Tuberculosis Co-infection and Associated Factors among People Living with HIV /AIDS Who are on Antiretroviral Therapy in Pastoral Community, Northeast Ethiopia. A Bayesian Analysis Approach

Abstract

: Tuberculosis is one of the major public health problems in Afar region where the majority of the population is pastoral. However, there are limited data and analyses upon which to guide public health interventions. Therefore, this study aimed to determine the prevalence of TB and its associated factors among people living with HIV in the pastoral community, Afar region, Northeast Ethiopia. A facility-based cross-sectional study was conducted on 443 participants selected by systematic random sampling technique. Bayesian multivariable logistic regression analysis was done and an adjusted odds ratio with its corresponding 95% credible interval was used to declare a statistical significance. The prevalence of TB-HIV con-infection was 26.7% [95% CI: (22.6%–31.2%)]. Patients without chronic diseases (AOR = 0.49, CI = 0.29–0.75) and who did not use substances (AOR = 0.44, CI = 0.29–0.64) were negatively associated with TB-HIV co-infection whereas living in rural residence (AOR = 2.93, CI = 1.55–5.08), having baseline CD4 count 200 to 499 cells/mm³ (AOR = 2.02, CI = 2.02–3.87), CD4 count less than 200 cells/mm³ (AOR = 11.23, CI = 6.09–18.73), and having a family size of greater than 5 (AOR = 3.36, CI = 1.49–6.62) were the positively associated with TB-HIV co-infection. The prevalence of TB-HIV co-infection was high in the study area. Residing in a rural site, using substances, having a chronic illness, having more than five family sizes, baseline CD4 count of 200 to 499 cells/mm³, and less than 200 cells/mm3 were factors that increase the probability of tuberculosis among people living with HIV. Thus, comprehensive interventions focused on the identified factors are recommended to decrease the burden of TB-HIV co-infection.

Keywords:

• Tuberculosis co-infection
• associated factors
• HIV
• Bayesian analysis approach
• pastoral community
• Northeast Ethiopia

PUBLIC INTEREST STATEMENT

Tuberculosis is one of the major public health problems in Afar region where the majority of the population is pastoral. Although several studies were conducted on TB-HIV co-infection in different parts of Ethiopia, there are no sufficient studies among the pastoralist community.

We have recently used a new statistical approach (i.e Bayesian analysis approach) to assess associated factors of TB-HIV coinfection. This research was conducted to determine the prevalence of TB-HIV coinfection and its associated factors in the pastoral community of Afar Regional State, Northeast Ethiopia. Therefore, this study provides important information to health-care professionals to give more attention and work on the factors that are responsible for TB-HIV coinfection. It would also be useful to input other researchers who want to conduct research in a similar area.

1. Background

The double burden of tuberculosis (TB) and human immunodeficiency deficiency virus (HIV) is one of the major global health challenges of the 21^st century. TB is the leading immuno-suppressing infection and the commonest cause of death among HIV-infected patients (Swaminathan & Narendran, 2008). It is responsible for more than a quarter of deaths in people living with HIV/AIDS (PLWHA; Damtie et al., ; Wondimeneh et al., 2012). The number of TB cases worldwide in 2018 was estimated at 10.0 million (with a range of 9 to 11.1 million), or 130 per 100,000 persons, and 8.6% of TB cases were HIV-positive individuals (W.H. Organization,).

Due to its increased vulnerability to other infections, HIV infection is the main risk factor for starting an active TB infection (Sinshaw et al., 2017) and not only increases the prevalence but also complicates the follow-up period and compromises the immunodeficiency response against TB. Because of stacked pharmacokinetic and pharmacodynamic disruptions, the coexistence of TB and HIV may increase the risk of drug-related issues in patients taking antiretroviral therapy (ART), such as organ damage and metabolic distress (Kiros et al., 2020). The increased pill burden of co-infection may lead to decreased adherence and consequently increased risk of drug-resistant TB(Alexander & De, 2007; Gassama & Kao, 2018)

Ethiopia is one of the nations that is severely impacted by both TB and HIV. According to reports, there are 164 TB cases per 100,000 people in the nation (W.H Organization, 2018). The nation is one of the 22 with the highest burden of TB in the world (WHO, 2017). Moreover, TB is one of the major public health problems among the pastoralist communities of the Afar region (Organization, 2009).

