Tuberculosis Outbreaks in the United States, 2009-2015

Abstract

Objectives:

The Centers for Disease Control and Prevention provides on-site epidemiologic assistance for outbreak response when the health capacity of state, tribal, local, and territorial health departments has been exceeded. We examined recent outbreaks of tuberculosis (TB) for which health departments needed assistance.

Methods:

We defined a TB outbreak as detection of ≥3 TB cases related by transmission, as suggested by routine genotyping and epidemiologic linkages. We conducted retrospective reviews of documentation from all 21 TB outbreak investigations in the United States for which the Centers for Disease Control and Prevention provided on-site assistance during 2009-2015. We abstracted data on patients’ demographic characteristics and TB risk factors, as well as factors contributing to the outbreak from trip reports written by on-site investigators, and we compared these with outbreaks investigated during 2002-2008.

Results:

The 21 TB outbreaks during 2009-2015 involved 457 outbreak patients (range, 3-99 patients per outbreak). Of the 21 outbreaks, 16 were first identified through genotype data. In sum, 118 (26%) patients were identified through contact investigations of other patients in the outbreak. Most outbreak patients (n = 363, 79%) were US born. Ninety-two (26%) patients had a mental illness, 204 (45%) had been homeless in the year before diagnosis, and 379 (83%) used alcohol excessively or used illicit substances. The proportion of patients experiencing homelessness doubled between 2002-2008 and 2009-2015; other characteristics were similar between the 2 periods. Delayed TB diagnosis contributed to unmitigated transmission in all but 1 outbreak.

Conclusions:

TB outbreaks challenge frontline public health resources. Genotyping and contact investigations are important strategies for detecting and controlling TB outbreaks, particularly among people experiencing homelessness or those with mental illness.

Outbreaks of tuberculosis (TB) are an ongoing problem in the United States,^1–4 hindering efforts to eliminate TB.⁵ Costly, labor-intensive activities, such as active case finding and large-scale treatment, must be sustained for prolonged periods to interrupt transmission.^2,5,6 Responding to unmitigated Mycobacterium tuberculosis transmission can rapidly deplete frontline public health resources.^3,5

State, tribal, local, and territorial health departments have primary responsibility for preventing and controlling TB in their jurisdictions. When health departments determine a need for additional epidemiologic or programmatic assistance for outbreak response, the Centers for Disease Control and Prevention (CDC) might be invited to assist with on-site investigations. During these investigations, CDC personnel work closely with frontline public health partners to describe the epidemiology of the outbreak, understand the transmission chain, identify and prioritize contacts (ie, people exposed to infectious TB cases), and devise strategies and interventions to stop the outbreak.^1,2,4

Contact investigations—which are systematic efforts to identify and evaluate people exposed to infectious TB cases—are fundamental to TB control efforts in the United States.^5,6 Because these contact investigations usually depend on the ability to gather contacts’ names or other identifying information during interviews with patients diagnosed with TB, these efforts are often ineffective for addressing outbreaks in congregate settings because patients might not know any identifying information about people whom they might have exposed or they might be unwilling to name contacts (ie, contacts encountered during illegal activities).

Genotyping provides another method for public health departments to assess relatedness of TB cases. Based on the principle that TB cases in the same chain of transmission have identical or nearly identical genomes, genotyping of culture-positive TB cases helps to identify instances of recent M tuberculosis transmission. In 2004, CDC introduced the National Tuberculosis Genotyping Service, offering routine spoligotyping and 12-locus mycobacterial interspersed repetitive unit–variable number tandem repeat (MIRU-VNTR) analysis of M tuberculosis isolates on every culture-positive TB case in the United States.⁷ To improve discriminatory power among strains, routine MIRU-VNTR analysis was expanded in 2009 to include 24 loci.^8,9 Starting in 2010, TB genotype results have been linked to standard surveillance variables in the secure TB Genotyping Information Management System, which is accessible to all authorized health departments and generates automated alerts based on geospatial concentrations of matching cases for further review and potential action.^8,9 Since its introduction in 2004, routine TB genotyping has revolutionized TB outbreak detection and response in the United States because it provides another way for programs to identify chains of transmission in the absence of contact investigations.^10,11 Having routine TB genotyping results facilitates outbreak detection,¹⁰ refines outbreak case definitions,² and enables postinvestigation surveillance.¹¹ Furthermore, the TB Genotyping Information Management System offers features that allow authorized state, local, and federal public health staff members to prospectively monitor for future cases caused by outbreak strains.^8,11

