Risk Factors for Rifampin-monoresistant Tuberculosis A Case-Control Study

Risk Factors for Rifampin-monoresistant Tuberculosis A Case-Control Study LAURIE SANDMAN, NEIL W. SCHLUGER, AMY L. DAVIDOW, and STANLEY BONK Division of Pulmonary and Critical Care Medicine, Department of Environmental Medicine, and Department of Pathology, New York University School of Medicine, New York, New York Rifampin is the cornerstone of short-course chemotherapy for the treatment of tuberculosis (TB). Rifampin monoresistance (RMR) is less common than resistance to isoniazid alone or in combination with other antituberculous medications. We conducted a retrospective case-control study to identify risk factors for RMR-TB. Complete records for 21 of a total of 26 RMR patients from 1990 to 1997 were available for review, and were compared with those of 48 patients with drug-susceptible TB, controlling for year of diagnosis. Cases more frequently had a history of TB than did controls (61% versus 22%, p 0.01), and were more often human immunodeficiency virus (HIV) positive (81% versus 46%, p 0.02). With control for HIV status, cases were more likely to have extrapulmonary involvement (47.6% versus 11.6%, p 0.05). Four cases (19%) and one control (2.1%) died (p 0.02) during hospitalization. Cases more often had a history of incarceration (71.4% versus 37.5%, p 0.09). Among the 13 cases with a history of TB, five had evidence of malabsorption (vomiting and/or diarrhea), versus none of the 11 controls with prior TB. These data support the hypothesis that RMR is seen primarily in individuals with a history of TB and who are HIV positive. Cases were frequently noncompliant with previous treatment for TB, had a history of incarceration, and had poor outcomes. Sandman L, Schluger NW, Davidow AL, Bonk S. Risk factors for rifampin-monoresistant tuberculosis: a case-control study. AM J RESPIR CRIT CARE MED 1999;159:468 472. (Received in original form May 29, 1998 and in revised form August 11, 1998) Supported in part by grant K07 HL03030 (Tuberculosis Academic Award) from the National Institutes of Health (N.W.S). Correspondence and requests for reprints should be addressed to Neil W. Schluger, M.D., Division of Pulmonary, Allergy, and Critical Care Medicine, Columbia University College of Physicians and Surgeons, VC 12-213, 630 West 168th Street, New York, NY 10032. E-mail: ns311@columbia.edu Am J Respir Crit Care Med Vol 159. pp 468 472, 1999 Internet address: www.atsjournals.org Rifamycins (rifampin in particular) are the cornerstone of short-course chemotherapy regimens for tuberculosis (TB) (1). Most treatment regimens that do not contain rifampin require at least 18 mo of treatment and/or include injectable antibiotics for prolonged periods, and are therefore difficult to administer, are associated with numerous adverse effects, and create substantial obstacles to compliance with therapy (1). For these reasons, the loss of rifampin from the treatment regimen has considerable implications both for individual patient outcomes and for TB control programs generally. Clinically, development of rifampin monoresistance (RMR) is not as common as development of resistance to isoniazid alone, and in the United States and elsewhere, resistance to isoniazid (INH) and rifampin appears to be more common than resistance to rifampin alone (2). Drug resistance may occur with a first TB infection (primary) or occur in a person with a history of drug-sensitive TB (acquired resistance). Exogenous reinfection with resistant organisms has been documented in patients with multidrugresistant disease, but is generally regarded as an uncommon cause of drug-resistant TB (3). Clinical factors associated with drug-resistant TB generally include nonadherence to medication regimens by patients with organisms susceptible to first-line drugs, prescribing errors, and malabsorption of TB medications, with subsequent subtherapeutic drug levels (4 6). Malabsorption may be the result of vomiting, diarrhea, or other gastrointestinal disease. Human immunodeficiency virus (HIV) disease, with or without