Although several studies were conducted on TB-HIV co-infection in different parts of Ethiopia, there are no sufficient studies among the pastoralist community. We have recently used a new statistical Bayesian analysis approach to assess associated factors of TB-HIV coinfection. This Bayesian analysis approach is a powerful statistical method in medical and public health research (Aychiluhm et al., ; Contreras et al., 2018) that is done based on the data at hand and pre-existing data. Therefore, the study was conducted to determine the prevalence of TB-HIV coinfection and its associated factors in the pastoral community of Afar Regional State, Northeast Ethiopia.

2. Methods

2.1. Study setting

Afar Regional State is one of the nine regional states of Ethiopia. It is administratively divided into five zones and 32 districts and Samara is the capital of the region. The Region is bordered by Eritrea to the Northeast, Djibouti to the East, Tigray region in the Northwest, Amhara to the West, Oromia to the Southwest, and Ethiopia Somali land region to the South East. The Region lies in the Great Rift Valley of East Africa about 120 meters below sea level around the Dallol depression.

There are six hospitals, 58 health centers, and 294 health posts that are owned by the regional government.

The study was conducted in Dubti General Hospital (DGH), which is found in Dubti town in Zone one, Afar regional state, North Eastern Ethiopia, at a distance of 602 km from Addis Ababa and 12 km from Semera, which is the regional capital city of the region.

The estimated projected population for 2013 in the district was 66,585 (CSA 2008). The district is subdivided into 14 kebeles (the smallest unit of public administration). There is one general hospital, two health centers, and nine health posts. There are also six private clinics and seven rural drug vendors that are privately owned

2.2. Study design and period

A facility-based cross-sectional study was conducted from January 1^st to February 1^st, 2021.

2.3. Source population and study participants

The source of the population of this study was all cases of HIV that are following care and treatment in DGH. All HIV-positive adults aged 18 years and above attending the ART center during the study period were included in the study participation.

2.4. Eligibility criteria

All adult HIV patients who were willing to participate in the study, who were in chronic ART care, and those who did not take anti-TB drugs either due to treatment missed, loss to follow-up, or naïve to anti-TB drugs for the last couple of weeks were included in the survey. However, HIV-positive patients who were severely ill, who had a serious mental problem or psychosocial disorder, HIV-positive patients who are already PTB diagnosed and already initiated co-treatment for TB-HIV co-infection are excluded from our survey, unable to give consent and those who were absent during the time of data collection were also excluded from the study.

Moreover, participants with incomplete secondary medical data records, particularly for laboratory (baseline CD4 + T-cell count) and other related missed clinical data were excluded from the study.

3. Sample size determination and sampling technique

The sample size for this study was determined using a formula for the estimation of a single population proportion (n = (Z α/2)² P (1-P)/d2) by using the prevalence, p = 28.6% (0.286) which was taken from previous study (Belay et al., 2015), using a 95% confidence interval, Zα/2 = 1.96, and the margin of error, d = 0.05. Then, the required sample size for the study by substituting the above values was approximately 490. We use correction formula and we found the final sample size Nf = ni/(1+ ni/N) 490/(1 + 490/3042) = 422. By adding a 5% non-response rate, the total sample size required for this study was 443.

We used a systematic random sampling technique and every individual had been selected according to their timing of follow-up to get the estimated sample size by computing K= $\frac{N}{n}$=$\frac{3042}{443}$= 6.87

We take K value = 7

A systematic random sampling technique was used to select study participants. The total number of PLWHA at the DGH ART center was 3042. The K^th value for the study was computed using the formula, K = $\frac{N}{n}$=$\frac{3042}{443}$= 6.87 = 7. Then, the first study participant was selected using the lottery method.

