Prevalence of and risk factors associated with latent tuberculosis infection in a Latin American region

Introduction

Tuberculosis (TB) is an infectious disease caused by the bacillus Mycobacterium tuberculosis. In the last 200 years TB has killed more than a billion people, and it is the main cause of death due to a single infectious agent (Yap et al., 2018). Antibiotic resistance, lack or poor vaccination campaigns, globalization and poverty have complicated the control of this disease (Dheda et al., 2014; Kargi et al., 2017). In 2018, the World Health Organization (WHO) reported 10 million new TB cases and 1.2 million TB deaths globally. Eighty percent of TB cases and 70% of deaths occurred in middle-income and low-income countries (WHO, 2019). Coevolution with humans has allowed this microorganism to develop mechanisms to persist within the host for decades (Huynh, Joshi & Brown, 2011). It is estimated that a quarter of the world’s population (~1.7 billion or 23% of people) has latent tuberculosis infection (LTBI) (Houben & Dodd, 2016), a state of continuous stimulation of the immune system by M. tuberculosis without evidence of clinical symptoms of active disease (Sharma, Mohanan & Sharma, 2012; Yap et al., 2018). It is thought that 5–10% of these individuals will progress to active TB within the first 2 years, and the risk increases in those patients with suppression of cellular immunity by HIV infection, the use of glucocorticoids, blood or organ transplantation, treatment with tumor necrosis factor α inhibitors, malnutrition or diabetes (Basera, Ncayiyana & Engel, 2017; Chen et al., 2015; Huynh, Joshi & Brown, 2011; Yap et al., 2018).

The WHO South-East Asia, Western Pacific and Africa regions have an LTBI prevalence above 20%, while the Eastern Mediterranean, Europe and The Americas have an LTBI prevalence lower than 17%. The Americas region has the lowest prevalence of LTBI, with 11% or approximately 108,000 million infected people (Houben & Dodd, 2016). However, in the Americas, TB persists as a major public health problem, approximately 282,000 new TB cases are reported every year, and approximately 18,000 people die from this cause (WHO, 2018b). Countries showing the highest morbidity and mortality are Brazil, Peru, Mexico, Haiti and Colombia, representing 68% of the TB cases in the region (PAHO, 2018).

Colombia is the third most populated country in Latin America with a population of 49 million inhabitants. Despite the country’s efforts to control TB, its incidence has increased in the last decade, from 26 cases per 100,000 inhabitants in 2008 (Rodríguez, Gil & Vera, 2010) to 33 cases per 100,000 in 2018 (WHO, 2018a). Local incidence rates highly differ between regions. Cali, is one of the cities with the highest TB incidence (41 cases per 100,000 inhabitants) which is higher than the national average (26.5 cases per 100,000 inhabitants) (Lesmes Duque & Reina, 2016; Sivigila, 2017), and Cali has an annual risk of tuberculosis infection of 1.3%, which is higher than the 0.1% risk reported for the world population (De la Pava, Salguero & Alzate, 2002; Hassan & Diab, 2014). Two contributing factors for this particular setting are (a) continued community transmission due to the delay in diagnosing cases of active TB, and (b) reactivation of LTBI in vulnerable populations (household TB contacts, prisoners, health care workers, immunocompromised patients, children under 5 years old and the elderly) (Barbosa et al., 2014; Del Corral et al., 2009).

Latent tuberculosis infection seems to be a reservoir from which active TB will emerge (Abubakar et al., 2018), representing a challenge to the aims of the End TB Strategy (90% reduction in TB incidence by 2035) (Houben & Dodd, 2016). However, detecting LTBI in humans is difficult. Historically, LTBI detection has been performed by assessing the T-cell response against M. tuberculosis using the tuberculin skin test (TST) or interferon-gamma release assay (IGRA). These tests are useful to identify people who could benefit from prophylaxis and to control TB incidence and transmission (Abubakar et al., 2018). Nevertheless, both the TST and IGRA have sensitivity and specificity limitations (Muñoz, Stagg & Abubakar, 2015), and a challenge to understanding how to best use these test is the lack of a gold standard (Stout et al., 2018). Although the evidence shows that both the TST and IGRA can be used for LTBI detection, the availability and affordability of the tests will determine which one should be chosen (WHO, 2018c). In Colombia, the test of choice is the TST, as it requires fewer resources than the IGRA and is more familiar to professionals in clinical settings (Ministry of Health of Colombia, 2015).

