Trends in Tuberculosis/Human Immunodeficiency Virus Comorbidity, United States, 1993-2004 FREE

After more than 30 years of steady decline, tuberculosis (TB) in the United States increased by 20% from 1985 to 1992.^1- 2 The human immunodeficiency virus (HIV) epidemic was an important factor in this resurgence.¹ The geographic and demographic distributions of increased TB incidence were highly correlated with the prevalence of AIDS cases.² Epidemiologic studies have suggested that HIV infection is the strongest predisposing factor for TB developing in persons with latent TB infection (LTBI).^3- 5 Infection with HIV increases the risk of progression from LTBI and more rapid development of TB disease among persons recently infected with Mycobacterium tuberculosis.⁶ The risk of TB developing in HIV-negative persons with LTBI is estimated to be as high as 10% over an entire lifetime.^7- 8 In contrast, 1 study showed that the annual risk for TB was 8% in persons with LTBI and HIV.⁴

A number of important changes have occurred in the epidemiology of both TB and HIV since 1993. With increased federal resources for TB control programs in the early 1990s (from $24 million in 1990 to $112 million in 1994) and implementation of measures to address the resurgent epidemic, TB cases declined by an average of 5% annually from 1993 through 2004.⁹ Also, improvements in HIV-prevention strategies and the advent of antiretroviral therapy have been temporally associated with steady declines in AIDS cases and deaths, after peaking in 1993 and 1992, respectively.¹⁰

A number of studies have described local trends in HIV-associated TB in clinic- or hospital-based settings or have matched TB and HIV data at local and state levels.^11- 13 Concerns about data confidentiality, however, have limited systematic linking of national AIDS and TB surveillance systems. Also, data on HIV infection status were not included in the national TB surveillance system until 1993. This change permitted us to examine national trends in TB/HIV cases for selected demographic and clinical characteristics for 1993 through 2004. We also compared the characteristics of patients with TB/HIV with those of patients with TB but without HIV.

We analyzed data on all incident TB cases reported to the Centers for Disease Control and Prevention (CDC) national TB surveillance system from all 50 states and the District of Columbia from 1993 through 2004. The TB case reports record HIV status as positive, negative, or unknown (including indeterminate and missing results and testing refused or not offered). Cases of TB in persons reported as having a positive HIV test result at the time of TB diagnosis were classified as TB/HIV cases. Patients with TB and negative HIV results or unknown HIV status were classified as having non-HIV TB. We dichotomized HIV status into positive or “other” to avoid artificial inflation of HIV prevalence rates that would result from exclusion of patients with unknown HIV status, especially in earlier years. Persons at high risk for HIV were more likely to be tested for HIV infection and therefore more likely to be HIV positive than persons who were not tested.

As a sensitivity analysis, we adjusted for unequal probabilities of being tested for HIV infection during the study period by using a subclass adjustment based on propensity stratification.¹⁴ In this method, a numerical adjustment factor (or propensity score), which is the probability of nonmissing HIV status conditional on covariates, is applied to each patient with known HIV-infection status. We first applied the subclass adjustment to patients from states other than California. California was excluded because it reports only positive HIV results to the CDC and does not distinguish negative from unknown HIV status. We subsequently calculated estimates for California by exploring a range of plausible assumptions about the testing rate in California compared with New York City, New York. (An annex with more information on the methods is available from the corresponding author.)

Trends in TB/HIV cases were analyzed according to selected demographic and clinical characteristics. Race/ethnicity is self-reported by the patient and assigned to predefined census categories in the case report form. Regions of the country (South, West, Northeast, and Midwest) are defined based on the national surveillance reports.⁹ High TB/HIV-burden states were arbitrarily defined in the analysis as those states with 100 or more TB/HIV cases in 1993. All others were low-burden states. States defined as having low TB case rates were those with fewer than 3.5 cases/100 000, the interim year 2000 target for TB elimination. States with case rates above the national average for 1993 or 2004 were considered to have high TB case rates.⁹

Data on HIV testing rates by selected demographic and clinical characteristics are shown in Table 1. The percentage of persons with TB tested for HIV steadily increased and doubled from 36% in 1993 to 72% in 2004. With propensity stratification, the estimated comorbidity rate compared with that observed was statistically higher during the earlier years of this analysis than in more-recent years.

