Robert M. Jasmer

A Molecular Epidemiologic Analysis of Tuberculosis Trends in San Francisco, 1991–1997

During the late 1980s and early 1990s, when tuberculosis case rates were increasing in many parts of the United States, it became apparent that considerable transmission of Mycobacterium tuberculosis was occurring and that the infection often progressed rapidly to tuberculosis (1-4). Transmission of M. tuberculosis followed by rapid progression to clinical disease was directly shown by many instances of institutional transmission and was indirectly shown by population-based studies using genotyping of the organism to infer epidemiologic links among patients with tuberculosis (5-14). Case rates are again decreasing in many urban areas and in the United States as a whole (15, 16). In New York City, the reduction in the tuberculosis case rate has been attributed mainly to the use of directly observed therapy, but other control measures have also been instituted (15). Many of the control interventions focused on preventing transmission of M. tuberculosis in high-risk groups and in such settings as hospitals, prisons, and shelters. Prevention of M. tuberculosis transmission would decrease the rate of tuberculosis cases that result from recent infection followed by a short latency period and rapid progression to clinical disease. Other measures, such as increased use of isoniazid preventive therapy, would decrease tuberculosis cases caused by reactivation of latent infection. In San Francisco, California, we identified a high proportion of clustered cases by using both conventional and molecular epidemiologic methods (9, 12); cases were considered to be clustered if the isolates had identical DNA fingerprints. These findings indicated that considerable transmission of M. tuberculosis was occurring within the population. On the basis of these observations, tuberculosis control measures directed toward reducing transmission of M. tuberculosis in high-risk groups were instituted beginning in early 1991. We hypothesized that these measures would reduce the frequency of tuberculosis resulting from recent transmission (clustering) as well as decrease the overall incidence of the disease. In this observational study, we describe the trends in tuberculosis incidence and the molecular epidemiologic pattern of the disease in San Francisco from 1 January 1991 through 31 December 1997. Methods Study Sample Tuberculosis was diagnosed in 2051 persons in San Francisco between 1 January 1991 and 31 December 1997. Of these cases, 1761 (86%) were confirmed by isolation of M. tuberculosis. Deoxyribonucleic acid fingerprinting using IS6110-based restriction fragment length polymorphism analysis was done on isolates from 1548 (88%) of the 1761 culture-positive cases. Six or more IS6110 hybridizing bands were found in 1238 (80%) of the tested isolates, and the remaining 310 isolates had five or fewer IS6110 bands. We excluded isolates from 25 cases (1.6% of cases for which DNA fingerprinting was done) that met criteria for laboratory cross-contamination. Polymorphic guanine-cytosine-rich sequence (PGRS)-based restriction fragment length polymorphism analysis was done on 257 of the 310 (83%) non-laboratory contaminant isolates with five or fewer IS6110 bands. Thus, 1495 of 1761 patients (85% of all culture-positive cases) had isolates with complete DNA fingerprinting data. No significant year-to-year differences were seen in the percentage of isolates for which IS6110-based and PGRS-based DNA fingerprinting was done. At the time of diagnosis, tuberculosis control personnel entered standard demographic information into the Centers for Disease Control and Preventions Report of a Verified Case of Tuberculosis form. Annual case rates of tuberculosis in San Francisco were calculated per 100 000 persons by using 1990 U.S. Census data on the population of San Francisco residents according to age, sex, place of birth, and race or ethnicity (17). The prevalence of HIV infection in San Francisco was obtained by using data from the 1997 Consensus Report on HIV Prevalence and Incidence in San Francisco (18). DNA Fingerprinting Methods IS6110-based DNA fingerprinting was performed by using the internationally standardized method, and the resulting patterns were compared by using BioImage Whole Band Analyzer software (version 3.3, BioImage Corp., Ann Arbor, Michigan) (19, 20). Isolates that had identical IS6110 fingerprints that contained five or fewer hybridizing bands were further analyzed with PGRS-based fingerprinting (21-24). Genomic DNA was digested with Sma1 and was probed by using a consensus sequence of the PGRS (22). The resulting patterns were examined visually and were defined as matching if the number, relative intensity, and molecular weights of the bands were identical. Definition of Laboratory Cross-Contamination Laboratory cross-contamination was considered to be the cause of a positive culture when all of the following criteria were met: 1) A patient had only one positive culture