clinical evidence of malabsorption or HIV-related enteropathy, has been documented as a risk factor for multidrug-resistant TB (7, 8). Single drug resistance in TB was first documented in the earliest trials of streptomycin, and is believed to occur when dividing organisms are exposed to a single drug. Presumably, RMR should develop in the same way, although rifampin is never prescribed as a single agent in the treatment of active TB. (Rifabutin, given as a single agent for the prevention of bacteremia due to Mycobacterium avium complex in patients with HIV infection, has been associated with the development of rifampin-resistant TB [9].) The clinical circumstances in which RMR may occur have not been well described, although recent reports have suggested that HIV infection is associated with rifampin monoresistance, and have used restriction fragment length polymorphism (RFLP) analysis to show that RMR-TB is rarely the result of reinfection (10 12). Recently, we have noted a number of cases of TB caused by isolated organisms resistant to rifampin alone. To more rigorously identify clinical characteristics associated with RMR-TB, we conducted a case-control study of all cases seen at our institution

Sandman, Schluger, Davidow, et al.: Rifampin-Resistance Risk Factors 469 (a hospital with a large TB service) over a period of several years. METHODS We conducted a retrospective case-control study of patients with RMR-TB at Bellevue Hospital Center, an 811-bed municipal hospital in lower Manhattan (New York City), between 1990 and 1997. A case of RMR was defined as one in which a patient with a culture positive for Mycobacterium tuberculosis from any body site, and resistant to rifampin at either 5 g/ml on solid media by the method of proportions or at 2 g/ml in liquid media using the BACTEC (Becton Dickinson, Sparks, MD) method, and without documented resistance to INH, pyrazinamide (testing for this at our institution began in February 1994), streptomycin, or ethambutol. All cases with complete charts available for review were included in the study. Controls were matched by year of diagnosis, and were randomly selected from all patients at Bellevue Hospital with culture-proven TB sensitive to INH, rifampin, streptomycin, pyrazinamide, and ethambutol. A 2:1 ratio of controls to cases was used to increase the ability of the study to detect differences between groups. Demographic and clinical data were abstracted from inpatient and outpatient hospital records. Demographic data included age, sex, race, and birthplace. Social history included history of homelessness, defined as living on the street, in a shelter, or in a single-room occupancy (SRO) hotel at any time in the preceding 5 yr; previous or current incarceration; and past or present intravenous drug use (IVDU). Lengths of stay and deaths during hospitalization for the defining episode of TB were compared in the RMR-TB and control groups. For cases and controls, a positive history of TB was defined as having a culture positive for M. tuberculosis from any body site 6 mo before the index episode. For cases and controls with a history of TB, previous drug-susceptibility data and compliance with treatment regimens were verified by review of medical records and the New York City Department of Health (NYC DOH) TB registry. Noncompliance was defined as attendance at less than 75% of scheduled directly observed therapy (DOT) visits. For patients treated without DOT (or before DOT was available in New York City), a determination of noncompliance was made if the NYC DOH Bureau of Tuberculosis Control records indicated that the patient had either refused therapy or had been lost to follow-up. For cases, the number of months between the last drug-susceptible TB culture and the first RMR-TB culture was determined. Site of disease, chest radiograph-based reports of significant findings (cavitary or noncavitary lesions), and sputum smear findings (positive or negative) for patients with pulmonary TB were compared. HIV status was documented; for HIV-positive patients, CD4 cell counts and opportunistic