Factors Associated With Diagnostic Evaluation for Tuberculosis Among Adults Hospitalized for Clinical Pneumonia in Thailand
Objective. Thailand is one of 22 countries designated by the World Health Organization as “high burden” with regard to tuberculosis. Preventing nosocomial tuberculosis transmission remains an important, unmet need. We investigated the adequacy of current practices to evaluate hospitalized patients for tuberculosis, which is critical in preventing delayed diagnosis and nosocomial tuberculosis transmission.
Methods. Thailand conducts active, population-based surveillance for pneumonia in 2 rural provinces. Case report forms are completed for all persons who are hospitalized and meet a case definition of having clinical pneumonia. We analyzed how frequently patients had an adequate diagnostic evaluation for infectious pulmonary tuberculosis, in accordance with national guidelines. We conducted multivariate analyses to determine patient and health-system factors associated with an inadequate diagnostic evaluation for tuberculosis and with tuberculosis disease.
Results. Of 8,853 cases of clinical pneumonia between September 2003 and March 2006, 73% were in patients not adequately evaluated for tuberculosis. Acid-fast bacilli (AFB)–positive tuberculosis was diagnosed in 188 cases, which was 2% of all pneumonia cases and 12% of pneumonia cases in patients adequately evaluated for tuberculosis. Diagnostic evaluations for tuberculosis were less commonly performed among those who were younger than 25 years of age, were female, and lacked cough, sputum production, hemoptysis, and dyspnea. Among patients adequately evaluated, a clinical syndrome of no cough, no hemoptysis, and normal chest radiography findings had a 95% negative predictive value.
Conclusions. The prevalence of AFB-positive, pulmonary tuberculosis was high among adults hospitalized with clinical pneumonia in Thailand. Most patients were not adequately evaluated for tuberculosis. Efforts are needed to improve identification and diagnosis of infectious tuberculosis cases in hospitalized patients.
Received October 29, 2007; accepted March 26, 2008; electronically published June 19, 2008.
Tuberculosis is a major cause of illness and death globally. In 2004, the World Health Organization (WHO) estimated that 8.8 million new cases of tuberculosis (136 per 100,000 population) occurred, including 3.9 million new acid-fast bacilli (AFB) smear–positive, pulmonary cases (60 per 100,000 population).1 Nosocomial transmission of Mycobacterium tuberculosis to healthcare workers and other persons is a major contributor to the tuberculosis epidemic, particularly in developing countries where rates of tuberculosis are high, healthcare facilities are overcrowded, and infection control resources are limited.2 The WHO recommends that healthcare facilities in resource-limited settings implement a standard package of tuberculosis control measures: (1) early diagnosis for potentially infectious patients with suspected tuberculosis, (2) prompt respiratory isolation of potentially infectious patients with tuberculosis or suspected tuberculosis, and (3) prompt initiation of antituberculosis treatment.3 The first of these 3 measures targets delays in diagnosis, the factor that is responsible for the highest risk of nosocomial transmission.3
Thailand is one of 22 countries designated by the WHO as “high burden” with regard to tuberculosis; it has an estimated annual incidence of 142 tuberculosis cases per 100,000 population, including 63 AFB-positive tuberculosis cases per 100,000 population.4 One study of patients with tuberculosis in 7 public hospitals found that diagnostic delays were common in Thailand, with a median time of 2.8 weeks between first presentation to a healthcare facility and tuberculosis diagnosis.5 In this referenced study, patients who presented first to primary-care facilities had significantly longer delays in diagnosis than did those who presented to higher-level facilities.5 Among persons admitted with active tuberculosis disease to hospitals in Thailand, transmission to healthcare workers and other patients occurs frequently.6 As drug resistance emerges globally, the consequences of such delays might be devastating, as documented in nosocomial outbreaks of multidrug-resistant tuberculosis in the United States and Europe in the late 1980s and extensively drug-resistant tuberculosis in South Africa in 2006.7-9
Previous studies in the United States and Canada have identified individual patient factors (eg, advanced age and absence of cough) and health-system factors (eg, low tuberculosis patient volume and private hospitals) associated with delayed diagnosis of tuberculosis in hospitalized persons, but less is known about these factors in high-burden countries, such as Thailand.10 In the present analysis, involving 2 provinces in Thailand, we sought to determine how frequently persons hospitalized with pneumonia are adequately evaluated for tuberculosis and to identify the clinical and health-system factors associated with adequate tuberculosis evaluation.
