Secondary Logo

Journal Logo

Original Contribution

Antiretroviral Therapy for HIV-Infected Tuberculosis Patients Saves Lives but Needs to Be Used More Frequently in Thailand

Sanguanwongse, Natpatou MD*; Cain, Kevin P MD; Suriya, Patcharin MS; Nateniyom, Sriprapa MD§; Yamada, Norio MD; Wattanaamornkiat, Wanpen MS; Sumnapan, Surin MD#; Sattayawuthipong, Wanchai MD**; Kaewsa-ard, Samroui BS*; Ingkaseth, Sakon MD††; Varma, Jay K MD*‡

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: June 1, 2008 - Volume 48 - Issue 2 - p 181-189
doi: 10.1097/QAI.0b013e318177594e



Click on the links below to access all the ArticlePlus for this article.

Please note that ArticlePlus files may launch a viewer application outside of your web browser.


Tuberculosis (TB) is one of the most common causes of death among HIV-infected persons worldwide.1 In Southeast Asia, the mortality rate for HIV-infected TB patients is particularly high, ranging from 20%-50% during TB treatment.2-5 Several observational studies, most done in single center and/or resource-rich settings, have documented that HIV-infected TB patients receiving antiretroviral therapy (ART) have higher rates of survival during and after TB treatment compared with patients who do not receive ART.2,6-12 The World Health Organization (WHO) currently recommends that HIV-infected TB patients who are eligible for ART begin receiving it within 2 weeks to 2 months after beginning anti-TB treatment.13

Even though ART has been documented to save lives, important questions remain about its use as part of routine, public health programs in resource-limited settings. First, it has been challenging to quantify the impact of ART at a population level. Most studies were conducted at single centers, making it difficult to generalize findings to public health programs. All these studies have been biased toward demonstrating a benefit for ART because physicians may have avoided starting ART in patients at high risk of death, that is, survival may have predicted ART use rather than ART use predicting survival.2 Second, it is not known whether ART benefits patients regardless of immune-suppression, particularly those with very low CD4+ T-lymphocyte (CD4) counts who are at high risk of death even after receiving ART.14 Third, population-level data are needed to assess the frequency with which HIV-infected TB patients actually receive ART and to identify factors associated with not receiving ART.

Thailand has a severe TB/HIV syndemic, that is, 2 diseases acting synergistically to cause excess illness and death.15,16 In 2005, approximately 580,000 persons in Thailand were living with HIV infection, and an estimated 91,000 persons developed TB.17,18 Fully one quarter of HIV-infected persons in Thailand were diagnosed with TB before learning their HIV status, and nationally, at least 12% of TB patients in Thailand are HIV infected, with some provinces reporting HIV prevalence in TB patients as high as 40%.19 HIV-infected persons in Thailand now have free, widespread access to ART in public facilities.

Using data collected prospectively from public and private health facilities in 5 provinces in Thailand, we sought to determine the following: (1) whether ART saves lives for all strata of HIV-infected TB patients across a diverse range of health facilities in a resource-limited setting; (2) factors associated with why some patients were not started on ART; and (3) the overall magnitude of benefit of ART for HIV-infected TB patients, decreasing biases encountered in previous studies by controlling for a person's propensity to receive ART.


Data Collection

In 2003, the US Centers for Disease Control and Prevention began collaborating with the Thai Ministry of Public Health, Japan's Research Institute for Tuberculosis, 4 provinces in Thailand (Bangkok, Ubon-Ratchathani, Phuket, and Chiang Rai), and the National Infectious Diseases Hospital (in Nonthaburi province) on the Thailand TB Active Surveillance Network, a project to enhance surveillance, monitoring, evaluation, and treatment of TB in Thailand.20

For all patients with a diagnosis of TB in the National Infectious Diseases Hospital or any public or private facility in the 4 provinces, staff recorded standardized epidemiologic data, collected sputum specimens for microbiologic testing, and offered HIV counseling and testing. Details about the data collection system and microbiologic testing have been previously reported.20

Patient Population

All persons registered for TB treatment were considered TB patients, consistent with WHO guidelines.21 We restricted our analysis to patients who were documented to be HIV infected and who registered for TB treatment between October 1, 2004, and March 31, 2006. Patient outcomes were recorded through the end of TB treatment, which was usually about 6 months after registration; no patient data were recorded after the end of TB treatment.

