Schmaltz, Carolina Arana Stanis MD*†; Sant'Anna, Flávia Marinho MD*; Neves, Simone Carvalho MSc*; Velasque, Luciane de Souza PhD*; Lourenço, Maria Cristina MSc‡; Morgado, Mariza Gonçalves MD, PhD§; Rolla, Valéria Cavalcanti MD, PhD*; Lopes, Guilherme Santoro MD, PhD†
Tuberculosis (TB) is one of the leading causes of death around the world.1 HIV pandemic has had a major impact on the incidence and outcome of TB, particularly in resource-limited countries. It was estimated, in the year 2000, that 9% of TB cases and 12% of TB deaths around the world were attributable to HIV infection.2
Highly active antiretroviral therapy (HAART) has been associated with significant reduction in TB incidence3-9 and in TB-associated mortality among HIV-infected patients.10-15 However, the management of TB in HIV patients (TB/HIV) is associated with complex issues such as drug interactions with antiretroviral (ARV) drugs, the occurrence of immune reconstitution syndrome (IRIS),16-19 and a high risk of drug toxicity.16-21 These factors may partially counteract the expected beneficial effect of access to HAART on the excess risk of mortality among HIV-infected patients treated for TB. Data on the influence of HIV infection on TB-associated mortality in settings with access to HAART are scarce. The results of 2 studies conducted in developed countries22,23 suggest that HIV infection might be still associated with a considerable increase of the risk of death among subjects treated for TB. Although access to HAART is gradually scaling up in developing countries, we are not aware of any study assessing the influence of HIV serostatus on TB mortality, after HAART became available to patients living in these countries.
Brazil was the first country with high TB burden to provide universal access to HAART to eligible HIV-infected patients. Despite the considerable reduction of TB incidence among seropositive patients treated with HAART in this country,8 a recent study carried out in Rio de Janeiro suggested that TB still remains the leading cause of death among seropositive subjects.24 In this study, we analyzed the influence of HIV serostatus on mortality among patients with culture-proven TB managed at a referral center in Rio de Janeiro, Brazil.
Population and Design
This prospective cohort study was conducted at the Instituto de Pesquisa Clínica Evandro Chagas (IPEC, Fiocruz) in Rio de Janeiro, Brazil. IPEC is an institution for infectious diseases management, with integrated TB and HIV care programs. Patients with known HIV serologic status who started antituberculous therapy from April 14, 2000, to July 7, 2005, were included in the study if they had TB diagnosis confirmed by culture and signed a written informed consent. Patients were followed up until July 24, 2006. This study was approved by the Committee on Ethics in Research of IPEC.
Data Collection and Follow-Up
Follow-up start was determined by the date of the first prescription of antituberculous therapy. Data collected in this baseline visit included sex, age, monthly income, education, history of incarceration, psychiatric disease, alcohol abuse (screened with the CAGE questionnaire),25 use of illicit drugs, weight loss, clinical presentation of TB, and the results of the following laboratorial tests: microscopic examination of sputum; culture of sputum, blood, and other clinical specimens for mycobacteria and fungal pathogens; serologic tests for Histoplasma capsulatum, Paracoccidioides brasiliensis, viral hepatitis, toxoplasmosis, and syphilis; latex assay for cryptococcal antigen, blood cell counts, serum levels of creatinine, liver enzymes, and albumin. Additional baseline data collected for HIV-infected patients included date of first positive HIV serology, history of opportunistic diseases, current and previous ARV therapy regimens, and results of CD4 cell count and HIV viral load within the preceding 6 months.
Follow-up visits were scheduled on day 15, 30, and 60 after the initial visit and bimonthly thereafter. Data recorded in these visits included: clinical manifestations associated with TB or HIV infection, the occurrence of IRIS, adverse events related to HAART or antituberculous therapy, and the results of blood cell counts and serum levels of liver enzymes.
Ziehl-Neelsen staining was used for the detection of acid-fast bacilli in sputum smears and biopsy specimens. Sputum and biopsy samples were cultured in Löwenstein-Jensen medium. Culture of blood specimens used the lysis-centrifugation method. Drug susceptibility tests for rifampicin, isoniazid, pyrazinamide, ethambutol, streptomycin, and ethionamide were performed by the agar proportion method.
The primary outcome of the study was death related to TB. The secondary outcome was death due to any cause.