Methods
Setting
Our analysis involved 2 provinces in Thailand: Nakhon Phanom (estimated population, 732,000) and Sa Kaeo (estimated population, 520,000). These provinces have a total of 20 acute-care hospitals, including 2 general hospitals, 16 community hospitals and 2 Ministry of Defense hospitals. Each district with a population of at least 10,000 persons has 1 community hospital (10-150 beds) that offers primary-care services, and each province has at least 1 general hospital (200-500 beds) that offers some primary-care services but also serves as the referral center for complicated medical conditions outside the scope of the community hospitals.11 All of these public facilities are able to perform and interpret chest radiography and sputum smear microscopy for AFB.
Surveillance for Pneumonia
The International Emerging Infections Program, a collaboration between the Thailand Ministry of Public Health and the US Centers for Disease Control and Prevention, conducts active, population-based surveillance for hospitalized patients with pneumonia in these 2 rural Thai provinces.12 Active surveillance provides an accurate measure of the incidence of radiographically confirmed pneumonia requiring hospitalization, with estimated incidences ranging from 1- to 3-fold higher than those based on passive surveillance, despite the limitation that not all patients with suspected pneumonia undergo chest radiography. Access to health care is good in Sa Kaeo province: a community health survey conducted to assess healthcare-utilization behavior for pneumonia found that 80% of patients with pneumonia sought care at a hospital facility.12
A case of clinical pneumonia is defined as evidence of acute infection (at least 1 of the following: reported fever or chills, documented temperature higher than 38.2°C or lower than 35.5°C, white blood cell [WBC] count higher than 11,000 cells/mm3 or lower than 3,000 cells/mm3, or abnormal differential WBC count) and signs or symptoms of lower-respiratory-tract illness (at least 1 of the following: abnormal breath sounds on chest auscultation, tachypnea, cough, sputum production, hemoptysis, chest pain, or dyspnea) in a resident of one of the 2 surveillance provinces. Standardized case report forms are completed for all persons whose conditions meet the case definition and are hospitalized for this illness in one of the hospitals in the surveillance provinces.12 Physicians are encouraged, but not required, to perform chest radiography and microbiological tests—for example, AFB smear microscopy. Diagnostic test results, including smear and chest radiography findings, in the medical records are recorded in the case report form.
Study Population
We included patients in the pneumonia surveillance database who were admitted from September 1, 2003, through March 30, 2006. We excluded records of patients younger than 15 years of age (14,008 of 25,566 patients), because the recommended diagnostic approach to tuberculosis differs for children and because infectious tuberculosis is uncommon in children.13 We excluded duplicate records and records without sufficient information to determine whether the record was unique (1,718 patients). We restricted our analysis to 1 hospital admission record per person per year. A discrete period of 1 year for each patient was chosen on the basis of public health practice in the United States, in which a case of tuberculosis cannot be counted twice within any period of 12 consecutive months.14 If a patient was confirmed to have tuberculosis during a given hospital stay, that hospital stay was used, and any subsequent hospital stays were excluded. If a patient was never confirmed to have tuberculosis, then the hospital stay during which they were first adequately evaluated for tuberculosis (up to 1 hospital stay per year) was included. If a patient was never adequately evaluated for tuberculosis, or if the adequacy of the evaluation was not clear, then their first hospital stay per year was included.
The Centers for Disease Control and Prevention and the Thailand Ministry of Public Health have determined that this program is surveillance—that is, public health practice—and not human subject research.
Definitions
We used the WHO definition to classify cases of bacteriologically confirmed tuberculosis: 2 or more AFB-positive sputum smears or 1 AFB-positive sputum smear and abnormal chest radiography findings.13 We defined an “adequate evaluation for tuberculosis” on the basis of the minimum number of tests needed to diagnose pulmonary tuberculosis by use of this definition of tuberculosis [Table 1].
Patients were categorized into age groups (15-24, 25-34, 35-44, 45-54, 55-64, and 65 years and older), for consistency with Thailand's national tuberculosis surveillance program. The 3 different hospital types were combined into 2 groups by number of beds: general (more than 150 beds) and community plus military (fewer than 150 beds).