In public facilities, patients received standardized TB treatment regimens, consistent with WHO guidelines. Patients with no history of TB treatment received isoniazid, rifampin, ethambutol, and pyrazinamide.21 HIV-infected TB patients were referred for HIV-related care and treatment. Individual physicians exercised their own clinical judgment about measuring CD4+ T-lymphocyte count (CD4), providing opportunistic infection prophylaxis or ART, and managing other clinical conditions. When measured, CD4 counts were usually checked within the first month of TB treatment. Thai Ministry of Public Health guidelines recommend that HIV-infected patients with CD4 < 250 cells/mm3 receive co-trimoxazole and stavudine, lamivudine, and nevirapine; in patients with TB, efavirenz is recommended as a substitute for nevirapine.2 Data were not collected on the exact ART regimen prescribed for each patient.


We used standard WHO definitions to categorize patients according to previous TB treatment history, type of TB (pulmonary, extrapulmonary, or both), and treatment outcome.21 Consistent with WHO guidelines, we classified any death that occurred during TB treatment as a TB death.21

We classified use of ART, co-trimoxazole, and fluconazole as begun before TB diagnosis, after TB diagnosis, or never started. Consistent with WHO guidelines for monitoring and evaluating TB/HIV patients, any patient who took at least one dose of ART or co-trimoxazole was considered to be receiving it throughout TB treatment, regardless of adherence or interruptions; we applied this same standard to fluconazole.22 Because 21% of patients had unknown CD4 count, we analyzed CD4 count as a categorical rather than a continuous variable. We stratified CD4 count as <10, 10-24, 25-49, 50-99, 100-199, and ≥200. Because many patients had very low CD4 count, using fewer categories could have hindered our ability to accurately measure the association between CD4, ART use, and death. The Thai Ministry of Public Health recommends ART for HIV-infected persons with CD4 < 250.2 However, we used CD4 ≥ 200 as our highest category because (1) relatively few patients had CD4 count ≥ 200 and (2) it was consistent with international guidelines on initiation of ART in HIV-infected TB patients during TB treatment.23

Data Analysis

We analyzed data for all HIV-infected TB patients who had known ART status and a treatment outcome of cured, completed, failed, or died. Because one goal of our analysis was to identify potential interventions to improve outcomes of patients diagnosed with both TB and HIV, we excluded patients who received ART before TB diagnosis. We also excluded those with a TB treatment outcome of “transfer out” or “default” because we did not have data whether these patients were alive or dead at the end of TB treatment. Patients with an outcome of failure were combined with those who were cured or completed treatment because all 3 groups were known to have survived the first 6 months of TB treatment. In addition, we excluded patients with a history of TB treatment because retreatment patients are known to have substantially different treatment outcomes than patients never previously treated.24 To determine the impact of these exclusions on our analysis, we performed sensitivity analyses in which we included (1) patients who defaulted from treatment and combined them with patients who were known to be alive at the end of treatment, censoring them at the time of default and (2) patients with a history of TB.

In bivariate analysis, we calculated the relative risk (RR) for factors associated with death and for factors associated with the use of ART. For multivariate analysis, we calculated hazard ratios (HRs) for factors associated with death using a Cox proportional hazards model. The number of days from initiation of TB treatment to treatment outcome date was our dependent variable, and we censored treatment outcomes other than death. For patients who were still on treatment, we censored their outcome at 12 months. To determine the potential impact of missing CD4 counts on our findings, we also used a variety of methods to determine the HR for the association between ART use and death accounting for missing CD4, including the following: (1) excluding patients with missing CD4 and including CD4 as a continuous variable; (2) excluding patients with missing CD4 and including CD4 as a log-transformed variable; (3) assigning a value for all missing CD4 counts equal to the median CD4 for the study population and then including CD4 as a continuous variable; (4) recoding all missing values as CD4 < 10; and (5) recoding all missing values as CD4 > 200. Because TB diagnosis was not confirmed bacteriologically for all patients, we also calculated a HR for the association between ART and death for the subset of patients who had the following: (1) bacteriologically confirmed diagnoses (culture or smear positive) and (2) culture-confirmed diagnoses. We calculated adjusted RRs (aRRs) for factors associated with ART using log-binomial regression. In each case, variables were chosen for inclusion in the multivariate analyses based on one or more of the following: P < 0.20 in bivariate analysis, biologic plausibility, or previously published evidence.