The number of TB/HIV cases decreased from 3681 (15% of 25 108 TB cases) in 1993 to 1187 (8% of 14 515 TB cases) in 2004, accounting for 23% of the overall decrease in TB cases during this period (Table 2). The number of TB/HIV cases decreased by 18% during 1993 through 1995, by 34% during 1996 through 1999, and by 19% during 2000 through 2004. In 1993, TB/HIV cases as a percentage of all TB cases (hereafter referred to as TB/HIV comorbidity rates) were greater than 10% in 14 states (range, 11%-28%); the number dropped to 8 states in 2004 (range, 11%-21%) (Figure 1). In states with a high burden of TB/HIV, the number of TB/HIV cases decreased from 2969 (19% of 15 985 TB cases) in 1993 to 802 (9.2% of 8699 TB cases) in 2004. In the low TB/HIV-burden states, the number of TB/HIV cases decreased from 712 (7.8% of 9123 TB cases) in 1993 to 385 (6.6% of 5816 TB cases) in 2004. From 1993 through 2004, the TB/HIV case rate decreased by 71% (from 1.4/100 000 to 0.4/100 000); the non-HIV TB case rate decreased by 46% (from 8.3/100 ;17U>000 to 4.5/100 000) (Table 3).

Table 3 shows TB/HIV case rates and non-HIV TB case rates for 1993 and 2004 for selected demographic characteristics. The percentage of decline in the number of TB/HIV cases was greatest in persons aged 25 to 44 years (75%, from 2788 to 683) (Figure 2A). The comorbidity rates of TB/HIV remained highest in this age group (1993, 29% [9586]; 2004, 14% [4942]). From 1993 through 2004, there was a 71% decline in the number of cases in males (from 2948 to 866) and a 56% decline in females (from 733 to 321) (Figure 2B). From 1993 through 2004, the number of TB/HIV cases and TB/HIV comorbidity rates remained stable among persons born outside the United States (1993-2004 average, 6.6%) and steadily decreased (from 18% in 1993 to 11% in 2004) in US-born persons (Figure 2C). The number of TB/HIV cases among US-born persons declined by 77% (from 3143 to 708). Persons born in the United States accounted for all of the decline in TB/HIV cases and non-HIV TB cases. From 1993 through 2004, the number of TB/HIV cases declined by 84%, 63%, and 66% for non-Hispanic whites, Hispanics, and non-Hispanic blacks, respectively. The TB/HIV comorbidity rates were substantially higher in non-Hispanic blacks than in the other racial/ethnic groups for all years (Figure 2D). The TB/HIV comorbidity rates remained relatively stable in the South and declined in the Northeast, West, and Midwest (Figure 2E). The number of TB/HIV cases declined by 81%, 71%, 60%, and 53% for the Northeast, West, Midwest, and South, respectively. The percentage of TB cases in groups with the highest TB burden was considerably higher among TB/HIV cases than among non-HIV TB cases: for persons aged 25 to 44 years (58% vs 32%), males (73% vs 60%), non-Hispanic blacks (58% vs 25%), US-born persons (60% vs 45%), and persons from the South and Northeast combined (76% vs 58%). In 2004, about 13% of all TB/HIV cases occurred in non-Hispanic black males aged 25 to 44 years in the South.

The TB/HIV comorbidity rates declined among all TB high-risk populations except alcohol abusers but increased slightly during 2000 through 2003 in all but persons living in correctional facilities (Figure 3). The number of TB/HIV cases decreased by 46%, 80%, and 71% among the homeless, persons living in correctional facilities, and injection drug users, respectively. The number decreased by 7.1% among noninjection drug users and 20% among alcohol abusers. Table 4 shows the percentage of high-risk populations in TB/HIV and non-HIV TB cases. The percentage of TB/HIV cases in injection drug users steadily decreased, from 22% in 1993 to 7.3% in 2004, and remained stable among non-HIV TB cases (1993-2004 average, 1.6%) (Table 4).

Clinical and TB treatment information are presented in Table 5 and Table 6, respectively. Extrapulmonary TB and noncavitary pulmonary TB were more common among persons with TB/HIV than among those with non-HIV TB. Drug-resistant TB, especially multidrug-resistant (MDR) and rifampin-monoresistant TB, decreased more rapidly among coinfected persons. The percentage of persons whose treatment was entirely directly observed more than doubled from 1993 to 2002 in both TB/HIV and non-HIV TB cases. The percentage of persons who died during treatment decreased by more than 50% among TB/HIV cases but declined by only 28% among non-HIV TB cases.

Worldwide, the prevalence of HIV infection in persons with TB ranges from less than 1% to nearly 70% in certain parts of Africa.¹⁵ In many developing countries, TB is one of the leading causes of morbidity and mortality in HIV-infected persons. Globally, an estimated 11% of deaths in patients with TB are the result of HIV infection.¹⁵ In the countries that are most affected by HIV, up to 40% of TB deaths are attributable to HIV.^16- 17 In the United States, HIV infection contributed to the increases in TB rates in the late 1980s and early 1990s. The data reported herein show that since 1993 TB/HIV comorbidity has steadily decreased contemporaneously with the declines in TB rates, although the decrease has slowed over the last several years.