The TST interpretation should consider the presence of immunosuppressive conditions and TB prevalence in the setting, so different cut-off values have been recommended for defining a positive TST: ≥5 mm in immunosuppressed patients and close contact of TB cases and ≥10 mm in healthy individuals living in high-prevalence areas (ATS, 2000; Muñoz, Stagg & Abubakar, 2015). Considering a possible cross-reactivity with Bacillus Calmette-Guérin (BCG) vaccination and non-tuberculosis mycobacterial infections, some authors suggest using a threshold of 15 mm or more (Abubakar et al., 2018; Shero et al., 2014). In fact, TST reactivity stratified by BCG vaccination status have shown a positive predictive value for the development of TB similar to IGRA (Abubakar et al., 2018), and large TST reactions are more likely to indicate a higher risk of developing active disease (Cao et al., 2019; Moran-Mendoza et al., 2007; Watkins, Brennan & Plant, 2000). All these considerations must be taken into account when comparing the TST properties between different studies.

There are limited data on the epidemiology and risk of LTBI in the general population, particularly in developing countries. Additionally, LTBI screening is only recommended in target groups and extensive testing of LTBI status in the general population is not affordable (Chen et al., 2015). In Colombia, there is only one published study that describes TST reactivity in the source population (Del Corral et al., 2009). For this reason, the aim of this study was to estimate the prevalence of LTBI using two different TST thresholds (≥10 mm and ≥15 mm) and to identify the risk factors associated with a positive TST result and TST induration size. This may allow us to determine the potential size of the current reservoir of infection and to provide information to implement control measures.

Materials and Methods

Participants and data collection

A total of 1,079 volunteers with ages ranging from 14 to 70 years completed a standardized questionnaire including demographic, socioeconomic and clinical data and lifestyle habits. Considering an estimated 42.7% LTBI (Del Corral et al., 2009) with a 95% confidence level, 4% precision and 10% nonresponse rate, the sample size calculated was 647 individuals. Exclusion criteria for this study were a history of chronic diseases (hypertension, diabetes and cancer), previous TB disease or chronic respiratory symptoms, heart disease, advanced liver disease and immunosuppressive conditions (HIV/AIDS infection, transplant and lupus), pregnancy or lactation. These conditions were ruled out to reduce anergy to TST and the risk of severe arm swelling after TST application (Muñoz, Stagg & Abubakar, 2015). Thus, 629 (58.3%) subjects were subjected to a TST. However, 589 individuals were included for further study (Fig. 1).

Overcrowding was defined as homes with more than three to less than five people per room. BCG vaccination status was evaluated by visual inspection (vaccination scar). The household socioeconomic status (SES) was defined as a composite index developed by an analysis based on characteristics of the dwelling, source of drinking water, type of toilet facilities and features of the neighborhood. The household SES index was categorized into tertiles of (1) extremely-low, (2) low and (3) medium, as this was a population with low resources. A history of smoking was defined as tobacco use in the last 6 months. Alcohol consumption was stratified into four levels: high (consumption between 350 and 750 mL, ≥4 times/week), moderate (consumption between 250 and 750 mL, 2–3 times/week), low (consumption between 50 and 100 mL, once/week) and none. Physical activity was defined as yes (exercise every day or ≥3 times/week) or no (never exercise or <2 times/week). The body-mass index (BMI) was classified as underweight (<18.5 kg/m²), normal weight (≥18.5 to 24.9 kg/m²), overweight (≥25 to 29.9 kg/m²) and obesity (≥30 kg/m²) according to the classification proposed by the WHO (WHO, 2003). The employment situation was categorized as employe (self-employed workers and subordinate workers) and unemployed (housewives, student or retired).

