Trends in high-risk populations among patients with tuberculosis, National Tuberculosis Surveillance System, 1997-2006. HIV indicates human immunodeficiency virus.
Trends in high-risk populations among US-born patients with tuberculosis, National Tuberculosis Surveillance System, 1997-2006. HIV indicates human immunodeficiency virus.
Substance abuse–related barriers to tuberculosis (TB) control in the United States.
Oeltmann JE, Kammerer JS, Pevzner ES, Moonan PK. Tuberculosis and Substance Abuse in the United States, 1997-2006. Arch Intern Med. 2009;169(2):189-197. doi:10.1001/archinternmed.2008.535
Tuberculosis (TB) control efforts are often ineffective in controlling TB among patients who use illicit drugs or abuse alcohol (substance abuse). This study examined the prevalence of substance abuse among TB cases reported in the United States and assessed the relation between substance abuse and indicators of TB transmission.
A cross-sectional analysis was performed of data on US TB cases in patients 15 years or older reported from 1997 through 2006. Analyses included number and proportion of patients with TB characterized by substance abuse and associations between substance abuse, sputum smear status, treatment failure, and inclusion in a county-level genotype cluster.
Of 153 268 patients with TB, 28 650 (18.7%) reported substance abuse, including 22 293 of 76 816 US-born patients (29.0%). Multivariate analysis showed that, among patients negative for human immunodeficiency virus, odds of sputum smear–positive disease were 1.8 (99% confidence interval [CI], 1.7-1.9) times greater among those who reported substance abuse; this association was weaker among patients with human immunodeficiency virus infection (odds ratio [OR], 1.2; 99% CI, 1.1-1.4). Among female patients, odds of treatment failure were 2.4 (99% CI, 1.9-3.0) times greater among those who reported substance abuse. The association was weaker among male patients (OR, 1.5; 99% CI, 1.3-1.7). Patients who abused substances were more likely to be involved in a county-level genotype cluster (US-born: OR, 2.3; 99% CI, 2.0-2.7; foreign-born: 1.5; 1.2-2.0).
Substance abuse is the most commonly reported behavioral risk factor among patients with TB in the United States. Patients who abuse substances are more contagious (eg, smear positive) and remain contagious longer because treatment failure presumably extends periods of infectiousness. Increased transmission is consistent with our finding that patients who abuse substances were more likely to be involved in a localized genotype cluster, which can represent recent transmission.
The annual tuberculosis (TB) incidence in the United States (4.6 cases per 100 000 people) is among the lowest in the world,1 and the number of TB cases reported in the United States in 2007 (13 293) was the lowest ever recorded.2 However, risk of TB continues to be higher among certain demographic groups, including ethnic minorities, immunocompromised people, those who have a history of homelessness, and foreign-born persons.3- 6 The epidemiology of TB in the United States has been best summarized by Iseman7: “As the epidemic tide of tuberculosis recedes from the shores of America, small tidepools of disease remain behind: pools populated by immigrants, the elderly, and the immunocompromised.”
A less frequently discussed tidepool that complicates US TB control efforts is the illicit drug–using and alcohol-abusing population. Although several reports of TB transmission and outbreaks among those who abuse substances have been published,8- 14 no national summary of the impact of illicit drugs and alcohol abuse on TB control has been reported. We studied the prevalence of substance abuse among TB cases reported in the United States. The study also included a national assessment of substance abuse as a risk factor for TB transmission. Because substance abuse is associated with other routinely collected factors related to TB transmission, these data provide an opportunity to estimate the association between substance abuse and measures of transmission while controlling for the effects of other factors.
The study population consisted of all incident TB cases in persons 15 years or older reported to the Centers for Disease Control and Prevention National Tuberculosis Surveillance System (NTSS) from all 50 states and the District of Columbia from 1997 through 2006. Data collected included demographic, clinical, and behavioral information on each patient.
We obtained genotyping data from the Centers for Disease Control and Prevention National Tuberculosis Genotyping Service,15 which uses spacer oligonucleotide typing and mycobacterial interspersed repetitive unit polymerase chain reaction–based genotyping for molecular characterization.16 For the genotyping analysis, we included all incident, culture-positive TB cases from 2004 and 2005 for which valid genotyping results and matching NTSS data were available. We defined a genotype cluster as 2 or more patients with TB from the same county and state whose genotyping results exhibited identical spacer oligonucleotide typing and mycobacterial interspersed repetitive unit patterns.