and that specimen tested negative on microscopic examination for acid-fast bacilli; 2) the specimen had been processed in the laboratory within 28 days of a specimen from another patient who had a positive result on microscopic examination for acid-fast bacilli; and 3) the microscopically positive specimen had the same DNA fingerprint as the putative contaminated specimen (25). Definition of Clustering Clustering was defined as the occurrence, within 1 year, of two or more isolates from different patients that contained 1) six or more IS6110 bands in an identical pattern or 2) five or fewer IS6110 bands that matched identically and had an identical PGRS pattern. A 1-year period for defining clustering was used because we were especially interested in assessing transmission of M. tuberculosis that resulted in rapid progression to disease. Clustered cases were identified as follows. For every initial isolate of M. tuberculosis from an incident case of tuberculosis, the database of IS6110 fingerprints was searched to identify any matching patterns within the previous year. If a match was found in the previous year, the case was counted as being in a cluster. For example, if the fingerprint from a case diagnosed in March matched the fingerprint from a case diagnosed the previous August, the case diagnosed in March was counted as clustered. Because no match preceding the first case identified with a particular fingerprint (index cases) will be found, these cases were counted as not clustered. Data from the 1990 U.S. Census from California (17) and the 1997 Consensus Report on HIV Prevalence and Incidence in San Francisco (18) were used to calculate annual rates of clustered cases per 100 000 persons for each year from 1992 to 1997, both for the population as a whole and according to place of birth (United States or outside of the United States) and presence of HIV infection. Because we did not begin performing comprehensive DNA fingerprinting until 1 January 1991, we could not compare isolates from patients in whom tuberculosis was diagnosed in the first 1-year period (1 January 1991 through 31 December 1991) with isolates from previous cases. Thus, the comparison of strains began with isolates obtained after 1 January 1992. Of the 1495 cases with isolates that had complete DNA fingerprinting data, 225 were from patients whose diagnosis was established between 1 January 1991 and 31 December 1991, and 1270 were from patients whose diagnosis was established between 1 January 1992 and 31 December 1997. These 1270 patients made up the final study sample for the analysis of clustering over time. Intensification of Tuberculosis Control Measures Tuberculosis control measures were intensified in five areas beginning in 1991, partly on the basis of findings from our earlier molecular epidemiologic studies (12, 26). 1. Improved contact investigations. Disease control investigators were trained to improve communication with specific populations at risk, including homeless persons and persons with a history of substance abuse. In addition, the assessment of what constitutes significant contact was broadened. The median number of contacts per new case increased from 1 in 1991 to 4 in 1995 and was 3 in 1996 and 1997 (P<0.001; chi-square test). The percentage of cases with no contacts identified decreased from 25% in 1990 to 3% in 1997 (Table 1). The total number of contacts identified each year increased between 1991 (approximately 1100) and 1992 and continued to range from approximately 2000 to 3000 from 1993 through 1997. The number of contacts infected with M. tuberculosis identified through these investigations increased from 498 in 1991 to 956 in 1997, and approximately 80% of the contacts completed a course of isoniazid preventive therapy in each of these years. The number of cases of active tuberculosis found through contact investigations ranged from 19 to 25 per year during the study period. Table 1. Data on Contacts of Tuberculosis Cases and Number of Case-Patients with Tuberculosis Receiving Directly Observed Therapy per Year in San Francisco, California, 1991-1997 2. Expanded use of directly observed therapy. Overall, 28% (541 of 1958) of all patients with incident cases received directly observed therapy as part of their treatment during the study period (Table 1). The proportion of patients receiving directly observed therapy increased from 22% in 1991 to a peak of 35% in 1996; in 1997, the percentage was 22%. The percentage of all patients who had documented negative sputum cultures within 3 months of the initiation of therapy increased slightly, from 65% in 1990 to 68% in 1997. 3. Development of an HIV-related tuberculosis prevention program. Persons seen at any of six city-based HIV primary care and methadone maintenance clinics underwent tuberculin skin testing and were evaluated clinically for tuberculosis. Depending on HIV status, tuberculin reactors were treated with a 12- or 6-month course of isoniazid preventive therapy after exclusion of active tubercu