Results

Demographic, socioeconomic, lifestyle factors and their association with TST positivity (≥10 mm)

In the bivariate analysis present in Table 3, when the 14–19 years age category was taken as the reference group, it was found that after the age of 30 years, the risk of LTBI increased significantly, with ORs from 4.31 to 7.78. In this cross-sectional study, males had a 0.70-fold higher risk of LTBI than females (OR = 1.70, 95% CI [1.10–2.64]). Meanwhile, as shown in Fig. 2, TST-10 positivity increased in males and females with increasing age, a trend that was significant (p for trend <0.05). However, the LTBI prevalence in males was significantly higher than in females for the 40–49 years age category (OR = 3.72, 95% CI [1.38–10.03]) TST-10 (Table 4). The education distribution showed that a higher education level decreased the risk of LTBI when compared with a primary or lower level (OR = 0.59, 95% CI [0.39–0.86] for secondary). Being an employe significantly increased the LTBI risk (OR = 1.89, 95% CI [1.30–2.75]). When the risk of LTBI by zone was evaluated with reference to the high prevalence center zone (50%), it was observed that living in the north or suburb zone decreased the LTBI risk (OR = 0.29, 95% CI [0.17–0.48] and OR= 0.27, 95% CI [0.15–0.49], respectively). Low alcohol consumption significantly increased the risk of LTBI (OR = 2.28, 95% CI [1.13–4.59]) (Table 3). No association was observed between BCG scar and TST positivity.

Table 3:

Characteristics of the study population and its association with TST-10 and TST-15 positivity.

Characteristics	n	TST-10			TST-15
		n (%)	OR [95% CI]	p	n (%)	OR [95%CI]	p
Age (years)
14–19	31	2 (1.3)	Reference		2 (2.6)	Reference
20–29	74	10 (6.7)	2.26 [0.46–11.0]	0.300	4 (5.1)	0.83 [0.14–4.77]	1.00
30–39	96	22 (14.8)	4.31 [0.95–19.5]	0.042	11 (14.1)	1.87 [0.39–8.97]	0.424
40–49	145	45 (30.2)	6.52 [1.49–28.5]	0.005	20 (25.6)	2.32 [0.51–10.5]	0.262
50–59	180	48 (30.2)	5.27 [1.21–22.9]	0.014	30 (38.5)	2.90 [0.65–12.8]	0.143
60–69	63	22 (14.8)	7.78 [1.69–35.7]	0.003	11 (14.1)	3.06 [0.63–14.7]	0.146
Sex
Male	118	40 (26.8)	1.70 [1.10–2.64]	0.016	21 (26.9)	1.57 [0.91–2.72]	0.103
Female	471	109 (73.2)	Reference		57 (73.1)	Reference
Household SES
1	212	48 (32.2)	0.66 [0.42–1.04]	0.071	20 (25.6)	0.56 [0.30–1.03]	0.062
2	198	46 (30.9)	0.68 [0.43–1.08]	0.101	30 (38.5)	0.96 [0.55–1.68]	0.895
3	179	55 (36.9)	Reference		28 (35.9)	Reference
Overcrowding
>3 person/room	32	8 (5.4)	0.98 [0.43–2.24]	0.968	5 (6.4)	1.23 [0.45–3.29]	0.683
≤3 person/room	557	141 (94.6)	Reference		73 (93.6)	Reference
Educational status
Postsecondary	53	11 (7.4)	0.55 [0.27–1.14]	0.106	4 (5.1)	0.35 [0.12–1.05]	0.052
Secondary	321	69 (46.3)	0.59 [0.39–0.86]	0.006	34 (43.6)	0.52 [0.32–0.85]	0.008
Primary or less	215	69 (46.3)	Reference		40 (51.3)	Reference
Employment situation
Employee	259	83 (55.7)	1.89 [1.30–2.75]	0.001	42 (53.8)	1.58 [0.98–2.55]	0.059
Unemployed	330	66 (44.3)	Reference		36 (46.2)	Reference
Zone
North	334	74 (49.7)	0.29 [0.17–0.48]	0.000	36 (46.2)	0.22 [0.12–0.41]	0.000
Suburb	183	39 (26.2)	0.27 [0.15–0.49]	0.000	17 (21.8)	0.19 [0.09–0.38]	0.000
Central	72	36 (24.2)	Reference		25 (32.1)	Reference
Physical activity
No	412	95 (63.8)	0.68 [0.46–1.01]	0.057	52 (66.7)	0.83 [0.50–1.39]	0.497
Yes	177	541 (36.2)	Reference		26 (33.3)	Reference
Ever smoked
Yes	89	26 (17.4)	1.27 [0.77–2.09]	0.356	13 (16.7)	1.14 [0.60–2.18]	0.680
No	500	123 (82.6)	Reference		65 (83.3)	Reference
Current smoking
Yes	68	20 (13.4)	1.27 [0.72–2.21]	0.407	11 (14.1)	1.31 [0.65–2.62]	0.448
No	521	129 (86.6)	Reference		67 (85.9)	Reference
Alcohol consumption
High	6	1 (0.7)	0.61 [0.07–5.25]	0.648	0 (0.0)	–	0.347
Moderate	36	6 (4.0)	0.61 [0.25–1.49]	0.273	3 (3.8)	0.62 [018–2.06]	0.428
Low	35	15 (10.1)	2.28 [1.13–4.59]	0.018	9 (11.5)	2.34 [1.05–5.22]	0.032
None	513	127 (85.2)	Reference		66 (84.6)	Reference
BMI (kg/m2)
<18.5	20	1 (0.7)	0.18 [0.02–1.38]	0.065	1 (1.3)	0.35 [0.04–2.74]	0.301
≥18.5–24.9	248	56 (37.6)	Reference		32 (41.0)	Reference
≥25–29.9	229	65 (43.6)	1.35 [0.89–2.05]	0.146	32 (41.0)	1.09 [0.65–1.86]	0.732
≥30	92	27 (18.1)	1.42 [0.83–2.44]	0.197	13 (16.7)	1.11 [0.55–2.22]	0.767
BCG Scar
No	53	13 (8.7)	0.96 [0.49–1.84]	0.893	8 (10.3)	1.18 [0.53–2.61]	0.677
Yes	536	136 (91.3)	Reference		70 (89.7)	Reference