4. Study variables

The dependent variable was the magnitude of TB among HIV-positive patients attending the ART clinic of DGH. Whereas the independent variables include sociodemographic characteristics (age, gender, marital status, level of education, residence, and current occupation), past and current clinical characteristics (ART duration, CD4+ status both baseline and current count, plasma viral load, level of ART adherence, previous history of TB, hemoglobin level), body mass index (BMI), and related personal behaviors like alcoholic drinking and smoking of study subjects.

5. Data collection tools and procedure

The questionnaire was adapted from studies (Alemu et al., 2020; Belay et al., 2015; Brunello et al., 2011; Dooley & Chaisson, 2009; Kiros et al., 2020; Negussie et al.,) conducted before and modified in context. The questionnaire was translated from English version to Amharic version and back to English to keep its consistency.

A pretest was done using 5% of the questionnaires outside of the study area before the actual study was made to ensure completeness, simplicity, and clarity based on the objective of the study.

6. Data management and analysis

Data entry and coding were done using Epi-data version 3.1. Data cleaning and analysis were carried out using Stata version 16. Markov Chain Monte Carlo (MCMC) simulation with Metropolis-Hastings sampling algorithm was carried out. To assess the convergence algorithm in our study, time series (history) plots, density plots, Autocorrelation plots, and Gelman-Rubin statistics was used.

Summary statistics were carried out from the posterior distribution and an adjusted odds ratio (AOR) with a 95% Bayesian credible interval in the Bayesian multivariable logistic regression model was used to select variables that have a statistically significant effect on the dependent variable.

7. Results

7.1. Socio-Demographic Characteristics of Study Participants

In this study, 430 HIV cases with a 97% response rate were involved, of those 260 (60.5%) were female. The majority, 366 (85.1%) of the respondents were urban dwellers. Regarding the educational status of the study, 213 (49.5%) had no education. The majority of respondents 397 (92.3%) had a family size of 1 to 5.

This study revealed that 242 (56.3%) respondents had normal BMI (18.5–24.5), and more than 5% of the participants in the study were overweight whereas 164 (38.1%) of respondents were severely malnourished (BMI 16 to 18.49). Besides, the study showed that 164 (38.2%) respondents of the study had baseline CD4 count 200 to 499 cells/mm³, and 44 (10.2%) had CD4 count from <200 cells/ mm³ (Table 1).

Table 1. Sociodemographic Characteristics of HIV Patients in Dubti General Hospital, Afar Regional State, Northeast Ethiopia

Characteristics	Category	Frequency	Percentage
Age	18 to 24	23	5.3
	25 to 34	117	27.2
	35 to 44	174	40.5
	44 and above	116	27
Sex	Male	170	39.5
	Female	260	60.5
Religion	Muslim	259	60.2
	Orthodox	151	35.1
	Protestant	20	4.7
Marital status	Single	46	10.7
	Married	307	71.4
	Divorced	52	12.1
	Widowed	25	5.8
Educational status	No education	213	49.5
	Primary education	152	35.4
	Secondary and above	65	15.1
Occupation	Housewife	138	32.1
	Pastoralist	17	4
	Farmer	24	5.6
	Merchant	81	18.8
	Civil servant	34	7.9
	Daily laborer	106	24.7
	NGO employee	30	7
Family size	≤5	397	92.3
	greater than 5	33	7.7
Current BMI	16–18.49	164	38.1
	18.5–24.9	242	56.3
	25and above	24	5.6
WHO stage	Stage1	89	20.7
	Stage2	291	67.7
	Stage3	50	11.6
Baseline CD4count(in cells/mm^3	>500	222	51.6
	200–499	164	38.2
	<200	44	10.2
ART regimen	D4t+3TC+NVP	44	10.2
	D4t+3TC+EFV	61	14.2
	AZT+3TC+EFV	86	20
	TDF+3TC+EFV	121	28.1
	TDF+3TC+NVP	91	21.2
	TDF+3TC+DTG	27	6.3
Substance use	Yes	146	34
	No	284	66
Chronic illness	Yes	99	23.1
	No	331	76.9
Kitchen availability	Yes	342	79.5
	No	88	20.5
Location of kitchen(n = 342)	Attached	103	30.1
	Separated	239	69.9
Use un boiled cow milk	Yes	256	59.5
	No	174	40.5
Knowledge about preventive methods for TB (at least one method)	Yes	289	67.2
	No	141	32.8

7.1. Prevalence of tuberculosis among people who live with HIV/AIDS

The prevalence of tuberculosis among HIV cases was 115 (26.7%) with 95%CI (22.6%–31.2%). The magnitude varies in sex with 57.4% of tuberculosis cases being in females. The prevalence of tuberculosis regarding respondents’ educational status was 27.2%, 24.3%, and 30.8% in those who have no education, primary education, and secondary and above respectively.