Substance abuse was defined as self-reported excessive alcohol use, non–injection drug use, or injection drug use during the year before TB diagnosis. Patients were asked whether they had used alcohol to excess during the year before TB diagnosis. We calculated the 10-year prevalence of substance abuse, as well as the prevalence of each of the substance abuse categories for the entire sample, and stratified by country of birth and sex.
We created graphs to illustrate the 10-year trend in the prevalence of reported substance abuse among all patients with TB and among patients with TB born in the United States. Trends in the prevalence of 5 other established risk factors for TB were plotted on the same graph to illustrate the prevalence of substance abuse relative to the prevalence of these other risk factors (ie, homelessness, human immunodeficiency virus [HIV] infection, employment in a high-risk occupation, residence in a congregate setting, and recent immigration to the United States). We defined homelessness as self-reported homelessness within 12 months before TB diagnosis; high-risk occupation as employment as a health care worker, correctional facility employee, or migrant worker; a congregate setting such as a long-term care facility or a correctional facility; and recent immigration as immigration to the United States within 2 years before TB diagnosis. We used a χ2 test for trend to assess whether the proportion of patients with TB reporting substance abuse remained stable during the 10-year study period.
We compared the distribution of demographic, clinical, and behavioral characteristics among patients who reported substance abuse with that among patients who did not. A significance level of .01 was used for all statistical tests.
Self-reported excess alcohol use had associations similar to those for illicit drug use on our outcomes of interest: TB treatment outcome and sputum smear result. Therefore, to assess the association between substance abuse and TB transmission, we created one composite substance abuse variable that reflected any of the following: injection drug use, non–injection drug use, or excessive alcohol use. We used 2 multivariate logistic regression models to determine the extent to which substance abuse was associated with sputum smear–positive disease and failure to complete therapy. We categorized patients as having failed to complete therapy if the reason therapy was stopped was reported as lack of cooperation, loss to follow-up, or treatment refusal. Patients who died during therapy or whose therapy completion status was unknown were removed from the analysis. We excluded patients who died because cause of death was not recorded and therefore may have been unrelated to TB. Because therapy outcome information had not yet been reported to the NTSS for many patients whose TB was diagnosed in recent years, we restricted the analysis of the relationship between substance abuse and treatment failure to patients whose cases were reported from 1996 through 2004.
We restricted the analysis of the relationship between substance abuse and sputum smear status to patients with pulmonary TB or both extrapulmonary and pulmonary TB. Patients whose sputum smear status was unknown or missing and those for whom a test was not done were excluded from this analysis. Covariates included in both original models were age, sex, race/ethnicity, HIV infection status, homelessness, and country of birth. Disease site (pulmonary vs extrapulmonary) was included only in the model designed to measure the association between substance abuse and treatment failure. After assessing for effect modification with the Wald statistic, we removed those variables that did not distort the association between substance abuse and the outcome in question by more than 10%. Because California reports only positive HIV results to the NTSS, we excluded California patients with TB from our examination of the potential confounding effects of patients' HIV status. Covariates were removed from the model by backward elimination.
After stratifying patients by country of birth, we calculated odds ratios (ORs) and 99% confidence intervals (CIs) to compare the proportion of substance-abusing patients whose isolate was part of a genotype cluster with the proportion of non–substance-abusing patients whose isolate was part of a genotype cluster.
From 1997 through 2006, 153 268 cases of TB among US residents 15 years or older were reported to the NTSS, of which 145 276 (94.8%) had information about substance abuse. Injection drug use was reported by 3972 of these patients with TB (2.6%), non–injection drug use by 11 616 (7.6%), excessive alcohol use by 23 138 (15.1%), and at least 1 category of substance abuse by 28 650 (18.7%) (Table 1). Among patients with TB born in the United States, 3499 (4.6%) reported injection drug use, 9697 (12.6%) reported non–injection drug use, 17 803 (23.2%) reported excessive alcohol use, and 22 293 (29.0%) reported at least 1 category of substance abuse. Among US patients with TB, 10 550 (13.7%) reported excessive alcohol use only, 603 (0.8%) reported injection drug use only, and 2756 (3.6%) reported non–injection drug use only (data not shown). Stratification by country of birth and sex showed large differences in the prevalence of substance abuse. The US-born males reported the highest prevalence of substance abuse (35.4%), followed by US-born females (16.9%), foreign-born males (12.9%), and foreign-born females (1.5%) (Table 1).
