American Journal of Epidemiology Vol. 155, No.

11
Copyright © 2002 by the Johns Hopkins Bloomberg School of Public Health Printed in U.S.A.
All rights reserved

Genetic and Environmental Risk Factors for Tuberculosis Lienhardt et al.

PRACTICE OF EPIDEMIOLOGY

Investigation of Environmental and Host-related Risk Factors for Tuberculosis

in Africa. I. Methodological Aspects of a Combined Design

C. Lienhardt,1,2 S. Bennett,3 G. Del Prete,4 O. Bah-Sow,5 M. Newport,6 P. Gustafson,7 K. Manneh,8 V. Gomes,9

A. Hill,10 and K. McAdam1

Host-related and environmental factors for tuberculosis have usually been investigated separately using
different study designs. Joint investigation of the genetic, immunologic, and environmental factors at play in
susceptibility to tuberculosis represents an innovative goal for obtaining a better understanding of the

Downloaded from aje.oxfordjournals.org by guest on September 14, 2010

pathogenesis of the disease. In this paper, the authors describe methods being used to investigate these points
in a West African study combining several designs. Patients with newly diagnosed smear-positive cases of
tuberculosis are recruited. The effect of host-related factors is assessed by comparing each case with a healthy
control from the case’s household. The role of environmental factors is estimated by comparing cases with
randomly selected community controls. The frequencies of candidate gene variants are compared between
cases and community controls, and results are validated through family-based association studies. Members of
the households of cases and community controls are being followed prospectively to determine the incidence of
“secondary” tuberculosis and to evaluate the influence of geographic and genetic proximity to the index case.
This type of design raises important methodological issues that may be useful to consider in studies
investigating the natural history of infectious diseases and in attempts to disentangle the effects of environmental
and genetic factors in response to infection. Am J Epidemiol 2002;155:1066–73.

epidemiologic methods; genetics; infection; Mycobacterium tuberculosis; research design; risk factors;
tuberculosis

The development of tuberculosis in humans is a two-stage ease and involves different physiologic mechanisms, the
process in which a susceptible person exposed to an infec- risk factors for infection are quite different from the risk fac-
tious case first becomes infected and second, after an inter- tors for development of disease following infection (1). This
val of years or decades, may later develop the disease, has important implications for tuberculosis prevention and
depending on a variety of factors. Since the acquisition of control (2).
infection is often far removed from the development of dis- Among persons exposed to someone with an infectious
case of tuberculosis, the risk of becoming infected is deter-
Received for publication June 19, 2001, and accepted for publi- mined primarily by the combined action of three factors:
cation February 1, 2002. 1) the infectivity of the source case (which is itself a function
Abbreviations: Th1, type 1 helper [cells]; Th2, type 2 helper [cells]. of microbial virulence and the density of bacilli in the spu-
1
MRC Laboratories, Fajara, The Gambia.
2
Institut de Recherche pour le Développement, Dakar, Sénégal. tum), 2) the intensity of the susceptible person’s exposure to
3
London School of Hygiene and Tropical Medicine, London, the case, and 3) the susceptibility of the exposed person to
United Kingdom.
4
infection (3–5). Factors reported to influence the risk of
Department of Internal Medicine, Faculty of Medicine, University mycobacterial infection include age, sex, crowding, socioeco-
of Florence, Florence, Italy.
5
Programme National de Lutte Anti-Tuberculeuse, CHU Ignace
nomic conditions, urbanization, racial/ethnic group, and
Deen, Conakry, République de Guinée. human immunodeficiency virus infection (6, 7). In patients
6
Wellcome Trust Centre for Molecular Mechanisms in Disease, infected with Mycobacterium tuberculosis, the disease can
Cambridge, United Kingdom. develop at any time through reactivation of a previously
7
Projecto de Saude de Bandim, Danish Epidemiology Science acquired (latent) infection or through exogenous reinfection
Centre, Bissau, Guinea-Bissau.
8
National Leprosy/Tuberculosis Control Programme, Ministry of (8, 9).
Health, Banjul, The Gambia. The time interval from infection to disease ranges from a
9
10
Hopital Raoul Follereau, Bissau, Guinea-Bissau. few weeks to a lifetime (9). The risk of developing disease
Wellcome Trust Centre for Human Genetics, Oxford, United after infection is strongly age and time dependent (10) and
Kingdom.
Correspondence to Dr. Christian Lienhardt, Institut de Recherche has been reported to be much greater in the 5 years follow-
pour le Développement, BP 1386, Dakar, Sénégal (e-mail: ing infection and to decline as the time interval increases
lienhardt@dakar.ird.sn). (11). In a study of young children who were strongly posi-