7.2. Bayesian Logistic Regression Analysis

The Random-walk Metropolis-Hastings sampling procedure was applied with 205,000 total iterations. After 5000 burn-in terms were discarded, 200,000 samples were generated from the full posterior distribution. Non-informative normal prior distribution with mean = 0 and variance = 10⁶ for the fixed effect and gamma distribution with scale = 0.1, shape = 0.1 for the variance of random effect was used. Before undertaking any inference from the posterior distribution, the convergence generated from Markov chains was proved by convergence-assessment plots.

7.3. Associated factors of Tuberculosis among People Live with HIV/AIDS(PLWH)

According to this study residence site, use of a substance, having a chronic illness, family size, and baseline CD4 count were significantly associated with TB among People Living with HIV.

Accordingly, the odds of having TB/HIV co-infection were 2.93 times higher in patients living in the rural site as compared to those in urban areas (AOR = 2.93, CI = 1.55–5.08).

The study declared that patients with no chronic disease have 0.49 times less likely to develop active tuberculosis than those who did not have chronic diseases (AOR = 0.49, CI = 0.29–0.75).

Study subjects with a baseline CD4 count of 200 to 499 cells/mm³ (AOR = 2.02, CI = 2.02–3.87) and less than 200 cells/mm³(AOR = 11.23, CI = 6.09–18.73) had two and eleven times higher risk of developing TB respectively.

The study explored that HIV patients who did not use substances had more than 0.44 times less likely of developing active tuberculosis than those who did use substances (AOR = 0.44, CI = 0.29–0.64).

The odds of having TB/HIV co-infection was 3.36 times higher in patients who have a family size of greater than 5 as compared to those who have less than 4 (AOR = 3.36, CI = 1.49–6.62; Table 2)

Table 2. Factors associated with Tuberculosis among HIV cases on multivariable logistic regression analysis in Dubti General Hospital, Afar Regional State, Northeast Ethiopia

Variable	Category	Tuberculosis		SD	MCSE	AOR (95%CI)
		Yes	No
Sex	Male(ref)	49(28.8)	121(71.2)
	Female	66(25.4)	194(74.6)	0.22	0.005	0.87(0.52–1.39)
Age	-	-	-	0.009	0.001	0.99(0.98–1.01)
Residence	Urban(ref)	90(24.6)	276(75.4)
	Rural	25(39.1)	39(60.9)	0.91	0.038	2.93(1.55–5.08)*
Chronic illness	Yes(ref)	37(37.4)	62(62.6)
	No	78(23.6)	252(76.4)	0.11	0.004	0.49(0.29–0.75)*
Substance use	Yes(ref)	56(38.4)	90(61.6)
	No	59(20.8)	225(79.2)	0.09	0.003	0.44 (0.29–0.64)*
Respondent Educational status	No education(ref)	58(27.2)	155(72.8)
	Primary	37(24.3)	115(75.7)	0.25	0.005	0.93(0.53–1.51)
	Secondary and above	20(30.8)	45(69.2)	0.29	0.011	1.32(0.85–1.98)
Family Size	<4(ref)	5423.7)	176(76.3)
	4 to 5	47(27.8)	122(72.2)	0.32	0.007	1.22(0.70–1.95)
	>5	14(42.4)	19(57.6)	1.32	0.033	3.36(1.49–6.62)*
BMI	16–18.49	49(29.9)	115(70.1)
	18.5–24.9	59(24.4)	183(75.6)	0.44	0.022	1.56 (0.87–2.64)
	≥25	7(29.2)	17(70.8)	18.09	0.894	31.14(0.40–75.69)
WHO Stage	I	25(28.1)	64(71.9)
	II	59(20.3)	232(79.7)	0.07	0.002	0.18(0.08–1.35)
	III	31(62.0)	19(38.0)	1.24	0.056	2.00(0.55–5.21)
Baseline CD4 count	>500(ref)	38(17.1)	184(82.9)
	200 to 499	52(31.7)	112(68.3)	0.47	0.015	2.02(2.02–3.87)*
	<200	25(56.8)	19(43.2)	3.25	0.092	11.23(6.09–18.73)*