The substantial decreases in the total number of TB/HIV cases and in the rate of TB/HIV comorbidity over the past decade were driven by decreases in US-born persons in the high-burden states, the states that were responsible for the resurgence during 1985 through 1992. This decrease may have been the result of more intense TB and HIV control efforts in these states, leading to substantial declines in both TB and AIDS cases. These efforts included (1) interrupting TB transmission by isolating infectious patients; (2) decreasing transmission of TB in congregate settings; (3) encouraging HIV testing and, with the advent of highly active antiretroviral treatment (HAART), improving HIV treatment; and (4) expanding testing and treatment for LTBI among HIV-infected persons.^18- 20 The introduction of HAART in 1996 likely contributed to the accelerated decline in TB/HIV comorbidity that was seen in the United States, as well as in other countries, during the period 1996 through 1999.^21- 22

In 2004, the highest TB/HIV comorbidity rates continued to be in those groups contributing most to the decline in TB cases (eg, US-born persons, persons aged 25-44 years, males, and non-Hispanic blacks). In other groups—notably 45- to 64-year-olds, foreign-born persons, persons from the South, and women—the TB/HIV comorbidity rate did not substantially decline, although case rates did. The slower decline in comorbidity rates and in case rates in persons aged 45 to 64 years, in women, and in the South mirrors trends in the HIV/AIDS epidemic in the United States. The fact that the South has not experienced the decreases in TB/HIV comorbidity that have been seen in the rest of the country is consistent with concerns voiced by the Advisory Council for the Elimination of Tuberculosis about the TB situation in the Southeast,²³ historically the region of the country with the highest rates of TB.^24- 25

Although non-Hispanic blacks contributed more than 50% of the decline in TB/HIV cases and experienced the greatest decline in the absolute number of TB/HIV cases compared with the other racial/ethnic groups, the TB/HIV comorbidity rate in this group remained more than double that of Hispanics and more than 4 times higher than that of non-Hispanic whites. The AIDS case rate of African Americans in 2003 was nearly 10 times higher than that of whites and nearly 3 times that of Hispanics, which may partly explain this finding.²⁶ The LTBI rates are also much higher among non-Hispanic blacks than among whites.²⁷

The substantial declines in TB/HIV comorbidity rates in injection and noninjection drug users and in persons living in correctional facilities may reflect intense efforts in targeting these groups for HIV and TB care and prevention and greater attention to infection control. Many correctional facilities, for example, are now conducting universal TB and LTBI screening among inmates,²⁸ routinely testing inmates for HIV, and providing inmates with medications and medical management.²⁹ A study in New York, New York, demonstrated a reduction in TB among methadone users after they were targeted for LTBI screening and treatment.³⁰ Nonetheless, the TB/HIV comorbidity rate in this group remains high (44% in 2003), suggesting the need for additional and continued control efforts. The decrease in the number of TB/HIV cases among the homeless is related to the overall decrease in the number of homeless people with TB. A number of communities have implemented more intense TB screening programs among the homeless.³¹ However, the relatively stable TB/HIV comorbidity rate among homeless persons may be attributable to the difficulty in providing HIV testing and care and LTBI treatment to this population.

The clinical findings in this report are consistent with what is known about the presentation of TB in HIV-infected persons. The higher rate of extrapulmonary TB in TB/HIV is well known.³² Among persons with pulmonary TB, cavitary disease is less common in HIV cases.³³ The higher rate of acid-free bacillus smear–positive pulmonary TB in TB/HIV cases in 1993 is surprising, as many published reports suggest that smear positivity is less common in TB/HIV cases.^34- 35 However, in 2004, the rates were about the same in TB/HIV cases and non-HIV TB cases. One explanation for the 1993 finding may be that there was less aggressive screening for active TB in HIV-infected persons, especially when results of the chest radiographs were atypical. The dramatic decrease in MDR TB in TB/HIV cases is a reflection of the successful effort to promptly identify drug resistance in HIV-infected patients with TB and measures to reduce MDR TB transmission after the well-publicized MDR TB outbreaks in New York and Florida.³⁶ Among TB/HIV cases, the decrease in rifampin-monoresistant TB, known to be associated with TB/HIV, may be partly explained by the increased use of directly observed therapy, leading to higher rates of successful treatment.⁹ Another contributing factor may be the increased use of HAART, resulting in less immunosuppression and less need for rifabutin therapy and antifungal prophylaxis, all known risk factors for rifampin monoresistance.^37- 38

The remarkable decrease in the number of persons with TB/HIV coinfection who died during TB therapy is attributable to the earlier identification of TB and the appropriate treatment, earlier diagnosis, and treatment of HIV combined with the use of HAART beginning in 1996. Increased use of directly observed therapy may also have contributed to higher TB treatment completion rates in both TB/HIV cases and non-HIV TB cases.