A review of outbreaks for which CDC provided assistance during 2002-2008 found that most of those outbreaks involved US-born people who used illicit drugs or consumed excess alcohol,² supporting the assertion that substance abuse challenges TB control and elimination efforts.¹² Our primary objective was to describe TB outbreaks that challenged frontline public health resources during 2009-2015, since the introduction of genotyping with 24-locus MIRU-VNTR, which provides better specificity for identifying chains of TB transmission. A secondary objective was to compare our results with those of our previous report.

Methods

Using methods such as those used in our previous report,² we conducted a retrospective review of TB outbreaks in the United States for which CDC provided on-site assistance during 2009-2015. Similar to the previous report, we included only outbreaks with ≥3 TB cases related by transmission, as suggested by routine genotyping (ie, spoligotyping and 24-locus MIRU-VNTR)^8,9 and epidemiologic linkages.

We systematically abstracted data from standard reports written by CDC staff members when they returned from on-site outbreak investigations. All reports included aggregate data on patients’ demographic characteristics, TB risk factors, and clinical outcomes, such as hospitalization or death. We also developed discrete categories based on recurring themes for key factors considered to be fueling the outbreak. We considered a jurisdiction to have a low incidence of TB if the number of TB cases based on population size was lower than the overall national TB incidence that year. We defined mental illness as an Axis I disorder other than a substance use disorder,¹³ as documented in a patient’s medical record. We replicated other definitions from those used in the previous report.² We defined a household as a residential location not known to be associated with illicit drug use; we defined a drug house as any venue characterized by the sale or use of illicit drugs. We estimated infectious periods according to national guidelines,⁶ and we defined prolonged infectiousness as >3 months between symptom onset and the date that effective treatment had been administered for 2 weeks.² We considered delayed diagnosis to be >2 months between symptom onset and the date that the patient began TB treatment, and we discriminated between delays in seeking medical attention for TB symptoms and delays in diagnosing TB after medical attention was sought.^2,3

Contact investigation is the systematic process through which people exposed during the infectious period of a TB patient are identified, located, and evaluated for TB infection and disease. Treating recently infected contacts is an established strategy to prevent additional TB cases.^5,6 As in the previous report,² we quantified the frequency of incomplete contact investigations; however, for this report we also determined the primary challenges in completing the contact investigations. We classified contact investigation strategies into 3 categories. The first category was the traditional name-based approach primarily driven by patient interview.⁶ The second was a location-based approach that was still name driven but based on third-party rosters, such as jail, prison, or shelter logs, to elicit a list of contacts present during infectious periods. The third category was a mass screening strategy that sometimes had to be implemented in the absence of an effective interview or available rosters. Mass screening attempts to target people present during infectious periods but might include people who were not actually contacts. In the previous report, this mass screening approach was described as location-based screening ²; however, in this report, we preserved the term location based for a contact investigation approach that was still name driven. Finally, we summarized the CDC team’s key recommendations for changes to the outbreak response strategy.

Two authors independently abstracted data from the same outbreak report and then together compared the 2 data abstractions, referring to the trip report to resolve discrepancies. We also referred to any subsequent presentations and publications about the outbreaks,^3,14–16 and we occasionally consulted with the original investigators to clarify any ambiguities. We then compared the 2009-2015 outbreaks with those investigated during 2002-2008, by the proportions of patients in each outbreak period with certain demographic, clinical, and social characteristics and the distribution of factors contributing to the outbreaks.² The data collected during outbreak investigations were considered public health practice. As such, CDC determined that this analysis did not constitute research involving human subjects and was not subject to institutional review board review.

