Published ahead of print on May 28, 2003, doi:10.1164/rccm.200212-1483OC
© 2003 American Thoracic Society
Risk Factors for Tuberculosis Infection in Sub-Saharan AfricaA Contact Study in The GambiaMedical Research Council Laboratories, Fajara; National TB Control Programme, Banjul, The Gambia; Institut de Recherche pour le Développement, Dakar, Sénégal; and London School of Hygiene and Tropical Medicine, London, United Kingdom Correspondence and requests for reprints should be addressed to Christian Lienhardt, IRD, BP 1386, Dakar, Sénégal. E-mail: lienhardt{at}dakar.ird.sn
Few studies have investigated the risk factors for tuberculosis (TB) infection in highly endemic countries. We conducted a household study in The Gambia, in which a tuberculin skin test (TST) was performed in members of the households of 315 smear-positive pulmonary TB cases and 305 community control subjects. The risk of being TST positive (10 mm or more) was higher in contacts of cases than in contacts of control subjects. It increased with age, male sex, and duration of stay in the household but was not associated with the presence of a bacille de Calmette-Guérin scar. Within the households of the TB cases, the risk of TST positivity was higher in males and was increased with age, social proximity to the case, and the radiologic extent of the disease in the case's chest X-ray. Adjusting on these, the risk of TST positivity was higher in first-degree relatives compared with more distant relatives and nongenetically related household members, but the effect was not statistically significant. In highly endemic areas, the risk of TB infection in contacts of TB infectious cases is associated with age, sex, intensity of exposure to the case, and severity of disease in the case, but it is possible that genetic factors contribute to the susceptibility to Mycobacterium tuberculosis infection.
Key Words: tuberculosis infection • tuberculin skin test • Mantoux test • contact investigation In many industrialized countries, the incidence of tuberculosis (TB) has declined significantly in the last decade, and elimination of TB has come back as a foreseeable goal, based on efficient treatment of overt TB cases and treatment of latent TB infection to prevent development of disease (1). In developing countries, however, the number of TB cases is still reported to increase steadily, especially in Africa south of Sahara, where TB is a leading cause of mortality (2). The development of TB in humans is a two-stage process in which a susceptible person exposed to an infectious TB case first becomes infected and after an interval of years or decades may later develop the disease. As the development of disease is often distant from the acquisition of infection, the risk factors for infection are different from the risk factors for the development of disease after infection (3). Numerous studies have been conducted on the risk factors for TB; however, most of them investigated the risk factors for the development of disease rather than infection, and they were mainly conducted in industrialized countries (4). Relatively few studies have investigated the factors influencing the risk of TB infection, especially in highly endemic countries. To identify the risk factors for TB infection as a separate entity from TB disease in resource-poor countries with a high reported prevalence of TB infection, we investigated the distribution of tuberculin skin test (TST) responses among contacts of smear-positive TB cases and contacts of matched healthy community control subjects in The Gambia, West Africa.
This study took place within the frame of a multicenter prospective household study investigating the role of environment and host-related risk factors for TB. This study was conducted in The Gambia, a small country situated at the edge of the sub-Saharan arid Sahel belt in West Africa and extending 20 to 50 km on either side of the River Gambia from the Atlantic ocean side to 400-km inland. According to the 1993 census, the population is over 1.3 million. TB control relies on passive detection and treatment of smear-positive cases through general and primary healthcare services. All TB cases, including those diagnosed in nongovernmental clinics or by private practitioners, are referred to the National TB Control Program for treatment. Since 1986, confirmed sputum smear-positive cases are treated with a 6-month short course regimen (2RHZE/4RH) delivered three times weekly under supervision by health staff in health centers and by village health workers in remote areas. The total number of patients detected in 1999 was 1,527, representing a case detection rate of 118/100,000 population. Infant immunization with bacille de Calmette-Guérin (BCG) is a well-established part of the national Expanded Program on Immunisation since 1984, and vaccination coverage is reported to be approximately 96%. The general design of the study is detailed elsewhere (5). TB cases were recruited at three major urban health centers in The Gambia. All newly detected smear-positive pulmonary TB patients older than 15 years who have been living at the same address for more than 3 months were eligible for inclusion in the study. Pulmonary TB was confirmed by two consecutive sputum smears positive for acid-fast bacilli and/or a positive culture. Sputum smears were graded as follows, adapting from the International Union Against Tuberculosis and Lung Diseases (IUATLD) recommendations (6): +, 1 to 99 acid-fast bacillus per 100 fields; ++, 1–10 bacilli per field; and +++, 10 or more acid-fast bacillus per field. A chest X-ray was also performed in all TB cases, and films were read by an independent chest specialist. Informed consent was obtained before enrollment. Households of each case were visited. Households were defined as the extended family living together in the same area and eating from the same pot. Information