Risk of tuberculous infection in adolescents and adults in a rural community in Ethiopia

BACKGROUND: The incidence of tuberculosis (TB) in sub-Saharan Africa is one of the highest in the world. OBJECTIVE: To evaluate the prevalence of TB, the annual risk of tuberculous infection (ARTI) and associated risk factors in rural Ethiopia. M E T H O D S : A tuberculin skin test was performed among 2743 individuals in a rural community of Ethiopia around Ginci town, west of Addis Ababa, to estimate the prevalence of tuberculin reactivity and to assess factors associated with tuberculous infection. RESULTS: Among 2743 volunteer participants, test results were available for 2640, 691 (26.2%) of whom had an identifiable bacille Calmette-Gu´erin (BCG) scar; 221 (8.3%) reported household contact with a known TB case. The overall prevalence of TST reactions of 7 10 mm was 29.7%. The ARTI was estimated at 1.7%. Tuberculin reactivity varied with age, sex, income and history of household contact with a TB case. Presence of BCG scar was not related to tuberculin reactivity. CONCLUSIONS: Our findings indicate that despite an effective TB control programme, TB transmission rates are still high in rural Ethiopia. Provision of isoniazid prophylaxis in close contacts of active TB cases among the poorest population groups may reduce TB incidence

221 (8.3%) reported household contact with a known TB case. The overall prevalence of TST reactions of 710 mm was 29.7%. The ARTI was estimated at 1.7%. Tuberculin reactivity varied with age, sex, income and history of household contact with a TB case. Presence of BCG scar was not related to tuberculin reactivity. C O N C L U S I O N S : Our findings indicate that despite an effective TB control programme, TB transmission rates are still high in rural Ethiopia. Provision of isoniazid prophylaxis in close contacts of active TB cases among the poorest population groups may reduce TB incidence. K E Y W O R D S : tuberculin; ARTI; TB; delayed type hypersensitivity reaction; Ethiopia TUBERCULOSIS (TB) remains a major public health problem worldwide, particularly in the developing world. Globally, the disease took 1.4 million lives in 2012. 1 The situation is worsened by the rising incidence of multidrug-resistant TB (MDR-TB) and coinfection with the human immunodeficiency virus (HIV) in the resource-poor countries of sub-Saharan Africa and South-East Asia. 2 HIV is currently the greatest known risk factor for the development of active TB, resulting from a greatly increased rate of reactivation of latent tuberculous infection and enhanced susceptibility to progression to active TB following new infection. 3,4 The tuberculin skin test (TST), one of the oldest and most widely used of all immunological tests, is the standard method employed to estimate prevalence, incidence and trends in Mycobacterium tuberculosis infection in populations. 5 Despite its widespread use, major controversies remain over the mechanism, interpretation and utility of the test. 6 HIV infection is one condition that leads to loss of TST reactivity. Other factors associated with a reduced response to tuberculin include underlying viral or bacterial infections, malnutrition, immuno-suppression, overwhelming infection with M. tuberculosis, incorrect antigen injection techniques and conscious or unconscious bias. 5 Tuberculin surveys are important to evaluate the annual risk of tuberculous infection (ARTI), defined as the probability that an individual not previously infected with tubercle bacilli will be infected during the ensuing year, which is the most informative index of the magnitude of the TB problem. 6 The ARTI at a specific time indicates the magnitude of the incidence and prevalence of infectious TB cases. It has been estimated that an ARTI of 1% corresponds to about 50 cases of smear-positive TB per 100 000 population. 7 It can also be used to estimate the magnitude of the TB problem several years into the future. A decline in the ARTI is the earliest indicator of a decline in the TB problem, and could serve as a measure of the success of a TB control programme. 8 In an earlier study in Ethiopia, in 1990, when TB incidence was 290/100 000, overall TST reactivity of 11.8% was reported in children. 9 No studies of TST reactivity have been conducted since in Ethiopia, although the country has one of the highest TB burdens in the world, with an estimated incidence of 230/100 000. 1 These data may be useful for the National Tuberculosis Programme. Here we report on a study in a rural community in central Ethiopia to establish the prevalence of TST reactivity, assess the effect of various sociodemographic variables on TST reactivity and estimate the ARTI.

