Multidrug-Resistant Tuberculosis  Outbreak among US-bound Hmong  Refugees, Thailand, 2005

Enhanced pre-immigration screening and program expansion decreased TB importation.


Materials and Methods
This investigation was deemed an urgent public health response. Under the federal regulation for the protection of human research participants, Code of Federal Regulations Title 45, part 46, this investigation was determined by CDC to not be human subject research.

Case Defi nition and Case Detection
Cases were defi ned by positive sputum smears or cultures or by a physician's decision to initiate TB treatment in the context of radiographic abnormalities and clinical features consistent with TB. From April 2004 through January 2005, pre-immigration TB screening in the camp detected TB cases among the refugees. The TB screening algorithm used initially consisted of a medical history, a physical examination, and, for applicants >15 years of age, a chest radiograph. Persons whose clinical or radiographic fi ndings suggested TB disease submitted 3 sputum specimens for acid-fast bacilli smear microscopy. Limited laboratory capacity was available, and mycobacterial culture was performed on sputum samples that were smear positive for acid-fast bacilli. Refugees with smear-positive results were allowed to travel to the United States after they had begun anti-TB treatment and had smear-negative results for 3 follow-up sputum specimens. In July 2004, after a site visit by CDC and because of concerns about potential high prevalence of TB, including drug-resistant TB, the pre-immigration screening algorithm was expanded to include mycobacterial culture and drug-susceptibility testing for both smear-negative and smear-positive specimens. From February 2005 (after the TB outbreak was detected) through April 2007, suspected TB cases were also identifi ed through contact tracing.

Patient Interviews
Hmong interpreters interviewed TB patients in the camp, using a 24-item questionnaire about history of previous TB diagnosis and treatment. When asked about previous TB treatment, participants were shown anti-TB medications and asked if they had ever taken any of the displayed pills in the past. Patient responses were stratifi ed by drug-susceptibility testing results. To assess the possibility of recent MDR TB transmission in the camp, MDR TB patients were asked additional questions regarding social links to other known MDR TB patients. A strong link was defi ned as sharing a household or having contact with another MDR TB patient at least 1×/week. A weak link was defi ned as having contact with another MDR TB patient <1×/week.

M. tuberculosis Genotyping
Available M .tuberculosis isolates were genotyped by spoligotyping (19,20), mycobacterial interspersed repeti-tive units (MIRUs) (21,22), and IS6110 restriction fragment length polymorphism (RFLP) analysis (23). A cluster was defi ned as >2 M. tuberculosis isolates that had identical spoligotyping and MIRU results and for which RFLP results were identical or differed by only 1 band. To assess whether the proportion of MDR TB isolates that clustered was greater than that among pansusceptible isolates, we used the Fisher exact test with a signifi cance level of 0.05.

Mapping Patient Households
To assess geographic clustering of cases, we used global positioning system (GPS) technology to map TB patient households in the camp according to drug-susceptibility testing results. GPS data were analyzed with a spatial scan statistic that uses a varying-sized cylinder to encapsulate cases within the radius of the cylinders and calculates a p value and log likelihood ratio to determine the statistical signifi cance of any clusters that may be detected (24).

Tuberculin Skin Testing
Refugees received tuberculin skin tests (TSTs) to evaluate latent TB infection. Induration >5 mm was considered a positive test result (25). To assess recent transmission in the camp, we summarized and compared TST results for 3 categories of contacts: 1) a housemate or family member of a TB patient with at least 1 sputum smear-positive result for acid-fast bacilli; 2) a housemate or family member of a TB patient with only sputum smear-negative results; or 3) not a housemate or family member of a TB patient. Using those who were not a housemate or family member of a TB patient as the referent group, we calculated prevalence ratios and 95% confi dence intervals (CIs) for each exposure group.

Patient Interviews
Of 272 TB patients, 241 (88.6%) were interviewed. Treatment for TB before pre-immigration screenings began in April 2004 was reported by 15 (6.2%); none had received directly observed therapy (DOT). Of the 15 previously treated patients, 3 (20.0%) had a current diagnosis of MDR TB, 1 (6.7%) had TB with isoniazid resistance only, 1 (7.0%) had TB that was pansusceptible, and 10 (66.7%) had received clinical diagnoses without culture confi rmation. All 3 MDR TB patients who had a history of previous treatment had received their treatment during the 3 years before their current diagnosis.
After pre-immigration TB screenings started, some camp residents visited healthcare providers outside the camp to seek treatment for conditions that might have precluded their passing the pre-immigration medical examination. Chest radiograph screenings were obtained outside the camp by 39 (16.2%) patients during the period between when medical screening began at the camp and when their own pre-immigration screening was scheduled. Of these, 32 (82.1%) took TB medications during this time; only 9 (28.1%) reported that they were told by a doctor or nurse that they had TB. Among the remaining 23 patients not reporting a diagnosis of TB, 1 (4.3%) had MDR TB, 1 (4.3%) had streptomycin-resistant TB, 2 (8.7%) had pansusceptible TB, and the remaining 19 (82.6%) were diagnosed clinically without culture confi rmation.
Of the 24 MDR TB patients, 17 (70.8%) responded to the questions regarding social links to other MDR TB patients ( Figure 1.) Among these, 9 (52.9%) reported having at least 1 strong link with another MDR TB patient, and 4 MDR TB patients (23.5%) reported having at least 2 strong links with another MDR TB patient. One patient who had sputum smear-positive TB (Figure 1, patient 11) was central to a social network that linked 13 (76.5%) patients. The 3 MDR TB patients that had been previously treated for TB ( Figure 1, patients 1, 2, and 3,) were all directly linked to patient 11 and included in the 13-patient network.