DOI: 10.7717/peerj.9429/table-3

Note:

Significant variables are denoted in bold

Table 4:

LTB prevalence between males and females by age groups using a TST-10 threshold.

Age (years)	Male				Female				OR (95% CI)	p
	n	TST-positive		p	n	TST-positive		p
		n	(%)			n	(%)
14–19	12	0	0		19	2	10.5		–	0.510*
20–29	11	0	0		63	10	15.9		–	0.155
30–39	17	6	35.3	0.001ǂ	79	16	20.3	0.026ǂ	2.15 [0.69–6.69]	0.181
40–49	19	11	57.9		126	34	27.0		3.72 [1.38–10.0]	0.007
50–59	44	16	36.4		136	32	23.5		1.86 [0.89–3.86]	0.094
60–69	15	7	46.7		48	15	31.3		1.93 [0.59–6.23]	0.274

DOI: 10.7717/peerj.9429/table-4

Notes:

* p value was calculated from Fisher’s exact test.

ǂp value for trend was calculated from Chi-square test.

Significant variables are denoted in bold.

In the multivariate logistic regression analysis (Table 5), being age 30 years and above (ORs from 4.29 to 8.30), being male (OR = 1.71, 95% CI [1.04–2.84]), being employed (OR = 1.56, 95% CI [1.02–2.38]) and having a low alcohol consumption (OR = 2.40, 95% CI [1.13–5.11]) were risk factors for LTBI, while living in the north or suburb zone of the city reduced LTBI risk (OR = 0.32, 95% CI [0.18–0.55] and OR = 0.28, 95% CI [0.15–0.52], respectively). The model had 12.8% sensitivity and 97.3% specificity with a Cox & Snell R-square coefficient of 0.098 (9.8%). The bootstrapping results confirmed the estimates obtained with the initial model and corroborated the significance of the predictor (age 30–39 years) within the model.

Table 5:

Risk factors associated with TST-10 positivity using a logistic regression.