Because some groups of patients were more likely to be on ART than others and some factors associated with ART use were also associated with death, we also analyzed the association between ART and death using propensity score analysis. Propensity score analysis is especially useful when the baseline characteristics of the patients in the 2 exposure groups (in this case, those on ART and those not on ART) are very different.25 In observational studies, propensity score analyses can produce a more accurate estimate of the true association between an intervention (eg, ART) and an outcome (eg, death) by combining factors associated with the intervention into a composite variable, known as the propensity score, and dividing the study population into strata that differ with respect to the propensity of receiving the intervention, but are mostly equal with respect to other covariates.25,26 In this study, we first developed a multivariate model of factors associated with ART using logistic regression, constructed a propensity score based on these factors, and then divided the patient population into equally sized quintiles based on their propensity to receive ART. We calculated HRs for the association between death and ART use, controlling for the ART propensity quintile, using a Cox proportional hazards model.

The protocol for this project was reviewed by the Thailand Ministry of Public Health and the US Centers for Disease Control and Prevention and found to be surveillance and public health program implementation, not human subjects research requiring oversight by an institutional review board.


From October 2004 to March 2006, 3105 HIV-infected TB patients were reported as part of the Thailand Active TB Surveillance Network. Of these, 1615 (52%) had no history of TB treatment, were not diagnosed with non-tuberculous mycobacteria, and were recorded as not on ART at the time of TB diagnosis, thus meeting our inclusion criteria. Of these, 346 (21%) were excluded because of a treatment outcome other than cured, completed, died, or failed (151 transfer out, 144 default, 42 change of diagnosis, and 9 missing outcome). We included patients who were still on treatment and censored them after 1 year of follow-up (Fig. 1).

Flow diagram of patients included in analyses, Thailand, from October 2004 to March 2006.

The median age of the 1269 HIV-infected TB patients included in the analysis was 35 years (range 1-71); 437 (34%) were female. Pulmonary TB was diagnosed in 680 (54%), extrapulmonary in 438 (35%), and both pulmonary and extrapulmonary in 150 (12%). Of the 830 patients with pulmonary disease (including those with both pulmonary and extrapulmonary TB), 530 (64%) were confirmed by sputum smear or culture, including 397 who were at least culture positive, and 133 who were smear positive but culture negative (n = 54) or had cultures not done, contaminated, or missing results (n = 79). Multidrug-resistant TB was diagnosed in 9 (1%).

Directly observed therapy by a health care worker or village health volunteer was provided to 357 (28%) patients; the remaining patients were observed by either a family member or no one. Death was common with 363 (29%) dying during TB treatment. The median time from treatment initiation to death was 51 days (range 0-443 days) (Table 1).

Characteristics of HIV-infected Tuberculosis Patients, Thailand, October 2004-March 2006

