Greenwood and Warriner TB Paper PRPR

7/27/2019 Greenwood and Warriner TB Paper PRPR

1/23

1 23

Population Research and Policy

Review

in cooperation with the Southern

Demographic Association (SDA)

ISSN 0167-5923

Volume 30

Number 6

Popul Res Policy Rev (2011) 30:839-859

DOI 10.1007/s11113-011-9213-6

Immigrants and the Spread of Tuberculosisin the United States: A Hidden Cost of

Immigration

Michael J. Greenwood & Watson

R. Warriner


2/23


3/23

Immigrants and the Spread of Tuberculosis

in the United States: A Hidden Cost of Immigration

Michael J. Greenwood Watson R. Warriner

Received: 8 February 2011 / Accepted: 16 August 2011 / Published online: 30 August 2011 Springer Science+Business Media B.V. 2011

Abstract This panel-data study concerns the incidence of newly diagnosed

tuberculosis (TB) in specific U.S. metropolitan areas among immigrants and, in

turn, the possible transmission of the disease to the native-born population of these

same metropolitan areas. The study includes 50 U.S. Metropolitan Statistical Areas

as annual observations, 19932007. We find that a 10% increase in the number of

high-incidence immigrants results in a 2.87% increase in TB among the foreign-

born population, and that a 10% increase in the number of foreign-born TB casesincreases the number of new TB cases among the native-born by 1.11%. The study

concludes with a benefit/cost analysis of the societal cost of TB and suggests that

testing all immigrants for TB would be a cost-effective method to limit the amount

of TB that enters U.S. from abroad, thus limiting the transmission to both the

foreign- and native-born populations.

Keywords Tuberculosis Immigrants United States Cost of disease

Introduction

Tuberculosis (TB) has long been one of the worlds deadliest diseases. Moreover,

according to a recent article in The Lancet, human migration has had a major effect

on the spread of tuberculosis throughout the course of human history (Blumberg

et al. 2010, p. 2127). In our paper, using unique data, we focus on the United States

and uncover a link between the settlement in specific metropolitan areas of U.S.

immigrants and the incidence1 of TB among both the foreign and the native born in

M. J. Greenwood (&) W. R. WarrinerDepartment of Economics, University of Colorado, Campus Box 256, Boulder, CO 80309, USA

e-mail: [email protected]

1 Incidence rates refer to the number of newly reported cases per 100,000 persons and not to the overallprevalence of the disease.

123

Popul Res Policy Rev (2011) 30:839859DOI 10.1007/s11113-011-9213-6


4/23


5/23

etc.) due to its inability to spread quickly in rural areas. In fact, it was not until the

Industrial Revolution of the 18th and 19th centuries, and the subsequent rapid

urbanization, that western civilization became aware of the devastating impacts of

TB. In the 1700s, for example, annual TB mortality rates in the U.K. reached about

one percent of the population (Collins 1997).TB is a contagious disease; the bacteria are transferred via tiny water droplets

from either a cough or sneeze of an infected person, and once in contact, a person

can take months or years to develop symptoms (NHS 2010). Three possible

scenarios may occur once the bacteria are in contact with the host. For the majority

of cases, the body kills the infection and no further symptoms are experienced. If the

immune system cannot kill the bacteria, it may instead isolate the infection,

resulting in no further symptoms unless the disease is activated. This case is referred

to as a latent infection of TB; it cannot be transmitted in this form, but can be

reactivated if the immune system is compromised. Lastly, the immune system mayfail to kill and contain the infection, resulting in the movement of TB to the lungs.

This result is the active form of TB (NHS 2010).

If the TB infection becomes active, the patient should receive immediate and

precise care. If left untreated, TB results in a two-thirds mortality rate within the

first 5 years of infection (Laxminarayan et al. 2007). TB is usually treatable through

chemotherapy, which involves the administration of 34 drugs on a daily basis.

Although the average course of treatment is 6 months when the regimen is daily,

some treatments take longer to ensure that no reactivation occurs (CDC 2003).

Failure to complete the course of treatment may cause the strain to become drug-resistant; it is estimated that nearly 5% of all new global TB cases in 2008 were

multi-drug resistant, compared to only 1.2% in the U.S. for 2007 (WHO 2008)

(CDC, OTIS 2010b). The lower MDR rates in the U.S. are likely a result of the great

emphasis placed on direct observed therapy2 to ensure all treatments are completed.

One of the reasons TB chemotherapy is costly and time consuming is the fear of an

increase in the MDR TB rate. Complacency in TB treatment increases the likelihood

that the TB strain will become resistant, leading to stronger and costlier forms of the

disease (CDC 2010a).

Prevention of tuberculosis was first somewhat attainable with the invention of the

Bacillus Calmette-Guerin (BCG) vaccine in the early 20th century. The BCG

vaccine is typically administered only to children and remains effective for

approximately 15 years. Although rarely used in the U.S., billions of people have

been administered BCG; vaccination coverage exceeds 90% in countries such as

India, Vietnam, Mexico, China, and the Philippines (Weibrod 2010). The BCG

vaccine is popular mainly because of its documented protective effect against

Meningitis and Disseminated TB3; however, BCG does not protect against primary

TB infections or the reactivation of latent infections (WHO 2010a). Since the

2 Direct Observed Therapy (D.O.T.) is the goal of all TB elimination programs to ensure all treatmentcourses are successfully completed. D.O.T. consists of direct monitoring from a government or WHOaccredited doctor.3 Disseminated (or Miliary) TB is a rare occurrence (13% of all TB cases) in which the primaryinfection of TB (lungs) moves to other parts of the body, including the bones and joints, organ linings,

bronchus, and skin.

Immigrants and the Spread of Tuberculosis in the United States 841

123


6/23

efficacy of BCG begins diminishing 15 years after administration, older populations

are still unprotected. BCG vaccine is not sufficiently effective to be used as a

method for TB elimination, but to date remains the best way to protect children.

TB elimination efforts within the U.S. have been highly successful, mainly due to

adequate resources and diligence in TB treatment monitoring (American ThoracicSociety 2005). U.S. TB incidence rates were as low as 4.2 cases per 100,000 persons

in 2008, having decreased nearly 50% since 1992, and almost 90% since the 1950s

(CDC 2009a). Today the incidence of TB within the U.S. is falling increasingly on

the foreign-born (CDC 2010b). This changed incidence is due both to the continual

rise in global TB prevalence and to the fact that the U.S. accepts almost 20% of its

immigrants from high-incidence countries.4

The U.S. government recognizes the threat that immigrants pose for spreading

TB to the native-born population. Prior to 2007, a prospective U.S. immigrant

needed only a chest x-ray before being allowed to enter the country (Weibrod 2010).The shortcoming of the chest X-ray is that it identifies only active TB infections; a

person with a latent infection would pass through security and face possible

reactivation of the disease once within U.S. However, since 2007, all prospective

immigrants aged 214, from countries with TB incidence rates of 20 or more pe r

100,000 population, have been required to take a tuberculosis skin test (TST),5

which identifies both the active and latent forms of TB (Weibrod 2010).

