Isoniazid preventive therapy (IPT) has long been known to markedly reduce the risk of reactivation of latent Mycobacterium tuberculosis infection.1,2 More recently, IPT has proved effective for HIV-infected patients with latent tuberculosis infection (LTBI) who are at considerable risk of progressing to active tuberculosis (TB) disease.3,4 Despite the documented efficacy at the individual level, the effectiveness of IPT as a strategy to reduce TB incidence at the population level is not clear.
Mathematical models have suggested that preventive therapy could significantly reduce TB incidence in populations with high TB prevalence.5 Guidelines in the United States recommend targeted tuberculin skin test (TST) screening and IPT in TST-positive high-risk populations, including HIV-infected individuals.6 Globally, more widespread use of IPT for HIV-infected people at risk for TB has been endorsed.7,8 There are many logistical challenges with implementing a community-level IPT program; including the difficulty of placing and interpreting the TST, ruling out active TB, and promoting adherence to treatment. As a result, IPT is underutilized in high-risk populations.
Injection drug use is an important risk factor for HIV infection and for TB infection and disease in the United States.9,10 In populations where HIV infection overlaps with injection drug use, TB is an important opportunistic disease.11,12
In 1990, a campaign was launched to provide TSTs and IPT to over 2000 injection drug users (IDUs) who participated in the AIDS Linked to the Intravenous Experiences (ALIVE) cohort in Baltimore, MD. In 1996, a preliminary report found that expanded access to treatment for LTBI in this cohort of IDUs reduced TB incidence from 6 of 1000 (1991) to 0 over the 2-year period between mid 1992 and mid 1994.13 The campaign continued through May 1998. We investigated the long-term effectiveness of the TST/IPT campaign through its completion in 1998 and an additional 6 years of follow-up through December 31, 2004.
The ALIVE study recruited a cohort of IDUs from community settings in Baltimore, MD, beginning in 1988.13,14 In brief, both HIV-infected and HIV-uninfected study participants have semiannual visits to the clinic where they undergo a series of interviews, a clinical examination, and have blood drawn for laboratory tests. Included in these interviews are questions regarding history of diagnosis or treatment of TB. Any self-report of TB is confirmed through medical record review with standardized abstraction. Initial inclusion criteria for ALIVE were age, 18 years and older, no history of an AIDS-defining illness, and use of injection drugs within the previous 11 years. The current analyses include participants who were enrolled in the initial (1988-1989) and second (1994-1995) recruitment waves.
The TST/IPT campaign began in March 1990 and continued through March 1998. During this period, participants were invited to receive a TST and to return for a reading 2-3 days later.13,15 Repeat TST was offered during the testing interval to TST-negative participants on return visits. A positive test was defined as ≥5-mm-diameter induration for HIV-positive participants and ≥10-mm induration for HIV-negative participants. All participants with a positive result were screened for active TB disease with a symptom review and chest radiograph. All TST-positive participants not diagnosed with active TB disease were offered IPT. For the first year of the study, IPT was self-administered. However, as treatment expanded, participants received directly observed preventive therapy twice or thrice weekly. Treatment was considered complete if the participant was adherent to treatment for at least 26 weeks.
A registry match was conducted with the ALIVE cohort database and the Maryland State Department of Health and Mental Hygiene (DHMH) tuberculosis registry. All TB cases diagnosed in the state are reportable to DHMH, and reports are submitted by physicians, laboratories, pharmacies, and health departments. Local health departments perform evaluations to verify the accuracy of diagnoses before submitting reports to DHMH. Only cases reported through the DHMH tuberculosis registry were considered as TB cases in the current analysis.