Study site
This was a cross-sectional study conducted in a rural community in Ethiopia, around Ginci town, Dendi District, located 90 km west of Addis Ababa. In this mainly agricultural community, cultivating cereal crops and raising domestic animals is the backbone of the economy. The area has limited access to health care and education.

Study population
The community was informed about the study, and individuals aged 13-54 years willing to participate and able to comply with the study protocol were registered. The district has a population of 250 000; .40% are aged ,14 years and 55% are in the 14-55 years age range. The area is located at an altitude of 1700-1800 m above sea level. Of the 2743 volunteers, 94.3% came from within a 15 km radius of Ginci Health Centre. All subjects underwent a full clinical examination, and those with evident signs of chronic immunosuppression, acute illnesses, diabetes or those taking immunosuppressive medication within the last 3 months were excluded.

Methods
TST was performed among 2743 subjects in May-November 2003. Relevant socio-demographic data (date of birth, sex, income, family size, bacille Calmette-Guérin [BCG] vaccination status, history of household exposure to a known TB case, level of education and marital status) were collected in the local language using a pre-tested questionnaire. The TST was performed at the Ginci Health Centre by injecting 2 tuberculin units of purified protein derivative (Statens Serum Institute, Copenhagen, Denmark) intradermally on the dorsal side of the forearm according to World Health Organization/ International Union Against Tuberculosis and Lung Disease recommendations. 10 After 48 h, the transverse diameter of skin induration was measured using the ballpoint technique 11 and expressed in mm.
A cut-off value of 10 mm was used to define a positive TST reaction. 12 The annual ARTI was calculated using the formula 1 À (1 À p) 1/a , where p is the prevalence of TST reactivity and a is the average age of the study population. 13 The underlying assumption in this formula is that the prevalence of TST reactivity is stable over time. To limit the estimated ARTI to a relatively narrow period of time, we used data from subjects aged 13-20 years (mean 17.7, median 18, n ¼ 1667). As the TST reactivity (26.2% vs. 26.1%, P . 0.4) and the average age (17.6 vs. 17.7 years, P . 0.7) of subjects with or without BCG scar were similar, we pooled data from subjects with and those without BCG scar.
No data on actual household income were obtained in the study. As a proxy indicator for household wealth, data on household possessions of domestic animals such as cattle, horses, goats and sheep were used to construct an index of wealth for the study population. Each domestic animal was coded 1 if present in the household; otherwise, it is coded as 0. To obtain a composite index, dummy variables for each domestic animal were summed up and then collapsed to form three categories based on distribution. Index values that fell below 33% were categorised as low, those ranging between 33% and 66% were categorised as medium, and the remainder (.66%) fell under the high category.
The study was approved by the ethics committees of Armauer Hansen Research Institute/All Africa Leprosy Rehabilitation and Training Hospital (Addis Ababa, Ethiopia), the Karolinska Institute (Stockholm, Sweden) and the National Ethical Review Committee of Ethiopia (Addis Ababa, Ethiopia).
Participants were included in the study only if they provided informed written consent; participants aged ,18 years were included upon oral assent and written consent from their guardian.

Statistical analysis
Data were double-entered using Microsoft Excel spreadsheet (MicroSoft, Redmond, WA, USA). Stata, version 6 (StataCorp, College Station, TX, USA) was used for statistical analysis. Differences in proportions were evaluated using the v 2 test. Sociodemographic variables independently associated with TST reactivity were evaluated using multivariate logistic regression. The outcome variable, i.e., TST reactivity, was dichotomised based on the cut-off point (10 mm) and coded as 1 if positive, and 0 otherwise. Independent variables included age, level of education, sex, household contact history with a known TB case, presence of BCG scar, income, family size and marital status. P , 0.05 was considered significant.