The problem with the current security measures involving the TST is that they

apply only to children under the age of 15 (who most likely have been vaccinated)

and not older persons who no longer have immunity to TB. Given the highincidence of TB in many of the countries of origin for U.S. immigrants, the latent

form of TB will most likely continue entering the U.S. until further procedures are

implemented to stop it. This is cause for concern in that many active cases of TB

among the foreign-born are reactivations of the latent form of the disease (Dasgupta

and Menzies 2005), and according to some, all such cases are reactivations (Cohen

and Murray 2005). Reactivations of TB usually occur when the bodys immune

system becomes compromised, which can often happen during times of increased

stress from inter-country migration (Cohen and Murray 2005). Therefore, these

unscreened latent infections are, we believe, a major source of TB among the U.S.

foreign-born population.

Related Research

A number of papers have been published on TB transmission patterns within

developed countries. This research may be divided into two categories: (1) small-

scale case studies of individual cohorts and (2) large-scale epidemiological studies

of entire populations.

4 This 20% figure is based on the mean from 1993 to 2007 for the 50 MSAs included in our sample,which is discussed in more detail below.5 The Tuberculosis Skin Test involves injecting a small portion of Tuberculin into the arm. If a personhas some form of the TB infection, he/she will develop a red, bumpy rash in response to the shot within

4872 h.

842 M. J. Greenwood, W. R. Warriner

123


7/23


8/23

After including many socio-economic variables as controls, this study finds that

crowding has a positive effect on TB incidence; but more importantly, it finds little

evidence of transmission from the migrant to native populations. Large-scale studies

such as Myers et al. (2006) and Baker et al. (2008) use observations that are

geographic areas rather than actual individuals; thus, the variables are linkedthrough area-wide characteristics rather than on a person-to-person basis.

In our research on the United States, we also focus on geographic areas, but we

are able to identify new active TB cases and also identify whether the individual is

native born or foreign born. Such information provides a significant advance over

earlier work concerning the U.S. From prior research, we can conclude that socio-

economic conditions such as poverty, population density, and number of migrants

are generally positively correlated with TB rates within a given area. However, the

question of whether native-born populations are affected by immigrants is relatively

unexplored; evidence is available that immigration affects TB rates, but this is thefirst study to estimate the effect of immigration on TB among the native-born.

The Model

In this study an observation is an individual Metropolitan Statistical Area (MSA) in

a given year (19932007). Because most immigrants reside in metropolitan areas

and because these areas are most densely populated, if any transmission exists from

the foreign born- to the native-born population, we hypothesize that it most likelywould be found in an MSA rather than a rural area.

We employ a recursive model to explain the relationship between immigration

and TB incidence among the native-born. A recursive model is one in which X

causes Y, and in turn Y causes Z. X represents the number of U.S. immigrants from

the top 50 high-TB incidence countries that specify an MSA of intended residence

in a given year, whereas Y is the number of new TB cases among the foreign-born

population in a given MSA and year. Z refers to the number of new native-born TB

cases in a given MSA and year. Thus, the model explains how immigrants from

high-TB incidence countries affect TB among the foreign-born who locate in the

same MSAs, and in turn, how the number foreign-born TB cases affects the number

of TB cases in the native-born population within specific MSAs.

Due to the much higher incidence of TB among the foreign-born population (26.0

per 100,000) in our sample MSAs compared to the native-born population (5.0 per

100,000), we assume the transmission to be unidirectional from the foreign to the

native-born population. Other studies (e.g., Cohen and Murray 2005) report such a

finding, and we provide further evidence of such a relationship below. (Even when

we treat the transmission as two way in a simultaneous context, the link for foreign-

born transmission to the native-born remains statistically significant.) Of course, the

foreign-born also may infect other foreign-born persons; such transmission is

partially accounted for in the first stage of the model.

The following is an outline of the recursive model, in which the first line

represents X to Y, and the second portrays Y to Z:


123


9/23

1. FBTBit fHMIGit; LMIGit; DEMit; SOCECONit; and2. NBTBit gFBTBit; DEMit; SOCECONit;

where FBTBit = newly reported foreign-born TB cases in MSA i, year t;

HMIGit = flow of new immigrants from the top 50 high TB incidence countries

in MSA i, years t and t - 1 summed; LMIGit = flow of new immigrants from all

other countries in MSA i, years t and t - 1 summed; NBTBit = newly reported

native-born TB cases in MSA i, year t; DEMit = a vector of demographic control

variables in MSA i, year t; and SOCECONit = a vector of socio-economic control

variables hypothesized to have causal relationships with TB, for MSA i, year t.

The first equation of the model concerns the relationship between the number of

immigrants from both high-TB incidence and lower-TB incidence countries and the

number of TB cases among the foreign-born. We expect a positive relationship

between the settlement of immigrants from high-incidence countries and the number

of new TB cases among the foreign-born population in specific MSAs. We

anticipate a weaker relationship between the settlement of immigrants from lower-

incidence countries and the number of new TB cases among the foreign-born. Given

previous research regarding the relationship between certain socio-economic

conditions and TB, we expect positive relationships between poverty and population

density and the number of new TB cases among the foreign-born.

The second branch of the recursive model concerns the relationship between TB

among the foreign-born and TB among the native-born. We expect this relationship

to be positive as well, due to the assumption that a less healthy foreign-born

population may somehow affect the native-born persons living in that area.6

Specifically, we hypothesize transmission of TB from the foreign born to the native

born. Given the findings of previous research, we expect positive relationships

between the socio-economic variables poverty and population density and the

number of new TB cases among the native-born.

Data

This study incorporates the following five types of data: (1) tuberculosis data, (2)demographic and socio-economic data, (3) country-specific TB incidence data, (4)

MSA immigration data, and (5) HIV incidence data.

The TB data set consists of basic tuberculosis statistics for each MSA provided

by the Centers for Disease Control and Prevention (CDC), and can be foun d in their

archived reports, or in the WONDER OTIS Database (CDC 2010b).7 In this

6 Given the concentration of the foreign born in service occupations, at least some transmission could

occur through the work place. The following respective percentages show foreign-born versus native-bornemployment in selected service occupations in 2009: health support, 2.6 versus 2.3; food preparation and

serving, 8.0 versus 5.1; building and grounds cleaning and maintenance, 8.5 versus 3.0; and personal careand service, 4.5 versus 3.4 (Bureau of Labor Statistics, 2010).Transmission also could occur through theuse of public transportation and in numerous other ways.7 The WONDER OTIS Database (Wide-ranging Online Data for Epidemiological Research) is an onlinetool provided by the CDC to offer researchers easy access to downloadable information. The OTIS

(Online Tuberculosis Information System) portion of the database is specific to TB data.


