Comparison of secular trends in pneumococcal serotypes causing invasive disease in Denver, Colorado (1971-2004) and serotype coverage by marketed pneumococcal vaccines.

Invasive pneumococcal isolates from three hospitals in Denver, CO, USA were serotyped between 1971 and 2004. Serotype 14 was most common (13.2%), and other prevalent serotypes (3, 4, 6, 9 and 19) together accounted for 44.1% of the isolates. All prevalent serotypes and 91.3% of the total isolates were covered by pneumococcal polysaccharide vaccine, while 79.1% of prevalent serotypes and 56.7% of total isolates were covered by pneumococcal conjugate vaccine. Serotypes 6, 9 and 14 were more common in the final decade than in the first decade studied (37.3% vs. 20.2%), whereas serotypes 3 and 23 were more common in the first decade (18.5% vs. 11.0%).

Relationship between human immunodeficiency virus type 1 (HIV-1)-specific memory cytotoxic T lymphocytes and virus load after recent HIV-1 seroconversion.

Human immunodeficiency virus type 1 (HIV-1)-specific memory, or precursor, cytotoxic T lymphocytes (CTL) in 14 subjects who had recently experienced seroconversion were evaluated with respect to virus set point, defined as plasma HIV-1 RNA level 6 months after seroconversion. Env-, Gag-, Pol-, and Nef-specific precursor CTL were detected in (51)Cr-release assays, using antigen-stimulated peripheral blood mononuclear cells as effectors and B cell lines infected with HIV-1-vaccinia recombinants as targets. All subjects tested had precursor CTL specific to at least 2 HIV-1 antigens. Detection of Env-specific precursor CTL was associated with a high set point (P=.0221). The number of antigens recognized tended to be greater in subjects with higher set points (rho=.45621; P=.1171). Gag-specific precursor CTL frequency correlated inversely with set point (rho=-.8478; P=.0003). Two heterozygotes for a 32-bp deletion in CCR5 had the lowest set points (P=.0220) and highest Gag precursor CTL frequencies (P=.0128). These data suggest that host factors that restrict viral replication may be important determinants of the level of HIV-1-specific precursor CTL.

Gonorrhea in the HIV era: a reversal in trends among men who have sex with men.

OBJECTIVES: Gonorrhea cases among men who have sex with men (MSM) declined in the early years of the HIV epidemic. We evaluated more recent trends in gonorrhea among MSM through the Centers for Disease Control and Prevention's Gonococcal Isolate Surveillance Project. METHODS: Isolates and case information were collected from 29 US sexually transmitted disease (STD) clinics. Gonococcal urethritis cases among MSM were compared with those among heterosexual men, and cases among MSM in 1995 to 1999 were compared with earlier MSM cases. RESULTS: Of 34,942 cases, the proportion represented by MSM increased from 4.5% in 1992 to 13.2% in 1999 (P < .001). Compared with heterosexuals, MSM were older, more often White, and more often had had gonorrhea previously, although fewer had had gonorrhea in the past year. MSM with gonorrhea in 1995 to 1999 were slightly older than those with gonorrhea in 1992 to 1994, and a higher proportion had had gonorrhea in the past year. CONCLUSIONS: MSM account for an increasing proportion of gonococcal urethritis cases in STD clinics. Given recent evidence that gonorrhea may facilitate HIV transmission, these trends demand increased attention to safe sexual behaviors and reducing STDs among MSM.

Screening for Chlamydia trachomatis in an anonymous and confidential HIV counseling and testing site: feasibility and prevalence rates.

BACKGROUND: Chlamydia trachomatis is the most common bacterial sexually transmitted disease (STD) in the United States. The development of nucleic acid amplification tests for C trachomatis in urine specimens allows for screening outside traditional clinic settings. Persons visiting an HIV counseling and testing site may be at increased risk for STDs, including C trachomatis. GOAL: To measure the acceptance of C trachomatis urine screening and the prevalence of C trachomatis infection among clients at an HIV counseling and testing site. STUDY DESIGN: Site HIV counselors offered urine C trachomatis screening to clients, administered a questionnaire, and collected urine samples. RESULTS: Of 808 counseling and testing site clients approached for C trachomatis screening, 572 (71%) accepted. The most common reasons for declining screening were absence of symptoms (33%) and recent STD testing (32%). Men were more likely to accept urine screening than women (risk ratio, 1.31; 95% CI, 1.06-1.62), as were clients who practiced oral sex, had a history of STD, or who had never been screened for STD. Of 560 urine specimens processed, only 8 (1.43%; 95% CI, 0.66-2.91%) were infected with C trachomatis. CONCLUSIONS: Sites offering HIV testing and counseling are a feasible alternative to clinical settings for C trachomatis screening. Prevalence may be too low for screening to be cost effective unless higher-risk subpopulations can be identified.