HIV infection was diagnosed in patients with a positive anti-HIV enzyme-linked immunosorbent assay test confirmed by Western blot or immunofluorescence assay.
The clinical presentation of TB was classified as pleural-pulmonary (when restricted to the lungs and/or pleura), extrapulmonary localized (when just one extrapulmonary site was affected), or disseminated (involving spleen, liver, bone marrow, or at least 2 noncontiguous sites). Mycobacterium tuberculosis isolates resistant to at least rifampicin and isoniazid were defined as multidrug resistant.
IRIS was defined as a documented worsening of signs or symptoms of TB during appropriate antituberculous treatment and after the initiation of ARV therapy, not explained by any other disease or by an adverse effect of drug therapy.
Patients who failed to attend follow-up visits and to complete anti-TB therapy were defined as defaulted. Defaulted patients whose vital status at the study final date could not be ascertained were defined as lost to follow-up.
Death was defined as related to TB if it occurred during the course of TB treatment or after therapy default in patients with persistent clinical/microbiological evidence of TB and in the absence of any other known cause. All TB-related deaths occurred while patients were hospitalized at IPEC. Death classification as related or unrelated to TB was performed by one of the authors (C.A.S.S.), after thorough review of relevant clinical, microbiological, and pathological data.
Severe adverse events associated with drugs used for TB or HIV therapy were defined as those that required drug discontinuation.
Antituberculous and ARV Therapy
The first-line antituberculous regimen in Brazil, for both HIV-positive and HIV-negative patients, is the combination of rifampicin, isoniazid, and pyrazinamide during the 2 initial months “intensive phase”, followed by rifampicin plus isoniazid during 4 months (“continuation phase”). In cases with central nervous system TB, the continuation phase is extended to 7 months.26 Patients who developed severe hepatotoxicity with the first-line regimen were treated with a 3-month course of streptomycin, ethambutol, and ofloxacin followed by ethambutol and ofloxacin for 9 months. Cases that received antituberculous regimens without rifampicin were treated for at least 12 months. Most cases of multidrug-resistant infection were treated with the association of streptomycin (for 3 months), pyrazinamide (for 2 months), ethambutol, and ethionamide. In cases treated with HAART regimens that precluded the use of rifampicin, this drug was replaced by streptomycin (for 3 months) plus ethambutol.
ARV therapy was offered according to contemporary Brazilian National Guidelines for HIV Infection Therapy27 that were periodically updated during the study period.
Categorical variables are described by their absolute counts and percents. Numerical variables are described by their median and interquartile range (IQR). Mortality rates are described by their incidence per 100 patient-years. The rate ratio (RR) for TB-related mortality according to HIV serostatus was calculated with the EPITABLE program of Epi-Info 6.04 (Centers for Disease Control and Prevention, Atlanta, GA). The χ2 and Fisher exact test were used to compare the distribution of categorical covariates. The distribution of continuous covariates was compared by the Mann-Whitney test. Variables associated with a P value <0.10 were selected for further analyses using univariate and multivariate Cox regression models, with TB-related mortality as the outcome variable. In these analyses, HAART use was modeled as a time-dependent variable with an intention-to-treat approach. Seropositive patients who were already treated with HAART at baseline were considered to be on HAART throughout the study period. Patients who started HAART after being included in this study were considered to be on HAART from the first day of its prescription up to the end of their follow-up. Seropositive and seronegative subjects who completed antituberculous therapy had their follow-up censored at the end of treatment. Patients who did not complete therapy but whose vital status could be ascertained had their follow-up censored at the last date they had been contacted by the research team during the study period. The follow-up of subjects who were lost was censored at their last medical visit. Backward stepwise selection of covariates was used in the multivariate analyses. Variables associated with a P value ≤0.15 were retained in the final multivariate model. The calculated hazard ratios (HRs) and their 95% confidence intervals (95% CI) are presented. Two-tailed P values were calculated with the Wald test. These analyses were performed with the R version 2.7.1 statistical package (R Foundation for Statistical Computing).
Demographic and Other Baseline Characteristics
During the study period, 387 patients started antituberculous therapy. The diagnosis of TB was confirmed by culture in 208 cases. One patient who did not perform HIV serology was excluded from the study. Thus, 207 patients (106 HIV-positive patients and 101 HIV-negative patients) were included in the study.
