Latent tuberculosis infection and associated risk indicators in pastoral communities in southern Ethiopia: a community based cross-sectional study

A community-based cross-sectional study was carried out in South Omo, a Zone in southern Ethiopia, from May 2015 to February 2016 to estimate the prevalence of LTBI among pastoral communities. This Zone shares borders with Kenya, South Sudan, Gamo Gofa Zone and Oromia Regional State (Fig. 1) [19]. It has a total area of 21,056 km² with 573,435 population. South Omo has eight districts, the majority of the population live in Hamer District (25%) and pastoralists live in six of the eight districts. The inhabitants of the Zone are classified into 16 indigenous ethnic groups with the Dassanech as the most dominant group [20].

The available health facilities include one General Hospital, 32 health centers, and 225 health posts. Of these health facilities, the General Hospital and 23 health centers provide acid fast bacilli microscopy examinations and TB treatment, while the other nine health centers provide only TB treatment. In 2016, 58 additional health care posts started a DOTS programme. The TB detection rate was 64% in 2016 but TB was still not included in the list of top ten diseases in the Zone [21].

The sample size for the study was estimated by assuming a 63.7% prevalence of LTBI [12], 95% confidence interval, 5% margin of error and 1.26 design effect [22]. The non-respondent rate was estimated to be 10% resulting in a minimum required sample size of 497.

Participants were selected using a multi-stage sampling approach. In order to increase both the efficiency of sampling and the precision, the total sample size was distributed to the districts proportional to their population sizes. Accordingly, 128 participants were from Hamer, 116 were from Benatsemay, 105 were from Dasanech, 62 were from Selamago, 42 were from Malee and 44 were from Ngangatom. Several kebeles (sub-districts) corresponding to the number of study participants allocated to each district were randomly sampled using a computer-based random number generator. The list of participants was taken from health facility of each district. Study participants from each household were enrolled from those individuals who were at home during the data collection.

Individuals were eligible for participation if they did not have signs and symptoms of TB, were at least 18 years of age and not pregnant in case of females. Volunteers were informed of the study’s purpose, consented, provided three ml blood and underwent clinical and physical examinations including inspection for scars associated with bacilli Calmette-Guerin (BCG) vaccination and anthropometric measures to check nutritional status. Each study participant was interviewed face-to-face to gather information on her/his medical history including contact with TB patients, other acute or chronic disease and drug treatments for them, the number of family members, smoking, Khat chewing and alcohol consumption habits, prior imprisonment and hospitalizations, raw milk and meat consumption and sharing beverages. HIV test results were available for some individuals from health facility records. Socio-demographic characteristics of study participants were included in the questionnaire.

The QuantiFERON-TB Gold In-Tube test (GFT-GIT) was performed according to the manufacturer’s instructions (QuantiFERON-TB Gold In-Tube, Cellestis Ldt., Carnegie, Australia). In brief, 3 ml of a venous blood was collected from each study participant and1ml was put into each of three tubes labelled as ‘nil’, ‘TB-specific antigens’, and ‘mitogen’. The blood sample was transported to Jinka Regional Laboratory within 8 h of collection. Prior to incubation, all tubes were maintained at room temperature and re-mixed by inverting 10 times, then incubated for 24 h at 37 °C. The cultures were centrifuged for 15 min at 3000 relative centrifugal force (g) after which the plasma was harvested and stored at − 80 °C. Frozen samples were thawed and IFN-γ release was measured using the QuantiFERON-TB enzyme linked immunosorbent assay (ELISA) protocol. Sample absorbance was read at a lambda maximum of 450 nm with a reference wavelength of 620 nm using appropriate settings in a 96-well plate spectrophotometer. Results were interpreted as positive, negative or indeterminate with a cut-off value of interferon gamma (IFN-γ) > 0.35 international unite per milliliter (IU/ml) using QuantiFERON ®-TB Gold analysis software version 2.62 (Cellestis, Carnegie, Australia, http://​www.​cellestis.​com).