1066
 Genetic and Environmental Risk Factors for Tuberculosis 1067
tive reactors, the lifetime risk was reported to be as high as
10 percent (12). Any condition modifying the balance estab-
lished in the body between the tubercle bacilli and the host’s
immune defenses can have an impact on the risk of devel-
oping the disease. Factors that have been shown to influence
this balance include age, sex, human immunodeficiency
virus infection, immunosuppressive treatment, diabetes mel-
litus, malnutrition, alcoholism, and Bacillus Calmette-
Guérin vaccination (1, 6, 9). Any factor influencing the risk
of infection and/or the risk of breakdown after infection has
an effect on the incidence of tuberculosis (7).
Clustering of tuberculosis within families has been recog-
nized for a long time, but it is not clear whether this reflects
genetic factors predisposing people to infection and/or dis-
ease, shared environmental factors, or the facility of trans-
mission of infection within the home (13, 14). Epidemiologic
studies have demonstrated that risk of tuberculosis is
increased among close contacts of sputum smear-positive
patients and that the prevalence of active disease increases
with the intimacy of contact (15–17). High rates of transmis-
sion were found within households of smear-positive tuber-
culosis patients living in areas of high prevalence (18).
However, it has been suggested that the majority of cases in
the community are acquired from an unknown nonintimate
contact (19). Molecular epidemiologic studies have found
patients with no obvious epidemiologic relation to each other
to be infected with the same strain of M. tuberculosis, which
suggests that tubercle bacilli can be transmitted during brief
contacts between persons who do not live or work together
(20, 21). Thus, it remains unclear whether most transmission
of tuberculosis takes place within households or outside of
households (9).
Various lines of evidence indicate that genetic factors
partly determine differences in host susceptibility to
mycobacterial infection and that such factors might con-
tribute to the pattern of clinical disease (14). Studies in twins
have found higher concordance rates for monozygous twins
than for dizygous twins, which suggests that genetic factors
are important in susceptibility to tuberculosis (22, 23). A
number of genetic studies have demonstrated an association
between human leukocyte antigen haplotypes and suscepti-
bility to the disease (24). A case-control study carried out in
The Gambia showed that polymorphisms in the NRAMP1
gene were significantly associated with susceptibility to
tuberculosis, although it was not possible to distinguish
between susceptibility to acquisition of M. tuberculosis
infection and susceptibility to disease progression (25).
Recent case-control studies conducted among Gambians
(26) and among Gujarati Asians in London, United
Kingdom (27), found that a polymorphism in the VDR gene
is associated with susceptibility to tuberculosis. Lastly, a
genome-wide scan of sibling pairs from The Gambia and
South Africa identified potential susceptibility loci on chro-
mosomes 15q and Xq (28). Thus, genetic factors appear to
play a significant role in susceptibility to tuberculosis,
although their level of action and the physiologic pathways
involved remain to be fully understood.
Experimental evidence indicates that interferon-γ, a
cytokine produced by type 1 helper (Th1) lymphocytes,
Am J Epidemiol Vol. 155, No. 11, 2002
plays a central role in immune defense against tuberculosis
(29). Mutations in the interferon-γ gene or the interleukin-12
receptor gene have been identified among children with dis-
seminated nontuberculous mycobacterial disease (30–32).
The fine balance between the production of Th1 and type 2
helper (Th2) cytokines appears to be important for the deter-
mination of host resistance or susceptibility to mycobacteri-
al infection. Disruption of this balance in favor of Th2 cells
could affect the ability of the host to express a protective
immune response against M. tuberculosis and has been pro-
posed to play a role in the progression of human immuno-
deficiency virus disease in Africa (33). Thus, in countries
where parasite worms (such as helminths) are prevalent, the
differentiation of Th2 cells induced by parasite antigens
could produce cytokines that would affect the phenotype of
T lymphocytes that recognize third-party antigens and ham-
per their differentiation into Th1 effector cells (34, 35).
Thus, important questions remain regarding the respec-
Downloaded from aje.oxfordjournals.org by guest on September 14, 2010
tive roles of environmental and host-related factors in the
response to infection with M. tuberculosis, especially in
Africa. A large and complex study is required in order to
address these questions comprehensively. In this paper, we
describe the approach taken in a multicenter study of tuber-
culosis that was recently initiated in West Africa. We outline
the design of the study, describe the methodological issues
that arose in the planning of the study, and indicate how we
dealt with those issues. In a companion paper (36), we pre-
sent in detail the investigation of genetic factors and discuss
methodological aspects of the research. Although these
papers deal with the specific aspects of investigation of the
natural history of tuberculosis, we believe that the issues
considered will be relevant to studies of the etiology and
natural history of other infectious diseases, especially those
in which risk factors for infection might be different from
risk factors for disease—such as leprosy, African trypano-
somiasis, and even hepatitis B, which can lead (in high-
transmission areas) to hepatoma or cirrhosis decades after
the original infection.
STUDY DESIGN
Principles
Our principal objectives were to determine the risk fac-
tors for active tuberculosis and to investigate the relative
contributions of environmental and host-related factors. To
do this thoroughly, we decided to adopt a design that
included both a retrospective component and a prospective
component, based on the recruitment of infectious cases
with tuberculosis.
The study commenced in January 1999 and is being car-
ried out in three countries in West Africa (The Gambia, the
Republic of Guinea, and Guinea-Bissau), using consistent
methodology and standardized questionnaires. Laboratory
studies are being conducted in collaboration with European
institutions. Baseline data on demographic factors and
tuberculosis control in the three countries are shown in table
1. The general design of the study is shown in figure 1. The
study design combines two case-control studies, a prospec-
tive household study, and family genetic studies (table 2). In
1068 Lienhardt et al.