RESULTS
TST was administered to 2743 subjects, and the results of the test were obtained from 2640 subjects. The remainder (103, 3.7%) did not return for TST reading. The mean age of the population was 22.6 years (range 13-54); 58% were male. BCG scar was clearly identifiable in 691/2640 (26.2%). Two hundred and twenty-one (8.3%) reported household contact with a known TB case ( Table 1). The overall prevalence of TST reactivity was 29.7%. The results of multivariate logistic regression analysis showed that the risk of TST reactivity increased independently with age (P , 0.001) (Figure 1). There was a .40% increased risk of TST reactivity with each 10year increase in age (odds ratio [OR] 1.4, 95% confidence interval [CI] 1.3-1.6). The odds of having a positive TST reaction were nearly two-fold higher among those individuals who reported household contact with a known TB case than in those with no such contact (OR 1.8, 95%CI 1.3-2.3). Males were found to be at increased risk of being TST reactive compared to females (OR 1.2, 95%CI 1.01-1.5). Low-and middle-income individuals had a respectively 50% and 20% greater risk of being TSTreactive than those in the high-income category ( Table 2). Moreover, individuals with post high school education were found to have a nearly twofold greater risk of being TST-reactive than individuals with no formal education (OR 1.9, 95%CI 1.1-3.2) ( Table 2). However, history of BCG vaccination, as evidenced by a BCG scar, marital status and family size, was not significantly associated with TST reactivity.
Reducing the cut-off for a positive reaction from 10 to 5 mm slightly increased the prevalence of TST positivity, from 29.7% to 33.4%. More than 60% of the population showed no TST reaction (0 mm), while the remaining 39% showed reactions ranging from 2 to 50 mm (Figures 1 and 2). The ARTI was estimated at 1.7%.

DISCUSSION
Although Ethiopia is classified as one of the 22 high TB burden countries, with an estimated incidence of all forms of TB of 230/100 000, 14 no recent data on TST reactivity are available to estimate the extent of the distribution of the disease and to assess risk factors for tuberculous infection in rural communi- BCG ¼ bacille Calmette-Guérin; TB ¼ tuberculosis. Figure 1 Correlation between age of subjects and tuberculin reactivity. There was significant linear correlation between age and tuberculin reactivity (R ¼ 0.91, P , 0.0001).  ties. Most of the available estimates rely on unreliable routine health service reports of clinical cases. A survey conducted in the late 1980s in Ethiopia evaluated TST reactivity in children in the 6-10 years age group, showing overall TST reactivity (710 mm) to be 11.8%. 9 The current study, involving subjects aged 13-54 years in a rural area, showed a TST reactivity of 29.7%. The TST is an important tool for studying the prevalence of M. tuberculosis infection. 5 Alternative methods to estimate TB prevalence rely on the notification of clinical cases, which is highly influenced by the quality of reporting, particularly in countries where the health care infrastructure is weak. Knowledge of the prevalence of the infection is important for national TB programmes to be able to plan.
The current study revealed a prevalence of TST reactivity of 29.7% and an ARTI of 1.7%. Both figures indicate a high rate of TB transmission in Ethiopia. The ARTI computed in this study is lower than that observed in 1949 (4.1%), 8 but higher than that reported in 1989 (1.4%) in Ethiopia. 9 The increase from the 1989 rate may be a reflection of the rising incidence of TB as a result of deteriorating socio-economic conditions and/or the HIV pandemic in the country in the last two decades.
TST reactivity was significantly associated with age, sex, income, history of household exposure to a known TB case and post-high school education, but was not associated with history of BCG vaccination, marital status or family size in this rural population.
The lack of association between TST reactivity and BCG vaccination reported in this paper has also been observed in many other studies. 15,16 BCG vaccination is known to result in TST reactivity, but this has been shown to wane with time. 12,17,18 Some authors have suggested that BCG-induced TST reactivity could completely disappear within 10 years after vaccination in high TB incidence settings. 5 Considering the age of the subjects in this study (mean 22.6 years), BCGinduced tuberculin reaction would be expected to wane, as the population is BCG-vaccinated at birth. However, a meta-analysis of studies from 16 countries showed an increased risk of TST reaction in BCGvaccinated individuals if tested within 15 years following vaccination, 19 as was also reported by a study conducted in Switzerland among adults. 20 The reason for this discrepancy in BCG-induced TST responses in different regions is not clear, but we hypothesise that long and intense exposure to environmental mycobacteria in some regions may mask BCG vaccine-induced responses; environmental exposure is limited in the Western world.
Sex differences in the prevalence of TST reaction have been reported in several studies in different communities, with males being more likely to be TSTpositive than females. 21,22 Similarly, in the current study, the odds of being TST-positive was significant-ly higher in males than in females, confirming previous reports. It has been hypothesised that both socio-economic and cultural factors could account for the sex differentials in infection rates for diseases such as TB. It has been suggested that the male preponderance in TST reactivity might be due to the higher rate of exposure to TB, as males tend to have more social contacts than women, particularly in rural communities, where women generally stay at home. While such social factors may play a role in determining exposure to infection, it is unlikely that it occurs uniformly in all societies. 23 This leaves the possibility that biological differences are an explanation for this finding.
Another unexpected observation in this study was that there was a significantly higher rate of TST reactivity in subjects with post-high school education than in those with lower educational backgrounds. This difference persists after controlling for all other confounding variables considered in this study. This may be because these subjects are more mobile and may have travelled to and/or lived in more crowded cities-where transmission of TB is higher-to study or work. The increase in prevalence of TST positivity with advancing age observed in this study has also been reported by others, 17,24,25 and is probably a result of the longer exposure time in older subjects.
The HIV status of the subjects was not evaluated in this study. Several reports have shown that TST reactivity decreases significantly during the later stages of HIV disease. 26,27 However, a survey involving over 70 000 subjects in Ethiopia reported relatively low (2-3%) HIV infection rates in rural Ethiopia, 28 and a decline in HIV prevalence in rural Ethiopia, even in the most sexually active segments of the population. 29 Moreover, the subjects included in the current study were apparently healthy working individuals, with no clinical evidence of chronic disease. It is therefore unlikely that the number of subjects with HIV disease would be great enough to significantly affect the findings reported in this paper.
In this study, we observed that the poorest population groups were significantly more likely to be TST-positive than those with higher incomes; this may be because the poorest are more likely to live in crowded houses with less ventilation, and are less likely to seek health care when sick. Active TB cases in the poorest households are thus more likely to spread the disease than those who are now wealthy.
TST surveys in children are commonly used to estimate the ARTI. Choosing younger age groups allows the calculated average annual risk to be confined to a relatively narrow period of time between the average date of birth of the group and the date of the survey. 30 The present study, however, was conducted in adolescents and young adults; the estimated ARTI therefore indicates average exposure over a longer period of time. No recent data are available on the risk of TB infection in Ethiopia; however, such data would reveal valuable information on the extent of the disease in a rural community in the country, and may provide invaluable information for TB control in resource-poor countries in sub-Saharan Africa. Our data showed that household exposure to a TB case among the poorest is a significant factor for TB transmission. Targeted contact tracing and provision of prophylaxis to the poorest may therefore help to reduce the risk of transmission.
In conclusion, the current survey of TST reactivity in adolescents and adults in a rural setting in Ethiopia shows a prevalence of TST positivity of 29.7% and an ARTI of 1.7%. Increasing age, male sex, low income, post-secondary education and history of close contact with a known TB case were independent predictors of TST reactivity in this population.