Statistical Analysis
We performed all analyses using Stata software, version 9.2 (StataCorp). Statistical significance was defined as P < .05. We calculated unadjusted and adjusted odds ratios for characteristics associated with an inadequate tuberculosis evaluation. Suspecting that hospital type was an effect modifier, because of known differences in patient population, physician expertise, and hospital resources, we performed the Mantel-Haenszel test for homogeneity on all factors potentially associated with an inadequate evaluation, stratifying by hospital type. The null hypothesis was that the odds ratio for a given factor and hospital type is the same as the odds ratio for that factor not stratified by hospital type. We rejected the null hypothesis for 7 factors, thereby concluding that hospital type was an effect modifier and presenting the analysis of inadequate tuberculosis evaluation stratified according to hospital type (general hospitals vs community and military hospitals). In multivariable analyses, we performed logistic regression including all factors from the single-variable analyses. We developed parsimonious models by use of backward elimination from full models, removing variables stepwise, first with a threshold P value of .10 or greater, then with a threshold P value of .05 or greater. Parsimonious and full models were compared using the likelihood ratio test. In all cases, the parsimonious model fit the data as well as the full model.
We also calculated unadjusted and adjusted odds ratios for factors associated with bacteriologically confirmed tuberculosis in patients who were adequately evaluated for tuberculosis; for this analysis, no effect modification was found, and odds ratios were calculated without stratification by hospital type. In multivariable analysis, the same steps as described above were taken. Positive and negative predictive values were calculated, incorporating the prevalence of exposure (Bayes theorem).
Results
Frequency of and Characteristics Associated With Inadequate Tuberculosis Evaluation
All study hospitals. The analyses include 8,853 cases of clinical pneumonia in 8,685 unique patients. Of the 8,853 cases, only 1,558 (17.6%) were in patients who were found to have been adequately evaluated for tuberculosis during hospitalization. In 850 cases (9.6%), we were unable to determine the adequacy of the patient evaluation because of missing information. Demographic, clinical, and health-system factors all differed significantly (summary χ2 P < .01) between patients adequately and inadequately evaluated for tuberculosis (Table 2). Of 6,550 patients hospitalized at a community or military hospital, 1,111 (17.0%) were adequately evaluated for tuberculosis, compared with 447 (19.4%) of 2,303 patients hospitalized at general hospitals (P < .01).
General hospitals. We found several characteristics statistically associated with an inadequate evaluation for tuberculosis in general hospitals (Table 3). In a parsimonious multivariable model, characteristics that remained statistically associated (P < .05) with an inadequate evaluation for tuberculosis were female sex; age of 15-24 years; being hospitalized only 1 time during the 2.5-year study period; length of hospital stay of 1, 2, or 3 days; and absence of cough, sputum production, hemoptysis, or dyspnea.
Approximately one-third of pneumonia cases (n = 598) occurred in patients referred to the general hospital from another hospital. When these cases were excluded, the characteristics associated with an inadequate tuberculosis evaluation did not change, except that female sex and absence of dyspnea were dropped from the multivariable model (P > .05).
Community and military hospitals. Characteristics associated with an inadequate evaluation in community and military hospitals were similar to those in general hospitals; in addition, the absence of tachypnea, chest pain, and abnormal breath sounds were significantly associated with an inadequate evaluation (Table 3). We also found that the larger the hospital (ie, the more beds a hospital had), the greater the odds of receiving an adequate evaluation, even when patient characteristics, such as cough or hemoptysis, were controlled for (odds ratio for an increase of 30 beds, 1.4). Excluding cases in referred patients (n = 90) did not change the characteristics associated with an inadequate tuberculosis evaluation or their statistical significance.
Tuberculosis Among Patients Adequately Evaluated for Tuberculosis
Frequency. Bacteriologically confirmed tuberculosis was diagnosed in 188 (12.1%) of the 1,558 pneumonia cases in patients adequately evaluated for tuberculosis and in 155 (12.2%) of the 1,270 cases in patients who were both adequately evaluated and hospitalized only once during the 2.5-year study period. The proportion of patients with tuberculosis diagnosed did not vary significantly by hospital type or province. If we assume that all 7,295 patients not adequately evaluated did not have tuberculosis, then the minimum prevalence of AFB-positive tuberculosis among hospitalized patients with pneumonia would be 2.1% (188 of 8,853 cases). If we assume that patients not adequately evaluated had the same prevalence of AFB-positive tuberculosis (12.1%) as patients adequately evaluated, approximately 883 AFB-positive tuberculosis cases were not diagnosed during hospitalization.