Data were entered into EpiData version 3.1 and statistical package for the social sciences (SPSS) version 20.1 software tools. The primary outcome was LTBI status recorded as present or not present, defining a concentration of IFN-γ ≥ 0.35 IU/ml as presence of LTBI. The overall prevalence of LTBI was estimated by dividing the number of participants with the concentration of IFN-γ ≥ 0.35 IU/ml by the total number of study participants who had undergone the QFT-GIT test. Frequencies and percentages were used to summarize characteristics of study participants. The Pearson Chi-square (χ²) was used to test the difference in proportions of LTBI across the categorical variables. Bivariate logistic regression analysis was performed to obtain crude odds ratio (OR) with corresponding 95% confidence intervals (95% CI). Multiple logistic regression analysis was performed to assess simultaneously the association between multiple risk factors and the log odds of being positive for LTBI. From this model, adjusted odds ratios (AOR) and 95% CI were obtained. The linear regression method was used to assess the effects of individual-level confounders like gender, age, BMI, health status, any drug treatments during blood sample collection and vaccination with BCG on IFN-γ response.

Ethical approval for the study was obtained from Addis Ababa University, Aklilu Lemma Institute of Pathobiology Research and Ethics Committee as well as from the National Research Ethic Review Committee of Ethiopia (Ref No:3.10/785/07). Written consent was obtained from each study participant after a clear explanation of the study objectives. Blood sample collection was undertaken under aseptic conditions by licensed medical laboratory professionals. Volunteers with any sign and symptom of active TB or any other diseases during data collection time were transported at the project’s expense to nearby health facilities to undertake a complete examination. Individuals who had LTBI were advised to consult nearby health facilities regarding the development of symptoms of active TB.

As shown in Tables 1, 497 study participants (the age ranged from 18 to 80, mean age 37.2) were enrolled 50.5% of whom were males and 50.4% of whom were within the age range of 25 to 44 years. The majority (76.7%) were married and 84.5% were not formally educated.

The concentration of IFN-γ (TB antigen minus the nil) was more than 10 IU/ml in 194 individuals, in the range of 5-10 IU/ml in 18 individuals, in the range of 1-5 IU/ml in 18 individuals, in the rage of 0.35-1 IU/ml in 21 individuals and < 0.35 IU/ml in 230 individuals. The results were indeterminate in 16 (3.2%) of the cases. None of the individual-level variables investigated during the analysis had significant effect on the IFN-γ response to Mtb specific antigens before and after adjusting for the remaining variables (P > 0.05) (Table 2).

The prevalence of LTBI was 50.5% (95% CI: 46%, 55%) with no significant gender difference (49.8% among males and 51.2% among females; χ² = 0.10; P = 0.41). Although the prevalence of LTBI increased from 45.2% among those below 24 years to 54.5% in the age range of 45-64 years, this increase was not statistically significant (χ² = 6.91; P = 0.075). The prevalence was higher in the Dasanech District than in the Benatsemay District (64.8% vs 41.4%; χ² = 15.17; P = 0.010) and among those who reported eating raw meat frequently versus those who did not (66.7% vs 52.4%; χ² = 8.1; P = 0.042) (Table 3). Two hundred and six individuals reported previous HIV testing; three of them were HIV positive. Two of the three HIV positive subjects had a positive LTBI result.

The results from a logistic regression taking log-odds of LTBI positive as an outcome variable are summarized in Table 3. The odds of having LTBI were higher among individuals living in the Dasanech District (OR = 2.35; 95% CI: 1.36, 4.05; AOR = 2.62, 95% CI: 1.30, 5.28; P = 0.007) compared to those living in the Benatsemay District and among individuals who frequently ate raw meat (OR = 1.8 95% CI: 0.87, 3.81; AOR = 2.89, 95% CI: 1.09, 7.66; P = 0.033) compared to those who did not. The odds of LTBI positivity was smaller among individuals with a small family size (OR = 0.64 95% CI: 0.44, 0.94; AOR = 0.65, 95% CI: 0.42-0.99; P = 0.045) compared to those part of larger families (Table 3).