Are US populations appropriate for trials of human immunodeficiency virus vaccine? The HIVNET Vaccine Preparedness Study.

Questions exist about whether testing of preventive human immunodeficiency virus (HIV)-1 vaccines, which will require rapid recruitment and retention of cohorts with high HIV-1 seroincidence, is feasible in the United States. A prospective cohort study was conducted in 1995-1997 among 4,892 persons at high risk for HIV infection in nine US cities. At 18 months, with an 88% retention rate, 90 incident HIV-1 infections were observed (1.31/100 person-years (PY), 95% confidence interval (CI): 1.06, 1.61). HIV-1 seroincidence rates varied significantly by baseline eligibility criteria--1.55/100 PY among men who had sex with men, 0.38/100 PY among male intravenous drug users, 1.24/100 PY among female intravenous drug users, and 1.13/100 PY among women at heterosexual risk-and by enrollment site, from 0.48/100 PY to 2.18/100 PY. HIV-1 incidence was highest among those men who had sex with men who reported unprotected anal intercourse (2.01/100 PY, 95% CI: 1.54, 2.63), participants who were definitely willing to enroll in an HIV vaccine trial (1.96/100 PY, 95% CI: 1.41, 2.73), and women who used crack cocaine (1.62/100 PY, 95% CI: 0.92, 2.85). Therefore, cohorts with HIV-1 seroincidence rates appropriate for HIV-1 vaccine trials can be recruited, enrolled, and retained.

Slow diffusion of home HIV-specimen collection: provider concerns at odds with client preferences.

BACKGROUND: Home specimen collection and telephone counseling (HSCTC) may be a convenient new method for detection of HIV infection among cohorts at high-risk for HIV. GOAL: To evaluate attitudes about HSCTC among participants, HIV counselors, and community advisory board members associated with a national multisite study of persons at high risk for HIV. STUDY DESIGN: Twelve focus groups and surveys were conducted at six sites among 126 counselors, community advisory board members, and cohort participants. RESULTS: Staff and community advisory board members raised concerns about the acceptability, feasibility, safety, and effectiveness of HSCTC. In contrast, participants (92%) reported a willingness to collect blood and oral samples on a frequent basis, and preferred telephone (73%) to office-based counseling. CONCLUSION: Home specimen collection and telephone counseling appear to be preferred by study participants at high risk of HIV infection. Staff and community advisory board members had stronger reservations than prospective users.

Home collection for frequent HIV testing: acceptability of oral fluids, dried blood spots and telephone results. HIV Early Detection Study Group.

OBJECTIVE: To assess the feasibility and acceptability of bimonthly home oral fluid (OF) and dried blood spot (DBS) collection for HIV testing among high-risk individuals. DESIGN: A total of 241 participants [including men who have sex with men (MSM), injecting drug users (IDU), and women at heterosexual risk] were recruited from a randomly selected subset of study participants enrolled at four sites in the HIV Network for Prevention Trials (HIVNET) cohort, and assigned at random to bimonthly home collection of OF or DBS specimens over a 6 month interval. Participants could select telephone calls or clinic visits to receive HIV test results. METHODS: Bimonthly specimens were tracked for adherence to the schedule, were evaluated for adequacy for testing, and tested using antibody assays and polymerase chain reaction (PCR) for DBS. The acceptability of bimonthly home OF and DBS collection and telephone counseling was assessed in an end-of-study questionnaire. RESULTS: The laboratory received 96 and 90% of expected OF and DBS specimens, respectively; 99% of each specimen type was adequate for testing. Almost all (95%) participants chose results disclosure by telephone. The majority of participants (85%) reported that bimonthly testing did not make them worry more about HIV, and almost all (98%) judged that with bimonthly testing their risk behavior remained the same (77%) or became less risky (21%). CONCLUSION: Bimonthly home specimen collection of both OF and DBS with telephone counseling is acceptable and feasible among study participants at high risk. These methods will be useful for the early detection of HIV infection and remote follow-up of research cohort participants in HIV vaccine and prevention trials.

Comparative prevalence of infection with Trichomonas vaginalis among men attending a sexually transmitted diseases clinic.