Twelve patients were admitted for clinical pneumonia 1 or 2 times before being admitted and having tuberculosis diagnosed. Of these, 5 patients were adequately evaluated for tuberculosis but did not have tuberculosis diagnosed during their previous hospitalizations for pneumonia. One patient had missing information about the tuberculosis diagnostic evaluation during their first hospitalization; nearly 12 months later, they were rehospitalized and had tuberculosis diagnosed. Six patients were inadequately evaluated for tuberculosis during their first hospitalization, discharged within 2-10 days, and subsequently readmitted 1-5 months later and had tuberculosis diagnosed. Of these 6 patients, 5 were women between 34 and 83 years of age and either HIV uninfected or of unknown HIV infection status, and 1 was male, 24 years of age, and HIV uninfected at the time of tuberculosis diagnosis.
Associated characteristics. Among patients with pneumonia who were adequately evaluated for tuberculosis, several characteristics were independently associated with having tuberculosis: age of 25-34 years or 55-64 years, length of hospital stay of 1 day or less, presence of hemoptysis, presence of abnormal breath sounds, and absence of dyspnea (Table 4). Cough and sputum production were not associated with having tuberculosis, possibly because 94% of patients in the subset analyzed had a cough, resulting in a lack of heterogeneity of those symptoms.
Associations with WBC count and HIV infection status. WBC count and HIV infection status were recorded if the information was available in the medical records but were not included in the above analysis, because of insufficient information (WBC count was either not performed or missing for 24% of cases, as was HIV infection status for 75% of cases). For the subset of cases for which data were available, a WBC count of less than 11,000 cells/mm3 and being HIV uninfected were found to be associated with an inadequate evaluation for tuberculosis. Among patients who were adequately evaluated, an elevated WBC count (higher than 11,000 cells/mm3) was associated with tuberculosis. Being HIV infected appears to have been positively associated with tuberculosis, but the association was not statistically significant (odds ratio, 1.3 [95% confidence limits, 0.6, 2.8]; P > .4).
Patient Characteristics as Diagnostic Tests
We calculated the sensitivity, specificity, and predictive values of various combinations of patient characteristics for predicting tuberculosis (Table 5). No combination had a sensitivity exceeding 95%, and the highest positive predictive value was not substantially greater than the proportion of adequately evaluated cases in patients with confirmed tuberculosis (12%). The presence of cough had the highest sensitivity (93.6%) but lacked specificity (6.1%); 12 case patients who did not report cough had tuberculosis diagnosed. The largest negative predictive value was for the absence of a combination of classical pulmonary tuberculosis symptoms and chest radiography findings—that is, patients with no cough, no hemoptysis, and normal chest radiography findings had a 95% chance of not having tuberculosis.
Discussion
In this study, we found that more than 73% of patients hospitalized with clinical pneumonia were not adequately evaluated for tuberculosis, even though an estimated 2%–12% of them had AFB-positive, pulmonary tuberculosis. Patients who were young and female and lacked cough, sputum production, hemoptysis, and dyspnea were the least likely to be evaluated for tuberculosis, in both large and small hospitals. Physicians might be less likely to evaluate such adults for tuberculosis, because tuberculosis disease rates are lower among younger patients and women throughout the world, and the WHO recommends that cough be used as the primary symptom when screening persons for tuberculosis.1,15,16 Nevertheless, of the cases of pneumonia in patients eventually confirmed to have tuberculosis in this study, 38% occurred in women, 5% occurred in persons 15-24 years of age, and 6% occurred in persons without a cough. Failure to evaluate specific subgroups could lead to delayed diagnosis and increase the risk of nosocomial tuberculosis transmission from untreated, AFB-positive patients.