infections concurrent with the defining episode of TB were determined. For all patients with a history of TB, we reviewed inpatient and outpatient records for a history of vomiting or diarrhea lasting 2 wk and use of antifungal drugs (ketaconazole, fluconazole) and rifabutin during the previous episode of TB. Malabsorption was deemed to have been present if vomiting or diarrhea were documented for 2 wk while a patient was receiving therapy. Cases of RMR were categorized as having acquired (or secondary) resistance when there was a history of culture-positive TB with a documented susceptible organism. Primary resistance was defined as RMR without a previous history of TB or without TB treatment verified by both hospital records and the NYC DOH TB registry. When no susceptibility results were available for cases with a history of culture-positive TB, patients were classified as unknown. Analysis of categorical variables in the case control study was done with the Mantel Haenszel chi-square test, McNemar s test, and exact methods, including Fisher s exact test. Continuous data (mean age and mean length of stay [log transformed]) were analyzed with analysis of variance (ANOVA) models. Median CD4 cell counts were compared through the Kruskal Wallis test. Analysis of cases was done with Fisher s exact test. RESULTS During the period included in this study, 26 cases of RMR-TB were diagnosed. These 26 cases represented 1.4% (26 of 1,921 TABLE 1 DEMOGRAPHIC CHARACTERISTICS OF CASES AND CONTROLS Cases (n 21) Controls (n 48) Age, mean (range) 38 (23 63) 40 (19 76) Male sex 15 (71%) 40 (83%) Race Black 15 (71%) 27 (56%) Hispanic 6 (29%) 13 (27%) Other 0 (0) 8 (17%) Foreign-born 6 (29%) 19 (40%) History of homelessness 12 (57%) 29 (60%) History of incarceration 15 (71%) 18 (37%) History of IVDU 10 (48%) 20 (42%) Definition of abbreviation: IVDU intravenous drug use. All differences between cases and controls were nonsignificant (p 0.05). cases) of culture-positive TB patients seen at our hospital during this interval. Case-Control Analysis Demographics. Complete charts for 21 of 26 (81%) RMR patients and 48 of 52 controls (92%) were available for review. Cases and controls were well matched and similar with regard to sex, age, homelessness, and IVDU (Table 1). Cases were more likely than controls to have a history of incarceration (71.4% versus 37.5%). However, the difference was not statistically significant, possibly because of the small numbers of subjects; incarceration history was unknown for six (28.6%) cases and 21 (43.8%) controls. Among the 45 subjects with no known history of TB, five of the eight cases (62.5%) and nine of the 37 controls (24.3%) were prisoners at the time of hospitalization (p 0.03, odds ratio [OR] 8.7, 95% confidence interval [CI] 1.2 to 63.2). TB history and disease. Thirteen cases (62%) and 11 controls (23%) had a history of at least one previous episode of TB (p 0.01, OR 7, 95% CI 2.04 to 24.02). Of these 24 patients, eight cases (62%) and seven controls (64%) had been noncompliant with previous treatment regimens (Table TABLE 2 CHARACTERISTICS OF PATIENTS WITH RIFAMPIN- MONORESISTANT TUBERCULOSIS Cases (n 21) Controls (n 48) p Value TB history History of prior tuberculosis 13 (62%) 11 (23%) 0.01 Compliant with therapy for the prior episode of tuberculosis 4 (31%) 2 (18%) NS Clinical characteristics Cavitation on chest radiograph 2 (10%) 12 (25%) NS Positive sputum smear 14 (67%) 29 (60%) NS Pulmonary tuberculosis only 11 (52%) 39 (81%) 0.001 Days of hospitalization, mean (range) 52 (3 178) 34 (2 107) 0.07 Deaths during hospitalization 4 (19%) 1 (2%) 0.02 HIV status HIV serostatus Positive 17 (81%) 22 (46%) 0.03 Negative 1 (5%) 12 (25%) Unknown 3 (14%) 14 (29%) Median (mean) CD4 cell count/mm 3 57 (119) 83 (194) NS Prior opportunistic infection 8 (47%) 3 (14%) 0.01 Concurrent opportunistic infection 10 (59%) 11 (50%) NS