As shown in Figure 1, the overall percentage of patients with TB who abused a substance (18.7%) was greater than the percentage who reported other established known risk factors for TB during the study period, including having recently immigrated to the United States (12.9%), having HIV infection (9.5%), residing in a congregate setting (6.6%), being homeless (6.3%), or being employed in a high-risk occupation (4.3%). The prevalence of substance abuse among all patients with TB declined slightly from 19.6% in 1997 to 17.2% in 2006 (χ2 test for trend, P < .001).
Figure 2 demonstrates that the proportion of US-born patients with TB who abused a substance (29.0%) during the study period was greater than the proportion having HIV infection (12.8%), residing in a congregate setting (9.7%), being homeless (10.1%), or being employed in a high-risk occupation (3.2%). There was a slight upward trend in the overall proportion of substance abuse among US-born patients (χ2 test for trend, P = .002).
The proportion of patients who reported substance abuse differed significantly across demographic and clinical characteristics (Table 2). The majority of patients who reported substance abuse were male (82.9%) and born in the United States (77.8%) and did not report homelessness during the past year (75.6%). Although the homeless patients accounted for only 22.7% of the substance-abusing patient population, 6496 of the 9723 homeless patients included (66.8%) reported substance abuse. The largest proportion of patients who reported substance abuse were black (44.2%), followed by Hispanic (24.4%) and white (23.9%) patients. Prevalence rates of substance abuse within racial subgroups showed that 39.0% of black patients, 26.2% of white patients, and 22.7% of Hispanic patients reported substance abuse. The majority of patients who reported substance abuse were not HIV infected (61.2%), were sputum smear–positive (53.9%), and had pulmonary TB (82.7%).
Of the 123 303 reported pulmonary cases, 105 688 (85.7%) had information about both sputum smear status and substance abuse. The crude association between substance abuse and positive sputum smear status was 1.6 (99% CI, 1.6-1.7). The presence of HIV infection modified the effect of substance abuse on sputum smear status (Table 3). Among the HIV-infected patients, the odds of being diagnosed with sputum smear–positive disease were 1.2 (99% CI, 1.1-1.4) times greater among those who reported substance abuse. Among the HIV-negative patients, the association was even greater (OR, 1.8; 99% CI, 1.7-1.9).
Of 126 455 age-eligible patients whose TB was reported from 1996 through 2004, 100 755 (79.7%) had information about treatment outcome and substance abuse history. The crude association between substance abuse and treatment failure was 1.9 (99% CI, 1.8-2.1). This association was modified by sex (Table 4). Homelessness and country of birth confounded the relationship between substance abuse and treatment failure and therefore remained in the final model as control variables. Among male patients, odds of treatment failure were 1.5 (99% CI, 1.3-1.7) times greater among those who reported substance abuse. The association was even greater among female patients (OR, 2.4; 99% CI, 1.9-3.0).
Of 14 157 TB isolates collected in 2004 and 2005, complete genotyping results and valid patient-level information were available for 11 891 (84.0%). Among US-born patients, odds of involvement in a county-level genotype cluster were 2.3 (99% CI, 2.0-2.7) times greater among patients who reported substance abuse. Among foreign-born patients, odds of involvement in a county-level genotype cluster were 1.5 (99% CI, 1.2-2.0) times greater among those who reported substance abuse (Table 5).
Rates of TB have been shown to be high among persons who abuse alcohol and use illicit drugs.17,18 Our results suggest that substance abuse is the most commonly reported modifiable behavior impeding TB elimination efforts in the United States. We found that approximately 1 in 5 patients with TB either use an illicit drug, drink alcohol to excess, or both, and that nearly 1 in 3 US-born patients with TB do so. The prevalence of substance abuse among patients with TB was far greater than the prevalence of other established risk factors for TB, including homelessness, HIV infection, employment in a high-risk occupation, residence in a congregate setting, and recent immigration to the United States. The proportion of US-born patients with TB who reported substance abuse has increased since 1997, suggesting that, relative to other US-born populations with TB, rates of TB among those who abuse substances are declining more slowly. The prevalence of substance abuse among patients with TB varies greatly according to country of birth, sex, and race. Substance abuse among patients with TB is most prevalent among the US-born (29.0%), male (32.8%), and black (39.0%) patients (data not shown). Prevalence of substance abuse was also relatively high among white (26.2%) and Hispanic (22.7%) patients (data not shown).