TB Genotyping
Genotyping results were available for 46 (80.7%) of the 57 culture-confi rmed cases. Of these, 30 (65.2%) belonged to 1 of 5 clusters (clusters A-E), which ranged from 2 to 20 matching isolates. The remaining 16 (34.8%) isolates were unique ( Table 2). The largest culture, cluster C, had 20 cases, all of which were MDR TB; they represented 87.0% of the 23 MDR TB isolates with genotyping results. The cluster C spoligotype pattern was 000000000003771, and the MIRU pattern was 223325173533. Isolates in cluster C had a 21-band RFLP result. Of the MDR TB isolates that were not included in cluster C, 2 had spoligotype and RFLP results that matched those included in cluster C, but according to MIRU results, they differed at 1 locus. The third MDR TB isolate not included in cluster C differed according to both MIRU and RFLP results. MDR TB isolates were more likely than non-MDR TB isolates to cluster (Fisher exact p = 0.02).
Among the 17 MDR TB patients who responded to the questions regarding social links to other MDR TB patients (Figure 1), 15 (88.2%) were included in cluster C. Isolates from patients 5 and 12 matched cluster C according to spoligotype and RFLP results but differed at 1 locus according to MIRU results.   As of 2007, genotyping results were available for 7 additional patients from this camp who had received a diagnosis of MDR TB after arrival in the United States (California). According to spoligotyping and MIRU results, isolates collected from 6 of these patients matched the cluster C strain. The seventh isolate had the same spoligotype result but differed at 1 locus according to MIRU.

Patient Household Maps
TB patients were widely distributed throughout the camp living quarters (≈0.5 km 2 ) (Figure 2). The spatial analysis showed 3 nonsignifi cant spatial clusters (p>0.05), which suggests no signifi cant geographic clustering of TB patient households, either among all patients or among subsets with similar drug-susceptibility testing patterns.

TST Results
Of the 5,637 camp residents that had TSTs performed and results read, 1,624 (28.8%) had positive results. Among those who were family members or housemates of a camp patient with a sputum smear-positive TB diagnosis, 96 (44.0%) had positive TST results. These contacts were 1.6× (95% CI 1.4-1.9) as likely as the referent group to have a positive result. There was an increased risk (not statistically signifi cant) for a positive result associated with being a family member or housemate of a patient with sputum smear-negative disease (relative risk = 1.1, 95% CI 1.0-1.2). After controlling for patients' sputum smear status, we found that household contacts of patients with MDR TB, pansusceptible TB, and TB with some drug resistance had the same risk for a positive TST result.

Modifi cations and Enhancements to Pre-immigration TB Screening and Treatment
After the investigation in February 2005, recommendations for pre-immigration TB screening and treatment for Hmong refugees in Thailand were again modifi ed and enhanced (Table 3). These modifi cations required that all refugees >6 months of age be screened with chest radi-ography. Persons with suspected TB had 3 sputum specimens collected for smear microscopy, culture, and drugsusceptibility testing with rapid, liquid-based methods. All patients were required to show documentation of having received DOT for the duration of their TB treatment before they were permitted to travel to the United States. In addition, a TB culture laboratory was built within the camp, mechanisms were developed to import quality-assured second-line TB medications, and nurses and physicians in the camp received advanced training in TB diagnosis and treatment. Finally, although TSTs were performed for all camp residents during this investigation, the enhanced TB screening algorithm required that only those 6 months to 10 years of age receive TST.
After implementation of the fi nal enhanced TB screening and treatment requirements, 97 additional TB cases, including 2 MDR TB, were diagnosed in Thailand, resulting in an overall total of 369 TB cases. As of April 2007, health departments in the United States reported 46 cases of TB, 6 of which were MDR TB, among 9,455 Hmong refugees who immigrated to the United States before implementation of enhanced screening (487 cases/100,000 persons). In Hmong refugees who immigrated after implementation of enhanced screening (70 cases/100,000 persons). The proportion of cases diagnosed in the United States after the enhanced screening was signifi cantly lower than the proportion diagnosed before the enhanced screening (Fisher exact test p<0.001).