*statistically significant variables at 95% credible interval, ref = reference

8. Discussion

HIV infection increases the individuals’ susceptibility to Tuberculosis by impairing the immune response of the body to mycobacterium infection (Gesesew et al., 2016). In this study, the prevalence of TB-HIV co-infection was 26.7% which is higher than studies conducted in Motta hospital, Ethiopia (18.1%; Fenta et al., 2020), Wolaita Sodo Hospital, Ethiopia (17.4%; Fite et al.,), Addis Ababa, Ethiopia (25.8%; Alemu et al., 2020), and southern Ethiopia (18.2%; Fekadu et al., 2015).

This prevalence is also higher than in a national study conducted in Ethiopia (0.8%; Alene et al., 2019), survey in Ethiopia (7.3% (Gelaw et al., 2020), Arba Minch Hospital, Ethiopia (7.2%; Mama et al., 2018), Butajira, South-central Ethiopia (20.3%; Mohammed & Gebremariam, 2015), systematic review of 30 studies (22%; Teweldemedhin et al., 2018), University of Gondar Hospital (7.5%; Wondimeneh et al., 2012), and in northwest Ethiopia (10.1%; Alemayehu Belay & Berie, 2013). However, the finding of this study is lower than the studies conducted in Debre Markos, Ethiopia (40%; Alemu et al., 2021), Yirgalem General Hospital, southern Ethiopia (36.9%; Negussie et al.,).

The differences could be explained by the existence of socio-cultural and socioeconomic differences across Ethiopian regions regarding health service utilization, and differences in health facilities set-ups (i.e. drugs availability and trained health professionals). Besides, there will be differences in awareness of the community upon utilization of available health services like taking prophylaxis for the prevention of TB. Pastoralists and developed communities use health services in significantly unequal ways. Furthermore, there may be differences in adherence to the prescribed prophylaxis which has a direct impact on the development of TB among HIV-positive people.

This study indicated that patients with lower CD4 counts were more likely to have TB-HIV co-infection compared to those who have higher CD4 counts. This finding is similar to studies conducted in Motta hospital, Ethiopia (Fenta et al., 2020), Addis Ababa (Alemu et al., 2020), southern Ethiopia (Fekadu et al., 2015), Arba Minch Hospital, Ethiopia (Mama et al., 2018), Butajira, South-central Ethiopia (Mohammed & Gebremariam, 2015), Gondar University Hospital (Wondimeneh et al., 2012), Yirgalem General Hospital (Negussie et al.,), and systematic review of 21 studies (Tesfaye et al., 2018). People with low CD4 cells have impaired immune responses to any infections in the body system.

This study revealed that patients without chronic illness were 51% less likely to have TB-HIV co-infection compared to their counterparts. This finding is consistent with studies conducted in southern Ethiopia (Fekadu et al., 2015), and other studies conducted in different parts of the globe (Dooley & Chaisson, 2009; Seegert et al., 2017).

Those with HIV/AIDS are susceptible to a number of co-morbidities (Brunello et al., 2011; Gesesew et al., 2016). However, if HIV-positive people have the potential to prevent and control other comorbidities, they are less susceptible to opportunistic infections. Thus, HIV-positive patients without chronic morbidities have a lesser risk of developing opportunistic infections including TB.