Results

CDC provided on-site epidemiologic assistance for 26 TB investigations during 2009-2015. Of these, 2 occurred outside the United States, and 3 did not have ≥3 culture-positive cases; therefore, 21 TB investigations met the inclusion criteria for this report. Of the 21 outbreaks, 15 occurred in jurisdictions with low TB incidence; 5 involved American Indians/Alaska Natives; and 3 were characterized by TB strains resistant to isoniazid, a first-line TB medication. The subsequent investigations of these 21 outbreaks identified a total of 457 TB patients (median, 9 patients per outbreak; range, 3-99 patients per outbreak) and 40 229 contacts (median, 402 contacts per outbreak; range, 73-17 954 contacts per outbreak). Sixteen outbreaks were first identified according to genotype data; the remaining 5 were first detected locally, on the basis of there being more cases than what the health department expected to see in a given population group.

Characteristics of Patients in TB Outbreaks

Of the 457 patients in the 21 outbreaks during 2009-2015, 363 (79%) were US born, 365 (80%) were male, and 51 (11%) were aged <15. With the exception of homelessness, the prevalence of known risk factors for TB was similar to that reported for 2002-2008 (Table 1). Since the previous report, the prevalence of homelessness doubled; among 457 patients diagnosed during the 2009-2015 outbreaks, 204 (45%) had recently been homeless (ie, in the year leading up to TB diagnosis). In the 8 outbreaks during 2009-2015 in which the mental health history of patients was systematically ascertained, the prevalence of mental illness was 26% (92 of 351). Tobacco use (202 of 457, 44%) was common across these outbreaks.

Table 1.

Demographic, clinical, and social characteristics of patients in TB outbreaks for which the Centers for Disease Control and Prevention provided on-site assistance, United States, 2002-2015

	No. of Cases (%)
Characteristics	2009- 2015	2002- 2008a
Total	457 (100)	398 (100)
Demographic
US born	363 (79)	364 (91)
Foreign born	94 (21)	30 (9)
Male	365 (80)	259 (65)
Female	92 (20)	136 (35)
Age <15 y	51 (11)	50 (13)
Age ≥15 y	406 (89)	348 (87)
Race/ethnicity
Non-Hispanic black	202 (44)	265 (67)
Non-Hispanic white	73 (16)	66 (17)
Hispanic	42 (9)	31 (8)
American Indian/Alaska Native	40 (9)	0 (0)
Asian American/Pacific Islander	51 (11)	0 (0)
Other or unknown	49 (11)	36 (9)
Medical risk factorsb
HIV infection	69 (17)	46 (12)
Diabetes mellitus	42 (10)	23 (6)
Previous TB disease	14 (4)	16 (4)
Other immunocompromising conditions	27 (6)	14 (4)
Mental illnessc	92 (26)	—d
Social risk factors
Tobacco use at time of outbreak	201 (44)	—d
Any substance use within past year	379 (83)	233 (59)
Excess alcohol use	180 (47)	204 (51)
Noninjection drug use	95 (25)	117 (29)
Injection drug use	16 (4)	19 (5)
Any history of incarceration	135 (30)	126 (32)
Homelessness within past year	204 (45)	78 (20)
Method of case detection (reason evaluated)
TB symptoms	277 (60)	—d
Contact investigation	118 (26)	—d
Other	62 (14)	—d
Clinical symptoms and outcome
Pulmonary site of disease	408 (89)	333 (84)
Cavitary on chest radiographe	132 (32)	122 (37)
Acid-fast bacilli sputum smear   positivee	230 (56)	204 (61)
Hospitalized	113 (25)	99 (25)
Died	41 (9)	23 (6)

Abbreviations: HIV, human immunodeficiency virus; TB, tuberculosis.

^aData for 2002-2008 were previously reported.²

^bPercentages for some characteristics were based on the subset of patients with this variable systematically ascertained. Medical risk factors with complete data for 2009-2015 outbreaks: HIV results for 411 patients from 18 outbreaks, diabetes status for 410 patients from 15 outbreaks, and TB history for 368 patients from 13 outbreaks. The total number of patients with any substance abuse in 2009-2015 includes some for whom the particular substance was not ascertained in the investigation trip report (ie, some do not appear in the subsequent 3 rows).