Of the 1012 patients with CD4 recorded, the median was 54 cells/mm3 (range 1-1169). Most patients received co-trimoxazole during TB treatment, but only 664 (52%) received fluconazole. ART was started after TB diagnosis in 626 (49%) patients (Table 1). In bivariate analysis of factors associated with ART use, patients with CD4 < 200 were more commonly prescribed ART than those with CD4 ≥ 200, but the frequency with which ART was prescribed varied by CD4 count: 57% of patients with CD4 < 10 were prescribed ART compared with 63% of patients with CD4 10-24, 74% of those with CD4 25-49, 72% of patients with CD4 50-99, and 60% of patients with CD4 100-199. Only 11/257 (4%) of patients with no recorded CD4 count were started on ART [RR 0.14, 95% confidence interval (CI): 0.08 to 0.3]. In multivariate analysis, ART use was more common in patients with CD4 < 200 cells/mm3, in patients who received fluconazole after TB diagnosis (aRR 1.3, CI: 1.1 to 1.4), and in patients treated in an urban area (aRR 1.2, CI 1.1 to 1.3) (Table 2). ART use was less common in patients diagnosed with TB in a private clinic or hospital (aRR 0.7, CI: 0.5 to 1.0) and those with missing CD4 count (aRR 0.14, CI: 0.08 to 0.27) (Table 2).

Factors Associated With ART Use Among HIV-infected TB Patients, Thailand, October 2004-March 2006

In bivariate analysis of the association between ART and death, 68 (11%) of 626 patients started on ART died compared with 295 (46%) of 643 patients who were never started on ART during TB treatment (RR 0.24, CI: 0.19 to 0.30). The Kaplan-Meier survival function is shown in Figure 2. ART was protective against death even for the group with CD4 < 10 cells/mm3; 12/56 (21%) of patients in this group who received ART died compared with 35/43 (81%) of patients who did not receive ART (RR 0.26, CI: 0.16 to 0.44). Among patients who were alive at the end of treatment, those taking ART less commonly defaulted from therapy than those not taking ART (RR 0.18, CI: 0.12 to 0.27).

Kaplan-Meier curve of survival stratified by ART use among HIV-infected TB patients, Thailand, from October 2004 to March 2006.

In multivariate analysis, controlling for factors associated with death, ART use during TB treatment remained strongly protective against death (HR 0.18, CI: 0.13 to 0.25). Patients with missing CD4 more commonly died than patients with nonmissing CD4 count (56% vs 22%, RR 2.6, CI: 2.2 to 3.0). The number of days to death for patients with missing CD4 count was also substantially lower than for those with nonmissing CD4 count (27.5 vs 73 days, P < 0.001). However, the HR for the association between ART use and death did not substantially change when we applied the methods described above to account for missing CD4 count, ranging from 0.15 to 0.20 in each case. Limiting the analysis to only those patients with a bacteriologically confirmed TB diagnosis yielded similar results (HR 0.15, CI: 0.09 to 0.24), as did limiting the analysis only to patients with culture-confirmed TB disease (HR 0.11, CI: 0.06 to 0.20). Other factors which remained independently associated with death in the multivariate model included Thai nationality, diagnosis of both pulmonary and extrapulmonary TB compared with only pulmonary TB, multidrug-resistant TB, increasing age, and decreased or missing CD4 count. Use of fluconazole after TB diagnosis remained statistically protective against death (HR 0.7, CI: 0.5 to 1.0) (Table 3).

Factors Associated With Death Among HIV-infected TB Patients, Thailand, October 2004-March 2006
(Continued) Factors Associated With Death Among HIV-infected TB Patients, Thailand, October 2004-March 2006

In propensity score analysis, the proportion of patients on ART varied from 3% to 82% across the 5 quintiles (Appendix 1 available at Within quintiles, covariates that were used to derive the propensity score were evenly distributed between those on ART and those not on ART (Appendix 2 available at However, in the group of patients with the lowest propensity to receive ART, just 7/253 (3%) were on ART, which is too few to allow for comparisons of patients on ART to those not on ART within that group, thus we excluded this group from the propensity score multivariate analysis. Including the other 4 quintiles, when controlling for propensity to receive ART and other confounding factors, the risk of death remained substantially reduced in those receiving ART (HR 0.17, CI: 0.12 to 0.24) (Table 4). In the group with the lowest propensity to receive ART, CD4 count was missing in 233 (92%). Likewise, time to death for those in this group who died was shorter than in the other groups. The median number of days to death was 28 compared with 72.5 in the other 4 quintiles (P < 0.001). If this group is included in the propensity score analysis, then the association between ART and death is essentially unchanged (HR 0.18, CI: 0.13 to 0.25). In a propensity score analysis in which patients who defaulted from therapy were included, ART remained associated with reduction in death (HR 0.20, CI: 0.14 to 0.28). Likewise, in a propensity score analysis including only patients with a history of TB treatment, ART was associated with reduced mortality (HR 0.28, CI: 0.16 to 0.46)