The distribution of baseline and follow-up covariates according to HIV serostatus is shown in Table 1. HIV-positive patients were significantly younger than HIV-negative subjects (P = 0.003). Disseminated TB (P < 0.001), weight loss >10% (P < 0.001), hemoglobin level ≤10 g% (P < 0.001), and serum albumin ≤3 g% (P = 0.032) were also significantly more frequent among HIV-positive patients. Multidrug resistance to antituberculous drugs tended to be more frequent among seropositive subjects (P = 0.07), whereas a previous diagnosis of psychiatric illness tended to be more frequent among HIV-negative patients (P = 0.06). The median time elapsed between the start of TB symptoms and the beginning of antituberculous therapy was long for both groups but tended to be shorter among HIV-infected subjects (84 days, IQR: 41-120 days) than among seronegative patients (90 days, IQR: 60-180 days, P = 0.06).
Twenty-two patients were not treated with rifampicin throughout the whole course of antituberculous therapy. Five of these cases switched to regimens without rifampicin because of an eventual diagnosis of drug resistance. Five seropositive patients did not use rifampicin from the beginning due to predictable interactions with ARV drugs. In other 12 subjects, rifampicin was not used throughout on account of severe adverse effects. The proportion of patients with susceptible TB infection not treated with rifampicin throughout was significantly larger among HIV-seropositive patients. (Table 1, P = 0.007).
The occurrence of severe adverse effects attributed to antituberculous drugs was also more common among coinfected patients (Table 1, P < 0.001). On the other hand, the frequency of treatment default was significantly higher in the HIV-negative group (P = 0.014).
At baseline, 34 patients were already on HAART. Of these, 27 had been treated with HAART for more than 3 months. Other 49 patients started HAART after a median follow-up of 43 days (IQR: 28-74 days) on antituberculous therapy. Thus, 83 patients (82.2%) received HAART during the course of antituberculous treatment. The most frequent regimens used were the combination of 2 nucleoside reverse transcriptase inhibitors with efavirenz (65 patients) or with saquinavir and ritonavir (12 patients). Two subjects received dual nucleoside reverse transcriptase inhibitor therapy. IRIS occurred in 7 coinfected patients. None of these patients died during the study.
Baseline CD4 lymphocyte count and viral load were available for 80 seropositive patients. CD4 cell counts tended to be higher among 24 patients treated with HAART for at least 3 months (median: 169 cells/mm3, IQR: 136-355) than in the remaining 56 patients (median: 125 cells/mm3, IQR: 64-214, P = 0.07). Viral load was significantly lower among patients on HAART for at least 3 months (median: 4.15 log copies/mL, IQR: 2.9-4.95) compared with the other patients (median: 5.36 log copies/mL, IQR: 4.99-5.98, P < 0.001). Baseline viral load was undetectable in only 5 patients treated with HAART.
Patients were followed for a total of 1976 patient-months. The median follow-up was 6.3 months for HIV-seronegative patients and 6.7 months for seropositive subjects (P = 0.64). Four patients were lost to follow-up. All these patients were HIV negative (P = 0.12).
Twenty-one HIV-positive subjects (20%) and 4 seronegative patients (4%) died during the study (P < 0.001). Deaths were unrelated to TB in 2 HIV-negative patients (pulmonary thromboembolism and disseminated paracoccidioidomycosis). Thus, there were 21 deaths related to TB in HIV-positive subjects (24.7 deaths per 100 patient-years) and 2 (2.5 deaths per 100 patient-years) among seronegative patients (RR = 9.76, 95% CI: 2.29 to 41.6, P < 0.001). Deaths related to TB occurred after a median follow-up of 3 months (IQR: 0.2-7.8 months). In a sensitivity analysis in which all patients lost to follow-up were considered as having died on the day after their last visit, TB-related mortality remained significantly higher among coinfected patients (RR = 3.25, P = 0.005).
Table 2 shows the results of univariate and multivariate Cox regression analyses. Because of the small number of patients with psychiatric disease, the influence of this variable on TB-related mortality could not be analyzed. In time-dependent univariate analyses, TB-related mortality was significantly lower in HIV-negative subjects as compared with seropositive subjects whether these were using HAART (P = 0.014) or not (P = 0.001). Mortality was also significantly higher among subjects who presented disseminated TB (P < 0.001), who had susceptible TB infection but did not use rifampicin throughout treatment period (P = 0.005), and who had baseline hemoglobin concentration ≤10 g% (P = 0.009) or serum albumin ≤3 g% (P = 0.016). The occurrence of severe adverse reactions to antituberculous drugs tended to be associated with a worse prognosis (P = 0.09).