^cDefined as an Axis I disorder other than a substance use disorder, as documented in a patient’s medical record. Percentage based on the 351 patients in 8 outbreaks for whom a history of mental illness was systematically ascertained during the 2009-2015 outbreak investigations.

^dData concerning these characteristics were not collected for the 2002-2008 outbreaks.

^ePercentages based on the proportion of 333 people with pulmonary disease diagnosed during the 2002-2008 outbreaks and 408 diagnosed during the 2009-2015 outbreaks.

A total of 277 (60%) TB diagnoses in the 2009-2015 outbreaks were made after patients sought medical attention for TB symptoms, whereas 118 (26%) were identified through active case finding associated with contact investigations. A further 113 (25%) patients were hospitalized, and 41 (10%) died during treatment.

Factors Contributing to TB Outbreaks

Delayed TB diagnoses were cited as contributing factors to all but 1 of the 21 outbreaks; these delays were more pronounced during 2009-2015 than during 2002-2008 (Table 2). In 16 of the 21 outbreaks investigated during 2009-2015, at least 1 patient did not seek care within 2 months of symptom onset. In 14 outbreaks (categories not mutually exclusive), at least 1 patient sought medical care within 2 months of symptom onset but remained contagious because of delays in diagnosis once under medical care. Nonadherence to treatment after TB diagnoses contributed to prolonged infectiousness in 7 of the 21 outbreaks during 2009-2015, compared with 5 of the 27 outbreaks during 2002-2008 (Table 2).

Table 2.

Factors contributing to TB outbreaks for which the Centers for Disease Control and Prevention provided on-site assistance, United States, 2002-2015

	No. of Outbreaksa
Characteristics	2009-2015	2002-2008b
Total	21	27
Delays in TB diagnosisc	20	12
Patient delay in seeking care for symptoms	16	6
Delay in diagnosis once under medical care	14	12
Nonadherence to treatment	7	5
Congregation of high-risk populationsd	12	7
Overnight homeless facilities	8	—e
Local jails	8	—e
Prisons	2	—e
Residence for people with mental disability	2	—e
Health care facilities	2	—e
Otherf	7	—e
Incomplete contact investigations	13	10
Contacts not identified	7	—e
Contact investigation not expanded	6	—e
Contacts identified but not evaluated	8	—e
Contacts evaluated but infection not treated	3	—e

Abbreviation: TB, tuberculosis.

^aCategories not mutually exclusive.

^bData for 2002-2008 were previously reported.²

^cData concerning delays in TB diagnoses considered to be >2 months between symptom onset and the date that the patient began TB treatment.

^dNumbers may exceed the number of outbreaks because several outbreaks involved >1 congregate setting.

^eData concerning these characteristics were not collected for the 2002-2008 outbreaks.

^fOther facilities included drug houses (n = 4 outbreaks), workplaces (n = 2 outbreaks), and a school.

Transmission venues in these outbreaks included households (n = 9 outbreaks), overnight homeless facilities (n = 8), local jails (n = 8), drug houses (n = 4), residential facilities for people with mental disability (n = 2), health care facilities (n = 2), workplaces (n = 2), a federal prison (n = 1), a state prison (n = 1), and a school (n = 1). Of the 9 outbreaks characterized by household transmission, lack of stable housing led to M tuberculosis transmission across households in 5 outbreaks; in each of these outbreaks, a person with undiagnosed TB disease transmitted M tuberculosis to people in multiple households. In 2 of the remaining 4 households, regular congregation in a private home for the purpose of alcohol consumption was also considered a key factor in fueling the outbreak.

Incomplete contact investigations were also cited as a challenge. In 7 of the 21 outbreaks during 2009-2015, the list of initially identified contacts did not name recently exposed vulnerable contacts. In 6 outbreaks, the initial list of people named as contacts lacked important locating information (eg, names, addresses). In 8 outbreaks, contacts had been identified but not yet medically evaluated for TB disease or infection. In 3 outbreaks, contacts had been identified and evaluated, but treatment for latent TB infection had not yet been offered at the time of CDC’s on-site involvement in the investigations.