123


10/23

database, the CDC believes that it reports 100% of all cases because if left

untreated, TB proves fatal in a relatively short period of time. The sample for this

study includes the top 50 MSAs by total cases of TB (19932007) because the

number of TB cases become too few and are suppressed for smaller MSAs.8 The 50

MSAs included in the study account for 69.5% (or 185,234 of 266,556) of all newlyreported TB cases in the U.S., 19932007. CDC data also provide TB cases broken

down by the native- and foreign-born populations. Nativity was not recorded for

1,176 cases in our sample (0.6% of the total observations) and these were allocated

based on known native- and foreign-born TB incidence rates. That is, the fractions

of the known native-born and foreign-born TB cases, by MSA, were applied to the

cases in which nativity was unknown, by MSA, and then included with the known

MSA cases. The 50 MSAs account for 60.1% (89,979 of 149,698) of the total U.S.

native-born TB and 81.5% of the total U.S. foreign-born TB (94,255 of 116,858)

over the study period.Due to confidentiality restrictions, the WONDER database releases MSA

TB statistics only in minimum 5-year increments. Therefore, TB data were

downloaded in 15- and 14-year blocks, and were subtracted to obtain TB statistics

by MSA per year. For example 19932006 data were subtracted from the

19932007 block to obtain 2007 TB statistics for each MSA. Finally, TB data are

made available in the OTIS Database by geographic MSA definitions that are

constant over time.9

The second data set includes demographic and socio-economic statistics broken

down by MSA. These data are available from the U.S. Census Bureau through theAmerican Community Survey in 2007 and the for 1990 and 2000 (U.S. Bureau of

8 The 50 sample MSAs and their respective annual average (19932007) native and foreign-born TBrates per 100,000, grouped by Census region/division, are as follows: Northeast: New EnglandBoston-

Cambridge-Quincy (1.7, 26.7), Providence-New Bedford-Fall River (1.8, 21.2); Middle AtlanticNewYork-Nothern New Jersey-Long Island (8.3, 26.8), Philadelphia-Camden-Wilmington (3.8, 28.5),Pittsburgh (2.1, 16.0); Midwest: East North CentralChicago-Naperville-Joliet (6.4, 17.2), Cleveland-Elyria-Mentor (3.3, 18.5), Columbus (1.9, 38.1), Detroit-Warren-Livonia (4.4, 13.6); West NorthCentralKansas City (2.3, 26.4), Minneapolis-St. Paul-Bloomington (1.3, 38.1), St. Louis (2.7, 25.3);

South: South AtlanticAtlanta-Sandy Springs-Marietta (6.6, 23.7), Baltimore-Towson (4.1, 27.1),Charlotte-Gastonia-Concord (5.9, 22.9), Jacksonville (8.5, 22.6), Miami-Fort Lauderdale-Miami Beach(7.1, 16.3), Orlando-Kissimee (7.7, 17.6), Raleigh-Cary (5.0, 26.4), Tampa-St. Petersburg-Clearwater(5.3, 13.9), Virginia Beach-Arlington-Alexandria (2.6, 31.5); East South CentralBirmingham-Hoover(7.0, 26.1), Louisville-Jefferson County (4.0, 26.7), Memphis (8.6, 31.7), Nashville-Davidson-Murfreeboro (6.6, 35.8); West South CentralAustin-Round Rock (4.9, 23.0), Dallas-FortWorth-Arlington (5.7, 22.2), El Paso (5.3, 25.8), Houston-Sugar Land-Baytown (10.3, 22.9), McAllen-Edinburg- Mission (9.3, 29.4), New Orleans-Matairie-Kenner (9.6, 21.8), Oklahoma City (4.7, 26.6), SanAntonio (4.9, 19.5); West: MountainDenver-Aurora (1.5, 19.2), Las Vegas-Paradise (3.5, 19.4),

Phoenix-Mesa-Scottsdale (3.0, 19.6); PacificBakersfield (5.0, 30.4), Fresno (5.1, 32.6), Honolulu (4.5,61.2), Los Angeles-Long Beach-Santa Ana (5.1, 28.0), Oxnard-Thousand Oaks-Ventura (2.9, 29.2),Portland-Vancouver-Beaverton (2.0, 29.8), Riverside-San Bernardino-Ontario (3.0, 16.4), Sacramento-

Arden-Arcade-Roseville (3.1, 35.5), San Diego-Carlsbad-San Marcos (4.9, 41.0), San Francisco-Oakland-Fremont (6.2, 36.7), San Jose-Sunnyvale-Santa Clara (2.4, 39.4), Seattle-Tacoma-Bellevue (2.3, 32.9),Stockton (5.4, 40.0).9 MSA geographic definitions, which are based on counties except in New England, change on an annualbasis; however, the OTIS system provides TB statistics for constant geographies, with respect to the 2007

MSA definitions provided by the Executive Office of the President (2007).


123


11/23

the Census 2010).10 Data were linearly interpolated for years 1991 to 1999 and 2001

to 2006, using 1990, 2000, and 2007 as endpoints. These data provide MSA

population controls, including total population broken down by the foreign born and

native born, as well as race/ethnicity controls. All race/ethnicity variables are

recorded for the entire population, and are therefore not distinguishable by foreignor native-born.

The Census Bureau data also provide socio-economic controls that have been

previously shown to have causal relationships with TB, including population density

(persons per square mile) and poverty. The poverty variable was created to measure

the annual number of families earning less than $25,000 per year based on 2007

dollar equivalents. According to the Bureau of Labor Statistics CPI Inflation

Calculator, $25,000 in 2007 had the same purchasing power as about $15,000 in

1990. The number of families earning under $15,000 in 1990 and the number of

families earning under $25,000 in 2007 were deemed to be under the same real-income poverty threshold, and the number of impoverished families between these

years was linearly interpolated.