Three periods were considered for the current analysis: pre-TST era (1988 to February 28, 1990), TST era (March 1, 1990 to May 31, 1998), and post-TST era (June 1, 1998 to December 31, 2004). Annual and era-specific incidence rates (IRs) for TB were calculated per 1000 person-years of follow-up from study enrollment through December 31, 2004, or from the beginning to the end of each era. Incidence rate ratios (IRRs) compared TB incidence in the different eras. Cox proportional hazards models were used to investigate factors associated with a diagnosis of TB. Participants who were diagnosed as TB cases more than 1 year after their last ALIVE visit were not included in the Cox models because their risk factors close to time of diagnosis could not be ascertained. All analyses were conducted using SAS (Version 9.1; SAS Institute, Cary, NC) and Stata (Version 9.1; STATA Corporation, College Station, TX).
A total of 2376 patients had visited the ALIVE clinic between 1988 and 1998. During the TST era (March 1990 to May 1998), 2010 (84.6%) participants were interviewed by a TB study nurse during a study visit. Among those who were interviewed by a TB study nurse, the majority were males (75%), median age = 36.3 (interquartile range [IQR] = 31.6-41.1), and blacks (94%). Six hundred eighty-two (34%) of the 2010 participants were HIV positive and 1327 (66%) were HIV negative at baseline. There were 118 (5.9%) participants who seroconverted for HIV antibodies during follow-up. Participants who were not interviewed did not differ in demographic or clinical characteristics compared with those who were interviewed (data not shown).
TST and IPT Results
Overall, among 2376 participants, 1753 (74%) had at least 1 TST placed and read and 536 (31%) had a positive TST result. Among 800 HIV-positive participants, 649 (81%) had a TST placed and read compared with 1104 (70%; P < 0.01) among 1576 HIV negatives (Table 1). TST positivity differed significantly between HIV-positive participants (16%) and HIV negatives (39%; P < 0.01). Among HIV positives, TST positivity had an inverse association with CD4 count, with 4.1% of those with CD4 <200/mm3 having a positive result at first visit compared with 11.0% among participants with CD4 >200/mm3 (P = 0.02).
Similar proportions of TST-positive HIV positives (59%) and HIV negatives (55%; P = 0.45) started IPT. Additionally, the proportions completing therapy were similar between HIV-positive (60%) and HIV-negative participants (54%; P = 0.28). Overall, 31% percent of all participants with a positive TST completed 6 months of IPT (35% HIV positive and 30% HIV negative; P = 0.28). The majority of participants starting (80%) and completing (78%) IPT were HIV negative. Participants with a positive TST who started IPT did not differ from those who did not start IPT by age, gender, race, HIV status, or drug use. Among those starting IPT, 20 (7%) self-administered all doses, 134 (45%) had all doses directly observed, and the remaining 145 (48%) received a combination of direct observation and self-administered therapy.
Tuberculosis IRs by TST/IPT Status
Among participants who visited the clinic at least once between 1990 and 1998, 28 cases of TB were recorded. Six of these were excluded because they had no visits in the TST era (4 of these 6 had a baseline visit only) and were therefore not given the opportunity to receive a TST or IPT. A total of 22 TB cases were included in the analysis.
Of the 22 patients who developed TB, 19 were HIV positive and 3 were HIV negative. Among the 19 HIV positives, 13 had a TST, of whom 3 were reactive and 10 were nonreactive. Two of the 3 HIV-positive reactors initiated IPT: 1 received IPT for 1 week and was diagnosed with TB 3 months later and the other received IPT for 20 weeks and was diagnosed with TB 3 years later. Two of the 3 HIV negatives had a TST and both were reactive. Neither of the HIV-negative reactors initiated IPT.
The incidence of TB was calculated for participants who never had IPT, those who started IPT, those who completed at least 30 days, and those who completed at least 6 months of IPT (Table 2). Among patients who completed at least 30 days of IPT, but did not complete the 6-month course, the median duration of therapy was 15 weeks. Among the full cohort, participants who never started IPT contributed 24,585 person-years and had 20 cases of TB diagnosed for an IR of 0.81 per 1000 person-years. Those who started IPT experienced 2 cases during 4185 person-years for an IR of 0.48 per 1000 person-years, and those with at least 30 days of IPT had an IR of 0.29 per 1000 person-years. Finally, among those who completed 6 months of IPT, there were no cases of TB in 2385 person-years of follow-up. These rates were higher when restricted to the HIV-positive subset; however, similar rates were seen in the no IPT group (1.92/1000 person-years), the starting IPT group (2.15/1000 person-years), and the 30-day IPT group (1.30/1000 person-years) versus no cases during 588 person-years in the group completing IPT (Table 3).