Key Recommendations to Stop the Outbreaks

During our reviews of the trip reports, we examined the CDC investigators’ recommendations to interrupt transmission in these outbreaks. For all but 1 of the 21 outbreaks, the CDC team recommended expanding the contact investigations to identify additional contacts to evaluate for TB disease or infection. For 16 outbreaks, the recommendations also cited a need to prioritize the highest-risk contacts for TB evaluation, rather than continuing to try to simultaneously find and evaluate all contacts. For 8 outbreaks, the CDC team recommended and facilitated a location-based contact investigation approach using existing rosters from congregate settings to identify contacts who were present during infectious periods.

In 9 outbreaks during 2009-2015, investigators recommended improved data management, often with the designation of data manager at the health department to help manage and prioritize the hundreds or thousands of identified contacts. This data management need was not for managing the cases of patients with TB disease; rather, it was for tracking patients’ overlapping and often large numbers of contacts to ensure that the highest-priority contacts were located, evaluated, and treated. Other recommendations included identifying someone to coordinate all outbreak response activities (n = 8 outbreaks).

Because lack of infection control in congregate settings with vulnerable populations (eg, emergency overnight homeless facilities, correctional facilities, mental health facilities) was cited as a contributing factor in 11 outbreaks during 2009-2015, investigators recommended administrative interventions designed to identify people with signs or symptoms of TB disease before or soon after entry. Examples of administrative measures included enforcing routine TB screenings at intake and monitoring for people who are coughing to ensure that these symptomatic people with TB are placed in respiratory isolation and receive diagnostic evaluations for TB and treatment if diagnosed.^17,18

Certain recommendations were documented in all 21 trip reports. One universal recommendation was that local clinicians be informed about recent transmission in their community and receive education about the signs and symptoms of TB disease, the components of an appropriate diagnostic evaluation, and the need to notify public health authorities about any suspected TB diagnoses. To address the social stigma often associated with TB and to educate the broader public about the role of the health department in controlling TB, speaking openly about the outbreak was a key recommendation for every outbreak. However, few frontline health departments desired outside attention to their challenges in responding to these outbreaks. For 2 outbreaks, the health departments proactively released press summaries and hosted news conferences.^3,19 An additional 7 outbreaks received unsolicited media attention. The remaining 12 outbreaks received no media attention.

Discussion

Despite a 29% decline in the overall national TB burden between 2002-2008 and 2009-2015,²⁰ the cumulative number of TB cases in the 21 TB outbreaks for which CDC provided on-site investigation assistance during the more recent 7-year period was higher than that during the previous study period, and the associated number of contacts was also substantially higher. Patients in these more recent outbreaks also had known risk factors, including a high prevalence of mental illness and a doubling in the prevalence of homelessness. In light of such challenges, TB outbreaks can easily overwhelm existing frontline public health resources to detect, investigate, and manage not only the outbreak cases but also the exposed contacts.

Of the 21 TB outbreaks, 15 occurred in low-incidence jurisdictions. Because of the persistent threat that M tuberculosis transmission poses to state, tribal, local, and territorial health departments, all jurisdictions must retain the ability to perform the essential components of TB control, including expertise and resources to identify cases, investigate transmission to contacts, and prevent progression to TB disease among infected people.^5,21,22 Maintaining such capacity will become increasingly difficult, however, as the incidence of TB is reduced.^21,23

The cornerstone of TB control is to identify cases as early as possible during the course of illness to promptly facilitate curative treatment for the patient and implementation of infection control procedures for the broader community.⁵ In all but 1 outbreak, however, delayed TB diagnoses contributed to unmitigated transmission. Patients had infectious TB for extended periods, exposing large numbers of contacts. These diagnostic delays occurred for 2 reasons: delays in seeking medical care for TB-related symptoms and delays in the TB diagnosis after the patients first presented for medical care. Both types of delays were more pronounced in this outbreak report than in the previous report. The large proportion of patients experiencing homelessness and its associated poverty and unfavorable social circumstances suggests that a lack of routine access to health care plays a role in TB outbreaks. In addition, because TB incidence in the United States is 3 cases per 100 000 people annually and primarily involves foreign-born populations,¹⁷ clinicians might not readily consider TB when evaluating patients with signs and symptoms compatible with TB,^5,21 especially US-born patients (ie, most patients described in this report).