The third data set is from annual reports of the World Health Organization and

provides annual incidence rates for all WHO countries during the study period. The

top 50 high-TB incidence countries were defined using the average incidence rate

from 1990 to 2007 (WHO 2010b).11

The fourth data set is from the Department of Homeland Security (DHS) (and

before the formation of this Department, the Department of Justice.) The DHS data

contain the intended MSA of residence of all legal U.S. immigrants, by country ofthe immigrants birth (DHS 2008).12 We compiled the data for the top 50 high-

incidence countries of birth and assigned these immigrants to their intended MSA of

residence. Due to the availability of TB data, immigration statistics need to

correspond to constant geographies with respect to the 2007 MSA definitions. Since

the DHS immigration data available for download corresponded to inconsistent

10 To remain consistent in geographic definitions with the CDC, data collected from the Census Bureau

are on a county-wide level. 2007 MSA definitions were used and the county level data were combined by

this distinction for both 1990 and 2000.11 The list of 50 countries and the corresponding average incident rate per 100,000, 19902007 is asfollows: AfricaDjibouti (691.4), Swaziland (679.6), Zimbabwe (608.2), Namibia (596.2), Botswana(568.8), South Africa (556.3), Zambia (535.9), Lesotho (451.9), Malawi (375.4), Togo (365.3), SierraLeone (361.4), Mozambique (335.4), Uganda (322.6), Congo (313.6), Rwanda (308.8), Cote dIvoire(306.9), Democratic Republic of the Congo (304.6), Kenya (303.4), Mali (296.4), Ethiopia (294.0),United Republic of Tanzania (290.5), Burundi (285.3), Mauritania (270.3), Central African Republic(268.3), Somalia (249.0), Angola (243.9), Nigeria (241.7), Liberia (235.8), Chad (232.3), Senegal (231.1),Gabon (230.3), Gambia (219.8), Ghana (212.7), Madagascar (212.1), Sudan (207.0); Asia and West

PacificCambodia (538.9), Kiribati (435.1), Bhutan (375.1), Democratic Peoples Republic of Korea(344.0), Philippines (339.3), Indonesia (281.8), Papua New Guinea (250.0), Bangladesh (242.7), Tuvalu(225.6), Solomon Islands (207.3), Nepal (206.1), Mongolia (205.0); Caribbean and South AmericaHaiti

(306.0), Peru (207.9), Bolivia (200.6). By contrast, the United States had a 2005 rate of 4.8.12 The immigration data include only legal permanent residents. Temporary residents (called non-immigrants) and illegal migrants may also spread TB. Illegal immigrants residing in the U.S. have beenestimated to number as high as 12 million in recent years. Although the illegal migrants themselves arenot included in the migration numbers used here, they (along with non-immigrants) could be included in

the TB data for which the only distinction is foreign born.


123


12/23

geographic boundaries, John Simanski, from the Office of Immigration Statistics of

the DHS, provided us with immigration statistics that are adjusted to correspond to

2007 MSA definitions. Immigration data are based on fiscal years (October through

September), whereas all other data are based on calendar years. Therefore, the

immigration data reflect a three-month lead over all other data in the model.

The final data set is comprised solely of HIV incidence data, and was obtained

from the CDC Division of HIV/AIDS Preventions archived reports (CDC 2005).

The data provide HIV incidence rates per 100,000 persons in every MSA, but arelimited in availability. Prior to 2003, the HIV incidence rates were not recorded by

2007 MSA definitions, and thus could not be used for the entire 19932007 period.

The HIV incidence control is used for 20032007 only.

Table 1 provides descriptive statistics, including means and standard deviations.

Econometric Approach

This study utilizes an ordinary least squares (OLS) double-log regression, which

allows the coefficients to be interpreted as estimated elasticities. The estimated

model is of the following form:

1. FBTBit a1HMIGb1it LMIG

b2it DEM

b3it SOCECON

b4it e1it

2. NBTBit a2FBTBc1it DEM

c2it SOCECON

c3it e2it:

Table 1 Descriptive statistics

Variable Mean Standarddeviation

Number of native TB cases 120 219Number of foreign-born TB cases 126 240

Total immigrants 14279 27,159

Immigrants from top 50 high TB countries 2604 4,874

Total population 2905694 3,086,156

Native-born population 2417600 2,278,708

Foreign-born population 488094 891,794

Non-hispanic whites 1740128 1,629,987

Non-hispanic blacks 409482 538,430

Asians 169176 299,398American Indians 10499 11,274

Other races 57931 70,131

Hispanics 518479 893,002

Population density (peopleper square mile)

644 551

Families earning under$25,000 per year

103040 106,687

HIV Incidence (per 100,000)for 20032007

15 9


123


13/23

Taking logarithms of the terms on each side of the model yields estimated

relationships that are linear in logarithms. The robust feature13 is added to all

regressions in case of any heteroskedasticity (variance in the error term) or large

outliers in the data set. The data for all MSAs and years are pooled into a single

panel data set with 750 observations (15 years and 50 MSAs per year), with an

MSA in a given year as an individual observation. In order to control for time and

location, the model employs temporal (annual) and cross-sectional (MSA) fixed

effects.

The basic idea that underlies the use of fixed effects is that the measurable

variables of the model do not capture all of the variance in the dependent variable.

Certain unobservables also influence this variable. However, differences across

various groups (e.g., MSAs or years) are reflected in differences in the constant

term. Thus, to control for unobserved variance, we introduce dummy variables (or 0

and 1 variables) for years and for different regions and, alternatively, groups ofregions. One example of an unobservable is illegal immigration, which tends to be

region specific and could affect exposure to TB.

The immigration variables are defined to include years t and t - 1. Thus, a single

observation of the high-incidence immigration variable includes all immigrants

from the top 50 high-incidence countries for MSA i in years t and t - 1. TB is

unlikely to spread immediately from the foreign-born population; some sort of

lagged transmission is more likely to occur. Cohen and Murray (2005) find that the

highest TB incidence rates among immigrants occur within the first year of arrival,

and that the incidence rates decrease dramatically in subsequent years. For example,an immigrant arriving in December has the greatest chance of reactivating any latent

TB before the following December, and therefore should be counted in the

following calendar year as well. Summing the immigration variable to include the

present and prior year should account for lagged transmission.

The race/ethnicity variables are omitted from the first model (in which foreign-

born TB is the dependent variable), due to the fact that these statistics do not reflect

the foreign-born population that is at high risk to contract TB.

Empirical Results

In Table 2, we report double-logarithmic estimates and t-ratios for five regressions

for the first equation of the recursive model: the first of the five with no fixed effects;

the second with only temporal fixed effects (with 1993 as the base year); the third

with temporal and cross-sectional (all MSAs) fixed effects (with Atlanta-Sandy

Springs-Marietta as the base MSA); the fourth with temporal and regional (four

geographic subgroups) fixed effects (with West as the base region); the fifth with

temporal and divisional (nine geographic subgroups) fixed effects (with Pacific as

13 The robust feature in STATA (the data analysis and statistical software used in this research) is amethod of econometrics that allows for the research to circumvent certain OLS limitations regarding

variance in the error term and large outliers in the data set.


123


14/23

the base division). The regional and divisional distinctions are Census Bureau

classifications and are reflected in footnote 8.

