF, Dorsinville GJ, Valway SE, Onorato IM, Frieden TR. Tuberculosis screening among homeless persons with AIDS living in single-room-occupancy hotels. Am J Public Health 1995; 85: 1556–9. 27. Mayo K, White S, Oates SK, Franklin F. Community collaboration: prevention and control of tuberculosis in a homeless shelter. Public Health Nurs 1996; 13:120–7. 28. Jones TF, Schaffner W. Miniature chest radiograph screening for tuberculosis in jails: a cost-effective analysis. Am J Respir Crit Care Med 2001; 164:77–81. 29. Hammett TM, Maruschak LM. 1996–1997 Update: HIV/AIDS, STDs, and TB in correctional facilities. Washington, DC: US Department of Justice, Office of Justice Programs, National Institute of Justice, 1999. 30. Klovdahl AS, Graviss EA, Yaganehdoost A, et al. Networks and tuberculosis: an undetected community outbreak involving public places. Soc Sci Med 2001; 52:681–94. 31. Fitzpatrick LK, Hardacker JA, Heirendt W, et al. A preventable outbreak of tuberculosis investigated through an intricate social network. Clin Infect Dis 2001; 33:1801–6. 32. Centers for Disease Control and Prevention. Update: fatal and severe

HIV/AIDS • CID 2003:36 (15 May) • 1311


and cannot be interviewed more specifically about his or her association with other persons in the community, then the full benefit of a known RFLP pattern may not be realized. As the incidence of TB continues to decline in the United States, it will be increasingly difficult to identify source cases and additional contacts who will benefit from treatment for LTBI. TB programs that rely mostly on contact investigations around the household and workplace may already find that their efforts are yielding diminishing returns. Strategies that systematically uncover epidemiological links, not only among TB patients and their named contacts (i.e., primary contacts), but also among the contacts (i.e., secondary contacts) of those primary contacts with LTBI, may add valuable insight. Analysis of data to find and describe interconnections among patients and contacts will reveal key persons and places essential to an outbreak network [36]. Contacts associated with these key persons and places may then be targeted for control activities. The CDC is currently collaborating with state and local TB control programs and academic centers to determine the utility of social network analysis in TB contact investigations.

liver injuries associated with rifampin and pyrazinamide for latent tuberculosis infection, and revisions in American Thoracic Society/ CDC recommendations: United States, 2001. MMWR Morb Mortal Wkly Rep 2001; 50:733–5. 33. Pilote L, Tulsky JP, Zolopa AR, Hahn JA, Schecter GF, Moss AR. Tuberculosis prophylaxis in the homeless: a trial to improve adherence to referral. Arch Intern Med 1996; 156:161–5.

34. Tulsky JP, Pilote L, Hahn JA, et al. Adherence to isoniazid prophylaxis in the homeless. Arch Intern Med 2000; 160:697–702. 35. Centers for Disease Control and Prevention. Update: fatal and severe liver injuries associated with rifampin and pyrazinamide treatment for latent tuberculosis infection. MMWR Morb Wkly Rep 2002; 51:998–9. 36. Wasserman S, Faust K. Social network analysis: methods and applications. New York: Cambridge University Press, 1994.


1312 • CID 2003:36 (15 May) • HIV/AIDS