Most TB outbreaks in our study involved congregate facilities that served vulnerable populations. Optimization of recommended infection control procedures and TB control strategies in congregate settings (eg, emergency overnight homeless facilities,¹⁷ correctional facilities¹⁸) is essential to prevent outbreaks. Another fundamental TB control activity is contact investigation, which focuses on identifying and evaluating people who had contact with a contagious patient.^5,6 Transient populations, such as people living in congregate settings, have a higher risk of being exposed to M tuberculosis, but traditional name-based contact investigations⁶ depend on patients identifying their contacts, which might not be feasible in congregate settings with large numbers of unrelated people. Nevertheless, most TB outbreaks described in this report involved congregate facilities whose administrative records could be examined to generate a list of contacts present during infectious periods. Where feasible, case-finding efforts during TB outbreaks should prioritize the identification of people with known exposures in congregate settings, rather than rely on mass screenings, to target those at greatest risk. Mass screenings strain limited resources, can attract people with little or no recent TB exposure, and often present challenges in data management and interpretation.

The prevalence of mental illness, potentially an underrecognized factor in TB outbreaks,¹⁴ was 26% in the subset of outbreaks in which mental health history was ascertained. Similar to TB, mental illness is associated with homelessness and substance use. Homeless facilities, local jails, and group homes have a disproportionate share of mentally ill people as compared with the rest of the community,²⁴ placing many people with mental illness at greater risk of exposure to M tuberculosis than that of their community counterparts without mental illness. Their impairments might also render them less able to articulate their symptoms or seek medical attention for symptoms, leading to delayed TB diagnosis.

Genotype clustering helped identify most of the outbreaks in our study, suggesting a shift toward better detection of TB outbreaks in transient populations. Without genotyping, links among patients in our study might have gone undetected. Outbreak detection via genotyping might explain the apparently more prominent role of homelessness with the outbreaks during 2009-2015 than among the outbreaks during 2002-2008: rather than an ecologic trend of more transmission among homeless people, our outbreak detection tools were more sensitive for homeless populations. Awareness by frontline public health staff members that patients in their jurisdiction had matching TB genotypes could have helped guide reinterviews and ultimately led to a better understanding of transmission links among TB patients, particularly in situations complicated by homelessness, incarceration, or mental illness.

Limitations

This study had several limitations. Our description of outbreaks for which epidemiologic assistance from CDC was provided is not generalizable to all TB outbreaks. Most outbreaks are handled locally. Other outbreaks that challenged state, tribal, local, and territorial TB program resources were managed with remote consultation or other targeted supplemental resources; requests for on-site assistance from CDC epidemiologists are relatively rare. Because there is no national surveillance system to track all TB outbreaks, we were unable to assess the representativeness of the 21 outbreaks in this report. Nonetheless, efforts to understand characteristics of these outbreaks might inform targeted efforts to stop transmission in these populations before outbreaks occur. Because most outbreaks included in this report involved congregate settings, we relied on less traditional contact investigation methods (ie, a location-based approach), which may have led to more identified contacts than what is typical. Nonetheless, documentation of these difficult outbreaks and the affected populations might not otherwise have occurred; the health departments that invited CDC to assist often did so on the condition that their experiences not be shared with the broader public health community, except through an aggregate review such as this study. Whereas 17 of the 27 outbreaks described in the 2002-2008 report resulted in a publication,² only 6 of the 21 outbreaks described in this report have resulted in a publication^3,14–16 or have reports under development.

Conclusion

TB outbreaks continue to challenge frontline public health resources, but genotyping has improved outbreak detection, particularly among people who are homeless or mentally ill. However, the number of exposed contacts in TB outbreaks can be high. Even as health departments must retain the capacity to execute the essential components of TB control, developing new and creative strategies to address the challenge of TB control in vulnerable populations is essential.