IRs were calculated according to HIV and TST status without considering IPT use. Among HIV-infected participants, those with a positive TST had an IR = 1.93 per 1000 person-years compared with 1.56 per 1000 person-years for TST negatives and 3.24 per 1000 person-years for TST unknowns. Among HIV negatives, TST-positive participants had an IR = 0.35 per 1000 person-years compared with 0 per 1000 person-years for TST negatives and 0.22 per 1000 person-years for participants with an unknown TST result.
Temporal Trends in Tuberculosis Incidence
Two cases of TB were diagnosed in the pre-TST era, 13 during the TST era, and 7 in the post-TST era (Fig. 1). Annual TB incidence fluctuated greatly throughout the 17 years, and the trends in the overall cohort and HIV-positive participants were quite similar. The overall IR for the full cohort was 0.76 per 1000 person-years. The IRs for the 3 periods were: pre-TST era = 0.72 per 1000 person-years; TST era = 0.88 per 1000 person-years, post-TST era = 0.62 per 1000 person-years. The IRR comparing the TST era with the pre-TST era was 1.21 [95% confidence interval (CI): 0.34 to 10.02], whereas the post-TST era saw a reduction in TB incidence when compared with the TST era (IRR = 0.71; 95% CI: 0.42 to 1.34), but this decrease was not statistically significant.
Among the subset of HIV positives, the overall TB incidence was 1.94 per 1000 person-years. The IRs in each era were 2.16 per 1000 person-years (pre-TST era), 1.84 per 1000 person-years (TST era), and 2.03 (post-TST era; Table 4). Though TB incidence in the TST era was lower than that in the pre-TST era, the observed reduction was not statistically significant (IRR = 0.85; 95% CI: 0.23 to 7.04).
Correlates of Incident Tuberculosis
Eighteen of 22 TB cases were included in the Cox proportional hazards models; 3 were excluded because TB was diagnosed more than 1 year after final visit, and one other had only 1 visit before TB diagnosis. The strongest risk factor for TB in the full cohort was HIV infection (relative hazard = 8.4; 95% CI: 2.4 to 28.9; Table 4). There was no association between risk of TB and sex, age, education, homelessness, or employment status. Prior incarceration was associated with a decreased risk of TB. Among those with a positive TST, receiving any IPT was protective of developing TB (Table 4); however, these associations were not statistically significant primarily because there was only 1 case of TB among participants who received at least 30 days of IPT. In multivariate analysis for all participants, HIV infection (adjusted Relative Hazard [aRH]: 6.0; 95% CI: 1.7 to 21.1) was associated with an increased risk and history of incarceration (aRH:0.24; 95% CI: 0.1 to 0.7) was associated with a decreased risk of TB.
Among the HIV-infected participants, there was no observed difference in TB risk by gender, age, education, employment, or incarceration. In multivariate analysis, CD4+ cell count <200 μL/mL was the strongest risk factor for TB (aRH = 13.6; 95% CI: 1.5 to 119.9) (Table 5). Although a potentially protective effect was suggested, antiretroviral (ARV) treatment did not significantly affect TB risk (aRH = 0.4; 95% CI: 0.1 to 1.4). Body mass index <21, measured only among the HIV-positive participants, was strongly associated with incident TB (aRH = 4.6; 95% CI: 1.6 to 13.6) even after adjusting for advanced immunosuppression.
