Public Health Response to Tuberculosis Outbreak Among Persons Experiencing Homelessness, Minneapolis, Minnesota, USA, 2017–2018

Kelzee K. Tibbetts; Randy A. Ottoson; Dean T. Tsukayama


Emerging Infectious Diseases. 2020;26(3):420-426. 

In This Article

Abstract and Introduction


Tuberculosis (TB) is a greater risk for populations experiencing homelessness. When a TB exposure occurs in a homeless shelter, evaluation of contacts is both urgent and challenging. In 2017, local public health workers initiated a response to a TB outbreak in homeless shelters in Minneapolis, Minnesota, USA. In this contact investigation, we incorporated multiple techniques to identify, evaluate, and manage patients, including the concentric-circle method to characterize amount of contact, identifying the most frequent sites of sporadic medical care, using electronic medical records, and engaging with medical providers treating this population. Of 298 contacts evaluated, 41 (14%) had latent TB infection and 2 had active TB disease. Our analysis indicated a significant relationship between duration of exposure and positive TB test result (p = 0.001). We encourage local public health departments to expand beyond traditional contact tracing techniques by leveraging partnerships and existing systems to reach contacts exposed in shelters.


In 2017, a total of 9,093 new cases of active tuberculosis (TB) were reported in the United States, and ≈4.5% of these occurred in persons experiencing homelessness (PEH) in the year preceding their diagnosis.[1] The incidence of TB in PEH populations is >10 times that of the general population (i.e., 36–47 vs. 2.8 cases/100,000 population during 2006–2010)[2] because risk factors, such as HIV infection, mental illness, substance abuse, and barriers to accessing healthcare, put them at higher risk. In addition, PEH often use homeless shelters and congregate in environments where the risk for TB transmission is greatly increased.[3]

A priority for TB control and prevention is the screening of persons exposed to transmissible TB.[4] Locating and fully evaluating contacts, essential components of a contact investigation, is difficult but especially urgent for controlling TB among PEH populations. Homelessness at the time of diagnosis indicates the need for a prompt contact investigation; however, guidance or consensus on how to identify contacts in these situations is lacking. Interviews of PEH persons with active TB are not always reliable sources for contact information. An analysis of ≈3,000 PEH and non-PEH TB patients in New York, New York, USA, demonstrated that experiencing homelessness in the year before diagnosis predicted the likelihood of that person's contacts not being identified during interviews.[5] Shelter rosters can also be incomplete and unreliable, often changing over the course of a single night.

Guidelines for contact prioritization in congregate settings have been established but are impractical to apply to large, complex populations because priority is assigned primarily by likelihood of infection and progression to active disease, which is difficult to determine in these settings.[6] Mass screenings for latent TB infection (LTBI) eliminate the need for contact prioritization but are resource intensive and decrease the public health value of positive test results. In the absence of readily available information about a person's susceptibility, duration of exposure to an infectious person has been used to prioritize contacts. Although this benchmark for exposure is linked to risk for transmission,[7,8] the measure is not well defined.

In 2017, a total of 70 cases of TB were diagnosed in Hennepin County, the largest county in Minnesota, USA (population of 1.3 million). This area has experienced an ongoing cluster of genotypically linked TB cases among long-term PEH. The cluster, confirmed by whole-genome sequencing, could have been circulating in Hennepin County as early as 1992[9] but was not identified until the Centers for Disease Control and Prevention (CDC) began providing TB genotyping services to state health departments in 2004. During 2008–2016, a total of 18 cases of this genotype cluster were identified, resulting in multiple complicated contact investigations (Figure 1, panel A).

Figure 1.

Whole-genome sequencing map of Hennepin County tuberculosis (TB) case cluster, Minneapolis, Minnesota, USA, including cases identified in Texas, USA, in 2016. A) 2008–2016 case cluster, which included 18 cases; B) updated 2008–2018 case cluster, totaling 24 cases. Isolates with the same genome sequence are displayed together in 1 node. Nodes are connected by lines proportional in length to the number of single-nucleotide polymorphism differences between isolates (n = 1, for all). No epidemiologic link to Minnesota was identified for the cases in Texas. Node A contains 10 cases diagnosed during 2008–2015 and the most recent common ancestor reference point. In panel A, node B contains 4 cases diagnosed during 2014–2016, and in panel B, node B contains 8 cases diagnosed during 2014–2018.

In 2016, Hennepin County Public Health (HCPH) was notified of 3 newly identified active TB cases matching the cluster previously found in its PEH population. The most infectious person had experienced coughing, night sweats, and weight loss, and testing revealed acid-fast bacilli (AFB)–positive sputum samples and a cavitary lesion on chest radiograph. The contact investigation for this person identified 180 shelter residents at high risk for infection in need of testing, 85% of whom were evaluated.

In 2017, the HCPH was notified of another set of 3 newly identified active TB cases, which prompted another contact investigation. In this report, we review our experience conducting this contact investigation to prevent and control for a potential TB outbreak in Minneapolis homeless shelters during 2017–2018. We incorporated several methods that are not well described in the literature to identify, evaluate, and treat contacts. We specifically examined whether shelter rosters can be used to meaningfully quantify TB risk when multiple persons with TB have overlapping exposure periods and how outreach and partner strategies could be used to identify persons in need of TB testing.