All regressions have relatively high R2 s, with the R2 s in the first relationship of

the recursive model ranging from 0.914 when no fixed effects are used to 0.960

when the temporal and all-MSA cross-sectional fixed effects are employed, and in

the second relationship ranging from 0.749 when no fixed effects are used to 0.954

when temporal and all MSA cross-sectional fixed effects are employed. Even

though all estimations are similar in sign and significance, our focus will be on the

fifth regression with the temporal and divisional fixed effects. We focus on this

regression because when the more detailed cross-sectional fixed effects are

included, most of the coefficients on the individual MSAs are not significantly

different from zero.

In the first branch of the recursive model, foreign-born TB is the dependent

variable and both high-incidence and lower-incidence immigration are the

independent variables. The key immigration variables in this regression are both

consistent in their hypothesized relationships with foreign-born TB. The elasticity

for immigrants from high-TB incidence countries is statistically significant with a

value of 0.285, meaning that an increase in high-incidence immigrants of 10%

results in an increase in the number of new foreign-born TB cases by 2.85%, other

Table 2 Number of TB cases among the foreign-born population in U.S. metropolitan areas, 19932007:double logarithmic estimates (b) and t-ratios (t)

Variables (1) (2) (3) (4) (5)ln(FBTB) ln(FBTB) ln(FBTB) ln(FBTB) ln(FBTB)

ln(Immigrants from top 50 Inc. countries)

b: 0.173 0.258 0.241 0.269 0.285

t: (5.769) (7.744) (3.414) (7.851) (7.926)

ln(Immigrants from all other countries) 0.139 0.178 0.0544 0.143 0.0313

(2.936) (3.338) (0.779) (2.601) (0.579)

ln(Foreign-born population) 0.819 0.734 1.500 0.741 0.845

(16.39) (12.66) (6.764) (12.36) (12.76)

ln(Native population) -0.121 -0.267 3.313 -0.323 -0.233

(-

1.885) (-

3.799) (2.950) (-

4.632) (-

3.593)ln(Population density) 0.0391 0.0270 -4.815 0.0277 0.0803

(1.700) (1.299) (-3.540) (1.123) (3.254)

ln(Families in poverty) -0.23 -0.117 0.449 -0.0472 -0.141

(-3.700) (-1.757) (0.955) (-0.761) (-2.189)

Temporal fixed effects N Y Y Y Y

Cross-sectional fixed effects N N Y N N

Regional fixed effects N N N Y N

Divisional fixed effects N N N N Y

Observations 750 750 750 750 750R2 0.914 0.924 0.960 0.925 0.940

Robust t-statistics in parentheses. Y used in regression, N not used in regression


123


15/23

factors held constant. Furthermore, the elasticity for immigrants of lower incidence

countries is not statistically significant. The stark contrast in the elasticities for both

key variables shows the direct and unique connection between immigrants from

high incidence countries and TB within the foreign-born population. Although this

may seem obvious, it is important to realize that the foreign-born population iscontracting TB through immigrants from the 50 high-TB incidence countries, and

not the 156 remaining low-incidence countries. This finding provides further

evidence of a unidirectional relationship that runs from the foreign born to the

native born. If immigrants from countries other than the top 50 do not affect the

foreign-born incidence, then the native U.S. population, with considerably lower

rates than those of countries below the top 50, are unlikely to affect the incidence

among the foreign born.14

As expected, most of the control variables also have the appropriate sign and are

highly significant. For example, increased population density results in increasedTB cases among the foreign-born population, holding all else constant. Foreign-

born TB is negatively associated with the poverty rate with an elasticity of-0.141.

This finding suggests that the higher rates of TB among the foreign-born population

are most likely not a result of conditions associated with lower income areas but

rather the disease is carried by the immigrant population itself.

Table 3 shows the regression results for the second segment of the recursive

model with native-born TB cases as the dependent variable and foreign-born TB

cases as the key independent variable. As predicted, the number of TB cases among

the foreign-born positively affects the number of native-born TB cases with anelasticity of 0.111, which is also highly significant. In other words, a 10% increase

in the number of new foreign-born TB cases increases the number of TB cases

among the native-born population by 1.11%.

Also shown in Table 3 are the hypothesized key socio-economic control

variables of race/ethnicity and poverty. Poverty has the most prominent relationship

with native TB with a statistically significant elasticity of 0.759. Population density

also positively and significantly influences TB among the native-born, but with a

lower elasticity of 0.077. Blacks, Asians, and Hispanics are the races/ethnicities

most likely to have TB within the native-born population, a finding consistent with

previous research.

Tables 4 and 5 provide additional estimations of all regressions, but include an

HIV/AIDS control, due to the large number of co-infections between HIV and

Tuberculosis. In 2006, nearly 12% of the newly reported TB cases in the U.S. were

also co-infections with HIV (CDC 2007); co-infection rates reach as high as 80% in

sub-Saharan Africa, where a large portion of the high-incidence countries are

located (Infectious Diseases Society of America 2007). Omitting this control will

likely leave a positive omitted variable bias (an overestimation) on the foreign-born

TB coefficient in the second branch of the recursive model. This is due to the likely

14 As noted above, we also estimated the model using 2 stage least squares to account for any possiblesimultaneity between native- and foreign-born TB. When the first-stage regression contains all of thetemporal and cross-sectional fixed effects, the coefficient on the foreign-born TB variable is highlysignificant in the regression for native-born TB, although the magnitude of the coefficient declines

somewhat (to 0.531 with the t statistic of 2.326).


123


16/23

positive relationships between HIV/AIDS and native-born TB, and HIV/AIDS and

immigrants, since most high-TB incidence countries are also in high-HIV/AIDS

regions.

Because the data are available only after 2003, the estimations including HIV

incidence contain a panel with 250 observations. Both branches of the recursive

model were run again, but only for the five available years. The fifth and sixth

regressions in each table are with the temporal and divisional fixed effects; however,

the fifth omits HIV incidence and the sixth includes the HIV incidence as a control.

The estimations in all regressions are different given the smaller sample, but the

Table 3 Number of TB cases among the native-born population in U.S. metropolitan areas, 19932007:double logarithmic estimates (c) and t-ratios (t)