A limitation of this analysis is the self-reported nature of the substance abuse data. It is therefore difficult to determine the validity of these data. However, the ascertainment of substance abuse information may be reasonably accurate because the treatment course of at least 6 months, usually with directly observed therapy, provides more opportunities to identify substance abuse than a single interview would. Given the stigma associated with excess alcohol use and illicit drugs, we suspect that any misclassification would lead to an underestimate of the prevalence of substance abuse. It is also important to note differences in the way information regarding certain covariates are collected because it influences the prevalence of the covariate. For example, homelessness was recorded if present at any time during the year before TB diagnosis, whereas correctional institution residence and long-term care facility residence were recorded only if present at diagnosis.
Alcohol consumption is related to Mycobacterium tuberculosis transmission,19 developing TB disease,10,20,21 poor TB treatment outcomes, and mortality.22,23 Most substance abuse described in this report is attributed to excessive alcohol use. Although not an illegal substance, alcohol when used to excess has effects similar to those of illicit drug use on TB treatment outcomes and degree of infectiousness. We found that 15.1% of our study participants used alcohol to excess. For comparison, the Behavioral Risk Factor Surveillance System provides nationwide estimates of health-related behaviors for adults 18 years and older and reported that, in 2002, 5.9% of respondents reported long-term drinking of alcohol.24
Substance abuse has been associated with TB transmission and the generation of secondary TB cases.25 Our results suggest that those who abuse substances were more likely to transmit M tuberculosis because they were more likely to have sputum smear–positive disease and to experience treatment failure, which could lead to a longer period of infectiousness. The association between substance abuse and county-level cluster involvement supports this hypothesis because membership in a localized genotype cluster can be a marker for recent transmission in regions with a low prevalence of TB, such as the United States.26 A potential limitation with this analysis is the lack of epidemiologic data to confirm transmission among patients in a cluster. However, the probability that a genotype cluster represents recent transmission is likely similar among those who do and do not abuse substances and therefore would not bias our measure of association. Our calculation of clustering rates based on county-level data rather than traditional state-level or country-level data means that the observed association with substance abuse is all the more likely to reflect recent transmission.
The relation between substance abuse and increased transmission of M tuberculosis can be explained in several ways, some of which are indirect and revolve around delayed diagnosis and difficulties identifying at-risk contacts, screening them for TB, and treating patients with positive findings (Figure 3). Persons who abuse substances may have less access to routine medical care, potentially leading to delayed diagnoses. As the disease progresses, patients tend to become more contagious. During a TB outbreak investigation among long-term marijuana users, despite feeling ill and reporting increased coughing, patients delayed seeking care until their cases of TB were advanced and primarily cavitary.13
Efforts to reduce TB transmission in the United States are based largely on contact investigations. Contact investigations are multistep interventions in which persons exposed to patients with contagious TB are systematically examined on the basis of the amount of time they spent with the contagious person, the environment in which the exposure took place, and their own host susceptibility.27 During these investigations, patients with TB are interviewed by public health workers to elicit the names of other people who were exposed to them and venues in which the exposure took place. Although contact investigations are an important component of the overall tuberculosis elimination strategy in the United States, for many reasons, they are not conducted well even in successful TB control programs.28,29 The limited success of TB contact investigations is exacerbated when patients who abuse substances are involved14 because they are often reluctant to name fellow substance-using contacts or the venues of substance-abusing behaviors.12,13,30- 32
Even when venues are successfully identified, substance abuse can increase the likelihood of TB transmission. Often substance abuse occurs in enclosed spaces with deliberately limited or poor ventilation and high volumes of human traffic, increasing the likelihood of TB transmission.9,13,30,31 The prolonged use of many inhaled or smoked drugs, particularly crack cocaine, leads to increased coughing and other adverse pulmonary effects.33 Furthermore, persons who abuse substances may be less likely to seek treatment early during the course of their illness, resulting in an extended period of infectiousness and advanced disease at diagnosis.12,14
Patients who abuse substances are also less likely to complete TB screening34 or to initiate and complete treatment either for latent TB infection32,35,36 or for tuberculosis disease.30,37- 39 The results of this poor adherence to treatment are compounded by substance abusers with latent TB infection being at increased risk of progressing to active disease because of immunologic impairments.33,40 Persons who abuse alcohol,41 crack cocaine,42,43 heroin,44 marijuana,43 and methamphetamines45 have all been shown to experience significant impairment to one or more important immunologic mechanisms.
