We conducted a retrospective review of TB outbreak investigations in the United States for which CDC provided onsite assistance during 2002–2008. Included in the review were outbreaks having documented evidence of Mycobacterium tuberculosis transmission with >3 culture-confirmed TB cases linked by both genotype and epidemiology. Genotyping methods included spoligotyping (all years) and either restriction fragment length polymorphism (2002) or 12-locus mycobacterial interspersed repetitive units (MIRU) (2003–2008). Linkage by epidemiology meant known exposure to another outbreak patient by sharing enclosed airspace in the same period. Linkage by genotype required matching results by whichever genotyping methods were used for that outbreak (generally by spoligotype and either restriction fragment length polymorphism or 12-locus MIRU).

We systematically abstracted data on case characteristics, outbreak contributing factors, transmission sites, and interventions (including contact investigation) from reports written by CDC teams upon return from outbreak investigations. All reports were uniformly written (with background, methods, results, discussion, and conclusion sections) and included aggregate data on demographic, clinical, and social risk factors of cases, epidemiologic linkages, genotyping results, and contact investigations. These data covered the period of the onsite investigation and were either provided by the health departments or collected by CDC as part of the investigation. The reports also discussed the cause of the outbreak and were sent back to the respective local and state health departments with recommendations to further assist in achieving control of the outbreak. We referred to any subsequent presentations and publications describing the investigation (3–19) to check data quality and accuracy of outbreak contributing factors and transmission sites. For any discrepancy in number of outbreak cases, we deferred to the CDC reports to ensure consistency in the period of data collection across all outbreaks included in the review.

We defined outbreaks based on CDC guidelines for contact investigation as detection of TB disease among >2 persons exposed to a person with infectious TB (20), i.e., >3 linked cases within 2 years. The standard National TB Surveillance System variables were abstracted for all patients (21). Demographic, medical, and social characteristics were ascribed to an outbreak if >50% of the patients involved in the outbreak had that particular characteristic. For outbreak contributing factors, hotspots, and interventions, discrete categories were created based on recurring themes in the reports to enable systematic abstraction and quantification of these variables. Because multiple outbreak contributing factors and a large number of interventions were documented in CDC reports for each outbreak, we abstracted 2 factors considered by the onsite investigators to be key to fueling the outbreak and 3 interventions not already being used that were either used by the CDC team onsite or recommended to be pivotal to the overall control effort. The frequency that each category was encountered was totaled to quantify these outbreak variables. Whenever feasible, a member of the original team investigating the outbreak was consulted to review the accuracy of abstracted data.

Prolonged infectious period was defined as >3 months between symptom onset and the date that effective treatment had been administered for 2 weeks (20). Delayed diagnosis was defined as >2 months between symptom onset and date that the patient first sought care for TB symptoms or that TB treatment was initiated (22,23). Incomplete contact investigations meant inability to either locate or complete evaluation of contacts because of limited resources or a hard-to-reach population. Hotspots were defined as transmission sites where >2 outbreak patients had spent substantial time together, as determined by local public health investigators. Drug house was defined as a venue characterized by the sale or use of illicit drugs. Household was defined as a residential location without documented illicit drug use.

The 27 outbreaks included in the review involved 398 patients (median 10 patients/outbreak, range 3–35 patients). Of these patients, 364 (91%) were US-born and 50 (13%) were <15 years of age (Table 1). Three hundred thirty-three (84%) had pulmonary disease, including 204 (61%) and 284 (85%) with smear-positive and culture-positive TB, respectively. Eighty-nine percent of isolates (253 of 284) were susceptible to first-line TB medications (Table 1). Of the 197 patients for whom the reason for initial TB evaluation was documented, 74 (38%) were evaluated because of TB symptoms, and 57 (30%) were detected as part of a contact investigation. Ninety-nine (25%) patients required hospitalization, and 23 (6%) died. Infectious periods were documented for 172 patients; the mean and median infectious periods were 6 and 5 months, respectively (range 1–56 months).