470 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 159 1999 2). Only four patients were enrolled in a DOT program; one case of TB completed treatment and three controls were lost to follow-up. At the time of diagnosis, there were no differences between the cases and controls in terms of the presence of cavitation on chest radiography (9.5% versus 25%) or positive sputum smears (66.7% versus 60.4%). With control for HIV disease, cases were more likely than controls to have extrapulmonary TB, with or without pulmonary involvement (47.6% versus 11.6%, p 0.05, OR 2.9, 95% CI 1.02 to 8.1), and were less likely than controls to have only pulmonary TB (p 0.05, OR 0.3, 95% CI 0.1 to 0.98). HIV serostatus. Seventeen cases (81%) and 22 controls (45.8%) were known to be HIV positive (p 0.03, OR 10.9, 95% CI 1.1 to undefined). HIV status was unknown for three cases (14.3%) and 14 controls (29.2%). Median CD4 cell counts were similar for cases and controls (57/mm 3 versus 83/mm 3 ). Of the 17 HIV-positive cases, eight (47%) had had prior opportunistic infections, as compared with three of 22 (14%) of the HIV-positive controls (p 0.01, estimated OR 8.3, 95% CI 1.5 to 44.7). As shown in Table 2, concurrent opportunistic infections, including thrush, were no more likely in HIV-positive cases (10 of 17, 59%) than in controls (11 of 22, 50%). Concurrent opportunistic infections, excluding thrush, were present in six of 17 (35%) cases and six of 22 (27%) controls (p 0.046, estimated OR 10.7, 95% CI 0.04 to 4.7). Oral thrush without any other concurrent opportunistic infection was seen in four of 17 (24%) cases as compared with five of 22 (23%) controls (p NS). No HIV-related neoplasms were diagnosed in cases or controls. None of the study patients had a documented history of rifabutin use prior to the diagnosis of TB. Hospital stay and survival. All of the patients and all but three of the controls had been hospitalized; the latter three controls were treated exclusively as outpatients. The mean length of stay for cases was 52 d and that for controls was 34 d. With control for HIV status, the difference still approached statistical significance (p 0.07). Four of 21 cases (19%) and one of 48 controls (2.1%) died during this hospitalization (p 0.02, OR 14.3, 95% CI 1.6 to 127). All five of these cases were HIV positive, with CD4 cell counts 60, and all had extrapulmonary involvement. Four of the five had both pulmonary and extrapulmonary TB. The four patients with pulmonary disease had positive sputum smears but did not have cavitary lesions. Malabsorption/medication interactions. For the 24 patients with a prior history of TB, information about medications and gastrointestinal symptoms recorded during the prior episode of TB was incomplete; only eight (62%) cases and five (45%) controls were diagnosed at Bellevue Hospital. The remaining patients received care at nine other New York City hospitals and jails. In spite of this, five of the nine cases with acquired RMR-TB were identified as having evidence of malabsorption, and one case had been treated with fluconazole and ketaconazole during the previous treatment period. No controls having these risk factors were identified. Analysis of Cases of TB Of the 21 cases of TB, eight (38%) were classified as having primary resistance, nine (43%) had acquired (secondary) resistance, and four (19%) had an unknown mode of acquisition of resistance. The three groups were similar in terms of demographics and HIV status. Among the 13 cases with a history of TB, RMR emerged from 6 to 468 mo (median: 14 mo) after the previous diagnosis of TB. Eight of the 13 cases (53.8%) with a history of TB were noncompliant with previous treatment. Diarrhea and/or vomiting during treatment for the previous episode of TB were reported in five of the 13 (38.4%) cases. One noncompliant patient was documented as having taken fluconazole and ketaconazole during previous treatment for TB. Primary RMR. Among the eight cases with primary RMR, all had a history of living in at least one type of congregate setting (jail, prison, homeless shelter, SRO hotel). Seven had a history of incarceration, and four had a history of IVDU. Of the six patients whose HIV status was known, all were HIV positive. Acquired (secondary) resistance. Of the