Substance abuse is also one of the factors that affect co-infection status. Accordingly, those HIV‑ infected patients who were cigarette smokers were more likely to develop co-infection as compared to those who were not. The finding is consistent with studies conducted in different parts of Ethiopia (Alene et al., 2019; Fenta et al., 2020; Mama et al., 2018; Mitku et al., 2016), China (Soh et al., 2017), and 14 high-burden countries (Patra et al., 2014). The patients on ART are expected to be free from any substance abuse, which affects their level of adherence to their prescribed medication including TB prophylaxis. Thus, this study indicated that patients who are substance abusing have higher odds of TB-HIV co-infection compared to their counterparts.

This study indicated that the likelihood of TB-HIV coinfection was three times higher among patients who have more than five family size compared to those who have less than four family sizes. This finding is similar to studies conducted in Addis Ababa, Ethiopia (Alemu et al., 2020). This might be due to overcrowding, which increases the transmission of TB among family members. Besides, food insecurity is also more likely in a larger family size that leads the immunocompromised people to undernutrition. In turn, these conditions increase the vulnerability of HIV-positive people to opportunistic infections including TB infection.

In this study, people living in rural areas have three times higher odds of TB-HIV co-infection compared to HIV-positive patients in urban residences. In developing nations, like Ethiopia, people in rural areas are inaccessible (Gesesew et al., 2016) to the available treatment centers for both TB and HIV compared to people in urban areas. Besides, people in rural residences have low exposure to media, which advocates the prevention of TB among HIV-positive people. Thus, people living with HIV and those who are living in rural dwellers have higher odds of TB-HIV co-infection compared to their counterparts.

9. Conclusions

This study declared that the magnitude of TB among HIV cases was high in the study area. According to this study residing in a rural site, use of a substance, having a chronic illness, having more than five family sizes, baseline CD4 count of 200 to 499, and having less than 200 cells/mm³ increased the likelihood of TB among HIV in the study area. Despite the study’s endeavour of identifying factors associated with TB among HIV cases, as the study is a facility-based study undertaken in a single hospital, generalizations will not be possible and future community-based studies are recommended.

Abbreviations

Table

AOR	Adjusted Odds Ratio
ART	Antiretroviral Therapy
BMI	Body Mass Index
CD4	Cluster of Differentiation 4
CI	Credible Interval
DGH	Dubti General Hospital
HIV	Human Immunodeficiency Virus
HIV	Human Immunodeficiency deficiency Virus
MCMC	Markov Chain Monte Carlo
MCSE	Monte Carlo Standard Error
MDR-TB	Multi-Drug Resistant Tuberculosis
PLWHA	People Living with HIV or AIDS
PTB	Pulmonary Tuberculosis
SD	Standard Deviation
TB	Tuberculosis
WHO	World Health Organization

Ethics approval and consent to participate

Ethical clearance was obtained from the Ethical Review Committee of the College of Medicine and Health Sciences (CMHS), Samara University via permission letter of CMHS/0133/2020. The letter was submitted to the concerned bodies. Oral informed consent was obtained from the patients before enrolment. Privacy and confidentiality were kept.

Authors’ contributions

SBA, EM, KU, HA, KU, FY, and AWT had substantial contributions to the conception, design, and analysis of this research. They were involved in the interpretation of data and drafted the manuscript. SBA revised and finalized the manuscript. All authors read and approved the final manuscript.

Availability of data and materials

The datasets used during the current study are available from the corresponding author on reasonable request

Consent for publication

Not applicable.

Acknowledgements

We would like to thank Samara University, Afar regional health bureau and Dubti District General Hospital staff. We also thank the study participants, data collectors, and supervisors for their cooperation in the study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

No specific fund was received for this work.

Notes on contributors

Setognal Birara Aychiluhm

Setognal Birara Aychiluhm holds a Master of Public Health (MPH) in Biostatistics from University of Gondar. Currently, he is an Assistant Professor of Biostatistics in the Department of Public Health, College of Medicine and Health Sciences, Samara University, Ethiopia. He is a Coordinator of the Publication and Editorial Office of the University.

His research interests are maternal, neonatal and child health, biostatistical methods, and communicable and non-communicable diseases.

The current study is part of the team’s efforts to provide essential information on TB-HIV coinfection in pastoral communities.