OBJECTIVE: Although established in women as a common cause of vaginal discharge, the prevalence of Trichomonas vaginalis (TV) in men compared with other classic urethral pathogens has not been well characterized. To assess this issue, the authors compared the prevalence of Neisseria gonorrhoeae (GC), Chlamydia trachomatis (CT), and TV in consecutive men attending a sexually transmitted diseases (STD) clinic. METHODS: From June 1, 1998 to July 27, 1998, 454 consecutive men presenting to the Denver Metro Health Clinic with a new problem were tested for GC by urethral swab culture, for CT by polymerase chain reaction of urine, and for TV by urine sediment culture. RESULTS: GC, CT, and TV were detected in 23 (5.1%), 34 (7.5%), and 13 (2.8%) of men, respectively. There were significant differences by age for both CT (11.3% in men younger than 30 years versus 3.3% in men 30 years and older, P < 0.05) and TV (0.8% in men younger than 30 years versus 5.1% in men 30 years and older, P < 0.05). In 50 men 30 years or older with symptoms of urethral discharge, TV prevalence (12.0%) rivalled that of GC (12.0%) and CT (14.0%). In 45 men 30 years and older with nongonococcal urethritis, the prevalence of TV and CT were each 13.3%. Multivariate logistical regression analysis showed the presence of discharge and nongonococcal urethritis in men 30 years and older to be an independent predictor of TV. CONCLUSIONS: TV is common in men attending sexually transmitted disease clinics, especially in those 30 years or older, in whom it may account for as much urethritis as GC or CT. These findings suggest that in older men with nongonococcal urethritis, diagnostic evaluation, empiric treatment, and partner management should include the possibility of TV infection.

The Denver school-based adolescent hepatitis B vaccination program: a cost analysis with risk simulation.

OBJECTIVES: This study sought to compare the cost-effectiveness of a school-based hepatitis B vaccine delivery program with that of a vaccine delivery program associated with a network health maintenance organization (HMO). METHODS: The vaccination program enrolled 3359 sixth-grade students from 18 middle schools in Denver, Colo. Immunization status and direct and indirect program costs were compiled. The sensitivity of the outcomes was assessed by simulation methods. RESULTS: The per-dose cost-effectiveness ratio for the school-based delivery system was $31. This cost-effectiveness ratio remained stable when the model was simulated with costs that were underestimated or overestimated by 20%. In the network HMO, the direct cost per dose was $68 and the societal cost was $118 when the child's father worked full-time and the mother worked part-time. There is less than a 5% chance that the network HMO-based vaccination program could be more cost-effective than the school-based program. CONCLUSIONS: The cost per dose of the school-based program was significantly less than that of the network HMO-based program, because in the school program government-purchased vaccine was available at a lower cost and parents did not incur work-loss costs.

Non-injection substance use correlates with risky sex among men having sex with men: data from HIVNET.

Associations between substance use and sexual behavior were examined among 3220 seronegative men who have sex with men (MSM) in a HIV vaccine preparedness study. Relationships between current and past substance use and current sexual risk were evaluated using crude odds ratios and logistic regression to adjust for confounding variables. Heroin and injection drug use were uncommon (< 2%). Substances most often used were alcohol (89%), marijuana (49%), nitrite inhalants (29%), amphetamines or similarly acting stimulants (21%), cocaine 14% and hallucinogens (14%). Increased adjusted odds for unprotected sex were significantly associated with current heavy alcohol use (OR 1.66; CI 1.18, 2.33), past alcohol problems (OR 1.25; CI 1.05, 1.48), and current drug use (OR 1.26; CI 1.08, 1.48). When associations with specific drugs and nitrite inhalants were examined separately, current use of cocaine and other stimulants (OR 1.25; CI 1.01, 1.55), hallucinogens (OR 1.40; CI 1.10, 1.77), and nitrite inhalants (some (OR 1.61; CI 1.35, 1.92); heavy (OR 2.18; CI 1.48, 3.20)), were independently associated with unprotected sex. Those with past drug use or past heavy alcohol use but not currently using demonstrated no increase in sexual risk, suggesting an important role for substance-focused interventions in risk reduction efforts among MSM.

Feasibility and suitability of targeting young gay men for HIV vaccine efficacy trials.