We sought to identify a combination of characteristics that could assist hospital staff in stratifying hospitalized, adult patients with pneumonia into groups at high or low risk for bacteriologically confirmed tuberculosis. Unfortunately, no combination of signs and symptoms performed sufficiently well for this purpose. Even the best-performing combination of characteristics (no cough, no hemoptysis, and normal chest radiography findings) missed 5% of AFB-positive tuberculosis cases. Our data, therefore, suggest that hospitals in Thailand should consider performing an evaluation for tuberculosis for all adults hospitalized with clinical pneumonia, by use of smear microscopy. International guidelines recommend that tuberculosis programs find 1 AFB-positive tuberculosis case for every 10 patients evaluated, suggesting that smear microscopy should be performed universally in any population that has a prevalence of patients with AFB-positive sputum of at least 10%.17 In our population of adult, hospitalized patients with pneumonia, a minimum of 2% and maximum of 12% have AFB-positive tuberculosis. The 12% estimate may be biased upward if physicians had reason to suspect tuberculosis among patients who received an adequate tuberculosis evaluation. Even if the true rate of AFB-positive tuberculosis is lower than 10%, it could be argued that a lower threshold is warranted, given the hazards of nosocomial transmission. In fact, actual rates of pulmonary tuberculosis, including AFB-negative tuberculosis, in this population could be substantially higher if more-sensitive assays, such as sputum culture, were used routinely.18,19
During the period covered in the present analysis, public hospitals in Thailand did not routinely have isolation rooms, and there was no uniform practice of isolating patients with infectious tuberculosis in healthcare facilities. The continuing absence of sufficient isolation facilities leaves early tuberculosis diagnosis and treatment as the best strategy for reducing nosocomial tuberculosis transmission. Research is needed to determine the effectiveness and cost-effectiveness of screening hospitalized patients with pneumonia for accelerating tuberculosis diagnosis and treatment in Thailand.
This analysis has several important limitations. We relied on data collected for surveillance, not for research. Clinicians may have used additional information to inform their decisions, but we were unable to adequately analyze these data, because they were not collected (duration of cough) or not sufficiently complete to analyze (HIV infection status and WBC count). Among these factors, HIV infection status is the most critical, because it is the strongest risk factor for developing tuberculosis disease, and Thailand has an estimated HIV infection prevalence of 7.6% among adults with new cases of tuberculosis.1,19-21 For inadequate evaluations, we were unable to determine whether AFB sputum smear analyses were not performed because of provider-related factors (eg, failure to order the test), health-system factors (eg, failure to perform a test that was ordered), or patient-related factors (eg, inability to produce sputum).
It is possible that patients known to have tuberculosis or patients treated empirically were included in the surveillance system. Because patients with known tuberculosis status do not require a complete diagnostic evaluation for tuberculosis when they are admitted to the hospital, their inclusion in the data set would reduce the estimated proportion of persons who have been adequately tested. We were not able to assess accurately how many patients this represented; on the basis of discussions with hospital staff, we think that the number of such patients is likely to be small, because the vast majority of patients with tuberculosis receive diagnoses and treatment as outpatients. In contrast to this possible misclassification, our approach to patients with multiple admissions would tend to increase the estimated proportion of persons who have been adequately tested, by counting any patient with more than 1 hospitalization within 1 year as having an adequate evaluation as long as he or she was adequately evaluated during at least 1 of the hospital stays. Patients with a history of treatment for tuberculosis are at risk of relapse and reinfection, and, therefore, such patients were appropriately included in the denominator of patients who should be considered for diagnostic evaluation for tuberculosis.
Preventing nosocomial tuberculosis transmission remains an important, unmet need in the developing world. This study demonstrates that clinicians frequently fail to perform tuberculosis diagnostic evaluations for adults hospitalized with respiratory infections and that the burden of disease is substantial in rural Thailand. Possible interventions to address this problem include education of healthcare workers to facilitate prompt identification of potential tuberculosis cases, expansion of tuberculosis laboratory diagnostic capacity to improve speed and accuracy of diagnosis, and engagement of hospital directors and public health officials to ensure that effective hospital infection control policies are implemented.
Acknowledgments
We thank the following collaborators in Thailand: Prabda Prapasiri (International Emerging Infections Program [IEIP], Nakhon Phanom), Sathapana Naorat (IEIP, Sa Kaeo), Somrak Chantra (Crown Prince Hospital, Sa Kaeo), and Denchai Sornkij (Provincial Chief Medical Officer, Nakhon Phanom). In addition, we thank Dr Arthur Reingold and Dr Constance Wang (University of California at Berkeley School of Public Health) for their guidance in the analysis portion of the article.
Financial support. International Emerging Infections Program, Centers for Disease Control and Prevention (CDC). Persons involved in this project are employees of or project staff working for the CDC or the Thai Ministry of Public Health.
Potential conflicts of interest. All authors report no conflicts of interest relevant to this article.
References
- 1. World Health Organization (WHO). Global tuberculosis control: surveillance, planning, financing. WHO report 2007. Geneva: WHO; 2007. Available at: http://www.who.int/tb/publications/global_report/2007/download_centre/en/index.html. Accessed September 3, 2007.
- 2. Harries AD, Maher D, Nunn P. Practical and affordable measures for the protection of health care workers from tuberculosis in low income countries. Bull World Health Organ1997;75:477–489.