nine cases with acquired resistance, all had from one to three risk factors for RMR. Eight were HIV seropositive and seven were noncompliant with previous TB treatment. The two compliant patients had documented vomiting and/or diarrhea during treatment of the episode of susceptible TB. Unknown mode of acquisition of resistance. Three of the four patients whose mode of acquisition of resistant TB was unknown had been compliant with previous treatment, and two of these three patients were HIV positive. The fourth patient was noncompliant and HIV positive, with a CD4 cell count of 41/mm 3. The HIV-negative patient had no apparent risks for RMR, but his medical history was significant for multiple hospitalizations for complications of paraplegia; he received multiple rotating antibiotics for complications of a neurogenic bladder. DISCUSSION We have shown that RMR-TB is strongly associated with HIV infection. Furthermore, characteristics of patients with TB associated with RMR were extrapulmonary TB and a history of TB. There was a trend toward a more frequent history of incarceration in patients with RMR than of controls with drugsusceptible disease. The gravity of the problem of RMR-TB is underscored by the longer hospital stays of patients with RMR-TB and their greater likelihood of dying than of persons with drug-susceptible disease. Although a non-rifampin containing regimen of INH, streptomycin, and pyrazinamide achieved a high cure rate and a relapse rate in the 5 to 6% range in the era before acquired immune deficiency syndrome (AIDS), it is unclear whether similar results are achievable in HIV-infected persons (13). Survival from multidrug-resistant tuberculosis (MDR-TB) in patients with AIDS is much poorer than in immunocompetent hosts (14); our current study indicates that the inability to use rifampin, even when INH is effective, may be responsible for a good part of this excess mortality. Although there were no differences in our study between cases and controls in terms of treatment noncompliance, cases with RMR were highly noncompliant. Very few of the cases or controls had been enrolled in programs of DOT, because this treatment strategy was not as widely used in New York City in the early 1990s as it is today. This may account for the similarly high levels of noncompliance with tuberculosis medication regimens of cases and controls. TB patients at Bellevue Hospital during the study period were frequently homeless and had a history of IVDU. These and other demographic characteristics were not predictive of RMR-TB. It is likely that the mechanisms of action of the risk factors that we identified as being associated with RMR-TB are multifactorial and complex. There are several ways in which HIV infection could lead to a greater incidence of RMR-TB. One mechanism for primary drug resistance is likely to result from the more rapid development of active TB in HIV-seropositive patients infected with M. tuberculosis (15). If an HIV-positive cohort of persons is infected with an RMR strain of the organ-

Sandman, Schluger, Davidow, et al.: Rifampin-Resistance Risk Factors 471 ism, relatively more cases of RMR-TB will develop than would have developed if an HIV-seronegative cohort had been exposed. In this manner, RMR strains may propagate through the community, and the disease is likely to persist unless prompt institution of effective chemotherapy is widely available. This will obviously require not only rapid diagnosis of TB but also rapid drug-susceptibility testing of M. tuberculosis (at least for rifampin). Recent studies indicate that this may well be feasible. Several mechanisms whereby HIV-infected persons develop secondary drug resistance are also apparent; these include malabsorption of drugs as well as drug drug interactions that could lead to altered levels of antituberculous medications. Malabsorption of drugs in HIV-seropositive patients has been well described in persons with apparent diarrhea and weight loss, but recent evidence suggests that HIVinfected patients without clinical malabsorption or wasting may also have