Results

Cases. The number of TB cases per year from 1990 to 1999 is shown in figure 1; 41 TB cases were reported from 1996 through October 1999. Of these, 4 cases were not linked to the outbreak: the isolate recovered from 1 person had a discordant DNA fingerprint, and 3 persons had known source cases outside of the community, but no isolates were available for fingerprinting for these 3 cases. Of the remaining 37 potential outbreak-related cases, 24 (65%) were male, 32 (86%) were black, all had been born in the United States, and the median age was 39 years. Twenty-eight cases (76%) were HIV negative, 2 were HIV positive, and 7 had unknown HIV status.

Thirty-one (84%) of the 37 potential outbreak-related cases were confirmed to be outbreak related. An outbreak-related case was associated with the outbreak by a known epidemiologic link, a DNA fingerprint matching the outbreak strain, or both. No epidemiologic link or DNA fingerprint could be established for 6 of the 37 potential outbreak-related cases. Of the 31 confirmed outbreak-related cases, 16 (43%) had an epidemiologic link and matching DNA fingerprint, 13 (35%) had epidemiologic link only and the isolates had been discarded, and 2 (6%) had a matching DNA fingerprint and no epidemiologic link. Six patients had unknown epidemiologic links and no isolate was available for DNA fingerprinting. Twenty-nine of the 31 cases were culture confirmed. All isolates of M. tuberculosis were susceptible to first-line antituberculous agents. Twenty-nine cases completed treatment, 1 person died of TB, 1 person died of a separate illness during TB treatment, and 6 cases were still receiving therapy at the time of the investigation.

At least 6 cases that were not diagnosed and treated in a timely manner. All 6 cases had abnormal findings on chest radiographs documented prior to their diagnosis of TB, but TB was not suspected until late in their illness. Two of these 6 patients were evaluated by physicians for cough and fever but were treated for community-acquired pneumonia. Both patients presented 6 months later with advanced TB and were sputum smear—positive at diagnosis. A review of patient hospital records revealed that 5 patients with suspected TB initially received inadequate antituberculous therapy. Two of these 5 patients were initially treated for latent infection with isoniazid alone, pending the outcome of laboratory results, and 3 patients were initially treated with isoniazid and rifampin only, rather than the recommended 3- or 4-drug therapy [12].

No source patient was identified who accounted for all potential outbreak-related cases. However, a 1991 case was named as a contact by several cases and potentially accounted for 22 (71%) of the known outbreak-related cases (figure 2). This case initiated treatment for active TB in 1978, but there was no isolate available for DNA fingerprint analysis and no contact investigation was conducted for this case.

Transmission and social networks. From information discovered through interviews with patients and contacts, we deduced that transmission among patients occurred in congregate social settings, including an apartment complex, a junkyard, a local bar, and a liquor store, because these were named by patients as areas of frequent social gathering. Common activities at these places included use of drugs and alcohol. One patient's apartment was a well-known place for social gatherings. The junkyard contains a ∼10 × ∼10 m building that at different times has been used as a primary dwelling and for various social purposes, including prostitution. At least 1 patient with TB listed the junkyard as a primary place of residence.

The discovery of epidemiologic links among patients was not straightforward. Only 8 adult cases initially provided names of contacts with TB to the health department. Three patients were children who had source-case investigations performed. Through retrospective questioning of cases and their contacts, we determined that 29 (94%) of the patients with known outbreak-related cases knew, either directly or indirectly, another outbreak case. Epidemiologic links for 18 cases (58%) were only discovered through complex social networks (figure 2). Although all of the 1999 cases had DNA fingerprints that matched the outbreak strain, no epidemiologic links were discovered among these cases; all but 2 of the cases that presented in 1999 had an epidemiologic link to ≥1 cases from previous years.