was collected from the head of household on various variables, including the household size, the number of rooms and the structure of the house, hygiene conditions, water supply, and sanitation, as well as indicators of socioeconomic status. Detailed demographic information was collected from the members of the household, who were fully interviewed on their duration of residence in the compound, their relatedness and their exposure to the TB case, their past disease history, and the presence of symptoms of TB. Field workers checked the presence of a BCG scar on the left or right deltoid region of the arm. A TST was performed on the volar surface of the left forearm of each household member, using 2 TU of RT 23 (Statens Serum Institut, Copenhagen, Denmark). Induration diameters were measured along and across the arm within 48 to 72 hours by trained field workers using the pen method. For analysis purposes, the average of the width and length diameter was considered. Various criteria for skin tests positivity were explored, and cutoff points of 5 and 10 mm were chosen (7, 8). To ensure validity of the TST reading and to reduce interobserver variability, TST reading by each field worker was tested regularly against the same reference reader, and those departing from standard reference reading were retrained. In the absence of a sampling frame, control households were selected at random in the neighborhood of the TB case's household by choosing a random direction from the case's home and visiting the third dwelling on the right. If, as commonly observed in The Gambia, several households lived in the same dwelling, one household was selected by drawing lots. The study was explained to the members of the household, and after agreement, the household was recruited in the study. In case of refusal, the same procedure was repeated to select another household in the neighborhood. The household was then investigated in a similar way as for the household of the case. Information was collected on a wide range of variables in the questionnaires addressed to the cases, the control subjects, and the household members, with the view to cover as much as possible the large field of potential host-related and environment-related factors of TB. These variables were selected on the basis of the risk factors already known and accepted (reviewed in 4), adding on factors expected to predispose to TB in the particular situation of the developing countries, especially household-related and socioeconomic variables. The intensity of exposure of each household member to the index TB case was evaluated through the assessment of the social proximity of the individual to the case within the household at nighttime and the importance of activities shared with the case during the day. The social proximity of the individual to the case was quantified as follows: (1) sleeps in the same compound but in a different house, (2) sleeps in the same house but not in the same room, (3) sleeps in the same room but not the same bed, and (4) sleeps in the same room and in the same bed. Contact with the case at daytime was quantified as "occasional," "part of the day" (less than half a day in contact), or "most part of the day" (more than half a day). The genetic proximity of each contact to the TB case was categorized using the coefficient of relationship (r), which measures the proportion of genes shared identical by descent with the case (9). All genetically related contacts of the TB case were categorized as either first-degree relatives (parents, offspring, and full siblings), second-degree relatives (half siblings, grandparent, grandchild, uncle/aunt, nephew/niece), third-degree relatives (first cousins), fourth-degree relatives (half first cousins, first cousins once removed), or fifth-degree relatives (second cousins). Data were double entered and checked using Epi-Info software and analyzed using Stata software (version 7; Stata Corporation, College Station, TX). A random effects logistic model, which takes into account the clustering of contacts within households, was used to assess the relationship between the TST response of the contact (5 or more mm and 10 or more mm) and risk factors. The TB cases and their matched community control subjects were excluded from the analysis, and only household contacts were considered. Results are reported as unadjusted and adjusted odds ratios and their 95% confidence intervals. The likelihood ratio test was used to assess the overall significance of risk factors, tests for trend, and tests for interaction. The study was approved by the Medical Research Council (MRC)/Gambian Government Ethics Committee.
Data were collected on 315 case and 305 control households between March 1999 and January 2001 (Table 1) . All TB cases were smear positive, and 293 of them (93%) were culture positive. The number of contacts in the TB case households was 2,870 (median household size of 8, range of 2–43) and in control households 2,377 (median household size of 8, range of 1–31). The age distribution of contacts was similar in case (median of 18, interquartile ratio of 8, 28) and in control (median of 16, IQR of 7, 28) households. Male to female sex ratios were similar in case and control households.
The general distribution of TST response size in households of cases and control subjects follows a bimodal pattern (Figure 1) , with an excess of negative responses in community control households as compared with TB case households. The proportion of individuals with no palpable induration in case households was 1,151 of 2,664 (43%) compared with 1,376 of 2,124 (65%) in control households. In community control households, the proportion of positive TST responses was found to increase with age, and this increase was similar in males and females up to the age of 15 to 20, after which it was higher in males than in females (Figure 2) .