A program aimed at providing TSTs and IPT to a cohort of IDUs contributed to reduced TB incidence for some participants but did not eliminate TB from this high-risk population. Although there were no cases of TB diagnosed among participants who completed at least 6 months of IPT, the overall incidence of TB within the cohort did not significantly change over time. Within the HIV-infected subjects, however, the expected rate of TB without IPT would have increased over time as cellular immunodeficiency progressed,16 so the program most likely was beneficial at the population level. Nonetheless, although the intent of the program was to test and treat all participants infected with M. tuberculosis, only 32% of those found to be TST positive completed the recommended course of treatment. It is quite likely that with an increase in the proportion of participants starting and completing 6 months of IPT, TB could have been reduced more substantially than found in this analysis. Of note, however, is that 10 of 22 TB cases were in HIV-infected individuals with a negative TST. These people may have had false negative tests or could have had new M. tuberculosis infections that progressed to disease after testing.17
In the initial report of this intervention in 1996, it was stated that TB was eliminated from this population for at least a 2-year period, only 2 years after implementation.13 Our analysis revealed that some TB cases occurred during that period but were only captured by the registry match with DHMH. The methodology in this analysis differs from the initial report, which included TB diagnoses reported by patients but not verified or matched to the state TB registry.
In the current analysis, we observed a reduction in TB incidence in the final 2 years of the intervention and temporary elimination in the 2 years immediately after completion of the intervention. However, incidence increased again, and 7 cases were diagnosed between 2000 and 2004, all among HIV-infected participants. Six of these participants received a TST at least 7 years before developing TB. Thus, it is possible that the cumulative effect of detecting and treating LTBI in almost 300 participants had an impact on reducing TB incidence over a 4-year period, and that the observed rates of TB may have been even higher had the intervention not been initiated. Also, a prolonged period of TB elimination may have been observed if all these 7 patients had been detected and treated. Moreover, repeat TST may have detected latent infection in these (and other) patients.
Of the 19 cases of TB diagnosed among HIV-infected participants in the ALIVE cohort, 6 (31%) never received a TST and 3 (16%) had a positive TST but did not receive the recommended course of IPT. Thus, about half of HIV-infected cases may be considered missed opportunities of the program. Of the 3 cases of TB among the HIV-negative participants, 1 never received a TST and the other 2 had a positive TST but never started IPT. False negative reactions to the TST are also a concern, especially among HIV-infected participants, leading to missed opportunities for prevention. During the TST period of this study, the TST was the only available tool for diagnosing LTBI. New interferon-γ release assays may have greater sensitivity and specificity for LTBI,18 though the performance of these assays in HIV-infected individuals needs to be more fully evaluated. Four of 7 TB cases who were coinfected with HIV at TB diagnosis had CD4 lymphocyte counts less than 200 at the time of their negative TST result, a level that has been shown to have reduced sensitivity for detection of latent infection.19 In our population, participants with CD4 counts <200/mm3 were less likely to have a positive TST compared with those with higher CD4 counts and HIV negatives. This difference was likely due to the decreased sensitivity of the TST in HIV-infected patients.
In the initial analysis,13 there were no TB cases in 418 person-years of follow-up among participants receiving at least 6 months of isoniazid. The extended follow-up in this study reveals no cases occurring among 2385 person-years of follow-up among participants completing therapy and no cases occurring among 588 person-years of follow-up among the HIV-infected subset. It is noteworthy that among all participants who started IPT in this study, only 2 TB cases developed after 4185 person-years of follow-up, and among those who took at least 30 days of IPT, only 1 case of TB developed, suggesting benefits among IPT initiators even if the entire course is not completed.20-22 Of course, the goal of any IPT program should be completion of therapy to increase the efficacy of the treatment. Further, it should also be noted that this partial protection might differ by HIV status. Among HIV-infected participants who started IPT but did not complete at least 6 months, the IR was similar to those who did not receive any IPT. Thus, for HIV-infected patients, it is imperative to ensure adherence to IPT. Importantly, isoniazid resistance was not detected among TB cases who had received any duration of IPT.