Variables (1) (2) (3) (4) (5)ln(NBTB) ln(NBTB) ln(NBTB) ln(NBTB) ln(NBTB)

ln(Foreign-born TB cases)

c: 0.279 0.136 0.198 0.129 0.111

t: (5.081) (3.292) (5.599) (3.328) (2.685)

ln(Native population) 0.188 0.114 1.812 -0.301 -0.0422

(1.017) (0.819) (1.133) (-1.992) (-0.275)

ln(Foreign-born population) -0.407 -0.305 0.652 -0.349 -0.241

(-4.407) (-4.358) (2.185) (-4.804) (-2.577)

ln(Non-Hispanic Whites) -0.263 -0.368 0.366 -0.0133 -0.0997

(-

2.406) (-

4.727) (0.769) (-

0.171) (-

1.210)ln(Non-Hispanic Blacks) 0.369 0.415 0.545 0.232 0.234

(14.13) (20.30) (2.377) (9.315) (7.173)

ln(Asians) 0.271 0.123 -0.919 0.300 0.179

(5.213) -3.043 (-4.187) (5.972) (3.040)

ln(American Indians) 0.149 -0.0344 0.0160 -0.154 -0.0579

(4.361) (-1.155) (0.135) (-5.300) (-1.795)

ln(Other races, including mult-racial) -0.549 -0.0835 0.131 -0.0107 -0.00405

(-10.91) (-1.829) (0.935) (-0.246) (-0.0906)

ln(Hispanics) 0.104 0.182-

0.491 0.129 0.133(2.277) (5.142) (-2.356) (3.582) (3.409)

ln(Population density) 0.191 0.106 -1.972 0.0537 0.0771

(4.312) (2.964) (-1.261) (1.555) (2.089)

ln(Families in poverty) 0.675 0.69 0.747 1.018 0.759

(7.155) (8.529) (1.914) (11.04) (7.246)

Temporal fixed effects N Y Y Y Y

Cross-sectional fixed effects N N Y N N

Regional fixed effects N N N Y N

Divisional fixed effects N N N N YObservations 750 750 750 750 750

R2 0.749 0.853 0.954 0.879 0.879



123


17/23

importance of these tables lies in the difference between the fifth and sixth

regressions, which shows the change in coefficients with the inclusion of the HIV

control.

The sixth regression in Table 4 shows the regression results when foreign-born

TB is the dependent variable and high- and lower-incidence immigrants are the key

independent variables. The results show that the HIV incidence of an area does not

have a significant relationship with the number of foreign-born TB cases, given the

associated small and negative elasticity and t-statistic of-1.496. This makes sense

given that the coefficients for the other key variables do not change much once the

HIV control is included (comparing the fifth and sixth regressions). Therefore, it

seems that most of the foreign-born persons who get TB are not getting it because of

the HIV-incidence of an area, but are getting it either because of where they come

from or who they associate with.The inclusion of the HIV-incidence variable portrays a much different picture in

the second branch of the recursive model with native-born TB as the dependent

variable. As shown in Table 5, an increase in HIV incidence positively and

significantly increases TB within the native population, with an elasticity of 0.323.

Table 4 Number of TB cases among the foreign-born population in U.S. metropolitan areas (HIVincidence included), 20032007: double logarithmic estimates (b) and t-ratios (t)

Variables (1) (2) (3) (4) (5) (6)ln(FBTB) ln(FBTB) ln(FBTB) ln(FBTB) ln(FBTB) ln(FBTB)

ln(Imms. of top-50 Inc. countries)

b: 0.227 0.312 0.196 0.316 0.294 0.307

t: (4.521) (5.688) (1.067) (5.715) (4.913) (5.120)

ln(Imms. of all othercountries)

-0.130 -0.0659 0.0698 -0.0718 -0.0854 -0.0918

(-1.642) (-0.753) (0.621) (-0.787) (-0.886) (-0.951)

ln(Foreign-born population) 0.955 0.84 2.35 0.872 0.917 0.919

(11.74) (8.902) (2.083) (8.438) (8.045) (7.990)

ln(Native population) -0.144 -0.303 0.511 -0.317 -0.236 -0.248

(-1.376) (-2.620) (0.0912) (-2.681) (-1.951) (-2.092)

ln(Population density) 0.0929 0.0836 -3.049 0.0729 0.109 0.109

(2.497) (2.314) (-0.456) (1.911) (2.531) (2.530)

ln(Families in poverty) -0.0732 0.0519 0.742 0.0277 -0.115 -0.0897

(-0.741) (0.492) (0.360) (0.259) (-1.022) (-0.796)

ln(HIV incidence rate) -0.108 -0.138 -0.0885 -0.142 -0.0732

(-2.564) (-3.314) (-1.062) (-2.890) (-1.496)

Temporal fixed effects N Y Y Y Y Y

Cross-sectional fixed effects N N Y N N N

Regional fixed effects N N N Y N N

Divisional fixed effects N N N N Y Y

Observations 250 250 250 250 250 250

R2 0.927 0.933 0.975 0.933 0.940 0.941



123


18/23

Most importantly, the elasticity of foreign-born TB decreases when HIV is included,

from 0.172 to 0.136, but still remains positive and statistically significant. Giventhat this estimate decreases when HIV incidence is included, a positive omitted

variable bias probably is associated with not including HIV incidence for the

previous years. Therefore, the estimate in Table 3 for foreign-born TB is probably

overestimated. Furthermore, the elasticity on Blacks changes from positive and

Table 5 Number of TB cases among the native-born population in U.S. metropolitan areas (HIV inci-dence included), 20032007: double logarithmic estimates (c) and t-ratios (t)

Variables (1) (2) (3) (4) (5) (6)ln(NBTB) ln(NBTB) ln(NBTB) ln(NBTB) ln(NBTB) ln(NBTB)

ln(Foreign-born TB cases)

c: 0.211 0.162 0.124 0.155 0.172 0.136

t: (3.123) (2.591) (1.267) (2.505) (2.607) (2.111)

ln(Native population) 0.424 0.461 -5.669 -0.221 0.0676 0.213

(1.949) (2.136) (-0.768) (-1.046) (0.306) (1.003)

ln(Foreign-born population) -0.726 -0.671 3.837 -0.687 -0.366 -0.502

(-5.920) (-5.565) (2.125) (-6.247) (-2.327) (-3.546)

ln(Non-Hispanic Whites) -0.391 -0.411 1.610 -0.0124 -0.142 -0.151

(-

3.220) (-

3.372) (0.620) (-

0.113) (-

1.171) (-

1.304)ln(Non-Hispanic Blacks) 0.203 0.217 2.064 0.0594 0.239 0.0835

(4.285) (4.653) (1.774) (1.162) (4.305) (1.389)

ln(Asians) 0.297 0.329 0.351 0.533 0.264 0.367

(4.088) (4.702) (0.267) (6.136) (2.569) (3.550)

ln(American Indians) 0.0902 0.0796 -0.131 -0.0757 -0.0182 0.0441

(1.869) (1.636) (-0.340) (-1.624) (-0.336) (0.802)

ln(Other races, includingmulti-racial)

-0.193 -0.232 0.445 -0.110 -0.0942 -0.142

(-2.380) (-2.755) (1.022) (-1.410) (-1.211) (-1.769)

ln(Hispanics) 0.255 0.254 -4.234 0.205 0.136 0.168

(3.914) (4.026) (-2.776) (-3.326) (1.932) (2.737)

ln(Population density) 0.185 0.182 -1.073 0.0956 0.139 0.137

(3.190) (3.184) (-0.139) (1.734) (2.304) (2.333)

ln(Families in poverty) 0.621 0.596 4.046 1.054 0.653 0.662

(4.812) (4.852) (1.877) (7.752) (4.168) (4.020)

ln(HIV incidence rate) 0.405 0.377 0.0506 0.276 0.323

(6.057) (5.623) (0.451) (3.562) (4.177)

Temporal fixed effects N Y Y Y Y Y

Cross-sectional fixed effects N N Y N N N

Regional fixed effects N N N Y N N

Divisional fixed effects N N N N Y Y

Observations 250 250 250 250 250 250

R2 0.838 0.850 0.956 0.882 0.871 0.883



123


19/23

significant to insignificant upon the HIV inclusion. This result suggests that the

Black population within the native-born is contributing importantly to the HIV/TB

co-infections, and the omitted variable bias may be picked up by the Black

population and not the foreign-born TB variable.