Overall, the risk of being TST positive was found to be higher in household contacts of TB cases than in household contacts of community control subjects (odds ratio = 3.46 [95% confidence interval, 2.80–4.27] and odds ratio = 4.17 [95% confidence interval, 3.34–5.20] for TST positivity defined as 5 mm or more and 10 mm or more, respectively). As results were similar when cutoff points of TST positivity were set at 5 and 10 mm, results are shown for the latter only. On univariate analysis, the risk of positive TST response in household contacts was found to increase with age and to be higher in males (Table 2) . It also increased with the duration of stay of the individual in the household. The risk of TST positivity increased in the presence of a family or an individual history of TB. It was, however, not associated with the presence of a BCG scar. The risk of TST positivity was higher in small households compared with large ones but was not associated with the average number of persons per room in the household (data not shown). Regarding socioeconomic variables, TST positivity increased if the house did not belong to the TB case/control or their family. It was not associated with occupation or with the ownership of specific assets in the household (fan, radio, television, bicycle, motorbike, refrigerator, air conditioner, car). Finally, the distribution of TST responses was not influenced by structural characteristics of the house (type of floor, walls, roof, number of windows, presence of stove, etc.) (data not shown) or by hygiene indicators (source of water, latrines, presence of animals).
A multivariate model was fitted to assess the potential confounding effect of the previously mentioned variables on the risk of being TST positive when being a contact of a case as compared with being a contact of a community control. None of the above variables confounded the association. However, the effect of being a contact of an infectious case was modified by age (p < 0.001) and household size (p = 0.0016), indicating that the increased risk of positive TST response in contacts of TB cases compared with contacts of community control subjects was even higher in children and in members of small households (Table 3) . A multivariate analysis restricted to the members of the control households showed that the only factors significantly related to TST positivity were age and sex and the interaction between them (p < 0.001 for both TST positivity defined as 5 mm or more and 10 mm or more) (as shown in Figure 2), and these effects were not confounded by BCG scar status.
Further analysis was undertaken among contacts of TB cases to identify the risk factors for TB infection given exposure. Univariate analysis showed that within the households of the TB cases, the risk of TST positivity was associated with sex, age, former history of TB, duration of stay in the household, and household size (Table 4) . There was no association between the presence of a BCG scar and the prevalence of a positive TST reaction for either cutoff point. The assessment of the intensity of exposure of the household members to the case showed that the risk of positive TST response increased with the social proximity of the individual to the TB case at nighttime and the importance of activities shared with the case during the day (although the latter was not statistically significant). The risk of positive TST response in household contacts was also found to increase with the presence of a cavity on the case's chest X-ray, with the number of zones involved (test for linear trend, p = 0.001), as well as with the density of acid-fast bacillus in the sputum of the case (test for linear trend, p = 0.03). It was not associated, however, with the age, sex, and human immunodeficiency virus status of the TB case or with the duration of cough in the case. Looking at the genetic proximity of the household contacts to the case, the risk of positive TST response appeared reduced in second-degree compared with first-degree relatives but was similar in more remote relatives as well as in those contacts not genetically related to the case.
A multivariate model was fitted to investigate the effect of potential confounders and effect modifiers on the risk of positive TST response in contacts of TB cases. An interaction between age and sex was observed (p < 0.001), indicating that the effect of age on a positive TST response was stronger in males than in females (Table 5) . There was some evidence that the risk of TST positivity was higher in small households compared with large households, although the effect was not statistically significant (odds ratio for household size 1–5 vs. more than 10 = 1.69; 95% confidence interval, 0.99–2.88; overall p = 0.11). The effect of severity of disease (expressed as the number of affected zones on the chest X-ray) remained after adjustment for age and sex (test for linear trend p = 0.003) but not the apparent effect of microbial density in the sputum of the case. The effect of social proximity with the case within the household also remained after adjustment, the risk of positive TST reaction increasing with the proximity of contact with the case (test for trend, p < 0.001).
To evaluate the potential role of genetic factors in susceptibility to Mycobacterium tuberculosis infection among persons in contact with an infectious TB case, we included the genetic proximity with the case in the model (Table 5). As can be seen, the risk of positive TST response is still reduced in second-degree compared with first-degree relatives, adjusted for age, sex, household size, severity of disease, social proximity with the case, and the interaction between age and sex. In more genetically remote relatives, however, as well as in household members who are not genetically related with the case, the effect appears reduced compared with first-degree relatives, but confidence intervals overlap 1, and the overall effect is not statistically significant (p = 0.15). Altogether, these results indicate that within the households of infectious TB cases, there is a strong effect of social proximity with the infectious case on TST positivity in contacts, adjusted on age, sex, household size, and severity of disease and that there might be an independent effect of genetic factors, although probably not very strong.