Participants with more advanced HIV disease (CD4 < 200) had a much greater incidence of TB.23,24 However, we did not observe a significant reduction in TB incidence among HIV-infected participants who received ARV therapy. Though several cohort studies have shown that TB incidence is reduced in HIV-infected patients receiving highly active antiretroviral therapy, rates among these patients still remain alarmingly high.23,25-27 Although TB incidence did not differ by ARV therapy use, the small number of cases in the study may have limited the ability to make this observation. Further, the majority of person-time, and almost all the intervention time, was before the start of the highly active antiretroviral therapy era. Thus, the cohort as a whole had increasingly progressive immune suppression.
Our study had several limitations. Though all participants in the ALIVE cohort were crossmatched with the Maryland State TB registry, it is possible that ALIVE participants were diagnosed with TB outside the state of Maryland and were therefore not captured. If TB cases were missed, our reported incidence would be an underestimate of the true incidence. However, our study population is known to be restricted in movement out of the city and rarely seeks health care elsewhere. If any misclassification occurred, it was unlikely to be differential. Secondly, this intervention was at the program level and therefore did not include a control group. However, the primary analysis in this study was conducted to compare TB incidence by era among all ALIVE participants during this period regardless of participation in the intervention. This conservative approach allows for a more realistic assessment of population-level effectiveness of a programmatic intervention that was intended to include as many participants as possible. Though a high proportion (74%) had a TST reading, we were limited by having significant numbers of participants who did not initiate or complete IPT. When analyzed further, the positive effects of isoniazid were considerably more pronounced when limited only to those receiving a TST and preventive therapy (data not shown). Further, our primary intention was to describe the effectiveness of the TST/IPT intervention and its potential effectiveness despite the relatively low uptake of the intervention. It is possible that those individuals who adhered to their IPT were those at greatest TB risk, but this is difficult to assess with these data. Finally, despite the >30,000 person-years of follow-up and a high prevalence of LTBI, the number of TB cases detected among the ALIVE participants was relatively low.
There are several strengths to this analysis. The study population is a unique high-risk population in the United States and one that is of high priority for TB control programs. IDUs are at high risk for both HIV and TB, and this analysis was able to determine the effectiveness of preventive therapy for both HIV-infected and HIV-uninfected populations. The ALIVE cohort is a long-standing cohort, thus long-term follow-up was able to be assessed because of the strong retention rate of study participants. Finally, the Maryland DHMH TB registry has a strong record of complete data collection.
Tuberculosis incidence can be reduced in a high-risk population through programmatic implementation of tuberculin skin testing and IPT. For a program to achieve its maximum benefit, a high proportion of patients must be tested and high rates of treatment completion must be achieved among the eligible population. HIV-infected IDUs will receive the greatest benefit and should be a high priority for targeted tuberculin testing and IPT in the United States. In addition, efforts to identify better tests for LTBI in HIV-infected individuals and to prevent transmission of M. tuberculosis infection are important strategies for TB control.
We would like to dedicate this article to Dr. George W. Comstock whose pioneering work as an epidemiologist has inspired generations of researchers.
1. Comstock GW, Baum C, Snider DE Jr. Isoniazid prophylaxis among Alaskan Eskimos: a final report of the bethel isoniazid studies. Am Rev Respir Dis
2. Comstock GW, Ferebee SH, Hammes LM. A controlled trial of community-wide isoniazid prophylaxis in Alaska. Am Rev Respir Dis
3. Woldehanna S, Volmink J. Treatment of latent tuberculosis infection in HIV infected persons. Cochrane Database Syst Rev
4. de Pinho AM, Santoro-Lopes G, Harrison LH, et al. Chemoprophylaxis for tuberculosis and survival of HIV-infected patients in Brazil. AIDS
5. Murray CJ, Salomon JA. Modeling the impact of global tuberculosis control strategies. Proc Natl Acad Sci U S A
6. American Thoracic Society/Centers for Disease Control and Prevention/Infectious Diseases Society of America. Controlling tuberculosis in the United States. Am J Respir Crit Care Med
7. World Health Organization. Interim Policy on Collaborative TB/HIV Activities
. WHO/HTM/TB/ 2004 330; WHO, Geneva, Switzerland; 2004.