Multivariate Analysis of the Association Between ART Use and Death Among HIV-infected TB Patients, Thailand, October 2004-March 2006, Controlling for Propensity Score and for Factors Which Remained Confounders Even After Inclusion of Propensity Score


In a resource-limited, public health program setting, HIV-infected TB patients who received ART had approximately one sixth the risk of death of those not receiving ART. The survival benefit persisted even for those with extremely low CD4. Despite the documented benefits of ART in HIV-infected patients with or without TB, we found that over half of patients were not prescribed ART during TB treatment.

We found that the mortality rate during TB treatment was high for HIV-infected TB patients and that most patients had markedly advanced immune-suppression, consistent with other studies done in Southeast Asia.2-5,10 In sub-Saharan Africa, mortality rates are reported to be lower, and HIV-associated TB occurs across a broad range of immune-suppression.27-30 Reasons for this different epidemiology are not known but may be related to differences in host susceptibility, pathogen virulence, or intensity of TB transmission. Mortality rates increased as CD4 declined in our study, yet ART was associated with increased survival at all levels of CD4, an important finding given that the majority of HIV-infected TB patients have extremely low CD4 in Thailand. Although our study was observational rather than a randomized clinical trial, we believe that its findings are valid and relevant for public health policy in Thailand and, possibly, other developing countries. With sufficient sample size, covariate measurement, and statistical rigor, observational studies of data from public health programs can generate reliable estimates of the impact of an intervention in a population.31 Moreover, by studying a heterogeneous population-one derived from diverse geographic and clinical settings and exposed to all the inefficiencies of real-world health care services-observational studies such as this provide data that is representative of the impact an intervention is likely to have when applied across the general health system.31 Using both conventional and propensity score multivariate analysis, we found convincing evidence that most, if not all, HIV-infected TB patients in Thailand should receive ART during TB treatment to reduce mortality.

ART, unfortunately, is underused in HIV-infected TB patients, with over half of the patients in our study not receiving it. Anecdotally, physicians do not prescribe or delay initiating ART during TB treatment because of concerns about overlapping toxicity of medications, the number of pills to be ingested, and risk of immune reconstitution syndrome. That both default and death were less common among HIV-infected TB patients receiving ART should reassure providers that the benefits of ART outweigh the risks associated with toxicity. In our analysis, we found several factors associated with not receiving ART. First, patients being treated in the private health care facilities were less likely to receive ART, a finding that is not surprising given that private sector providers rarely follow practice guidelines for TB treatment in Asia.32-35 Second, patients with missing CD4 were less likely to receive ART. This may be a marker of quality of care or, perhaps more commonly, an indicator of medical futility. Patients with no documented CD4 had a mortality rate even higher than those with CD4 < 10 cells/mm3, and the median time to death in these patients was 27 days.

In addition to ART, we found other biomedical factors associated with death. More extensive TB disease, for example, both pulmonary and extrapulmonary TB, was associated with higher risk of death. Although WHO recommends that co-trimoxazole be provided to all HIV-infected TB patients, we were unable to measure the effect of co-trimoxazole in multivariate analysis because only 16 patients were documented to have never received co-trimoxazole and only one of these patients died. We did find, however, that patients prescribed fluconazole died less frequently than those not prescribed the same. Fungal infections, such as cryptococcal meningitis, are common causes of death in HIV-infected patients in Thailand.36 Further studies should be done to explore the benefits of routinely prescribing fluconazole, in addition to ART and co-trimoxazole, to all HIV-infected TB patients.