We evaluated factors affecting the feasibility of including young high-risk HIV-negative gay and bisexual men in preventive HIV vaccine trials using data from the U.S. Centers for Disease Control and Prevention Collaborative HIV Seroincidence Study. Of 2189 men enrolled in this study, 17% were <25 years of age. HIV seroincidence was 4.2/100 person-years (95% confidence interval [CI], 2.6-7.0) in young men compared with 2.0/100 person-years (95% CI, 1.4-2.6) for older men. Compared with men 25 and older, young men were more likely to report several high-risk behaviors, to perceive themselves to be at risk for HIV infection, and to report that their risk behavior might be increased by participation in an HIV vaccine trial. The majority of both young men (69%) and older men (74%) expressed willingness in participate in HIV vaccine trials. Young men were less likely to answer questions about vaccine concepts correctly and were more likely to be lost to follow-up. Young gay and bisexual men are important candidates for future HIV vaccine trials, but they may need targeted approaches to recruitment, retention, education about trial concepts prior to enrollment, and behavioral interventions during the trial.

Quinolone-resistant Neisseria gonorrhoeae: the beginning of the end? Report of quinolone-resistant isolates and surveillance in the southwestern United States, 1989 to 1997.

BACKGROUND: Fluoroquinolones are one of the most widely used treatments for gonorrhoeae. Changes in the susceptibility of Neisseria gonorrhoeae to these agents may threaten their use. GOAL OF THIS STUDY: To report several resistant strains (>1 mcg/ml) isolated in the western United States and to evaluate the prevalence of strains with reduced susceptibility (ofloxacin 0.25 mcg/ml, ciprofloxacin 0.06 mcg/ml). STUDY DESIGN: The microbiology and epidemiology of three resistant strains were characterized and 12,761 other strains were evaluated for fluoroquinolone susceptibility as part of the Gonococcal Isolate Surveillance Project of the Centers for Disease Control and Prevention. RESULTS: Fluoroquinolone-resistant strains may appear sporadically. The prevalence of isolates with reduced susceptibility to fluoroquinolones remains low in the Southwest region of the United States. CONCLUSIONS: Continued active surveillance is needed to detect and control the spread of quinolone-resistant Neisseria gonorrhoeae.

The diverse patterns of hepatitis A epidemiology in the United States-implications for vaccination strategies.

Hepatitis A is the most frequently reported vaccine-preventable disease in the United States. Hepatitis A incidence and risk factors during 1983-1995 were examined among cases reported to the study's Sentinel Counties: Denver County, Colorado; Pierce County, Washington; Jefferson County, Alabama; and Pinellas County, Florida. Of 4897 serologically confirmed cases, 611 patients (13%) were hospitalized and 9 (0.2%) died. The average incidence was 14.7/100, 000 (range, 0.6-100.7/100,000, depending on county and year). The frequency of reported sources of infection varied by county, but the largest single group overall (52%) did not report a source. During 3-year communitywide outbreaks in Denver (1991-1993) and Pierce (1987-1989) Counties, rates increased 4- and 13-fold, respectively, and increased in all age, racial/ethnic, and risk groups. During communitywide outbreaks, hepatitis A is not limited to specific risk groups; sustained nationwide reductions in incidence are more likely to result from routine childhood vaccination than from targeted vaccination of high-risk groups.

Tobacco use and other risk behaviors among adolescents in an STD clinic.

BACKGROUND: This study examined the potential for tobacco use and other health risk behavior interventions in the context of an urban sexually transmitted diseases (STD) clinic. METHODS: A cross-sectional survey of two populations. Adolescents seen at an STD clinic or at the teen clinic of a community health center completed a self-administered computer survey in 1996. Risk behaviors, attitudes, and readiness to stop smoking were analyzed for 225 patients at the STD clinic and 248 patients at the teen clinic. RESULTS: Compared with adolescents in the teen clinic, adolescents in the STD clinic were more likely to have smoked frequently (OR 1.7, 95% CI 1.1, 3.0), used any illegal drug (OR 2.7, 95% CI 1.3, 5.5), recently binged on alcohol (OR 1.7, 95% CI 1.0, 2.8), and had more than 10 lifetime sexual partners (OR 1.9, 95% CI 1.0, 3.4). Weapon carrying, readiness to stop smoking, and attitudes toward smoking did not differ between sites. CONCLUSIONS: Cigarette smoking and other health risk behaviors are more prevalent among adolescents in an STD clinic than among adolescents in a community health center. STD clinics are potential sites for cigarette, alcohol, and drug use interventions among "hard to reach" adolescents.

Assessment of the changing willingness to participate in phase III HIV vaccine trials among men who have sex with men.