TABLE 1. Baseline information on demographic factors and tuberculosis control in three West African
countries

Republic
The Gambia* of Guinea-Bissau‡
Guinea†

Population 1,380,000§ 7,557,023§ 1,096,224¶

Ratio of urban residents to rural residents 25:75 30:70 5:95
Bacillus Calmette-Guérin vaccination coverage (%) 94 76 countrywide 92 in Bissau
94 in Conakry
Tuberculosis case detection rate (per 100,000
population) 84 countrywide 70.2 countrywide 153 countrywide
136 in Conakry 389 in Bissau
Length of antituberculosis treatment (months) 6 8 8
Outcome of treatment (%)
Cured/completed treatment 76 78 50
Defaulted 14 9 31
Died 6 7 11

Downloaded from aje.oxfordjournals.org by guest on September 14, 2010

Other (failure, transferred out of study, unknown) 4 6 8
Relapse rate (%) 2 2.7 6.7
Prevalence of human immunodeficiency virus in
the general population (%) 2.3 2 8.1
Prevalence of helminthiasis in the general
population (%) 17 25 Not applicable

* Data were obtained from the National Leprosy/Tuberculosis Control Programme, Ministry of Health, Banjul,
The Gambia, 1999.
† Data were obtained from the Programme National de Lutte Anti-Tuberculeuse, Conakry, République de
Guinée, 2000.
‡ Data were obtained from the Programma Nacional de Luta contra a Lepra e a Tuberculose, Ministerio de
Saude, Bissau, Guinea-Bissau, 1996.
§ 1998.
¶ 1996.

each country, newly diagnosed smear-positive cases with
tuberculosis are recruited. Their households are visited, and
demographic information is collected from all household
members. The duration of the study allows 12–18 months
for recruitment of index cases and controls and 2 years for
follow-up of their household contacts. In each country, the
study has been reviewed and approved by the national ethics
committee.
tion of the effect of household factors (environmental and
socioeconomic factors). These factors include: household
size, number of people per room, type of house, hygiene,
water supply, sanitation, presence of animals in the house-
hold, and socioeconomic status. These community controls
also contribute to the genetic association study and provide
a further resource for study of the effect of host risk factors,
in addition to the household controls.

Case-control studies Household cohort study

Two types of controls are being recruited. Within the
household of each case, a healthy control is recruited to
allow investigation of individual (host-related) risk factors
while controlling for differences in household environment.
This control is age matched to the case within 10 years. The
host-related factors under investigation are: sex, Bacillus
Calmette-Guérin vaccination, previous history of tuberculo-
sis, infection with M. tuberculosis, smoking, alcohol intake,
drug use, nutritional status, intercurrent infectious diseases
(including human immunodeficiency virus infection), and
conditions triggering a Th2-type lymphocyte response (such
as parasitic infections, asthma, or atopy). A second control,
also age matched to the index case within 10 years, is
recruited at random from a nearby household for investiga-
In this study, a household is defined as members of an
extended family living together in the same area and eating
from the same pot. At baseline, extensive information is
collected from all individuals who have lived in the house-
hold of the case and the household of the community con-
trol for more than 3 months, to assess their relatedness to
the index case/control and their level of exposure to the
case/control. Every member of the household is inter-
viewed regarding past disease history, examined for the
presence of a Bacillus Calmette-Guérin scar, skin-tested
with tuberculin (2 tuberculin units of RT-23; Statens Serum
Institut, Copenhagen, Denmark), and screened for signs
and symptoms of tuberculosis. To avoid the influence of the
“boosting” phenomenon when the tuberculin skin test is