Our study is subject to important limitations. It is possible that other factors, not measured in our surveillance system, are independently predictive of ART use and/or death. Propensity score analysis can only control for measured confounding; unmeasured or hidden factors can only be controlled for through prospective randomization. Second, exclusion of patients who transferred out may bias our results slightly. However, it is not possible to include these individuals and accurately ascribe them a treatment outcome because their final treatment outcome is not known. Overall, the proportion of patients excluded was small, making it unlikely that inclusion of these patients would substantially alter our conclusion that universal access to ART should be scaled up for HIV-infected TB patients in Thailand.

Risk of death was over 80% lower in HIV-infected TB patients treated with ART compared with those not treated with ART. The survival benefit persisted even for those with a very low CD4 count and for patients with a history of TB and through multiple sensitivity analyses. Expanding use of ART is critical to improving the survival of HIV-infected TB patients.


The authors thank the US Agency for International Development for assistance in funding this project. We thank Michael Chen for his assistance with the propensity score analysis, and we also thank Phil LoBue, Jose Becerra, and Allyn Nakashima for their review of this manuscript.


1. World Health Organization. Interim Policy on Collaborative TB/HIV Activities. Geneva, Switzerland; 2004.
2. Akksilp S, Karnkawinpong O, Wattanaamornkiat W, et al. Antiretroviral therapy during tuberculosis treatment and marked reduction in death rate of HIV-infected patients, Thailand. Emerg Infect Dis. 2007;13:1001-1007.
3. Cain KP, Kanara N, Laserson KF, et al. The epidemiology of HIV-associated tuberculosis in rural Cambodia. Int J Tuberc Lung Dis. 2007;11:1008-1013.
4. Quy H, Cobelens F, Lan N, et al. Treatment outcomes by drug resistance and HIV status among tuberculosis patients in Ho Chi Minh City, Vietnam. Int J Tuberc Lung Dis. 2006;10:45-51.
5. Trinh TT, Shah NS, Mai HA, et al. HIV-associated TB in An Giang Province, Vietnam, 2001-2004: epidemiology and TB treatment outcomes. PLoS ONE. 2007;2:e507.
6. Breen RA, Miller RF, Gorsuch T, et al. Virological response to highly active antiretroviral therapy is unaffected by antituberculosis therapy. J Infect Dis. 2006;193:1437-1440.
7. Burman W, Benator D, Vernon A, et al. Acquired rifamycin resistance with twice-weekly treatment of HIV-related tuberculosis. Am J Respir Crit Care Med. 2006;173:350-356.
8. Dean GL, Edwards SG, Ives NJ, et al. Treatment of tuberculosis in HIV-infected persons in the era of highly active antiretroviral therapy. AIDS. 2002;16:75-83.
9. Dheda K, Lampe FC, Johnson MA, et al. Outcome of HIV-associated tuberculosis in the era of highly active antiretroviral therapy. J Infect Dis. 2004;190:1670-1676.
10. 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. 2006;43:42-46.
11. Mocroft A, Ledergerber B, Katlama C, et al. Decline in the AIDS and death rates in the EuroSIDA study: an observational study. Lancet. 2003;362:22-29.
12. Palella FJ Jr, Delaney KM, Moorman AC, et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators. N Engl J Med. 1998;338:853-860.
13. Harries A, Maher D, Graham S. TB/HIV: A Clinical Manual. 2nd ed. Geneva, Switzerland: World Health Organization; 2004.
14. Severe P, Leger P, Charles M, et al. Antiretroviral therapy in a thousand patients with AIDS in Haiti. N Engl J Med. 2005;353:2325-2334.
15. Centers for Disease Control and Prevention. Spotlight on syndemics. Available at: Accessed May 5, 2007.