A molecular epidemiological assessment of extrapulmonary tuberculosis in San Francisco.

The epidemiology of extrapulmonary tuberculosis (TB) is not well understood. We studied all cases of extrapulmonary TB reported in San Francisco during 1991-2000 to determine risk factors for extrapulmonary TB and the proportion caused by recent infection. Isolates were analyzed by IS6110-based restriction fragment-length polymorphisms analysis. There were 480 cases of extrapulmonary TB, of which 363 (76%) were culture positive; isolates were genotyped for 301 cases (83%). Multivariate analysis identified young age, female sex, and HIV infection as independent risk factors for nonrespiratory TB (excluding pulmonary, pleural, and disseminated TB). Pleural TB was less common in HIV-seropositive persons and women than were nonrespiratory forms of extrapulmonary TB. Pleural TB is different from other forms of extrapulmonary TB and is associated with the highest clustering rate (35% of cases) of all forms of TB. This high rate of clustering occurs because pleural TB is often an early manifestation of recent infection.

Adverse Events and Treatment Completion for Latent Tuberculosis in Jail Inmates and Homeless Persons

BACKGROUND Recently, a short-course treatment using 60 daily doses of rifampin and pyrazinamide was recommended for latent tuberculosis (TB) infection (LTBI). STUDY OBJECTIVES To determine the acceptability, tolerability, and completion of treatment. DESIGN Observational cohort study. SETTING Five county jails and TB outreach clinics for homeless populations in three cities. PATIENTS Study staff enrolled 1,211 patients (844 inmates and 367 homeless persons). INTERVENTIONS Sites used 60 daily doses of rifampin and pyrazinamide, an approved treatment regimen for LTBI. MEASUREMENTS Types and frequency of drug-related adverse events and outcomes of treatment. RESULTS Prior to treatment, 25 of 1,178 patients (2.1%) had a serum aminotransferase measurement at least 2.5 times the upper limit of normal. Patients who reported excess alcohol use in the past 12 months were more likely than other patients to have an elevated pretreatment serum aminotransferase level (odds ratio, 2.1; 95% confidence interval, 1.1 to 6.1; p = 0.03). Treatment was stopped in 66 of 162 patients (13.4%) who had a drug-related adverse event. Among 715 patients who had serum aminotransferase measured during treatment, 43 patients (6.0%) had an elevation > 5 times the upper limits of normal, including one patient who died of liver failure attributed to treatment. In multivariate analyses, increasing age, an abnormal baseline aspartate aminotransferase level, and unemployment within the past 24 months were independent risk factors for hepatotoxicity. Completion rates were similar in jail inmates (47.5%) and homeless persons (43.6%). CONCLUSIONS This study detected the first treatment-associated fatality with the rifampin and pyrazinamide regimen, prompting surveillance that detected unacceptable levels of hepatotoxicity and retraction of recommendations for its routine use. Completion rates for LTBI treatment using a short-course regimen exceeds historical rates using isoniazid. Efforts to identify an effective short-course treatment for LTBI should be given a high priority.

Short-Course Rifampin and Pyrazinamide Compared with Isoniazid for Latent Tuberculosis Infection: A Cost-Effectiveness Analysis Based on a Multicenter Clinical Trial

Two months of treatment with rifampin-pyrazinamide (RZ) and 9 months of treatment with isoniazid are both recommended for treatment of latent tuberculosis infection in adults without human immunodeficiency virus infection, but the relative cost-effectiveness of these 2 treatments is unknown. We used a Markov model to conduct a cost-effectiveness analysis to assess the impact on life expectancy and costs based on the results of a recent clinical trial that compared the rates of adverse events and completion of the 2 treatment regimens. Compared with no treatment, both regimens increased life expectancy by 1.2 years, but RZ cost 273 dollars more per patient. Sensitivity analyses showed that, assuming equal efficacy between the 2 regimens, there was no threshold completion rate for RZ at which the 2 treatments would be of equal net cost. Under most circumstances, treatment of latent tuberculosis infection with isoniazid is cost-saving than treatment with RZ.

Clinical and radiographic predictors of the etiology of computed tomography-diagnosed intrathoracic lymphadenopathy in HIV-infected patients.

&NA; In HIV‐infected patients with intrathoracic lymphadenopathy, it is not known whether clinical and radiographic findings are useful in predicting a specific diagnosis. We determined the etiology and predictors of the etiology of computed tomography (CT)‐diagnosed intrathoracic lymphadenopathy in HIV‐infected patients evaluated from June 1993 through April 1999. Multivariate analyses were performed to determine clinical and radiographic predictors of the three most common diagnoses. Of 318 patients, 110 (35%) had lymphadenopathy on chest CT. Among these 110 patients, tuberculosis/nontuberculous mycobacterial disease (n = 31), bacterial pneumonia (n = 26), and lymphoma (n = 21) were the most common diagnoses. Multivariate analysis identified cough and necrosis of lymph nodes on chest CT as independent predictors of tuberculosis/nontuberculous mycobacterial disease. African‐American race, symptoms for 1 to 7 days, dyspnea, and presence of airways disease on chest CT were independent predictors of bacterial pneumonia; symptoms for >7 days, absence of cough, and absence of pulmonary nodules on CT independently predicted lymphoma. Intrathoracic lymphadenopathy is a frequent chest CT finding in HIV‐infected patients. Opportunistic infections and lymphoma are the most common causes, and specific clinical and radiographic features can suggest these particular diagnoses.

Reaching the Limits of Tuberculosis Prevention among Foreign-Born Individuals: A Tuberculosis-Control Program Perspective

Analysis of whether assiduous implementation of American Thoracic Society/Centers for Disease Control and Prevention/Infectious Diseases Society of America guidelines for targeted testing and treatment of latent tuberculosis infection could have prevented any of 223 cases of active tuberculosis in foreign-born persons in San Francisco during the period 2002-2003. We report that 62% of these cases were not preventable and conclude that a further reduction in the incidence of tuberculosis among foreign-born persons will be modest without modification of current guidelines.

Perspectives on the Electronic ICU

Electronic ICUs (eICUs) have arisen as an effort to improve the intensive care medical coverage still unavailable to many patients. The overriding issue is the ongoing shortage of critical care med...