The illicit activity and clandestine social network of some patients presented a barrier to obtaining pertinent contact information. For example, a 16-year-old male patient who presented in 1998 (figure 2) had a DNA fingerprint that matched the outbreak strain, but he did not provide adequate contact information to determine an epidemiologic link to other outbreak cases. He denied a known exposure to TB. In 1999, a 15-year-old female patient had active TB diagnosed. Her contact investigation revealed that the aforementioned 16-year-male patient was her boyfriend and that they frequently purchased drugs from a 44-year-old patient who had TB diagnosed in 1998 and who lived at the apartment complex, which was one of the suspected areas of transmission.

In addition, the 59- and 77-year-old male 1998 cases lived in the same assisted-living complex and had known matching DNA fingerprint results but no epidemiologic links to other cases. A relative of the 77-year-old patient named a female commercial sex worker who made frequent home visits to both patients from 1996 to 1998. The commercial sex worker had TB diagnosed in 1996 and had no contact investigation performed initially. In 1998, a contact investigation was performed for the commercial sex worker, but the men were not named as contacts.

Twenty-four (77%) of the 31 known outbreak-related cases may have been prevented. For 12 patients (50%), no contact investigation was conducted for their source case, so there was no opportunity for these patients to be identified as a contact. Nine patients (38%) were not named as contacts during their putative source's contact investigation. Two patients (8%) had source cases that were not identified earlier during contact investigations, and 1 person who had renal disease and who was undergoing dialysis was immunocompromised and had a negative skin test result at the time of screening after exposure to a person with active TB. This person should have received treatment for latent TB infection (LTBI). In addition, one of the patients who had no opportunity to be identified as a contact had a prior positive TST result and did not receive preventive therapy, presumably because he was >35 years of age.

TB control and contact tracing. At the time of the outbreak, the local health department was staffed by 4 full-time and 2 part-time nurses and a nursing supervisor, all of whom were white and native to the community. Daily public health activities included a variety of public health interventions: childhood immunizations; environmental health issues, such as lead poisoning; HIV testing; and, occasionally, STD assessments. Prior to 1997, there was minimal TB control activity in the community. Contact investigations were not performed regularly, and there was no consistent administration of treatment for latent infection. Organized TB control with performance of contact investigations and DOT for patients with TB did not begin until May 1997.

Progress during contact investigations was hampered by the failure of patients to provide the names of social contacts. Patients freely revealed the names of household contacts as well as the locations of their workplaces; however, they were not forthcoming about social contacts because of the types of illicit activity occurring within their social network and their concerns regarding anonymity. In addition, the community involved in the outbreak did not realize that the public health department was a separate and distinct entity from law enforcement agencies, and they did not understand the health department's role in controlling the outbreak. As a result, many epidemiological links among patients and infected contacts were not identified because of the reluctance of patients to cooperate in naming contacts.

Records for contact investigations conducted from May 1997 to November 1999 for 18 (49%) of the 37 potential outbreak cases were available for review. There were 445 contacts identified around the 18 cases, including 247 named contacts, which consisted of family members and household contacts, and 198 persons from 2 workplace screenings. Because the patients with TB at both workplaces worked in large, open areas with little direct or close contact with other persons, the risk of transmission was low. Chest radiographs were performed on all contacts who had a positive skin test result (≥5 mm). No additional cases of TB were found during contact screenings.

A targeted skin test screening was later performed at the junkyard. The junkyard was considered to be a place with a high-risk of exposure, because many patients reported associating with persons who visited the facility frequently. A comparison of the skin test screening results for named contacts, persons at the workplaces, and persons at the junkyard is shown in table 1. The positivity rate of 55% among named contacts was higher than the rate of 3% at the workplaces (OR, 33.7; P < 0.0001). The results of the junkyard screening revealed a skin test positivity rate of 42%, which was also considerably higher than that in the workplace (OR, 20.2; P < 0.0001). The positivity rate at the junkyard was similar when compared with that of the named contacts (OR, 1.32; P = 0.562). Health department personnel were able to achieve an 88% completion rate for treatment of latent infection in the named-contacts group.