In an analysis that included only the subset of HIV-infected subjects, TB-related mortality tended to be higher among patients not treated with HAART. There were 10 deaths (46%) among subjects not on HAART and 11 (13%) in patients treated with HAART (HR = 0.58; 95% CI: 0.33 to 1.03, P = 0.06). The frequency of baseline hemoglobin concentration ≤10 g% tended to be higher among patients treated with HAART (50 cases, 66%) as compared with the other seropositive patients (8 cases, 40%, P = 0.09). There were no significant differences between subjects treated with HAART and the untreated patients with regard to the proportion of patients who had disseminated TB (35% and 39%, respectively, P = 0.8), baseline serum albumin ≤3 g% (41% and 40%, respectively, P = 1.0), severe adverse events related to antituberculous drugs (35% for both groups, P = 1.0), or who did not use rifampicin throughout antituberculous therapy (16% and 12%, respectively, P = 0.63).
In the final adjusted regression model, mortality of coinfected patients not treated with HAART remained significantly higher compared with the mortality observed among HIV-negative patients (HR = 6.3, P = 0.024). TB-related mortality in coinfected patients treated with HAART was also higher than in seronegative patients (HR = 3.48), but the difference was not statistically significant (P = 0.14). Disseminated TB was associated with a significantly higher risk of mortality (P = 0.04). Patients not treated with rifampicin throughout (P = 0.11) also tended to have a worse outcome.
This study compared mortality between HIV seropositive and seronegative subjects treated for TB at a referral center with well-trained personnel, well-integrated HIV and TB programs, and good laboratorial support. By the end of the second month of antituberculous therapy, most seropositive patients had already started HAART. Despite these favorable conditions at the study site, HIV infection was still associated with a substantial impact on the outcome of patients treated for TB.
In keeping with the results of previous studies,10-15 mortality tended to be lower among seropositive subjects treated with HAART. However, these subjects still had a significantly worse outcome compared with HIV-negative subjects, with an estimated 6.6 increase in the risk of TB-related mortality in unadjusted analyses. In the final regression model, adjusted for the use of antituberculous regimens without rifampicin and the presence of disseminated TB, the estimated hazard ratio for TB-related mortality among seropositive patients treated with HAART compared with seronegative subjects was reduced to 3.48. The fact that this result lacked statistical significance was possibly due to the relatively small sample size analyzed. Nonetheless, these findings suggest that the persistent excess risk of TB-related mortality in HIV patients treated with HAART might be in large part explained by the higher baseline clinical severity among seropositive patients and, possibly, by the more frequent use of less effective antituberculous regimens.
The main reason for the more frequent use of antituberculous regimens without rifampicin among seropositive subjects was the increased incidence of drug toxicity in these patients, a fact that has been observed elsewhere.20,21 Nunn et al28 found that the use of antituberculous regimens without rifampicin was associated with shorter survival among HIV patients not treated with HAART. Although the increase in mortality observed among patients treated with antituberculous regimens without rifampicin did not reach statistical significance (P = 0.11) in the final adjusted model, possibly because the sample size was relatively small to address this specific issue, our results suggest that, even in the context of wide access to HAART, treatment with less effective antituberculous regimens may negatively influence the outcome of HIV-infected patients.
The significant association of disseminated TB with lower survival has been demonstrated in other studies.29-32 The prevalence of disseminated TB among HIV-seropositive patients in this study was relatively high as compared with other reports,10,14,33 probably because many of these patients were referred to our center after seeking for a long time medical attention in primary care units with limited diagnostic resources. The long duration of symptoms before the start of antituberculous therapy in both HIV-seropositive patients and HIV-seronegative patients was probably a consequence of limited access to health care in poor communities in Rio de Janeiro, Brazil, caused by structural and social factors, including the rise of urban violence.34 Further reduction in the impact of HIV infection on TB-related mortality in this population will possibly depend on continuing efforts to improve the quality of primary care for both TB and HIV infection, especially in underprivileged communities.