R E S U M E N M A R C O D E R E F E R E N C I A:
África subsahariana presenta una de las incidencias de tuberculosis (TB) má s altas del mundo. O B J E T I V O: Evaluar la prevalencia, el riesgo anual de infección tuberculosa (ARTI) y los factores de riesgo asociados en los entornos rurales de Etiopía. M É T O D O S: Se practicó la prueba cutá nea de la tuberculina (TST) a 2743 personas de una comunidad rural de Etiopía en los alrededores de la población de Ginci al oeste de Adís Abeba, con el fin de estimar la prevalencia de positividad a la reacción tuberculínica y evaluar los factores que se asocian con la infección tuberculosa. R E S U L T A D O S: De los 2743 voluntarios que participaron en el estudio, 2640 contaron con resultados de la TST. De estas personas, 691 presentaban una cicatriz evidente de vacunación antituberculosa (BCG) (26,2%) y 221 refirieron contacto domiciliario con un caso de TB conocido (8,3%). La prevalencia general de reacciones tuberculínicas de 710 mm fue 29,7%. Se estimó un ARTI de 1,7%. La reactividad tuberculínica varió en función de la edad, el sexo, los ingresos y el antecedente de contacto domiciliario con un caso de TB. La presencia de la cicatriz vacunal no exhibió ninguna correlación con la reactividad tuberculínica. C O N C L U S I Ó N: Los resultados del presente estudio indican que pese a la existencia del Programa Nacional contra la Tuberculosis, se observa aun una alta tasa de transmisión de la enfermedad en las zonas rurales de Etiopía. El suministro del tratamiento preventivo con isoniazida a las personas má s pobres que tienen un contacto estrecho con casos de TB activa podría disminuir la incidencia de la enfermedad.
Tuberculin reactivity in rural Ethiopia i