The primary responsibility for treating, and oftentimes diagnosing, patients with tuberculosis (TB) and latent tuberculosis infection (LTBI) falls to state and local TB control programs [1]. The number of current and future TB cases is an important measure of burden on TB control programs and has implications for setting priorities among TB prevention and control efforts. Health departments not accustomed to serving large foreign-born populations may experience an increase in TB cases as a result of increasing numbers of foreign-born residents, who may be arriving from countries with high rates of TB. As foreign-born persons make up an increasing proportion of TB cases in the United States, predicting TB trends will help TB control programs identify needs for cultural competency, including foreign language speakers and interpreters needed for translating materials or engaging in appropriate community outreach.

Several studies have predicted future TB trends [2], [3], [4], [5], [6], [7], [8], [9], [10] however we are not aware of any published studies that predict future TB case counts in the United States among foreign-born persons from specific countries of birth. As progress continues in decreasing the burden of TB in the United States [11], understanding TB trends among the foreign-born groups that make up the largest proportion of TB cases will assist TB control programs in concentrating limited resources where they can provide the greatest impact. The purpose of this study was to help guide TB control and program planning. We achieved this objective by predicting the numbers of U.S. TB cases among U.S.-born, foreign-born and foreign-born persons from selected countries of birth.

From 2000 to 2010 a total of 153,353 cases of tuberculosis were reported in the United States; 45% (69,189) were among U.S.-born persons and 55% (83,625) were among foreign-born persons. TB cases among U.S.-born persons declined 49% from 2000 to 2010. The number of TB cases among foreign-born persons remained stable from 2000 to 2008 (0.6% decline), with an 11% decline in observed counts from 2008 to 2010.

The top 5 countries of birth among foreign-born persons diagnosed with TB for each year from 2000 to 2010 were Mexico, the Philippines, Vietnam, India and China. The greatest numbers of TB cases among foreign-born persons diagnosed in the United States were consistently among persons from Mexico, followed by the Philippines (24% and 11% of all TB cases among foreign-born persons between 2000 and 2010, respectively).

TB cases are predicted to decline among all persons in the U.S. A steeper decline is predicted among U.S.-born than foreign-born persons, which has implications for state and local TB control programs and TB elimination in the U.S. Among the top 5 countries of birth of foreign-born TB patients diagnosed in the United States, our model predicted a significant decrease and 2010 observed case count above the upper bound of the 2020 prediction interval only for foreign-born persons from Vietnam. The remaining countries (Mexico, the Philippines, India, and China) exhibited flatter slopes and 2020 prediction intervals that included the value of the 2010 observed count; consequently our model results for 95% prediction intervals indicate that cases among persons from these countries have the potential to increase or decrease. Modeling case counts from a longer time period (1993 to 2010) resulted in a significant increase in predicted estimates of cases among persons born in India, but ignores the change in the deceleration rate of TB from the 1990s to the 2000s [16]. We primarily presented single slope regression models for comparability across groups. Using a segmented regression approach with more than 1 slope when indicated resulted in even greater predicted declines in TB case counts by 2020. Thus, our predicted estimates from single slope regression models are conservative compared to other types of prediction models.

TB cases among persons who were foreign-born and foreign-born from the top 5 countries of birth experienced significant changes in trend in 2007 resulting in steeper declines during the latter part of the decade. However, among the individual top 5 countries of birth a significant change in TB case count trend from 2000–2010 was indicated only for Mexico. Mexico’s significant changes in trends occurred in 2004 (for a single joinpoint model), or in 2003 and 2006 (for a double joinpoint model). The single joinpoint change in trend among foreign-born from Mexico preceded the significant change in trend among all foreign-born by at least 3 years. Therefore, we hypothesized that the change in cases among persons from Mexico, which accounted for nearly a quarter of all foreign-born TB cases in the United States, influenced the significant change to a decreasing trend among all foreign-born persons. When Mexico was removed from foreign-born, results were consistent with this hypothesis; though the slope of the regression line among foreign-born excluding Mexico decreased in a similar manner as foreign-born including Mexico, the change in trend among foreign-born excluding Mexico occurred in 2008, a year later than it occurred for all foreign-born (data not shown). Thus the decrease in TB cases among foreign-born persons from Mexico influenced the year in which the foreign-born change in trend occurred and, subsequently, will continue to influence TB trends among foreign-born persons in coming years.

In 2001, TB case rates were 5 times higher among foreign-born persons from Mexico than U.S.-born persons, and Mexican-born TB patients were more likely than U.S.-born TB patients to have multidrug-resistant TB [17]. Furthermore, in California and Texas TB morbidity among U.S.-born persons is higher in counties that border Mexico than in the rest of the state [17]. The number of persons from Mexico emigrating to the U.S. declined by 60% from 2006 to 2010 [18], however this decrease has had minimal impact on the percentage of foreign-born TB cases attributed to persons from Mexico [12]. The unique challenges associated with TB among foreign-born persons from Mexico need to be continually addressed if the decline of TB among foreign-born persons is to be accelerated, thereby continuing the decline of TB in the U.S. [17], [19]. Bi-national programs, such as California’s CureTb and Texas’ TB Net, as well as TB control programs for Mexico-born migrant workers and non-immigrants, are necessary to diagnose, treat, and thereby reduce TB cases among foreign-born persons from Mexico.