The fact that all deaths observed during antituberculous therapy among HIV-infected subjects were related to TB contrasts with necropsy data reported from several studies that have shown that TB/HIV patients frequently die as a consequence of other opportunistic diseases, especially after the first few months of antituberculous therapy.28,35-38 Necropsies were not performed in this study and, therefore, the primary cause of death in our patients could not be fully ascertained. Nevertheless, all deaths defined as TB related were associated with clinical or microbiological evidence of TB persistence indicating that this infection was at least a contributing cause. It is also important to consider that factors other than misdiagnosing the cause of death may justify our finding. In this study, most of HIV infected were already on HAART by the end of the second month of antituberculous therapy, cotrimoxazole prophylaxis was routinely prescribed, and the study site was well structured for the diagnosis and treatment of other opportunistic diseases, especially fungal diseases. Such conditions that were not present in the aforementioned studies28,35-38 might have decreased the probability of dying from other complications related to HIV infection.
In conclusion, we found that HIV infection still has a substantial impact on TB-related mortality in the context of wide access to HAART in a middle-income country. Apparently, the persistent excess of TB-related mortality among seropositive patients treated with HAART was in large part explained by the higher clinical severity at presentation and by the more frequent use of regimens without rifampicin owing to severe drug toxicity and predictable drug interactions. Further reduction on the impact of HIV infection on TB-related mortality may depend on continuing efforts to provide better access and improved quality of care for both TB and HIV infection and on progresses in the management of antituberculous therapy in coinfected patients.
1. Murray CJ, Lopez AD. Mortality by cause for eight regions of the world: Global Burden of Disease Study. Lancet
2. Corbett EL, Watt CJ, Walker N, et al. The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Arch Intern Med
3. Girardi E, Antonucci G, Vanacore P, et al. Impact of combination antiretroviral therapy on the risk of tuberculosis among persons with HIV infection. AIDS
4. Jones JL, Hanson DL, Dworkin MS, et al. HIV-associated tuberculosis in the era of highly active antiretroviral therapy. The Adult/Adolescent Spectrum of HIV Disease Group. Int J Tuberc Lung Dis
5. Badri M, Wilson D, Wood R. Effect of highly active antiretroviral therapy on incidence of tuberculosis in South Africa: a cohort study. Lancet
6. Lawn SD, Badri M, Wood R. Tuberculosis among HIV-infected patients receiving HAART: long term incidence and risk factors in a South African cohort. AIDS
7. Brinkhof MW, Egger M, Boulle A, et al. Tuberculosis after initiation of antiretroviral therapy in low-income and high-income countries. Clin Infect Dis
8. Miranda A, Morgan M, Jamal L, et al. Impact of antiretroviral therapy on the incidence of tuberculosis: the Brazilian experience, 1995-2001. PLoS ONE
9. Muga R, Ferreros I, Langohr K, et al. Changes in the incidence of tuberculosis in a cohort of HIV-seroconverters before and after the introduction of HAART. AIDS
10. Girardi E, Palmieri F, Cingolani A, et al. Changing clinical presentation and survival in HIV-associated tuberculosis after highly active antiretroviral therapy. J Acquir Immune Defic Syndr
11. Carvalho AC, Nunes ZB, Martins M, et al. Clinical presentation and survival of smear-positive pulmonary tuberculosis patients of a university general hospital in a developing country. Mem Inst Oswaldo Cruz
12. Garcia de Olalla P, Martinez-Gonzalez MA, Cayla JA, et al. Influence of highly active anti-retroviral therapy (HAART) on the natural history of extra-pulmonary tuberculosis in HIV patients. Int J Tuberc Lung Dis
13. Dheda K, Lampe FC, Johnson MA, et al. Outcome of HIV-associated tuberculosis in the era of highly active antiretroviral therapy. J Infect Dis