Variables	β	Wald	OR*	95% CI	p	B p
Intercept	−2.007	6.670			0.010	0.007
Sex (male)	0.540	4.420	1.71	[1.04–2.84]	0.036	0.027
Employment situation (employee)	0.445	4.272	1.56	[1.02–2.38]	0.039	0.040
Zone (north)	−1.153	16.185	0.32	[0.18–0.55]	0.000	0.000
Zone (suburb)	−1.267	16.547	0.28	[0.15–0.52]	0.000	0.000
Alcohol consumption (low)	0.874	5.139	2.40	[1.13–5.11]	0.023	0.032
Age (30–39 years)	1.455	3.414	4.29	[0.92–20.0]	0.065	0.034
Age (40–49 years)	1.986	6.719	7.28	[1.62–32.7]	0.010	0.008
Age (50–59 years)	1.514	3.929	4.55	[1.02–20.3]	0.047	0.025
Age (60–69 years)	2.116	7.160	8.30	[1.76–39.1]	0.007	0.004

DOI: 10.7717/peerj.9429/table-5

Notes:

* From a multivariate logistic regression model with age, sex, social level, overcrowding, educational status, employment situation, zone, exercise, ever smoking, current smoking, alcohol consumption, BMI and BCG scar.

B: Bootstrapped for 5,000 samples.

Significant variables are denoted in bold.

Discussion

The risk factors associated with the acquisition of LTBI in the general population are rarely reported, and such studies are scarce (Chen et al., 2015; Martinez et al., 2013; Ncayiyana et al., 2016; Yap et al., 2018). This is the first study that evaluated the prevalence of LTBI, and the associated factors in the general population in Colombia, using two thresholds to classify TST positivity. A cut-off point ≥10 mm was used as the main criterion to define the positivity to M. tuberculosis infection following CDC guidelines (ATS, 2000), and it was found that LTBI prevalence (25.3%) was lower than the reported prevalence in previous studies, which also used the TST, conducted in patients from a trauma unit (38%) and health care workers (36.8%) in Cali, Colombia (Alzate et al., 1993; Barbosa et al., 2014). Compared to population-based studies where TST-10 positivity was the only diagnostic test used to identify M. tuberculosis infection, the LTBI prevalence observed in this study was lower than that reported in the source population of Medellín-Colombia (42.7%) and in urban informal settlements of Lima-Peru (52%) and Johannesburg-South Africa (34.3%) (Del Corral et al., 2009; Martinez et al., 2013; Ncayiyana et al., 2016) but higher than the estimated prevalence by mathematical models based on estimates of ARI from TST surveys and smear positive TB prevalence for the Americas region (11%) (Houben & Dodd, 2016) and that of other countries with a high TB incidence such as China (20% by IGRA in adults of a rural area) and Singapore (12.7% by IGRA in urban area residents) (Chen et al., 2015; Yap et al., 2018). These observations suggest that the LTBI prevalence in urban settlements is high and variable. The differences found with respect to other studies can be partially explained by the type of population. This study excluded participants with any underlying chronic diseases that may predispose them to the infection, and most of the included participants had low and medium household SES.

Considering the possible effect of BCG vaccination on TST specificity, a higher threshold was used (≥15 mm) in individuals who had vaccinated showing a prevalence of 13.2% that coincides with the LTBI prevalence reported in Singapore (Yap et al., 2018). The use of a TST-15 threshold in some studies has shown a higher likelihood of detecting M. tuberculosis infection (Wang et al., 2002) and is a predictor of progression to TB comparable to IGRA (Abubakar et al., 2018). In the present study, it is thought that the effect of vaccination was minimal because, in Colombia, this vaccine is administered in a single dose at birth, and as has been demonstrated, its effect on the outcome of a TST decreases after 15 years (Wang et al., 2002), and 82% of the participants were over 30 years old. No association was found between the absence of a BCG scar and TST positivity; this observation is in line with that reported in other studies conducted in South Africa, a high TB burden setting where BCG is given at birth, and the TST is performed more than 10 years later (Farhat et al., 2006; Mahomed et al., 2011).