Discussion
An outbreak of MDR TB occurred among a population in which TB rates were already elevated; as a result, TB (some MDR TB) was imported into the United States. Several lines of evidence support the conclusion that this was an outbreak. First, 13 (76.5%) of 17 MDR TB patients interviewed reported having had recent and regular exposure to another MDR TB patient. Although GPS did not demonstrate geographic clustering, lack of clustering is not unexpected because the camp was small and its population density was high, making social networks, rather than absolute physical distance between dwellings, the most important facilitator of TB transmission. Second, 20 (87.0%) of 23 MDR TB isolates were strains that matched by 3 different molecular subtyping methods. It is possible that 2 additional isolates (22 total) were part of the outbreak as well because they matched the outbreak strain according to spoligotyping and RFLP results and differed by only 1 locus according to MIRU; therefore, this difference may represent a change in genotype in the same clone over time (26). Third, elevated rates of tuberculin reactivity in household contacts of smear-positive patients suggest substantial recent transmission, not simply endemic disease. As rates of tuberculin reactivity were elevated in household contacts of all smear-positive patients, transmission of pansusceptible TB, TB with drug-resistance but not MDR TB, and MDR TB likely occurred in the camp before and during the initial resettlement of refugees to the United States in 2004. Rates of TST positivity were not presumed to be infl ated by vaccination with M. bovis BCG, because there was no indication or documentation of BCG vaccination among this group of refugees.
Delayed recognition of TB transmission in the camp had several negative consequences: increased number and severity of cases among refugees and importation of TB, including MDR TB, into the United States. In addition, the costs incurred by the US government were substantial and resulted from halting immigration (e.g., cancellation of fl ights, emergency overseas investigations, and program enhancements), public health investigations in the United States and Thailand, and medical costs of diagnosis and treatment. Because half of the cases of MDR TB were sputum smear-negative for acid-fast bacilli, had sputum smear microscopy continued to be used as the only tool for initial screening, the outbreak of MDR TB would likely have gone undetected and many more MDR TB cases would have been imported into the United States.
Hmong refugees who had a positive TST result did not receive treatment for latent TB infection before immigrating to the United States because they were not contagious and thus were eligible to travel on commercial airplanes. No universally accepted standard therapy is available for latent TB infection with a suspected MDR TB strain. Rather, therapy for suspected latent infection from MDR TB is determined on a case-by-case basis in accordance with drug-susceptibility testing results from the putative source. After arrival in the United States, the receiving local health department jurisdictions decided how to evaluate, reevaluate, and treat patients with latent TB infection. Those who had no known contact with an MDR TB patient were treated with isoniazid; those who had had contact with an MDR TB patient, either overseas or in California, were treated with an MDR TB contact treatment regimen tailored to the source case isolate's susceptibility pattern. Most often, they were treated with fl uoroquinolone and pyrazinamide because TB isolates from Hmong refugees were resistant to isoniazid, rifampin, and ethambutol. If a patient refused medication for latent TB infection, that patient was closely clinically monitored for 2 years. This outbreak led to major changes in public health practice for this refugee group and in future pre-immigration medical screening policies. Enhancements to pre-immigration TB screening (Table 3) contributed to a reduction in the number of imported TB cases. As a result, CDC is working with the US Department of State, panel physicians, the International Organization for Migration, and other organizations to implement similar enhancements to general pre-immigration TB screening guidelines. These new technical instructions are being implemented fi rst in priority countries, as determined by immigration patterns and TB prevalence. Eventually, all refugees and immigrants entering the United States will be screened with a revised TB screening algorithm that includes mycobacterial culture and susceptibility testing. Since the end of 2007, applicants for US immigration who have been screened according to the new technical instructions have originated from Mexico, the Philippines, Nepal, and Thailand. CDC notifi es US state and local health departments when panel physicians in a specifi c country begin implementing this revised algorithm. Ongoing US national TB surveillance will help determine the effect of this effort on reducing the number of foreign-born persons with TB living in the United States. In this outbreak investigation, 3 MDR TB patients were identifi ed who had received treatment in the past 3 years at healthcare facilities outside the refugee camp; however, none reported having received DOT, the strategy recommended for reducing emergence of drug resistance (27). Our fi ndings support the goals of the World Health Organization's second Global Plan to Stop TB, which includes refugees as a high-risk group requiring attention by TB control programs. This outbreak also highlights the need for US public health preparedness efforts to focus on containment of threats of emergent diseases, such as MDR TB, at their source (28).
To control TB and prevent MDR TB, multiple organizations-including government agencies, multilateral agencies, and nongovernment organizations-must work together to provide high-quality TB diagnosis and treatment consistent with international standards of care in both host and receiving countries (29). For low-incidence countries, such as the United States, investing in global TB control is a cost-effective strategy for reducing TB, domestically and globally (30,31).