Participants were recruited in Tijuana, Mexico, during April 2007–July 2007, by street outreach, targeted advertising, and word-of-mouth. Persons were eligible to participate if they were >18 years of age, provided informed consent, planned on staying in Tijuana for the next 30 days, and reported at least 1 of the following characteristics in the 6 months before enrollment: use of noninjected illicit drugs other than marijuana, injection drug use, receipt of money/goods in exchange for sex, and homelessness or unstable housing. Unstable housing was defined as living primarily in a rented hotel room, migrant work camp, or medical/drug treatment facility.

Participant interviews were conducted and biological samples were obtained at PrevenCasa, a community-based harm-reduction and research facility located in the Zona Norte neighborhood of Tijuana, which abuts the commercial sex trade district. Computer-assisted personal interviewing technology (QDS; Nova Research Company, Bethesda, MD, USA) was used to facilitate participant interviews. Interviews were conducted in Spanish by trained Mexican interviewers experienced with the specific populations. Questions included sociodemographic characteristics, putative risk factors for TB and HIV infection, TB knowledge and exposure history, and presence of TB-related symptoms. The survey instrument was developed in English, translated into Spanish, and then back-translated into English to verify accuracy and meaning. Monetary reimbursement of US $20 was offered to participants to compensate them for time and transportation.

During the computer-assisted personal interview, persons who reported a persistent cough for >3 weeks and an additional symptom indicative of TB (fever or chills, night sweats, swollen lymph nodes, hoarseness, shortness of breath, joint pain, fatigue, and unexplained weight loss) or hemoptysis with or without other symptoms, were considered suspected active TB case-patients. After the interview, these persons were asked to provide 3 sputum samples for acid-fast bacilli (AFB) smears; the first of these samples was collected immediately. The remaining 2 samples were collected on subsequent days. These participants were also transported to a nearby radiologic center for chest radiography to identify evidence of pathologic changes consistent with TB. Incentives of US $5 were given for each additional sputum collection visit and upon completion of a chest radiograph, as recommended by FitzGerald et al. (19). Participants with >1 AFB-positive sputum smear or chest radiography findings consistent with TB were determined likely to have active TB and were referred to the central public health clinic (Instituto de Servicios de Salud Pública del Estado de Baja California [ISESALUD]) for clinical confirmation and treatment through the national TB program. The study protocol was reviewed and approved by ethics committees at the University of California, San Diego, and the Tijuana General Hospital.

M. tuberculosis infection was detected by using an IGRA (QuantiFERON TB Gold In-Tube [QFT] assay; Cellestis Ltd., Carnegie, Victoria, Australia), an in vitro assay that uses an ELISA to detect interferon-γ released by whole blood samples after introduction of 6-kDa early secretory antigenic target protein, culture filtrate protein 10, and TB7.7 protein, which mimic antigens specific to the M. tuberculosis complex present in patients with TB and LTBI (20). In 2005, the US Food and Drug Administration (Silver Spring, MD, USA) approved the QFT assay for detection of M. tuberculosis infection, and the US Centers for Disease Control and Prevention (Atlanta, GA, USA) determined that this assay may be used to detect M. tuberculosis in all situations in which the TST is used (17). Antibodies against HIV were detected by using the Determine Rapid HIV Antibody Test (Abbott Laboratories, Boston, MA, USA) on site. All positive test results were confirmed at the San Diego County Public Health Laboratory by using an ELISA and an immunofluorescent antibody assay.

After data collection, all persons who did not report symptoms consistent with active TB were given an appointment 4 weeks later to receive the results of the HIV test and QFT assay. At this visit, those who were HIV positive were referred to ISESALUD for medical evaluation, and those who were IGRA positive were offered chest radiography when they returned to obtain results that ruled out active TB. Persons who had active TB by AFB smear or radiography were referred to ISESALUD. Persons who were IGRA positive, but had an unremarkable chest radiograph, were determined to have LTBI.

Descriptive statistics were used to characterize participants by IGRA status. Associations between participant characteristics and TB infection were evaluated by using the Wilcoxon rank-sum test or Fisher exact test, followed by univariate and multivariate logistic regressions. Variables with p values <0.10 in univariate models were considered for inclusion in multivariate analysis. A backward stepwise model that selected for main effects in the final model was used. At each step, likelihood ratio testing was used to compare nested models until only variables with p values <0.05 remained in the final model.

A total of 527 persons were recruited, of whom 503 met eligibility criteria and had complete HIV and IGRA results. Overall, 57% were positive for TB infection by QFT assay (IGRA+), 4.2% were HIV+ by HIV test, and 2.2% were positive by both tests. Fifty-nine percent of participants were men, median age was 36 years (interquartile range 29–42 years), and median length of time participants had lived in Tijuana was 6 years (interquartile range 1.0–13.3 years). On the basis of nonmutually exclusive groupings, there were 232 (46%) IDUs, 311 (62%) NIDUs, 115 (23%) FSWs, and 280 (56%) homeless persons. The prevalence of IGRA+ results was 63% among IDUs, 58% among NIDUs, 49% among FSWs, and 52% among homeless participants. A total of 14 (2.8%) participants reported symptoms suggestive of active TB, of which 2 persons were diagnosed with active TB by AFB+ smears (prevalence 398/100,000). Because chest radiography was used after recruitment began and was offered at the results visit, only 79 of the 286 IGRA+ participants had chest radiography, of which 8 (10%) had signs of current or past pulmonary TB.

Univariate analysis identified several factors associated with TB infection (Table 1). When compared with IGRA– participants, IGRA+ participants were significantly older (median age 38 vs. 34 years; p<0.01), had lived in Tijuana longer (median 9.0 vs. 3.5 years; p<0.01), were more likely to be men (64.7% vs. 51.6%; p<0.05), had stable sleeping arrangements in the 6 months before data collection (49.3% vs. 37.8%; p<0.05), and had known someone with TB (46.2% vs. 35.9%; p<0.05). When compared with those not in each risk group, IDUs were more likely to be IGRA+, and FSWs and homeless participants were less likely to be IGRA+; no association was found for NIDUs. History of incarceration was associated with TB infection. Persons who had been incarcerated in Mexico were more likely to be IGRA+ (68%) than those incarcerated in the United States (55%) or both countries (62%). To maximize statistical power, in subsequent analyses we combined incarceration in Mexico and the United States into 1 incarceration category. When compared with IGRA– participants, IGRA+ participants were no more likely to report sexually transmitted infections (2.5% vs. 2.3%; p = 0.91) or be HIV+ (3.8% vs. 4.6%; p = 0.67). TB infection was not associated with any other drug administration practice or sexual behavior examined in this study.

Multivariate logistic regression analysis (Table 2) showed that TB infection was independently associated with age, increasing years of Tijuana residence, and a history of being incarcerated in Mexico or in Mexico and the United States. For each 1-year increase in age and time lived in Tijuana, participants were 1.03× (95% confidence interval [CI] 1.01×–1.05×) and 1.02× (95% CI 1.01×–1.04×) more likely to be IGRA+, respectively. Persons who had a history of being jailed in Mexico were 2.28× (95% CI 1.48×–3.51×) more likely to be IGRA+ than those who had never been incarcerated.