altered uptake and pharmacokinetics of several antituberculous medications (16). If in such cases, INH and ethambutol were not well absorbed, preferential exposure to rifampin might have occurred and led to acquired drug resistance. Drug drug interactions may also play a role in the development of RMR-TB. Compounds that prolong the half-life of rifampin might lead to RMR, particularly if serum levels of other agents are diminished (by malabsorption or still other drug drug interactions). Several compounds, such as the protease inhibitor ritonavir, raise rifampin levels (17). If other drugs reduce serum levels of INH, ethambutol, and pyrazinamide, a situation favorable for the development of RMR may occur. Such mechanisms may underlie the recently reported development of RMR in HIV-positive patients receiving rifapentine, a long-acting rifamycin (18). Because many patients with HIV infection receive multiple medications, the potential for drug drug interactions in this group of patients is enormous, and it is likely that many of these interactions have not yet been well characterized. The identification of extrapulmonary TB as a risk factor for RMR-TB was another outcome of our study. There are two possible mechanisms by which extrapulmonary TB may be linked to rifampin-resistant TB. It is possible that extrapulmonary disease in the setting of HIV infection signifies advanced immunosuppression, and the complex interactions between the cellular immune system, the invading mycobacterium, and antituberculous agents may be altered in such a way as to favor the development of drug-resistant TB. Certainly HIV has been associated with the development of MDR-TB in previous studies. Another possibility is that antituberculosis drugs are delivered at varying concentrations to extrapulmonary compartments, leading to local conditions that may favor the development of RMR-TB (19). Although extrapulmonary TB is generally considered no different than pulmonary TB in terms of drug therapy and duration of treatment, this may not be the case in patients with advanced immunosuppression. Our study confirms and extends the understanding of the development of RMR-TB gained in the recent studies of Munsiff and Ridzon and their coworkers (12, 20). We found, as did Ridzon and colleagues, that prior TB was strongly associated with RMR-TB. In addition, we confirmed the association between RMR-TB and HIV infection noted in these two prior studies. We also found that prior opportunistic infections had occurred more commonly in HIV-positive RMR-TB patients than in controls. Median CD4 cell counts in our study were similar to those reported by Ridzon and coworkers in their cohort of patients, although the difference in cell counts between cases and controls in our cohort did not achieve statistical significance. A novel finding in our study was that extrapulmonary TB, with or without pulmonary disease, was more common in RMR-TB cases than in controls, although as discussed earlier, the nature of the association between extrapulmonary disease and RMR is unclear. In sum, RMR-TB seems strongly associated with HIV infection and extrapulmonary TB. Though the exact mechanisms associated with the development of RMR-TB are not clear, we have described several possible routes to the development of this syndrome. The high levels of morbidity and mortality associated with RMR-TB underscore the need for its prompt diagnosis and for drug-susceptibility testing in this condition, so that appropriate treatment can be instituted rapidly. In combination with the provision of DOT, this strategy should lead to improved outcomes for individual patients as well as less spread of the disease in the community. References 1. Schluger, N., T. Harkin, and W. Rom. 1996. Principles of therapy of tuberculosis. In W. N. Rom and S. Garay, editors. Tuberculosis. Little Brown, Boston. 2. Moore, M., I. M. Onorato, E. McCray, and K. G. Castro. 1997. Trends in drug-resistant tuberculosis in the United States, 1993 1996. J.A.M.A. 278:833 837. 3. Small, P. M., R. W. Shafer, P. C. Hopewell, S. P. Singh, M. J. Murphy, E. Desmond, M. F. Sierra, and G. K. Schoolnik. 