For a century, the TST has been the classic method used to identify individuals infected with M. tuberculosis and to measure the prevalence of M. tuberculosis infection in populations (10). It is generally accepted that an individual becomes tuberculin positive approximately 6 weeks after infection with the tubercle bacilli, and both infection and sensitivity to tuberculin remain for life in the absence of given appropriate treatment or the occurrence of some form of immunosuppression (11). The use and interpretation of the TST are, however, hampered by several limitations and constraints, as the response to the intradermal injection of tuberculin can also witness sensitization by environmental non-TB mycobacteria or former vaccination with BCG (7, 8, 10). For these reasons, although sensitivity of the test has been shown to vary little in various populations (12, 13), the specificity of the TST is unpredictable, and the underlying distribution of TB infection can be difficult to ascertain (14). In our study, however, cases and control subjects originate from the same community, in which sensitization by non-TB mycobacteria and former BCG vaccination is assumed to distribute evenly. As the only assumed difference between case and control households is the notion of a proven exposure to infectious TB, the overall distribution of TST responses in our study population is expected to reflect the dynamics of TB infection in that community, and the potential bias in assessing TB infection arising from sensitization by non-TB mycobacteria and BCG vaccination is assumed to be limited. The prevalence of positive responses to TST in the general population has been reported to vary with age and sex (13, 15). Throughout our dataset, the prevalence of tuberculin sensitivity is similar among male and female children up to adolescence, after which it is higher among males, as described in the literature. This difference after adolescence may reflect greater exposure among adult males because of different social role and economic activities but may also reflect a genuine difference in susceptibility to TB infection or predisposition to delayed-type hypersensitivity responsiveness (16). Studies conducted in the 1960s and 1970s showed that the risk of TB infection was increased among contacts of TB cases as compared with the general population (12, 13, 17, 18) and that the risk of infection increased with the intimacy of contact with the case (18). This has been confirmed in recent studies conducted in children in New York City (19) and in Botswana (20), in which contact with a TB case came out as the strongest risk factor for TB infection. In our study, we further showed that the risk of TB infection among household contacts was associated with the intensity of exposure of the household member to the case, as assessed through the social proximity with the case within the household and the extent of activities shared with the case, thus confirming and extending former findings. As TB is an air-borne disease, the risk of an uninfected person becoming infected is strongly associated with the probability of coming into contact with an infectious TB case and the intimacy of that contact (21). For this reason, crowding has been traditionally associated with the risk of TB infection, as a greater degree of shared airspace increased exposure to M. tuberculosis. We did not observe a variation in risk of TST positivity with crowding, defined as the presence of more than two persons per room in the household, similar to studies performed in Canada (22) and in Botswana (20). We observed, however, a reversed association between household size and TST positivity, similar to other findings in Africa (13). This suggests that more than the number of people in contact with the TB case in a limited environment, it is the increased occurrence of contact and the proximity of contact with the case that are determinant in the transmission of infection. This was recently confirmed in a study examining the distribution of TB infection in households of TB cases using an ELISPOT assay that measures the T cell response to M. tuberculosis specific antigen (ESAT-6) (23). This study showed a clear gradient of ELISPOT response with the proximity of contact with the case within the household. In addition to the degree of exposure, the risk of TB infection depends on the infectivity of the source case. Several studies have shown that sputum smear-positive pulmonary TB cases are more likely to infect their contacts than sputum smear-negative TB cases (18, 24, 25). We found that the degree of TST positivity was closely related to the presence of a cavity on the chest X-ray of the index TB case and to the number of zones involved on the X-ray, reflecting both the capacity for the case to excrete bacilli and the severity of the disease. We found also an association between a positive TST and the density of acid-fast bacillus in the cases' sputa, although this effect disappeared on multivariate analysis. Vaccination with BCG has been reported to induce cross-reactivity with tuberculin PPD, but the degree of tuberculin sensitivity is highly variable, depending on the vaccine strain used, the dosage, the method of administration, the age at vaccination, and other factors known to influence the reaction to tuberculin (8, 26). There is no reliable method to distinguish tuberculin reactions caused by vaccination with BCG from those caused by natural mycobacterial infections (7, 14). In a large data set collected in Malawi, the prevalence of TST positivity was consistently higher over all ages in individuals