Am J Epidemiol Vol. 155, No. 11, 2002

 Genetic and Environmental Risk Factors for Tuberculosis 1069
FIGURE 1. General design of a multicenter study of tuberculosis
risk factors in West Africa, showing the links between the case-
control studies, prospective household cohort study, and family-
based genetic studies. TB, tuberculosis; HH, household; TST, tuber-
culin skin test; TDT, transmission disequilibrium test; ASP, affected-
sibling pair analysis.
later repeated to detect conversion, two-step testing is being
used at baseline (37).
The purpose of the cohort study is to identify “secondary”
cases of tuberculosis among the households of the cases and
community controls in order to estimate relative rates of
development of the disease in individuals, given infection
(estimated by tuberculin skin test response at baseline),
according to various environmental and host-related factors.
In this cohort analysis, members of the case’s household are
considered exposed to infection with M. tuberculosis, and
they will be compared with members of the control house-
hold, who are considered unexposed in the absence of a
known source of infection within the household. Within the
index case’s household, “secondary cases” will be compared
with other household members for measurement of tubercu-
losis risk according to their geographic proximity and
degree of genetic relatedness to the index case, for assess-
ment of the relative effects of genetic and environmental
factors on risk of disease.
The households of cases and controls are visited after 3,
6, 12, 18, and 24 months for detection of any incident case.
During these visits, field-workers check for the presence of
Am J Epidemiol Vol. 155, No. 11, 2002
all persons initially recorded as being household members,
record the names of those who have left or died, and assess
the clinical condition of persons present. For detection of
conversion, tuberculin skin testing is repeated in subjects
who have previously had a negative response (<5 mm) 6
months after the first test has been performed (37). Field-
workers also identify suspected cases of tuberculosis on the
basis of the following characteristics: clinical signs and
symptoms (cough for more than 3 weeks, fever, weight loss,
night sweats, chest pain, hemoptysis); a tuberculin skin test
response greater than 15 mm in diameter (38); and tuber-
culin skin test conversion over the first 6 months of follow-
up. Persons with suspected tuberculosis are referred to the
study physicians for clinical examination and further inves-
tigation: sputum smear and culture, chest radiograph, gastric
lavage, bronchoscopy, and appropriate biopsy when extra-
pulmonary disease is suspected.
Downloaded from aje.oxfordjournals.org by guest on September 14, 2010
Immunologic studies
To evaluate the Th1/Th2 balance in tuberculosis, specific
serologic markers of Th1 and Th2 cell activity are mea-
sured. On the basis of recent work, we decided to use
immunoglobulin E, soluble CD30 antigen, and macrophage-
derived chemokine as markers of ongoing Th2 activity in
vivo (39, 40) and the product of lymphocyte activation
gene-3 as a marker of Th1 activity (41). These markers of
Th1/Th2 balance are measured in the cases, in household
members, and in community controls at entry into the study,
after 2 months of treatment, and at the end of treatment.
METHODOLOGICAL ISSUES
Case-control studies
Case definition and inclusion criteria. Cases are being
recruited at major urban health centers in The Gambia, the
Republic of Guinea, and Guinea-Bissau. All patients over
age 15 years with newly detected smear-positive pulmonary
tuberculosis who have been living at their current address
for more than 3 months are eligible for inclusion in the
study. Pulmonary tuberculosis must be confirmed by two
consecutive sputum smears that are positive for acid-fast
bacilli and positive culture. Informed consent is obtained
prior to enrollment.
Selection of household controls. The choice of possible
controls within each household is limited, despite the aver-
age household size’s being 10–12. For families, the only eli-
gible person in the required age range may often be the
spouse, which results in “antimatching” on sex. We decided
that the confounding effects of age were likely to be consid-
erably greater than those of sex and that this disadvantage
was worth accepting. We set no condition on the genetic rela-
tion between the case and the control. Any confounding
effects of sex or familial relationship will be taken into
account in the analysis, using conditional logistic regression.
Selection of community controls. A control household is
selected by choosing a random direction from the case’s
home (by spinning a pen in the air) and visiting the fifth
dwelling on the right. If, as is common in Africa, several
1070 Lienhardt et al.

TABLE 2. Design components of a study of tuberculosis risk factors in West Africa, 1999

Study type Study method Objective/outcome

Case-control study Comparison of index cases with Investigation of host risk factors
household controls

Case-control study Comparison of index cases with Family history of tuberculosis

community controls Investigation of environmental risk
factors
Investigation of host risk factors
Candidate genes (association
study)

Household cohort study Follow-up of members of house- Comparison of “exposed” families

holds of index tuberculosis
cases and of community controls
to identify “secondary” cases of
tuberculosis
with “nonexposed” families
Relative risk of developing
tuberculosis according to
exposure and relatedness to the
index case
Investigation of host and environ-

Downloaded from aje.oxfordjournals.org by guest on September 14, 2010

mental risk factors

Family-based genetic studies Transmission disequilibrium test Risk associated with having an
allele of a candidate gene; fine
mapping of the candidate region