The effects of substance abuse have clinical implications for treatment of both latent infection and active disease. With few exceptions, antituberculosis agents are metabolized by the liver.46 The liver, which is central in drug metabolism and detoxification, is vulnerable to injury. Several studies have found that excessive alcohol use is a significant predictor of TB drug–induced liver injury among patients with TB.47- 49 Because substance-abusing patients with TB being treated for TB are at high risk for severe liver toxic effects, they require vigilant biochemical and clinical monitoring for hepatotoxic symptoms. The American Thoracic Society recommends that patients being treated for latent TB infection receive biochemical monitoring of alanine aminotransferase if they use alcohol to excess, take concomitant hepatotoxic drugs, or have viral hepatitis.50 In the treatment of patients with severe liver disease who abuse substances, reducing the number of potentially hepatotoxic antituberculosis medications normally given to patients with TB and extending the normal TB treatment period are generally accepted as reasonable treatment modifications.50- 52
Despite the numerous challenges associated with controlling TB among those who abuse substances, some strategies focusing on those who inject illicit drugs have been successful. For example, monetary incentives have been shown to produce higher rates of successful TB screening and completion of treatment for latent TB infection among illicit-drug users,34,36,53- 56 and targeted TB testing and directly observed therapy have been shown to be both successful and cost-effective among participants in needle exchange programs and clients of methadone maintenance clinics.56- 61 Although these public health interventions appear to help control TB among injection drug users, few successful efforts to control TB have been reported among those who abuse alcohol and noninjection drugs, and these were the patients who constituted most of the group who abused substances in this report.
In 1979, Reichman et al17 identified drug dependence as a new risk factor for developing TB disease. Since then, the TB control community has been slow to recognize that TB is difficult to control in isolation and to develop effective interventions for those who abuse substances. Tuberculosis control and substance abuse programs need to work together to simultaneously treat the diseases of addiction and TB. Many problems associated with controlling transmission among those who abuse substances arise from the reluctance of substance-abusing patients to name exposed contacts and to complete therapy. It is critical for health care workers to develop a trusting relationship with substance-abusing patients to ensure that contacts are identified and the patients and infected contacts complete treatment. Currently, there are no national guidelines that specifically address the treatment of TB among patients who abuse substances. The Centers for Disease Control and Prevention's National Center for HIV, Hepatitis, STD (sexually transmitted disease), and TB Prevention has commissioned a working group that is currently developing integrated guidelines to provide more comprehensive services for people addicted to substances.
Correspondence: John E. Oeltmann, PhD, Division of Tuberculosis Elimination, Centers for Disease Control and Prevention, 1600 Clifton Rd NE, Mailstop E10, Atlanta, GA 30333 (firstname.lastname@example.org).
Accepted for Publication: June 15, 2008.
Author Contributions: All authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Oeltmann, Moonan, and Pevzner. Acquisition of data: Oeltmann, Kammerer, and Moonan. Analysis and interpretation of data: all authors. Drafting of the manuscript: all authors. Critical revision of the manuscript for important intellectual content: all authors. Statistical analysis: all authors. Administrative, technical, and material support: Moonan. Study supervision: Oeltmann and Moonan.
Financial Disclosure: None reported.
Funding/Support: Centers for Disease Control and Prevention, Division of Tuberculosis Elimination.
Additional Contributions: Maryam Haddad, MSN, MPH, FNP, provided critical revision of the manuscript and writing and editing assistance. Thomas Navin, MD, and Ryan Wallace, MPH, provided critical revision of the manuscript. No compensation was received by any of the individuals named. We are also grateful to the anonymous peer reviewers for their insightful critical review.