with a BCG scar than in those without a BCG scar (11). In our dataset, however, we did not find a difference in the prevalence of TST positivity among individuals with and those without BCG scar, as was reported among children who were under 5 years of age in Botswana (20) and in New York City (19) and among children aged 1 to 15 in North Canada (22) and in Brazil (27). In addition, recent data from a large tuberculin survey in Korea showed that the prevalence of TB infection was similar among persons with a BCG scar and persons without a BCG scar among children and young adults (28). Some discrepancy between effective BCG vaccination and presence of scar is to be expected, as scars are not invariably present among all vaccinees (29). Thus, in a tuberculin survey among school-children aged 0 to 9 years old in The Gambia, a scar was found in only 71% of the children who had a health card record of a vaccination (30). In the Canadian study, 17% of subjects with a record of past BCG vaccination showed no visible scar (22). As BCG is given immediately after birth in The Gambia and vaccination coverage is high, the absence of an association between BCG scar and TST positivity can be due to tuberculin sensitivity induced by BCG waning with time since vaccination (26). It was thus reported in Malawi that BCG rarely induces very strong tuberculin sensitivity and that the majority of vaccinated individuals lost their BCG-induced tuberculin sensitivity shortly after vaccination (11). In addition, in many tropical countries, postvaccination sensitivity cannot be entirely attributed to BCG in the presence of naturally acquired low-grade sensitivity due to environmental mycobacteria (13, 31). It thus appears that history of BCG vaccination should not be a factor in decision-making process about treatment of latent TB infection in persons from high-burden countries, especially in children and human immunodeficiency virus–infected persons (32). Our data confirm the observation already made in Africa in the 1960s that tuberculin infection tended to aggregate in certain households (12, 13). Whether this "clustering" of TB infection within families reflects just the facility of transmission of infection within the intimacy of the home or the effect of shared genetic factors predisposing to infection is not clear. It has long been suggested that genetic factors contribute to differences in host susceptibility to infection with mycobacteria (33), and numerous studies have been conducted to assess the genetic aspects of susceptibility to TB (34). One of the main weaknesses of these studies, however, lies in the fact that genetic factors were examined in relationship to the presence of TB disease and did not allow distinguishing between susceptibility to infection with M. tuberculosis and susceptibility to disease progression, as most studies were conducted in confirmed TB cases. Some studies were performed to examine the differences in TB infection among various ethnic groups as a proxy to evaluating genetic determinants of susceptibility to infection (35, 36), but results were conflicting (4). The possibility of a genetic influence on tuberculin responses was reported in children who had received two BCG immunizations (37), but not in young twins vaccinated at birth (38). Recently, however, in a study investigating genetic regulation of response to specific M. tuberculosis antigens among twins in The Gambia, it was reported that the magnitude of the delayed-type hypersensitivity response appeared to be heritable and that cellular response to specific M. tuberculosis antigens were genetically restricted (39). In our study, we showed that the risk of TST positivity among individual contacts of the index TB case within the household increased with the closeness of contact to the case. We also found that the risk appeared higher in first-degree relatives compared with more distant relatives and nongenetically related household members, adjusted on age, sex, social proximity to case, household size, and age and sex interaction, although evidence was not strong. Altogether, these results show that closeness of contact with an infectious TB case is the overriding determinant of TST positivity within the household, but they also suggest a possible contribution of genetic factors to the susceptibility to M. tuberculosis infection. This genetic influence on TST positivity would have at least two components: an influence on susceptibility to TB infection and an influence on the host's ability to mount a delayed-type hypersensitivity response to tuberculin. Further investigations are necessary, using the newly developed immunologic markers of M. tuberculosis infection that are independent of delayed-type hypersensitivity (23), which could help to disentangle these two components as well as to bridge the gap of knowledge between susceptibility to infection and susceptibility to disease (40).
The authors thank the field assistants who carried out the study with much dedication.
Supported by a grant from the European Commission (contract IC18CT980375) and by the Medical Research Council, United Kingdom (S.B. and K.F.). Conflict of Interest Statement: C.L. has no declared conflict of interest; K.F. has no declared conflict of interest; A.T. has no declared conflict of interest; S.D. has no declared conflict of interest; D.W. has no declared conflict of interest; K.P.M. has no declared conflict of interest. S.B. is deceased. Received in original form December 17, 2002; accepted in final form May 25, 2003
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Deceased
TOP
ABSTRACT
METHODS