Affected-relative linkage analysis Coarse mapping of the genome or

of the candidate region

households share the same dwelling, field-workers select
one household by drawing lots. The study is explained to
members of the household, and if they agree to participate,
workers select at random a healthy adult, matched to the
case within 10 years of age, to serve as a control for the
index case. In case of refusal or lack of a suitable control,
the field-workers repeat the above procedure to select
another household. If the household agrees to cooperate but
the person selected is not present, an appointment is made
for another visit.
The nature of the sampling process for external controls
means that households rather than persons are selected with
equal probability. This is appropriate, since external controls
are selected mainly for the evaluation of household-level
risk factors. However, a person in a small household will
have a higher probability of being selected as a control than
a person in a large household. We considered simple proce-
dures designed to overcome this—for example, selecting a
random household to start with, going from household to
household until, say, 20 adults have been listed, and then
selecting one at random. However, pilot studies showed the
procedure to be too complex to implement effectively.
Matching the control with the case on household size and
sex as well as on age would also have precluded the evalu-
ation of these factors as risk factors.
When a control develops tuberculosis. Occasionally a
control might become a case. Because of exposure, it is
likely that this will be a household control rather than a com-
munity control. This subject will not be recruited into the
case-control study as a case because of a lack of indepen-
dence from the index case, but he or she will remain in the
study as a control. In this situation, an analysis matched on
time of selection will yield an unbiased estimate of relative
risk, provided that this is constant over the study period
(42).
Cohort study
Case definition for secondary tuberculosis. In all cohort
studies, the outcome of interest should be clearly defined in
order to avoid biased estimates of rates. In this study, the
“secondary” cases that will develop within the households
of index cases are more likely to occur among children
(because of direct transmission’s leading to primary tuber-
culosis) or young persons (mainly because of exogenous
reinfection) than among adults, for whom the time span
between infection and development of disease might be
much longer in the absence of specific factors such as
human immunodeficiency virus infection (8, 9). To account
for the difficulty of diagnosing tuberculosis in adults with
smear-negative or extrapulmonary disease and in children
(43, 44), we set up a certainty grading system that quantifies
the confidence with which a suspected case can truly be said
to be a case of tuberculosis, based on several published sys-
tems for adults (45–47) and children (48). Thus, depending
on the presence and combination of defined signs and symp-
toms, cases of tuberculosis will be graded as “possible,”
“probable,” or “certain.”
Follow-up. The duration and frequency of follow-up are
based on the estimation that 5 percent of tuberculosis cases
will develop within 2 years of infection (11, 12). Members
of the households of the case and the community control are