Societal Cost of TB Transmission

The economy incurs many costs from TB. Such societal costs are considered in

(Miller et al. 2010), who group the costs into the following four categories: (1)

infrastructure costs which are defined as the hospitals opportunity costs15 of

treating TB; (2) actual treatment expenditures, which range from diagnostic

procedures, to pharmaceuticals used in treatment, to the labor costs of the hospital

employees; (3) transmission costs, which represent all costs associated with furthertransmissions of a given case; and (4) patient costs, which include all lost wages due

to treatment, possible pulmonary impairment, or even death. Miller et al. (2010) use

Tarrant County, Texas (which is part of the Dallas/Fort Worth Metropolitan Area),

in 2002, as their sample area because it reflects the average U.S. county. Tarrant

County contained just over 1.5 million people in 2002, had 108 confirmed TB cases

(43.5% of which were foreign-born), a subsequent TB case rate of 7.2 per 100,000

persons, and had many of the risk factors associated with tuberculosis as deemed by

the researchers.

To determine the average cost per TB case, Miller et al. (2010) use reportedgovernment and local organization costs and QALYs to estimate the effect of all the

various costs described above. QALY (or Quality Adjusted Life Year) is a measure

used to determine if the loss of potential health a person incurs is affected by a

certain aspect of a disease. In this case, QALYs were used to calculate the

production that a worker could have contributed if not affected by tuberculosis

(either by death or disability) over his expected lifespan. Using the QALYs method,

the Miller study finds that the average societal cost of TB for Tarrant County, Texas,

in 2002, was $346,756 per case. These costs include transmission costs of future TB

which were estimated at $176,778 per case, but do not include TB testing costs

which were estimated at an additional $29,498 per case. Furthermore, the research

finds that if both impairment and death were avoided, and if the further transmission

of TB were prevented, the estimated societal cost would be only $32,248.

The cost of Tuberculosis testing was also estimated by the same study for Tarrant

County in 2002. Miller et al. (2010) find that the average cost of a TST

(Tuberculosis Skin Test, used to detect latent TB infections), including the actual

test and labor, was $18 per test, or $11,875 per actual TB case detected. This

number is less than the total cost of TB testing noted above ($29,498/case) because

the smaller cost excludes extra infrastructure, treatment, and correctional costs that

would not be associated with solely applying a TST to a specific group. Although

15 Opportunity cost is the value of the next best choice. In this case, the opportunity costs consist of the

resources a hospital could direct to other functions if they were not being used to treat TB patients.


123


20/23

still rather expensive, this cost is significantly less than that of cleaning up an active

TB case.

Based on the lower cost per case noted above, we extrapolate the costs measured

for Tarrant County, Texas, for the entire U.S. to provide an estimate of what TB is

costing the country as a whole. The national estimates do not include thetransmission costs because these costs occur over the lifespan of a TB strain and

thus cannot be estimated for a given year. Cost estimates are inflated by 19.68% due

to the CPI inflation from 2002 to 2008. In 2008, there were 12,904 newly reported

TB cases in the United States, of which 5,283 were native-born, 7,563 were foreign-

born, and 58 were not reported by nativity (CDC 2009a). Again, the unreported

cases were allocated to either foreign or native. We use the projected cost estimate

per case if the person suffers no disability or death from the TB infection. In this

case, the lowest bound for the total U.S. societal cost of TB for 2008 was $0.50

billion. If the additional costs of death and disability are included, our upper-boundestimate of the societal cost of TB in 2008 is $2.63 billion (or $203,428.42 per case)

for the U.S. economy as a whole. Compared to the 2008 total GDP estimate for the

U.S. economy of $14.61 trillion, the TB societal cost amounts to only 0.018% of

total GDP (CIA 2010).

The estimated cost of latent TB testing using TSTs can also be calculated for

testing all persons obtaining legal permanent residence (LPR) in the U.S. in 2008.

The average cost of a TST was $9 with an additional $9 for labor and surveillance

costs in Tarrant County in 2002. In 2008, 1,107,126 people obtained legal

permanent resident status (DHS 2008). If the cost statistics from Tarrant County areextrapolated to the entire U.S. population (accounting for inflation), then testing

every LPR immigrant for latent TB in 2008 would have cost the U.S. economy

$23.8 million.

The estimates provided in our study should be accepted only conditionally. Only

one study has attempted to calculate entire societal costs of TB and then for only

one county in Texas. In providing potential costs for the entire U.S., numerous

assumptions are made including the supposition that Tarrant County provides an

adequate representation of the rest of the U.S., as well as the assumption that no

major changes have been made since 2002 regarding the framework behind the

original studys calculated costs. However; the benefit/cost analysis of this section

does provide rough estimates and an overall idea of whether testing immigrants for

latent-TB would be cost effective.

If the above calculations are reasonable, testing all LPR immigrants for latent

and active TB infections using the TST would cost anywhere between 0.91 and

4.79% of TBs societal cost in 2008. Requiring all LPR immigrants to take a TST

would cost the U.S. very little compared to the cost it is incurring from current

levels of TB, which makes a TST requirement for all immigrants extremely cost

effective. Given our findings of TB transmission from high-incidence immigrants

to the foreign-born and then to the native-born populations, limiting the amount of

latent TB that crosses U.S. borders would benefit both groups in the U.S. and

probably would significantly reduce the number of latent TB infections that enter

the U.S.


123


21/23

Discussion

Previous research on TB transmission in the U.S. has failed to discover a clear link

between immigration and the native-born population. However, by defining the

observation as a Metropolitan Statistical Area and using unique data provided by theCDC, as well as other unique data, we uncover evidence of TB transmission from

U.S. immigrants to the foreign- and to the native-born populations. Our estimated

elasticities show that the number of immigrants from countries of high-TB

incidence positively affects the number of foreign-born TB cases, and that, in turn,

an increase in TB among the foreign-born causes an increase in the number of new

TB cases among the native-born population. Therefore, areas that experience

relatively much immigrant settlement are likely to have relatively more TB cases

within their native populations, a direct result of TB transmission from the foreign

to the native-born populations. Specifically, at the means in our data, a 10% increasethe settlement of immigrants from high-incidence countries (260.4 more immi-

grants, on average) results in an average of 3.6 more new TB cases among the

foreign born in an MSA. In turn, a 10% increase in the number of TB cases among

the foreign born of an MSA (12.6 more cases, on average) results in an average of

1.3 more new cases among the native born.