14. Manosuthi W, Chottanapand S, Thongyen S, et al. Survival rate and risk factors of mortality among HIV/tuberculosis-coinfected patients with and without antiretroviral therapy. J Acquir Immune Defic Syndr
15. Nahid P, Gonzalez LC, Rudoy I, et al. Treatment outcomes of patients with HIV and tuberculosis. Am J Respir Crit Care Med
16. Narita M, Ashkin D, Hollender ES, et al. Paradoxical worsening of tuberculosis following antiretroviral therapy in patients with AIDS. Am J Respir Crit Care Med
17. Navas E, Martin-Davila P, Moreno L, et al. Paradoxical reactions of tuberculosis in patients with the acquired immunodeficiency syndrome who are treated with highly active antiretroviral therapy. Arch Intern Med
18. Breen RA, Smith CJ, Bettinson H, et al. Paradoxical reactions during tuberculosis treatment in patients with and without HIV co-infection. Thorax
19. Lawn SD, Myer L, Orrell C, et al. Early mortality among adults accessing a community-based antiretroviral service in South Africa: implications for programme design. AIDS
20. Lanternier F, Dalban C, Perez L, et al. Tolerability of anti-tuberculosis treatment and HIV serostatus. Int J Tuberc Lung Dis
21. Pukenyte E, Lescure FX, Rey D, et al. Incidence of and risk factors for severe liver toxicity in HIV-infected patients on anti-tuberculosis treatment. Int J Tuberc Lung Dis
22. Cayla JA, Caminero JA, Rey R, et al. Current status of treatment completion and fatality among tuberculosis patients in Spain. Int J Tuberc Lung Dis
23. Sterling TR, Zhao Z, Khan A, et al. Mortality in a large tuberculosis treatment trial: modifiable and non-modifiable risk factors. Int J Tuberc Lung Dis
24. Saraceni V, King BS, Cavalcante SC, et al. Tuberculosis as primary cause of death among AIDS cases in Rio de Janeiro, Brazil. Int J Tuberc Lung Dis
25. Ewing JA. Detecting alcoholism. The CAGE questionnaire. JAMA
26. Castelo A, Kritski AL, Barreto AW, et al. II Consenso Brasileiro de Tuberculose. J. Bras Pneumol
. 2004;30(Suppl 1):S24-S37.
27. Ministério da Saúde do Brasil. Recomendações Para Terapia Anti-Retroviral Em Adultos Infectados Pelo HIV
. 7a ed. Brasilia, Brazil: Programa Nacional de DST e Aids; 2008.
28. Nunn P, Brindle R, Carpenter L, et al. Cohort study of human immunodeficiency virus infection in patients with tuberculosis in Nairobi, Kenya. Analysis of early (6-month) mortality. Am Rev Respir Dis
29. Elliott AM, Halwiindi B, Hayes RJ, et al. The impact of human immunodeficiency virus on mortality of patients treated for tuberculosis in a cohort study in Zambia. Trans R Soc Trop Med Hyg
30. Whalen C, Horsburgh CR Jr, Hom D, et al. Site of disease and opportunistic infection predict survival in HIV-associated tuberculosis. AIDS
31. Gonzalez OY, Adams G, Teeter LD, et al. Extra-pulmonary manifestations in a large metropolitan area with a low incidence of tuberculosis. Int J Tuberc Lung Dis
32. Kwara A, Roahen-Harrison S, Prystowsky E, et al. Manifestations and outcome of extra-pulmonary tuberculosis: impact of human immunodeficiency virus co-infection. Int J Tuberc Lung Dis
33. DeRiemer K, Soares EC, Dias SM, et al. HIV testing among tuberculosis patients in the era of antiretroviral therapy: a population-based study in Brazil. Int J Tuberc Lung Dis
34. Souza FB, Villa TC, Cavalcante SC, et al. Peculiarities of tuberculosis control in a scenario of urban violence in a disadvantaged community in Rio de Janeiro, Brazil. J Bras Pneumol
35. Greenberg AE, Lucas S, Tossou O, et al. Autopsy-proven causes of death in HIV-infected patients treated for tuberculosis in Abidjan, Cote d'Ivoire. AIDS
36. Murray J, Sonnenberg P, Shearer SC, et al. Human immunodeficiency virus and the outcome of treatment for new and recurrent pulmonary tuberculosis in African patients. Am J Respir Crit Care Med
37. Churchyard GJ, Kleinschmidt I, Corbett EL, et al. Factors associated with an increased case-fatality rate in HIV-infected and non-infected South African gold miners with pulmonary tuberculosis. Int J Tuberc Lung Dis
38. Gutierrez EB, Zanetta DM, Saldiva PH, et al. Autopsy-proven determinants of death in HIV-infected patients treated for pulmonary tuberculosis in Sao Paulo, Brazil. Pathol Res Pract
© 2009 Lippincott Williams & Wilkins, Inc.