Am J Epidemiol Vol. 155, No. 11, 2002

 Genetic and Environmental Risk Factors for Tuberculosis 1071
followed up regularly for detection of any suspected case of
tuberculosis. To encourage self-referral of household
members in the event of suspected tuberculosis, study par-
ticipants are actively trained in detecting the signs and
symptoms of the disease, and a card permitting free access
to the study clinic is provided to them.
Development of tuberculosis and proximity to the index
case. In addition to investigating the effects of specific
risk factors on the development of tuberculosis, the rate of
developing the disease will be estimated within the cohort of
household contacts of cases, according to the degree of
exposure of the individual to the index case and his or her
genetic proximity to the index case. The degree of exposure
is assessed through measurement of the contact’s geo-
graphic proximity to the case within the household and the
case’s infectiousness. This infectiousness is estimated
through the “bacterial load” of the case’s expectoration,
measured by the standard semiquantitative assessment of
the initial sputum smear positivity, the frequency of cough
measured over 24 hours using a specifically designed
“cough meter,” and the presence of a cavity on the patient’s
chest radiograph. The degree of genetic proximity is mea-
sured by the coefficient of relationship, the fraction of iden-
tical genes shared by descent by two individuals (49). The
ratio of the disease rate in siblings of the case living in the
household to the disease rate in unrelated members of the
household will provide an indication of the sibling recur-
rence risk ratio, λs (50), although this is usually measured
by comparing all siblings with the general population.
Immunologic studies
Use of the matched triplets constituted by the case, the
exposed contact, and the healthy community control pro-
vides a unique model for studying the natural history of
tuberculosis, representing various stages of the infection
from exposure to disease. Comparison of Th1/Th2 indica-
tors in the sera of 1) community controls with no known
recent exposure to tuberculous infection, 2) recently
exposed healthy individuals, and 3) smear-positive cases
allows one to “reconstruct” the natural history of tuberculo-
sis in individuals without embarking on long prospective
studies. Follow-up of cases during treatment allows investi-
gation of the impact of treatment on Th1/Th2 markers and
the association of changes in cases’ serum concentrations
with disease persistence or healing. Comparison of in-vitro
responses to defined antigens in these triplets can also rein-
force analysis of the contribution of host immune response
to the natural history of tuberculosis. Nested case-control
studies investigating in-vitro response to defined M. tuber-
culosis antigens will be conducted in a subsample of cases
and controls (51).
Sample size calculation
The sample size required to detect a given odds ratio fol-
lows a U-shaped distribution with respect to the prevalence
of the risk factor under investigation. Hence, the recruitment
of 800 cases and similar numbers of controls will provide at
Am J Epidemiol Vol. 155, No. 11, 2002
least 80 percent power to detect an odds ratio of 1.6 for any
risk factor (genetic, immunologic, or environmental) with a
prevalence between 10 percent and 85 percent in the con-
trols and to detect an odds ratio of 1.4 for any factor with a
prevalence between 20 percent and 70 percent.
The number of secondary cases expected in this study is
not large. The incidence of active tuberculosis in the general
population is estimated to be approximately 0.1 percent per
year (2, 11). In the control households, with an average of
10 members per household, follow-up of 800 households for
2 years will be expected to yield only 13 cases, assuming 20
percent loss to follow-up. Preliminary results indicate an
incidence among members of case households of 0.8 percent
per annum, such that we can expect to detect approximately
92 secondary cases in these households; this should be suf-
ficient to estimate the effects of specific factors on develop-
ment of tuberculosis in contacts. Lastly, since most of the
secondary cases will be relatives of the index cases, they
Downloaded from aje.oxfordjournals.org by guest on September 14, 2010
will contribute one affected pair to the allele-sharing study.
DISCUSSION
Tuberculosis is a multifactorial disorder in which the
environment interacts with host-related factors, contributing
to the overall phenotype. Improved understanding of the
individual balance between degree of exposure and inher-
ited genetic susceptibility to infection, as well as the respec-
tive effects of environmental and host-related factors on the
development of disease, will have strong implications for
tuberculosis control and prevention (7); increased insight
can help us redirect intervention efforts, such as contact
tracing and contact prophylaxis. It may contribute usefully
to research on new vaccines against tuberculosis through a
better understanding of host immune response (52, 53), and
the identification of genes that affect risk of disease can be
useful in generating new vaccine candidates. Lastly, the
establishment of cohorts of tuberculosis patients and their
contacts may provide useful baseline data for the develop-
ment of future phase III vaccine trials.
The present study design combining the case-control
method and the prospective household contact method may
be complex, but it offers the advantage of assessing together
the roles of environmental, immunologic, and genetic fac-
tors, both prospectively and retrospectively (36). In addi-
tion, it allows differentiation between the risk of becoming
infected and the risk of developing disease after infection,
adjusting for tuberculin reactivity and/or history of Bacillus
Calmette-Guérin vaccination—something that has not
always been possible in former studies. The innovative
aspect of assessing the triplet formed by the case, the
exposed contact, and the healthy community control pro-
vides an opportunity to investigate protective immunity to