- 3. Granich R, Binkin N, Jarvis W, et al. Guidelines for the prevention of Tuberculosis in health care facilities in resource-limited settings. Geneva: World Health Organization; 1999.
- 4. World Health Organization (WHO). Summary by WHO region: South-East Asia. 2006 Global Report. Geneva: World Health Organization; 2006. Available at: http://www.who.int/tb/publications/global_report/2006/pdf/sear.pdf. Accessed February 8, 2007.
- 5. Rojpibulstit M, Kanjanakiritamrong J, Chongsuvivatwong V. Patient and health system delays in the diagnosis of tuberculosis in Southern Thailand after health care reform. Int J Tuberc Lung Dis2006;10:422–428.
- 6. Yanai H, Limpakarnjanarat K, Uthaivoravit W, Mastro TD, Mori T, Tappero JW. Risk of Mycobacterium tuberculosis infection and disease among health care workers, Chiang Rai, Thailand. Int J Tuberc Lung Dis2003;7:36–45.
- 7. Menzies D, Fanning A, Yuan L, Fitzgerald M. Tuberculosis among health care workers. N Engl J Med1995;332:92–98.
- 8. Centers for Disease Control. Epidemiologic Notes and Reports: nosocomial transmission of multidrug-resistant tuberculosis among HIV-infected persons—Florida and New York, 1988-1991. MMWR Morb Mortal Wkly Rep1991;40:585–591.
- 9. Gandhi NR, Moll A, Sturm AW, et al. Extensively drug-resistant tuberculosis as a cause of death in patients co-infected with tuberculosis and HIV in a rural area of South Africa. Lancet2006;368:1575–1580.
- 10. Rozovsky-Weinberger J, Parada JP, Phan L, et al. Delays in suspicion and isolation among hospitalized persons with pulmonary tuberculosis at public and private IS hospitals during 1996 to 1999. Chest2005;127:205–212.
- 11. Thailand Ministry of Public Health (MOPH). Health service systems in Thailand. Thailand Health Profile, 1999-2000. Available at: http://www.moph.go.th/ops/thealth_44/index_eng.htm. Accessed February 8, 2007.
- 12. Olsen SJ, Laosiritaworn Y, Siasiriwattana S, Chunsuttiwat S, Dowell SF. The incidence of pneumonia in rural Thailand. Int J Inf Dis2006;10:439–445.
- 13. World Health Organization (WHO). TB manual: National Tuberculosis Programme guidelines. Geneva: World Health Organization; 2001. Available at: http://www.euro.who.int/document/e75464.pdf. Accessed May 17, 2007.
- 14. Centers for Disease Control and Prevention. Case definitions for infectious conditions under public health surveillance. MMWR Recomm Rep1997;46(RR-10):1–55.
- 15. Mathers CD, Lopez AD, Murray CJL. The burden of disease and mortality by condition: data, methods, and results for 2001. In: Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJL, eds. Global burden of disease and risk factors. New York: Oxford University Press; 2006:228–229.
- 16. Rieder HL. Epidemiologic basis of TB control. Paris: International Union against TB and Lung Diseases (IUATLD); 1999.
- 17. International Union Against Tuberculosis and Lung Diseases (IUATLD). Technical guide for sputum examination for tuberculosis by direct smear microscopy. 3rd ed. Paris: IUATLD; 1978.
- 18. Siddiqi K, Lambert ML, Walley J Clinical diagnosis of AFB-negative pulmonary tuberculosis in low-income countries: the current evidence. Lancet2003;3:288–296.
- 19. Meya DB, McAdam KP. The TB pandemic: an old problem seeking new solutions. J Intern Med2007;261:309–329.
- 20. Center for Disease Control and Prevention (CDC), the World Health Organization (WHO). Tuberculosis infection control in the era of expanding HIV care and treatment. Addendum to WHO guidelines for the prevention of tuberculosis in health care facilities in resource-limited settings, 1999. Geneva: World Health Organization; 1999. Available at: http://whqlibdoc.who.int/hq/1999/WHO_TB_99.269_ADD_eng.pdf. Accessed February 8, 2007.
- 21. Ruxrungtham K, Brown T, Phanuphak P. HIV/AIDS in Asia. Lancet2004;364:69–82.
-
Address reprint requests to Jay Varma, MD, CDC/HIV, Box 68 American Embassy APO, AP 96546.
-
The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.