Most patients did not have established medical risk factors for TB (Table 2). Although the total number of patients tested for HIV was not available, HIV infection was documented for 45 patients (12%). Multiple social risk factors for TB were documented: 233 (58%) patients reported alcohol abuse or use of illicit drugs, 126 (32%) had a history of incarceration, and 78 (20%) had a history of homelessness. Sixteen (4%) patients were documented to have a previous diagnosis of TB; of these, 7 (44%) had received incomplete TB treatment.

Spoligotype and MIRU genotype data were available for 22 of 27 outbreaks. Three outbreaks involved a Beijing strain (spoligotype 000000000003771, MIRU 223325173533); the remaining 19 had genotypes that differed from each other. The most frequent strain lineage in these outbreaks was EuroAmerican.

US birth and substance abuse were predominant features of outbreaks (Table 3). On the basis of >50% of outbreak patients having a particular characteristic, the criterion used to ascribe a characteristic to an outbreak, 24 of the total 27 outbreaks were characterized by US-born persons, and 18 outbreaks by patients with reported substance abuse. Fourteen (52%) outbreaks were characterized by US-born men who used alcohol to excess or illicit drugs, i.e., marijuana, cocaine, methamphetamine, or heroin. All 8 outbreaks characterized by patients with incarceration histories and the 4 characterized by homelessness had a predominance of patients with reported substance abuse.

Of the 24 outbreaks that occurred among predominantly US-born persons, 21 outbreaks involved substance abuse. For 17 (71%) outbreaks, >50% of patients reported substance abuse; for 2 additional outbreaks, >40% of patients reported substance abuse; and in 2 others, the source patients who reported substance abuse had prolonged infectious periods during which TB was transmitted. In all, 21 (87%) of 24 outbreaks of predominantly US-born patients could thus be characterized as being related to substance abuse. The remaining 3 outbreaks associated with US-born persons were characterized by delayed diagnosis that resulted in transmission in a health care facility (6), among family members of a recently incarcerated patient (7), and among family members of an undocumented worker (17).

Three of the 27 outbreaks occurred among predominantly foreign-born persons. In 1 of these outbreaks, all patients engaged in substance abuse, and although these patients were foreign-born, they had been in the United States for more than a decade (18). The other 2 outbreaks among foreign-born persons involved transmission in school, church, and household settings. In each outbreak, the foreign-born patients did not access health care (caused by, in 1 outbreak, fear of repercussions for being undocumented, resulting in multidrug-resistant TB transmission among family members [17]).

Table 4 describes the most common outbreak contributing factors, which for 24 outbreaks was prolonged infectiousness. In 4 outbreaks in which patients delayed seeking medical attention for their TB symptoms, in 7 where provider-related diagnostic delays occurred, and in 1 where both types of delay occurred, >40% of patients had reported substance abuse. Incomplete contact investigations because of limited resources or a hard-to-reach population contributed to 10 outbreaks.

Drug house was the most commonly identified hotspot (17 outbreaks), followed by homeless shelter (n = 5), correctional facility (n = 4), household (n = 4), workplace (n = 4), church (n = 3), bar (n = 2), school (n = 1), and automobile (n = 1). Ten of the 17 drug houses were primarily residences with extended family members, and the other 7 were largely venues where unrelated persons gathered strictly for the use of illicit substances. In the 4 household outbreaks, crowded living conditions among extended families (primarily foreign-born in 2 outbreaks) were the main TB risk factors.

The most frequently documented intervention to control the outbreak was to prioritize contacts based on risk for infection and progression to disease to ensure that the highest risk contacts were completely evaluated (14 outbreaks). This intervention was necessitated by the large number of contacts identified; contact investigation of 398 patients had generated 16,559 contacts. Of these contacts, 10,142 (61%) had been evaluated by the time of the onsite investigation; 2,128 (21%) were found to have latent TB infection (range 4%–65% per outbreak). Other frequently used interventions included educating community health care providers, e.g., emergency departments, to be vigilant for TB in patients seeking treatment at their facilities (13 outbreaks), and location-based screening, which involves offering TB screening to potential contacts at that particular outbreak’s hotspot or other convenient location (10 outbreaks).