1993. Exogenous reinfection with multidrug-resistant Mycobacterium tuberculosis in patients with advanced HIV infection. N. Engl. J. Med. 328:1137 1144. 4. Perlman, D. C., and N. Salomon. 1996. Selective malabsorption and the origin of mono-rifampin-resistant Mycobacterium tuberculosis (letter). Am. J. Respir. Crit. Care Med. 154:1213. 5. Mitchison, D. A. 1998. How drug resistance emerges as a result of poor compliance during short course chemotherapy for tuberculosis. Int. J. Tuberc. Lung Dis. 2:10 15. 6. Mahmoudi, A., and M. D. Iseman. 1993. Pitfalls in the care of patients with tuberculosis: common errors and their association with the acquisition of drug resistance. J.A.M.A. 270:65 68. 7. March, F., X. Garriga, P. Rodriguez, C. Moreno, M. Garrigo, P. Coll, and G. Prats. 1997. Acquired drug resistance in Mycobacterium tuberculosis isolates recovered from compliant patients with human immunodeficiency virus-associated tuberculosis. Clin. Infect. Dis. 25:1044 1047. 8. Frieden, T. R., T. Sterling, A. Pablos-Mendez, J. O. Kilburn, G. M. Cauthen, and S. W. Dooley. 1993. The emergence of drug-resistant tuberculosis in New York City. N. Engl. J. Med. 328:521 526. 9. Bishai, W. R., N. M. H. Graham, S. Harrington, C. Page, K. Moore-Rice, N. Hooper, and R. E. Chaisson. 1996. Rifampin-resistant tuberculosis in a patient receiving rifabutin prophylaxis. N. Engl. J. Med. 334:1573 1576. 10. Lutfey, M., P. Della-Latta, V. Kapur, L. A. Palumbo, D. Gurner, G. Stotzky, K. Brudney, J. Dobkin, A. Moss, J. M. Musser, and B. N. Kreiswirth. 1996. Independent origin of mono-rifampin-resistant Mycobacterium tuberculosis in patients with AIDS. Am. J. Respir. Crit. Care Med. 153:837 840. 11. Nolan, C. M., D. L. Williams, M. D. Cave, K. D. Eisenach, H. el-hajj, T. M. Hooton, R. L. Thompson, and S. V. Goldberg. 1995. Evolution of rifampin resistance in human immunodeficiency virus-associated tuberculosis. Am. J. Respir. Crit. Care Med. 152:1067 1071. 12. Munsiff, S. S., S. Joseph, A. Ebrahimzadeh, and T. R. Frieden. 1997. Rifampin-monoresistant tuberculosis in New York City, 1993 1994. Clin. Infect. Dis. 25:1465 1467. 13. Hong Kong Chest Service/British Medical Research Council. 1975. Controlled trial of 6- and 9-month regimens of daily and intermittent streptomycin plus isoniazid plus pyrazinamide for pulmonary tuberculosis in Hong Kong. Tubercle 56:81 96. 14. Park, M. M., A. L. Davis, N. W. Schluger, H. Cohen, and W. N. Rom. 1996. Outcome of MDR-TB patients, 1983 1993: prolonged survival with appropriate therapy. Am. J. Respir. Crit. Care Med. 153:317 324. 15. Selwyn, P. A., D. Hartel, V. A. Lewis, E. E. Schoenbaum, S. H. Vermund, R. S. Klein, A. T. Walker, and G. H. Friedland. 1989. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. N. Engl. J. Med. 320:545 550.

472 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 159 1999 16. Peloquin, C. A., A. T. Nitta, W. J. Burman, K. F. Brudney, J. R. Miranda- Massari, M. E. McGuinness, S. E. Berning, and G. T. Gerena. 1996. Low antituberculosis drug concentrations in patients with AIDS. Ann. Pharmacother. 30:919 925. 17. Center for Disease Control. 1996. Clinical update: impact of HIV protease inhibitors on the treatment of HIV-infected tuberculosis patients with rifampin. Morbid. Mortal. Wkly. Rep. 45:921 925. 18. Vernon, A., A. Khan, L. Bozeman, and Y. Wang. 1998. Update on U.S. Public Health Service (USPHS) study 22: a trial of once weekly isoniazid and rifapentine in the continuation phase of TB treatment (abstract). Am. J. Respir. Crit. Care Med. 157:A467. 19. Lipstich, M., and B. R. Levin. 1998. Population dynamics of tuberculosis treatment: mathematical models of the roles of non-compliance and bacterial heterogeneity in the evolution of drug resistance. Int. J. Tuberc. Lung Dis. 2:187 199. 20. Ridzon, R., C. G. Whitney, M. T. McKenna, J. P. Taylor, S. H. Ashkar, A. T. Nitta, S. M. Harvey, S. Valway, C. Woodley, R. Cooksey, and I. M. Onorato. 1998. Risk factors for rifampin mono-resistant tuberculosis. Am. J. Respir. Crit. Care Med. 157:1881 1884.