A major strength of the analysis is that it provides national population-based data from the TB national surveillance system. To our knowledge, it is the first analysis that examines the impact of the HIV epidemic on all TB cases over the past decade. There are, however, a number of limitations. First, we did not have temporal information on the relationship of HIV infection to the reported TB cases. The TB/HIV cases may therefore be underestimated because we do not know if the persons with TB were subsequently diagnosed as having HIV/AIDS. It is likely that a significant number of patients with TB are diagnosed as having HIV at the time of TB diagnosis.³⁹ Second, the percentage of patients with TB with reported HIV test results was only 36% in 1993. It increased to 72% in 2004. The results of the propensity stratification suggest that the increase in the percentage of patients with TB with reported HIV test results underestimates the decrease, because more HIV-infected patients with TB would have been identified in 1993. More important, the trends comparing the observed and estimated comorbidity rates are essentially the same and do not change interpretation.

The analysis confirms that general TB control efforts combined with advances in HIV treatment have likely contributed to the substantial decreases in TB/HIV comorbidity. The increasing number of persons living with HIV/AIDS as a result of HAART appears to have had little impact on TB/HIV comorbidity, perhaps because patients treated with HAART have a lower risk of TB disease progression. Continued progress in reducing TB/HIV comorbidity will depend on sustaining these successes. It will also require increasing access to HIV diagnosis and care and reducing HIV transmission among high-burden (ie, groups with high TB rates) and high-risk populations. An important element of these efforts will be better coordination of TB and HIV control programs, which in turn will allow improvements in TB prevention in HIV-infected persons. The CDC currently recommends that patients with TB be counseled about HIV prevention and tested for HIV. Those found to be HIV infected should be referred for HIV treatment. Newly identified HIV-infected patients should be screened for TB and LTBI and treated accordingly. Expanding HIV testing in HIV risk groups, followed by TB screening, would identify additional candidates for HAART and LTBI treatment and thus lead to an accelerated decline in TB/HIV. The provision of LTBI treatment has been correlated with marked decreases in TB/HIV in some settings.⁴⁰ As many TB/HIV cases now occur in persons whose HIV infection was diagnosed as a consequence of the development of TB,³⁹ these additional efforts to identify HIV-infected persons earlier will have the greatest impact on preventing progression to TB in coinfected persons.

Our findings suggest that the groups that contributed most to the decline in TB cases should remain the focus of TB control efforts, because they still account for the highest burdens of comorbidity. Special attention should be given to subpopulations, such as non-Hispanic black men aged 25 to 44 years living in the South, who account for a disproportionate burden of comorbidity. Groups that have experienced slower declines in comorbidity rates, such as those living in southern communities, women, persons aged 45 to 64 years, and persons born outside the United States, especially foreign-born blacks, should also be targeted for services.

The United States faces a changing epidemiologic landscape in the 2 epidemics compared with much of the rest of the world, where rates of TB/HIV comorbidity continue to increase. A remaining concern is the possibility that the successes in controlling TB/HIV comorbidity and TB morbidity overall will lead to a renewed cycle of neglect. This complacency must be avoided if we are to make continued progress toward the eventual elimination of TB in the United States.

Correspondence: Rachel Albalak, PhD, Division of Tuberculosis Elimination, Centers for Disease Control and Prevention, 1600 Clifton Rd NE, Mail Stop E10, Atlanta, GA 30323 (rka3@cdc.gov).

Accepted for Publication: July 3, 2007.

Author Contributions: Dr Albalak had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Albalak, Marks, Castro, and Moore. Acquisition of data: Albalak, Kammerer, and Moore. Analysis and interpretation of data: Albalak, R. J. O’Brien, Kammerer, S. M. O’Brien, Marks, Castro, and Moore. Drafting of the manuscript: Albalak, R. J. O’Brien, and S. M. O’Brien. Critical revision of the manuscript for important intellectual content: Albalak, R. J. O’Brien, Kammerer, S. M. O’Brien, Marks, Castro, and Moore. Statistical analysis: Kammerer and S. M. O’Brien. Administrative, technical, and material support: Albalak, S. M. O’Brien, Marks, and Moore. Study supervision: Albalak, Castro, and Moore.

Financial Disclosure: None reported.

Additional Contributions: The state and local health departments across this nation collected the data on which this article is based. Robert Pratt of Northrop Grumman assisted in preparing the dataset for analysis, and Lilia Manangan, MPH, Division of Tuberculosis Elimination, CDC, provided helpful comments on the analyses. Tom Navin, MD, Michael Iademarco, MD, and Jose Becerra, PhD, Division of Tuberculosis Elimination, CDC, critically reviewed the manuscript.