tuberculosis in humans, and recruitment of contacts offers a
framework for additional studies on immunologic markers
of tuberculosis (51).
However, this design entails some limitations. The work-
load of follow-up is high, especially in the control families,
where the risk of developing tuberculosis is likely to be very
small. In addition, as in all cohort studies, validity depends
1072 Lienhardt et al.
on the capacity to detect all events of interest (here, the “sec-
ondary” cases) that occur within the study population. This
depends on the logistics of the study but also on the case
definition. Since follow-up in our study is initially restricted
to 2 years, we are more likely to detect “primary” tubercu-
losis among child contacts (11, 54). The clinical picture of
“primary” tuberculosis is quite different from that of the
“reactivated” or “reinfected” form of the disease, and the
pathophysiologic pathways might differ as well (55).
Therefore, at the time of analysis, it will be important to
quantify the certainty of the diagnosis. In many countries,
preventive therapy with isoniazid is recommended for chil-
dren under age 5 years who are exposed to an active case
(56). However, most countries in the developing world do
not implement chemoprophylaxis of contacts because of the
cost involved and the fear that noncompliance might
increase drug resistance. For that reason, in this study,
follow-up of contacts in the case households will be inten-
sive, and the criteria used to detect tuberculosis in children
are very sensitive.
There are major differences in the distributions of tuber-
culous infection (as measured by tuberculin surveys) and dis-
ease (as measured by case-notification rates) in industrialized
and nonindustrialized countries (2, 11). However, research
on environmental factors affecting the incidence of infec-
tion/disease carried out over the past 50 years has been con-
ducted mainly in industrialized countries, and few studies
have tried to assess the role of these factors in the transmis-
sion of tuberculosis in resource-poor countries. In industrial-
ized countries, increased case notification in the early 1990s
triggered investigation of individual risk factors for identifi-
cation of high-risk groups in which to target control activities
(57). This led to a rapid decline in notification for tuberculo-
sis in some areas (58). However, this type of investigation
has not been extensively carried out in developing countries.
While we are waiting for new vaccines to be developed and
tested, investigation of the respective roles of environmental
and genetic factors may provide new insights that will
improve tuberculosis control through better prevention and
care adapted to specific situations (7).
ACKNOWLEDGMENTS
This study is funded by the UK Medical Research
Council and the European Commission (DGXII, contract
IC18CT980375).
The authors thank Drs. Robin Bailey, Richard Bellamy,
Arnaud Marchant, Jane Rowley, and Peter Smith for their
contribution to the study’s development.
REFERENCES
1. Comstock GW. Frost revisited: the modern epidemiology of
tuberculosis. Am J Epidemiol 1975;5:363–82.
2. Murray CJ, Styblo K, Rouillon A. Tuberculosis in developing
countries: burden, intervention and cost. Bull Int Union Tuberc
Lung Dis 1990;65:6–24.
3. Shaw JB, Wynn-Williams N. Infectivity of pulmonary tubercu-
losis in relation to sputum status. Am Rev Tuber 1954;69:
724–32.
4. Loudon AG, Spohn SK. Cough frequency and infectivity in
patients with pulmonary tuberculosis. Am Rev Respir Dis
1969;99:109–11.
5. Grzybowski S, Barnett GD, Styblo K. Contacts of cases of
active pulmonary tuberculosis. Bull Int Union Tuberc Lung
Dis 1975;5:90–106.
6. Enarson DA, Rouillon A. The epidemiological basis of tuber-
culosis control. In: Davies PD, ed. Clinical tuberculosis.
London, United Kingdom: Chapman and Hall Ltd, 1994.
7. Lienhardt C. From exposure to disease: the role of environ-
mental factors in susceptibility to and development of tubercu-
losis. Epidemiol Rev 2001;23:288–301.
8. Styblo K. State of the art: epidemiology of tuberculosis. Bull
Int Union Tuberc 1978;53:141–52.
9. Smith PG. Epidemiology of tuberculosis. In: Bloom B, ed.
Tuberculosis: pathogenesis, protection and control. Wash-
ington, DC: American Society for Microbiology, 1994.
10. Vynnycky E, Fine PE. Lifetime risks, incubation period and seri-
Downloaded from aje.oxfordjournals.org by guest on September 14, 2010
al interval to tuberculosis. Am J Epidemiol 2000;152:247–63.
11. Styblo K. Epidemiology of tuberculosis. (Selected papers, vol
24). The Hague, The Netherlands: Royal Netherlands Tuber-
culosis Association (KNCV), 1991.
12. Comstock GW, Liveay VT, Woolpertt SF. The prognosis of a
positive tuberculin reaction in childhood and adolescence. Am
J Epidemiol 1974;99:131–8.
13. Frost WH. Risk of persons in familial contact with pulmonary
tuberculosis. Am J Public Health 1933;23:426–32.
14. Fine PE. Immunogenetics of susceptibility to leprosy, tubercu-
losis and leishmaniasis: an epidemiological perspective. Int J
Lepr Other Mycobact Dis 1981;49:437–54.
15. Andersen S, Geser A. The distribution of tuberculous infection
among households in African communities. Bull World Health
Organ 1960;22:39–60.
16. Raj Narain, Nair SS, Ramanatha R, et al. Distribution of tuber-
culous infection and disease among households in a rural com-
munity. Bull World Health Organ 1996;34:639–54.
17. Rouillon A, Perdrizet S, Parrot R. Transmission of tubercle
bacilli: the effects of chemotherapy. Tubercle 1976;57:275–9.
18. Beyers N, Gie RP, Schaaf HS, et al. A prospective evaluation
of children under the age of 5 years living in the same house-
hold as adults with recently diagnosed pulmonary tuberculosis.
Int J Tuberc Lung Dis 1997;1:31–7.
19. Madico G, Gilman RH, Checkley W, et al. Community infec-
tion ratio as an indicator for tuberculosis control. Lancet 1995;
345:416–19.
20. Genewein A, Telenti A, Bernasconi C, et al. Molecular
approach to identifying route of transmission of tuberculosis in
the community. Lancet 1993;342:841–4.
21. Small PM, Hopewell PC, Singh SP, et al. The epidemiology of
tuberculosis in San Francisco: a population-based study using
conventional and molecular methods. N Engl J Med 1994;330:
1703–9.
22. Simmonds B. Tuberculosis in twins. London, United
Kingdom: Pitman Medical Publishing Company, 1963.
23. Comstock GW. Tuberculosis in twins: a re-analysis of the
Prophit survey. Am Rev Respir Dis 1978;117:621–4.
24. Brahmajoti V, Pitchappam RN, Kakkanaiah VN, et al.
Association of pulmonary tuberculosis and HLA in South
India. Tubercle 1991;72:123–32.
25. Bellamy RJ, Ruwende C, Corrah T, et al. Variations in the
NRAMP1 gene and susceptibility to tuberculosis in West
Africans. N Engl J Med 1998;338:640–6.
26. Bellamy R, Ruwende C, Corrah T, et al. Tuberculosis and
chronic hepatitis B virus infection in Africans and variation in
the vitamin D receptor gene. J Infect Dis 1999;179:721–4.
27. Wilkinson R, Llewelyn M, Toosi Z, et al. Influence of vitamin
D deficiency and vitamin D receptor polymorphisms on tuber-
culosis among Gujarati Asians in West London. Lancet
2000;355:618–21.
Am J Epidemiol Vol. 155, No. 11, 2002
 Genetic and Environmental Risk Factors for Tuberculosis 1073
28. Bellamy R, Beyers N, McAdam KP, et al. A genome-wide
search for tuberculosis susceptibility genes in Africans. Proc
Natl Acad Sci U S A 2000;97:8005–9.
29. Orme IM. Progress in the development of new vaccines
against tuberculosis. Int J Tuberc Lung Dis 1997;1:95–100.
30. Newport MJ, Huxley CM, Huston S, et al. A mutation in the
IFNγ receptor gene and susceptibility to mycobacterial infec-
tion. N Engl J Med 1996;26:1941–9.
31. De Jong R, Altare F, Haagen IA, et al. Severe mycobacterial
and Salmonella infections in interleukin-12 receptor-deficient
patients. Science 1998;280:1435–8.
32. Altare F, Durandy A, Lamas D, et al. Impairment of mycobac-
terial immunity in human interleukin-12 receptor deficiency.
Science 1998;280:1432–5.
33. Bentwich Z, Kalinkovich A, Weisman Z. Immune activation is
a dominant factor in the pathogenesis of African AIDS.
Immunol Today 1995;16:187–91.
34. Shirakawa T, Enomoto T, Shimazu S, et al. The inverse asso-
ciation between tuberculin responses and atopic disorder.
Science 1997;275:77–9.
35. Bentwich Z. Good worms or bad worms: do worm infections
affect the epidemiological patterns of other diseases? (Letter).
Parasitol Today 2000;16:312.
36. Bennett S, Lienhardt C, Bah-Sow O, et al. Investigation of
environmental and host-related risk factors for tuberculosis in
Africa. II. Investigation of genetic factors. Am J Epidemiol
2002:155:1074–9.
37. Menzies D. Interpretation of repeated tuberculin tests: boost-
ing, conversion and reversion. Am J Respir Crit Case Med
1999;159:15–21.
38. American Thoracic Society. Diagnostic standards and classifi-
cation of tuberculosis and other mycobacterial diseases (14th
edition). Am Rev Respir Dis 1981;123:343–58.
39. Del Prete G, De Carli M, Almerigogna F, et al. Preferential
expression of CD30 by human CD4+ T cells producing Th2-
type cytokines. FASEB J 1995;9:81–6.
40. Galli G, Chantry D, Annunziato F, et al. Macrophage-derived
chemokine production by activated human T cells in vitro and
in vivo: preferential association with the production of type 2
cytokines. Eur J Immunol 2000;30:204–10.
41. Annunziato F, Manetti R, Tomasevic I, et al. Expression and
release of LAG-3-encoded protein by human CD4+ T cells are
associated with IFN-gamma production. FASEB J 1996;10:
769–76.
42. Rodrigues L, Kirkwood B. Case-control designs in the study of
common diseases: updates on the demise of the rare disease
assumption and the choice of sampling scheme for controls. Int
J Epidemiol 1990;19:205–13.
43. Osborne CM. The challenge of diagnosing childhood tubercu-
Am J Epidemiol Vol. 155, No. 11, 2002
losis in a developing country. Arch Dis Child 1995;72:369–74.
44. Fourie B, Becker PJ, Festenstein F, et al. Procedures for devel-
oping a simple scoring method based on unsophisticated crite-
ria for screening children for tuberculosis. Int J Tuberc Lung
Dis 1998;2:116–23.
45. Falk A, O’Connor JB, Pratt PC, et al. Classification of pul-
monary tuberculosis. In: Diagnostic standards and classifica-
tion of tuberculosis. 12th ed. New York, NY: National
Tuberculosis and Respiratory Disease Association, 1969.
46. Whalen CC, Johnson JL, Okwera A, et al. A trial of three reg-
imens to prevent tuberculosis in Ugandan adults infected with
the human immunodeficiency virus. Uganda–Case Western
Reserve University Research Collaboration. N Engl J Med
1997;337:801–8.
47. Glynn JR, Warndorff DK, Fine PE, et al. Measurement and
determinants of tuberculosis outcome in Karonga district,
Malawi. Bull World Health Organ 1998;76:295–305.
48. Donald PR, Fourie PB, Grange JM. Tuberculosis in children.
1st ed. Pretoria, South Africa: J. L. Van Shaik Publications,
1999.
49. Hartl DL, Clark AG. Principles of population genetics. 3rd ed.
Sunderland, MA: Sinauer Associates, 1997.
Downloaded from aje.oxfordjournals.org by guest on September 14, 2010
50. Risch N. Linkage strategies for genetically complex traits. II.
The power of affected relative pairs. Am J Hum Genet 1990;
46:229–41.
51. Andersen P, Munk ME, Pollock JM, et al. Specific immune-
based diagnosis of tuberculosis. Lancet 2000;356:1099–104.
52. Fine PE. Vaccines, genes and trials. In: Cardew G, ed. Genetics
and tuberculosis. (Novartis Foundation Symposium 217). New
York, NY: John Wiley and Sons, Inc, 1998:57–72.
53. Doherty T, Andersen P. Tuberculosis vaccines: developmental
work and what’s on the horizon. Curr Opin Pulm Med 2000;3:
203–8.
54. Sutherland I. Recent studies in the epidemiology of tuberculo-
sis, based on the risk of being infected with tubercle bacilli.
Adv Tuberc Res 1976;19:1–63.
55. Piessens WF, Nardell EA. Pathogenesis of tuberculosis. In:
Reichman LB, Hershfield ES, eds. Tuberculosis: a compre-
hensive international approach. 2nd ed. New York, NY: Marcel
Dekker, 2000.
56. O’Brien R. Preventive therapy for tuberculosis. In: Porter JD,
McAdam KP, eds. Tuberculosis: back to the future. New York,
NY: John Wiley and Sons, Inc, 1994.
57. Prevention and control of tuberculosis in US communities with
at-risk minority populations. MMWR Morb Mortal Wkly Rep
1992;41:1–11.
58. Frieden TR, Fujiwara PI, Washko RM, et al. Tuberculosis in
New York City—turning the tide. N Engl J Med 1995;333:
229–33.