8. World Health Organization/Global Tuberculosis Programme and UNAIDS. Policy statement on preventive therapy against tuberculosis in people living with HIV. Policy Statement. Wkly Epidemiol Rec
9. Dworkin MS, Adams MR, Cohn DL et al. Factors that complicate the treatment of tuberculosis in HIV-infected patients. J Acquir Immune Defic Syndr
10. Riley ED, Vlahov D, Huettner S, et al. Characteristics of injection drug users who utilize tuberculosis services at sites of the Baltimore city needle exchange program. J Urban Health
11. Selwyn PA, Hartel D, Lewis VA, et al. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. N Engl J Med
12. Selwyn PA, Alcabes P, Hartel D, et al. Clinical manifestations and predictors of disease progression in drug users with human immunodeficiency virus infection. N Engl J Med
13. Graham NM, Galai N, Nelson KE, et al. Effect of isoniazid chemoprophylaxis on HIV-related mycobacterial disease. Arch Intern Med
14. Vlahov D, Anthony JC, Munoz A, et al. The ALIVE study, a longitudinal study of HIV-1 infection in intravenous drug users: description of methods and characteristics of participants. NIDA Res Monogr
15. Graham NM, Nelson KE, Solomon L, et al. Prevalence of tuberculin positivity and skin test anergy in HIV-1-seropositive and -seronegative intravenous drug users. JAMA
16. Corbett EL, Charalambous S, Moloi VM, et al. Human immunodeficiency virus and the prevalence of undiagnosed tuberculosis in African gold miners. Am J Respir Crit Care Med
17. Daley CL, Small PM, Schecter GF, et al. An outbreak of tuberculosis with accelerated progression among persons infected with the human immunodeficiency virus. An analysis using restriction-fragment-length polymorphisms. N Engl J Med
18. Pai M, Kalantri S, Dheda K. New tools and emerging technologies for the diagnosis of tuberculosis: part I. Latent tuberculosis. Expert Rev Mol Diagn
19. Rangaka MX, Wilkinson KA, Seldon R, et al. Effect of HIV-1 infection on T-Cell-based and skin test detection of tuberculosis infection. Am J Respir Crit Care Med
20. Comstock GW. How much isoniazid is needed for prevention of tuberculosis among immunocompetent adults? Int J Tuberc Lung Dis
21. Comstock GW, Baum C, Snider DE Jr. Isoniazid prophylaxis among Alaskan Eskimos: a final report of the bethel isoniazid studies. Am Rev Respir Dis
22. Comstock GW, Ferebee SH, Hammes LM. A controlled trial of community-wide isoniazid prophylaxis in Alaska. Am Rev Respir Dis
23. Elzi L, Schlegel M, Weber R, et al. Reducing tuberculosis incidence by tuberculin skin testing, preventive treatment, and antiretroviral therapy in an area of low tuberculosis transmission. Clin Infect Dis
24. Golub JE, Saraceni V, Cavalcante SC, et al. The impact of antiretroviral therapy and isoniazid preventive therapy on tuberculosis incidence in HIV-infected patients in Rio de Janeiro, Brazil. AIDS
25. Badri M, Wilson D, Wood R. Effect of highly active antiretroviral therapy on incidence of tuberculosis in South Africa: a cohort study. Lancet
26. 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
27. Girardi E, Sabin CA, d'Arminio Monforte A, et al. Incidence of tuberculosis among HIV-infected patients receiving highly active antiretroviral therapy in Europe and North America. Clin Infect Dis
Keywords:© 2008 Lippincott Williams & Wilkins, Inc.
HIV; latent tuberculosis infection; tuberculosis