Using a TST-10 threshold, it was found that age, male sex, being employed, and low alcohol consumption increased the risk of LTBI, while living in the north zone or suburb zone decreased it. A large number of TB cases reported in Cali are located in the central zone including districts 9 and 11, an area that is characterized by a vulnerable population that lives in conditions of overcrowding, malnutrition, and drug use, and living in zones distant from this point reduced the LTBI risk. On the other hand, our data indicate that after 30 years, the risk of LTBI increases significantly compared with the 14–19 years age group. This is in agreement with other studies where they report a high LTBI prevalence with advanced age (Belo & Naidoo, 2017; Chen et al., 2015; Gao et al., 2015; Lee et al., 2014; Yap et al., 2018). It is unknown whether the increase in age is the risk factor for acquiring LTBI (Lee et al., 2014) or if the increase in age from 30 to 39 years reflects a cumulative exposure to people with TB within the community, allowing the development of a detectable immune response against M. tuberculosis infection (Belo & Naidoo, 2017; Do Prado et al., 2017; Gao et al., 2015).

It was found that the LTBI prevalence in males was significantly higher than that in females, especially in the 40–49 years age group. This finding matches those observed a in rural population of China and community-based studies in South Africa and Peru showing epidemiological differences in LTBI prevalence based on sex (Chen et al., 2015; Martinez et al., 2013; Ncayiyana et al., 2016). This is consistent with the ratio (2:1) between males and females observed in TB epidemiological studies (Rhines, 2013). Two possible explanations are proposed: (1) a large proportion of females remain at home and are less likely to be exposed compared to males who have more active social responsibilities (Gao et al., 2015; Kizza et al., 2015), and (2) there is a differential susceptibility to M. tuberculosis infection or predisposition to delayed-type hypersensitivity responsiveness that is dependent on sex (Verhagen et al., 2012). Being employed was associated with an increased LTBI risk. One possible explanation is that the transmission of M. tuberculosis occurs in spaces where the working population is exposed, for example, public transport, which is overcrowded and has little ventilation, increasing the possibility of acquiring M. tuberculosis infection between people who often use this type of transport to travel long distances to their work and are repeatedly exposed (Oni et al., 2012). In support of these findings, sex and employment status showed a significant association (p = 0.000). In fact, 63% of females (mostly housewives) were unemployed, while 73% of males worked outside the home.

Our results show an increase in the risk of TST positivity in light drinkers compared to nondrinkers. Previous studies have shown that, dependent on the amount consumed, alcohol intake is associated with an impaired immune system, increasing susceptibility to respiratory infections such as pneumonia and tuberculosis (Happel & Nelson, 2005; Rehm et al., 2009; Silva et al., 2018). The relationship between low levels of alcohol consumption and TB risk remains unclear. Soh et al. (2017), in a cohort of middle-aged and elderly Chinese adults, found that low-dose intake of alcohol (monthly to weekly frequency) was associated with a lower risk of TB compared to nondrinkers, but their observation was limited only to nonsmokers. In contrast, in current smokers the consumption of alcohol at low levels did not show any protective effect for the development of TB, but the intake of two or more drinks daily acted synergistically with smoking to increase the TB risk. Similarly, in a study of household and community contacts in India, it was observed that male sex, alcohol consumption and smoking were risk factors for TST positivity. In the multivariate analyses, these three variables also showed an association with LTBI risk (OR = 3.93, 95% CI [1.3–11.9]) (Narasimhan et al., 2017). These studies suggest that alcohol and smoking are strongly correlated. In the present study, a low proportion of individuals smoked (11.5%), so the influence of this factor on the findings was ruled out. The small number of participants with moderate and high alcohol consumption could make comparisons between the categories difficult, hiding any possible association with TST positivity. Given the low rate of TST positivity, the lack of reactivity due to the immunosuppressive effects of alcohol cannot be ruled out.

When the TST-15 threshold was used, results similar to those observed with TST-10 were found. Being male and having low alcohol consumption increased the LTBI risk. In contrast, residing in the north zone or suburb zone and having a secondary education level decreased risk within the model. The association between a higher level of education and a reduction in the LTBI risk can be explained by an improvement in the quality of life of individuals and awareness of health risks dependent on lifestyle habits, and thus increased efforts to reduce their exposure to recognized TB risk factors such as poverty, overcrowding, smoking, and malnutrition (Lonnroth et al., 2009). These findings coincide with the results found in other population-based studies in China, Singapore and South Africa (Chen et al., 2015; Ncayiyana et al., 2016; Yap et al., 2018).