16. Freudenberg N, Fahs M, Galea S, et al. The impact of New York City's 1975 fiscal crisis on the tuberculosis, HIV, and homicide syndemic. Am J Public Health. 2006;96:424-434.
17. World Health Organization. Global Tuberculosis Control: Surveillance, Planning, Financing. WHO Report 2007. Geneva, Switzerland: World Health Organization.
18. Joint United Nations Programme on HIV/AIDS (UNAIDS). AIDS Epidemic Update: Special Report on HIV/AIDS December 2006. World Health Organisation, Geneva, Switerland; 2006.
19. Proceedings of the WHO HIV/TB Conference for the Mekong Sub-region, 2005. Ho Chi Minh City, Vietnam: World Health Organization; 2005. Available at Accessed April 24, 2008.
20. Varma JK, Wiriyakitjar D, Nateniyom S, et al. Evaluating the potential impact of the new Global Plan to Stop TB: Thailand, 2004-2005. Bull World Health Organ. 2007;85:586-592.
21. Treatment of Tuberculosis: Guidelines for National Programmes. 3rd ed. Geneva, Switzerland (WHO/CDS/TB/ 2003. 313): World Health Organization; 2003. Available at: Accessed April 24, 2008.
22. A Guide to Monitoring and Evaluationfor Collaborative TB/HIV Activities. World Health Organization 2004. Geneva, Switzerland. World Health Organization; 2004.
23. Antiretroviral Therapy for HIV Infection in Adults and Adolescents: Recommendations for a Public Health Approach. Geneva, Switzerland: World Health Organization; 2006.
24. Zignol M, Wright A, Jaramillo E, et al. Patients with previously treated tuberculosis no longer neglected. Clin Infect Dis. 2007;44:61-64.
25. Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med. 1997;127(pt 2):757-763.
26. Joffe MM, Rosenbaum PR. Invited commentary: propensity scores. Am J Epidemiol. 1999;150:327-333.
27. Abouya L, Coulibaly IM, Wiktor SZ, et al. The Cote d'Ivoire national HIV counseling and testing program for tuberculosis patients: implementation and analysis of epidemiologic data. AIDS. 1998;12:505-512.
28. Martin DJ, Sim JG, Sole GJ, et al. CD4+ lymphocyte count in African patients co-infected with HIV and tuberculosis. J Acquir Immune Defic Syndr Hum Retrovirol. 1995;8:386-391.
29. Morris L, Martin DJ, Bredell H, et al. Human immunodeficiency virus-1 RNA levels and CD4 lymphocyte counts, during treatment for active tuberculosis, in South African patients. J Infect Dis. 2003;187:1967-1971.
30. Teck R, Ascurra O, Gomani P, et al. WHO clinical staging of HIV infection and disease, tuberculosis and eligibility for antiretroviral treatment: relationship to CD4 lymphocyte counts. Int J Tuberc Lung Dis. 2005;9:258-262.
31. D'Agostino RB Jr, D'Agostino RB Sr. Estimating treatment effects using observational data. JAMA. 2007;297:314-316.
32. Greaves F, Ouyang H, Pefole M, et al. Compliance with DOTS diagnosis and treatment recommendations by private practitioners in Kerala, India. Int J Tuberc Lung Dis. 2007;11:110-112.
33. Lonnroth K, Lambregts K, Nhien DT, et al. Private pharmacies and tuberculosis control: a survey of case detection skills and reported anti-tuberculosis drug dispensing in private pharmacies in Ho Chi Minh City, Vietnam. Int J Tuberc Lung Dis. 2000;4:1052-1059.
34. Lonnroth K, Thuong LM, Linh PD, et al. Delay and discontinuity-a survey of TB patients' search of a diagnosis in a diversified health care system. Int J Tuberc Lung Dis. 1999;3:992-1000.
35. Uplekar M, Juvekar S, Morankar S, et al. Tuberculosis patients and practitioners in private clinics in India. Int J Tuberc Lung Dis. 1998;2:324-329.
36. Tansuphasawadikul S, Amornkul PN, Tanchanpong C, et al. Clinical presentation of hospitalized adult patients with HIV infection and AIDS in Bangkok, Thailand. J Acquir Immune Defic Syndr. 1999;21:326-332.

tuberculosis; HIV; antiretroviral therapy; Thailand; mortality; propensity scores

Supplemental Digital Content

© 2008 Lippincott Williams & Wilkins, Inc.