This paper describes the willingness of 1267 men who have sex with men (MSM) enrolled in a prospective HIV vaccine preparedness study from Chicago, Denver, and San Francisco to enroll in HIV vaccine efficacy trials. Respondents were interviewed at baseline and followed-up at 6, 12, and 18 months. At each visit respondents were tested for HIV antibodies using enzyme-linked immunosorbent assay (ELISA) testing with Western blot confirmation. Over 18 months, the annualized HIV seroincidence of this cohort was 2.4%. At baseline, 37% of the men reported that they would be "definitely" willing to participate in an HIV vaccine efficacy trial; however, this dropped to 21% at 12 months and remained stable at 18 months. Greater willingness to participate (WTP) was related to lower education, engaging in HIV risk behavior, living in Denver, white ethnicity, and older age. Changing WTP suggests that the decision to participate in HIV vaccine efficacy trials may be complex and dynamic and take an extended time. These data underscore the importance of informed consent and raise questions regarding the influence of decision-making processes on HIV vaccine efficacy trial design, compliance, and validity.

Prevalence of human immunodeficiency virus infection, mortality rate, and serogroup distribution among patients with pneumococcal bacteremia at Denver General Hospital, 1984-1994.

Pandemics of human immunodeficiency virus (HIV) type 1 infection and penicillin resistance highlight the urgency of preventing invasive pneumococcal disease with vaccination. We characterized pneumococcal serogroup distribution and the mortality rate among 460 patients with pneumococcal bacteremia from 1984 through 1994 at Denver General Hospital and the prevalence of HIV infection in patients for whom pneumococcal bacteremia was diagnosed from 1989 to 1994. Vaccine-related serogroups accounted for 426 isolates (92.6%), including 48 (92.3%) of 52 isolates from HIV-infected patients. Mortality among patients 15 years of age or older was higher during 1984-1988 (18[12.9%] of 140) than during 1989-1994 (10 [5.2%] of 191: rate ratio, 2.5; 95% confidence interval, 1.2-5.2). Of patients 15-59 years of age from 1989 to 1994, 44 (39.6%) of 111 men and three (7.3%) of 41 women were HIV-infected. Four (8.5%) of 47 HIV-infected patients and four (3.8%) of 105 other patients in this group died (age-weighted rate ratio, 1.8; 95% confidence interval, 0.5-6.2). We recommend routine screening of young adults with pneumococcal bacteremia for HIV infection and immunization of HIV-infected patients with pneumococcal vaccine (which includes most serogroups of infecting strains).

Feasibility and yield of screening urine for Chlamydia trachomatis by polymerase chain reaction among high-risk male youth in field-based and other nonclinic settings. A new strategy for sexually transmitted disease control.

BACKGROUND: Inner-city youth are at disproportionate risk for Chlamydia trachomatis infection. Identification of infected individuals is hampered by the often asymptomatic nature of infection and access and utilization barriers to clinic-based screening services. The feasibility and yield of screening urine for C. trachomatis by polymerase chain reaction was studied among high-risk male youth outside traditional clinic settings. METHODS: As part of a community-level sexually transmitted disease (STD) prevention program among high-risk youth in Denver, outreach workers enrolled subjects, administered questionnaires, and collected first-catch urine samples in nonclinical facility-based and field-based settings. Facility settings consisted of community/recreation centers, high-schools, and an STD/human immunodeficiency virus prevention storefront. Field settings included alleys, parking lots, parks, and residences. Individuals who tested C. trachomatis positive were contacted by program outreach workers and provided with standard treatment and partner notification services. RESULTS: Over a 20-month period, 486 urine specimens were collected, 32 (6.6%) of which were C. trachomatis positive. Rates were higher for subjects screened in the field than in facility settings (11.9% vs. 4.4%, P < 0.05). Subjects with chlamydial infection were more likely to have had vaginal intercourse in the previous 30 days (adjusted odds ratio: 2.9) and to have been recruited in field settings (adjusted odds ratio: 2.5). Of subjects with chlamydial infection, 31/32 (97%) were treated within a median of 8 days after urine collection. CONCLUSIONS: Urine chlamydial screening by polymerase chain reaction of sexually active male youth in nontraditional settings appears to be feasible and to provide yields similar to those reported in standard clinic settings. Evaluation of samples easily collected in nonclinic locations holds great promise as an additional strategy for the control of chlamydial infection and other STD among difficult-to-reach populations.

Short-term incarceration for the management of noncompliance with tuberculosis treatment.