Our study also solidifies previous knowledge by finding significant relationships

between TB and certain socio-economic conditions. In the U.S., areas that have

higher population densities, increased numbers of Blacks, Asians, and Hispanics,

increased poverty, and higher HIV-incidence rates also have more TB cases amongthe native-born populations.

Because immigrants are contributing to increased TB among both the foreign and

native populations, a case can be made for enacting policies targeting immigrant

tuberculosis. As previously mentioned, the U.S. currently tests only for latent TB

infections in children, aged 214, from countries with TB incidence rates of greater

than 20 per 100,000 persons. We recommend that this policy be extended to

immigrants of all ages.

The importance of implementing TB immigration policy has become even more

pressing. In 2009, President Barack Obama announced the lifting of the nations

22 year-old ban of HIV-positive immigrants, allowing admittance into the United

States regardless of HIV/AIDS status (Preston 2009). As mentioned above, a large

number of TB/HIV co-infections occur across the world. As of 2009, over one-third

of the 33.2 million HIV-positive persons were co-infected with tuberculosis. In fact,

people with HIV have roughly a 2030 times greater likelihood of contracting TB

(WHO 2009b). Therefore, allowing people who are HIV positive to enter the U.S.

will increase the number of persons who either have TB, or are more susceptible of

activating it in the U.S.

Additionally, many of the top 50 high-TB incidence countries in our study also

have an extremely high HIV prevalence. Since a large fraction of TB cases in thesecountries are also co-infections of HIV, many people infected with TB from these

countries have historically been denied access to the U.S. Thus, our data contain no

HIV-positive immigrants, showing that only HIV-negative immigrants positively

affect foreign and native-born TB in the U.S. Elasticities, and thus the magnitude of


123


22/23


23/23

Dasgupta, K., & Menzies, D. (2005). Cost-effectiveness of tuberculosis control strategies amongimmigrants and refugees. European Respiratory Journal, 25.6, 11071116.

Department of Homeland Security. (2008). Office of Immigration Statistics. Yearbook of ImmigrationStatistics: 2007. Washington, D.C.

Executive Office of the President. (2007). Office of Management and Budget. OMB Bulletin No. 0801.

Nov. 20, 2007. Washington, D.C.Gandhi, N. R., Nunn, P., Dheda, K., Schaaf, H. S., Zignol, M., van Soolingen, D., et al. (2010).

Multidrug-resistant and extensively drug-resistant tuberculosis: a threat to global control oftuberculosis. The Lancet, 375, 18301843.

Infectious Diseases Society of America, HIV Medicine Association, The Forum for Collaborative HIVResearch. (2007). HIV/TB Coinfection: Basic Facts.

Laxminarayan, R., Klein, E., Dye, C., Floyd, K., Darley, S., & Adeyi, O. (2007). Economic benefit oftuberculosis control. The World Bank: Research Working Paper 4295, 2007.

Mangtani, P., Jolley, D. J., Watson, J. M., & Rodrigues, L. C. (1995). Socioeconomic deprivation andnotification rates for tuberculosis in London during 19821991. British Medical Journal, 310,963966.

Miller, T. L., McNabb, S. J. N., Hilsenrath, P., Pasipanodya, J., Drewyer, G., & Weis, S. E. (2010). The

societal cost of tuberculosis: Tarrant County, Texas, 2002. Annals of Epidemiology, 20.1, 17.Myers, W. P., Westenhouse, J. L., Flood, J., & Riley, L. W. (2006). An ecological study of tuberculosis

transmission in California. American Journal of Public Health, 96(4), 685690.National Health Service, United Kingdom. (2010). Tuberculosis-Introduction. Tues. Sept. 2010. http://

www.nhs.uk/conditions/Tuberculosis/Pages/Introduction.aspx.Preston, J. (2009). Obama lifts a ban on entry into U.S. by H.I.V.-positive people. New York Times. New

York, NY Oct. 31, 2009; A9.Tornieporth, N. G., Ptachewich, Y., Poltoratskaia, N., Ravi, B. S., Katapadi, M., Berger, J. J., et al.

(1997). Tuberculosis among foreign-born persons in New York City, 19921994: Implications fortuberculosis control. The International Journal of Tuberculosis and Lung Disease, 1(6), 528535.

U.S. Bureau of the Census. (2010). American fact finder. Sept. 6, 2010. http://factfinder.census.

gov/home/saff/main.html.Weibrod, C. (2010). The melting pot: new immigration guidelines affect TB testing. Medical LaboratoryObserver, 42.2, 1416.

World Health Organization. (2008). Drug Resistance Surveillance. Anti-Tuberculosis Drug Resistance inthe World. Vol. 4. Geneva, Switzerland: WHO.

World Health Organization. (2009a). United Nations. Global tuberculosis control: epidemiology, strategy,control 2009. Geneva.

World Health Organization. (2009b). United Nations. TB/HIV facts 2009. Geneva.

World Health Organization. United Nations. (2010a). http://www.who.int/biologicals/areas/vaccines/bcg/en/html.

World Health Organization. (2010b). United Nations. TB data. Geneva 2010b. Sept. 6, 2010. http://www.who.int/tb/country/data/download/en/index.html.

http://www.nhs.uk/conditions/Tuberculosis/Pages/Introduction.aspxhttp://www.nhs.uk/conditions/Tuberculosis/Pages/Introduction.aspxhttp://factfinder.census.gov/home/saff/main.htmlhttp://factfinder.census.gov/home/saff/main.htmlhttp://www.who.int/biologicals/areas/vaccines/bcg/en/htmlhttp://www.who.int/biologicals/areas/vaccines/bcg/en/htmlhttp://www.who.int/tb/country/data/download/en/index.htmlhttp://www.who.int/tb/country/data/download/en/index.htmlhttp://www.who.int/tb/country/data/download/en/index.htmlhttp://www.who.int/tb/country/data/download/en/index.htmlhttp://www.who.int/biologicals/areas/vaccines/bcg/en/htmlhttp://www.who.int/biologicals/areas/vaccines/bcg/en/htmlhttp://factfinder.census.gov/home/saff/main.htmlhttp://factfinder.census.gov/home/saff/main.htmlhttp://www.nhs.uk/conditions/Tuberculosis/Pages/Introduction.aspxhttp://www.nhs.uk/conditions/Tuberculosis/Pages/Introduction.aspx

Greenwood and Warriner TB Paper PRPR

Documents