The average induration size, using both TST-10 and TST-15 thresholds, was above the cut-off points, allowing us to rule out any possible effects of BCG vaccination or cross-reactivity with nontuberculous mycobacteria (Borroto et al., 2011). This suggests that the threshold of ≥10 mm in the city is a useful tool to confirm M. tuberculosis infection in agreement with the recommendation of the CDC (ATS, 2000), while the threshold ≥15 mm can be used as an increased risk indicator for the development of TB in asymptomatic individuals (Ministry of Health, 2010). In support of this recommendation, a cross-sectional study in the Ethiopian population (TB incidence: 164/100,000 inhabitants) using a threshold TST-10 showed that the average TST induration size in community controls was 7.9 mm, which was less than that observed in our study (16 mm), while in household contacts and patients with TB, it was 13.6 mm and 18.1 mm, respectively (Shero et al., 2014).

The TST induration size in the generalized linear model was influenced by the increase in age from 40 years, a higher level of education, residing in the north zone or suburb zone, low alcohol consumption and being underweight, which coincides with the variables previously associated with TST positivity in the logistic models for TST-10 and TST-15, except for the association with BMI < 18.5 kg/m². The results found show a negative association between underweight and TST induration, explaining the low average TST induration size observed in these individuals compared with normal weight, overweight or obese individuals. These results contrast with previous studies where it has been observed that a lower BMI is an important risk factor for the development of TB (Hanrahan et al., 2010; Lonnroth et al., 2010; Patra et al., 2014). Additionally, overweight and obesity could decreased the risk of active TB (Hanrahan et al., 2010; Kim et al., 2018; Lin et al., 2018). The relationship between BMI and LTBI risk is not well described (Chen et al., 2015; Saag et al., 2018). Several population-based studies in rural areas of China have shown that overweight and obesity significantly increase the LTBI risk. These studies have also reported a nonsignificant negative association between a lower BMI and LTBI (Chen et al., 2015; Gao et al., 2015). Associations similar to those observed in our study, with crude ORs of 0.18 for TST-10 and 0.35 for TST-15, were found.

There were some limitations of this study. First, its cross-sectional nature did not allow the establishment of temporality or causality between LTBI and the associated factors. Second, individuals from some zones of the city (east and southeast) were not included, which implies selection bias, given that people living in the eastern area have a high TB risk. Third, since there is no gold standard for LTBI diagnosis, the estimation of its prevalence could be affected by TST performance. Fourth, there is a possibility of misclassification of drinkers by a self-report bias and the ability to remember. Fifth, as in any observational study, there could be a residual confounding effect of unknown or unmeasured factors on the associations observed. Despite these limitations, this study had an adequate sample size and statistical power and was the first population-based study of LTBI prevalence and associated risk factors in Colombia, so it provides valuable information in a country with an intermediate TB burden, where BCG is administered at birth.

Conclusions

The LTBI prevalence in our population without associated comorbidities, as measured using two thresholds TST (≥10 mm and ≥15 mm), was moderate (25.3% and 13.2%, respectively), reflecting a significant TB burden and the ongoing transmission of M. tuberculosis in the community. Several risk factors traditionally associated with TB (age, educational status, sex, employment situation, BMI and alcohol consumption) showed an association with the positivity and induration of TST in the three multivariate models. Unexpectedly, a lower BMI (<18.5 kg/m²) showed a negative and significant association with TST induration, and the LTBI prevalence in underweight individuals was low. In contrast, studies have shown an increase in TB risk among underweight individuals. Additional studies are required to validate our findings and identify other risk factors associated with LTBI. Given that BCG vaccination does not confer protection against TB in adults, and most people who develop TB in Colombia and other developing countries are vaccinated. The community identification of high-risk groups and prophylactic LTBI treatment to prevent progression to TB could be a cost-effective strategy of great impact.

Supplemental Information