STUDY OBJECTIVES: To review the use of incarceration for noncompliance with tuberculosis treatment. DESIGN: Retrospective review. SETTING: An urban tuberculosis control program. PATIENTS: Patients treated for active tuberculosis. MEASUREMENTS AND RESULTS: We reviewed the legal basis and practical application of quarantine for active tuberculosis, including the use of incarceration for noncompliance. The records of patients treated at the Denver Metro Tuberculosis Clinic during 1984 to 1994 were reviewed to identify patients who were incarcerated and to evaluate the effectiveness of this intervention. Of 424 cases of tuberculosis, 20 patients (4.7%) were incarcerated for noncompliance; an additional 21 patients (5.0%) were lost to follow-up prior to completing therapy. Incarcerated patients were predominantly men who were born in the United States and had a history of homelessness and alcohol abuse. The median duration of the initial incarceration was 20 days (range, 7 to 51 days). Of the 17 patients released prior to completing therapy, 13 (76%) were compliant with outpatient, directly observed therapy after one or two short-term incarcerations (<60 days); only three patients were incarcerated for the duration of treatment. Overall, 18 of 20 incarcerated patients (90%) were successfully treated. CONCLUSIONS: Approximately 5% of the patients treated through our program were incarcerated for noncompliance; an additional 5% were unavailable for follow-up and would have been candidates for incarceration if found. Homelessness and alcoholism were closely associated with the use of incarceration. Short-term incarceration followed by outpatient, directly observed therapy was relatively successful in the management of this difficult patient population.

Antimicrobial resistance in Neisseria gonorrhoeae in the United States, 1988-1994: the emergence of decreased susceptibility to the fluoroquinolones.

Antimicrobial susceptibilities of Neisseria gonorrhoeae have been prospectively determined in the Gonococcal Isolate Surveillance Project of the Centers for Disease Control and Prevention. From 1988 through 1994, susceptibilities were determined for 35,263 isolates from 27 clinics. Patients were demographically similar to those in nationally reported gonorrhea cases. In 1994, 30.5% of isolates had chromosomally or plasmid-mediated resistance to penicillin or tetracycline. Penicillin resistance increased from 1988 (8.4%) to 1991 (19.5%) and then decreased in 1994 (15.6%). Tetracycline resistance decreased from 1988 (23.4%) to 1989 (17.3%) and then increased in 1994 (21.7%). Most isolates (99.9%) were highly susceptible to broad-spectrum cephalosporins. Isolates with decreased susceptibility to ciprofloxacin increased from 1991 (0.4%) to 1994 (1.3%); 4 isolates were ciprofloxacin-resistant. Ciprofloxacin-resistant strains may not respond to therapy with recommended doses of fluoroquinolones, and the clinical importance of strains with decreased susceptibility is unknown. The emergence of fluoroquinolone resistance in N. gonorrhoeae in the United States threatens the future utility of this class of antimicrobials for gonorrhea therapy.

Noncompliance with directly observed therapy for tuberculosis. Epidemiology and effect on the outcome of treatment.

STUDY OBJECTIVES: To describe the epidemiology and clinical consequences of noncompliance with directly observed therapy (DOT) for treatment of tuberculosis. DESIGN: Retrospective review. SETTING: An urban tuberculosis control program that emphasizes DOT. PATIENTS: All patients treated with outpatient DOT from 1984 to 1994. MEASUREMENTS AND RESULTS: We defined noncompliance as follows: (1) missing > or = 2 consecutive weeks of DOT; (2) prolongation of treatment > 30 days due to sporadic missed doses; or (3) incarceration for presenting a threat to public health. Poor outcomes of therapy were defined as a microbiologic or clinical failure of initial therapy, relapse, or death due to tuberculosis. Fifty-two of 294 patients (18%) who received outpatient DOT fulfilled one or more criteria for noncompliance. Using multivariate logistic regression, risk factors for noncompliance were alcohol abuse (odds ratio, 3.0; 95% confidence interval, 1.2 to 7.5; p = 0.02) and homelessness (odds ratio, 3.2; 95% confidence interval, 1.5 to 7.2; p = 0.004). Noncompliant patients had poor outcomes from the initial course of therapy more often than compliant patients: 17 of 52 (32.7%) vs 8 of 242 (3.3%); relative risk was 9.9; 95% confidence interval was 4.5 to 21.7 (p < 0.001). CONCLUSIONS: In an urban tuberculosis control program, noncompliance with DOT was common and was closely associated with alcoholism and homelessness. Noncompliance was associated with a 10-fold increase in the occurrence of poor outcomes from treatment and accounted for most treatment failures. Innovative programs are needed to deal with alcoholism and homelessness in patients with tuberculosis.