Immigration, which is directly affected by national policies and socioeconomic factors [20], is an important consideration in the number of TB cases among foreign-born persons [21], [22]. Major changes in immigration patterns could alter our case count predictions. In order to assess the influence of the population base size on TB case count predictions, population denominators from the U.S. Census Bureau’s Current Population Survey (http://www.census.gov/cps/) were incorporated into analysis using 2000–2010 TB case rates. We predicted a decline in TB case rates between 2010 observed and 2020 predicted for all persons and persons who were U.S.-born, foreign-born, foreign-born from the top 5 countries, foreign-born from countries other than the top 5, and foreign-born from Mexico, the Philippines, Vietnam, India and China (data not shown). The predicted declines in TB rates are consistent with our TB case count prediction results.

TB prevention and control strategies, which focus on interrupting recent transmission, have been successful in reducing the number of U.S.-born TB cases but have had limited impact on preventing disease among foreign-born persons entering the United States from countries with high rates of LTBI [21]. LTBI testing and treatment of foreign-born persons from the top 5 countries of birth, particularly those persons who have recently arrived in the U.S., would likely increase the rate of decline of TB cases among foreign-born persons [21]. A newly recommended LTBI treatment regimen [23], which reduces treatment time from 9 months to 12 weeks, increased LTBI treatment completion rates among study populations [24], [25] and could increase LTBI treatment completion in the future as the new regimen is implemented. Unfortunately recent shortages of tuberculin skin test antigens, used for detection of LTBI, and isoniazid, an important drug in the treatment of LTBI and TB disease, have restricted LTBI testing and treatment efforts in some states [26], [27]. Continuation of these shortages could lead to underdiagnoses, incomplete administration of medication and treatment completion failure resulting in an increase in new, recurrent and drug resistant TB disease. Challenges such as these may continue to impact the successful diagnosis and treatment of LTBI and TB disease in the future.

In 2007, the Centers for Disease Control and Prevention issued updated Technical Instructions (TIs) for TB screening of foreign-born persons entering the United States [28]. The updated TIs include guidance on performing TB culture to enhance TB diagnosis and administering directly observed therapy to increase treatment completion. Mexico and the Philippines began screening all applicants to the United States using the updated TIs in 2007, Vietnam in 2008, China in 2009 and India in 2010. A preliminary study of persons immigrating to the United States from Mexico, the Philippines and Vietnam suggests that the updated TIs improved case detection prior to U.S. entry [29]. Evaluation of the newly implemented TIs will be important in determining their impact on TB cases among the foreign-born.

Foreign-born persons who arrive in the United States without going through the immigrant and refugee application process (for example nonimmigrant visitors such as tourists and students, and undocumented foreign-born persons) would not be evaluated using the TIs and thus not be screened for TB regardless of their country of birth. A recent study estimated that between 2001 and 2008 56% of TB cases among newly arrived foreign-born persons were among nonimmigrant visitors who did not undergo TB screening prior to arrival [30]. Furthermore, in 2008 approximately 30% of foreign-born persons in the U.S. were undocumented [31]. Targeted overseas screening of nonimmigrant visitors from countries with a high burden of TB who plan to stay in the U.S. for a long period of time could decrease the number of TB cases among foreign-born persons [30].

The TB prediction methodology described here did not take into consideration factors that affect TB trends among foreign-born persons in the United States such as TB prevalence in the country of birth, immigration patterns, socioeconomic and health status, and global TB control efforts, nor the potential for testing and treatment disruptions due to shortages noted previously. In addition, although the relationship between Human Immunodeficiency Virus (HIV) infection and TB is well described [32], we were not able to incorporate information on HIV status due to underreporting and lack of published HIV incidence and prevalence estimates in the United States by country of origin. We assumed current trends in TB cases will continue with cases declining in a linear pattern; however, trends can be affected by changes in behavior, policies, or other factors that influence population and immigration [20], [22]. Although the linear regression model was not a good fit for all countries, we showed that our prediction results were generally conservative in relation to more complex regression models.

The described prediction analysis had several strengths. The model can be customized for different data subsets, and can be applied to other diseases and conditions. The World Health Organization uses a similar methodology to forecast TB prevalence and mortality rates [10]. Recent data show that our model by country of origin accurately predicted 2011 TB case counts; 2011 observed case counts for all subgroups were within the 95% prediction intervals calculated by our prediction model [12]. Results from our model closely reflect U.S.-born and foreign-born findings from a deterministic mathematical model recently published [4], with the additional benefit that we provide detail by country of origin.