Risk factors for active tuberculosis among people living with HIV on antiretroviral therapy in Kinshasa, Democratic Republic of Congo: a case-control study

¹Regulatory and Oversight Authority for Universal Health Coverage, Kinshasa, Democratic Republic of Congo, ²Department of Medical Biology, Faculty of Health Sciences, National Pedagogical University, Kinshasa, Democratic Republic of Congo, ³Department of Internal Medicine, University of Kinshasa, Kinshasa, Democratic Republic of Congo, ⁴Department of Nutrition, Kinshasa School of Public Health, University of Kinshasa, Kinshasa, Democratic Republic of Congo, ⁵Department of Epidemiology and Biostatics, Kinshasa School of Public Health, University of Kinshasa, Kinshasa, Democratic Republic of Congo

^&Corresponding author
Guylain Katayi Ngoy, Regulatory and Oversight Authority for Universal Health Coverage, Kinshasa, Democratic Republic of Congo

Introduction    

Despite advances in medicine, the combined burden of tuberculosis (TB) and human immunodeficiency virus (HIV) remains a leading global health concern [1-3]. People living with HIV (PLHIV) are much more likely to develop TB than those with competent immune systems [4-9]. Their annual risk of being affected by active TB varies between 5 and 10%, and over their entire life, this risk can even rise up to 51% [10]. The coexistence of TB and HIV has exacerbated the global TB epidemic, posing major challenges to care and treatment programs across the globe [3]. HIV increases the patients’ vulnerability to primary infection, reinfection, and reactivation of TB, particularly in those with latent TB [1, 5, 6, 9,11]. In turn, Mycobacterium tuberculosis infection further weakens the immune response to HIV, accelerating the progression of the infection to the acquired human immunodeficiency syndrome (AIDS) stage, thereby increasing mortality in co-infected individuals [2, 4]. In 2022, approximately 7.5 million cases of TB were reported worldwide, marking the highest morbidity rate recorded by the World Health Organization (WHO) in nearly three decades of tracking the disease [12,13]. In 2023, 331,000 patients were diagnosed with HIV/TB co-infection while already taking antiretroviral treatment (ART). During the same period, anti-tuberculosis prophylaxis was prescribed to 1.5 million people who were considered at risk [14]. 

The Democratic Republic of Congo (DRC) is among the 30 countries most heavily affected in the world by HIV/TB co-infection, ranking 5th in Africa [15]. In 2018, the country recorded nearly 171,682 cases of TB, of which 31,000 (11.7%) concerned PLHIV. That same year, TB caused 10,000 deaths among this vulnerable population [15]. Scientific studies have made it possible to identify certain sociodemographic and clinical factors associated with the development of TB in HIV-positive individuals, thus contributing to a better understanding of the mechanisms underlying the occurrence of this serious comorbidity [16,17]. The repercussions of this devastating couple are innumerable, and they extend their devastation to many areas such as the health status of the victims, the economy at different levels (personal, family, community and national), the society and the culture [5,18]. Many households, already faced with limited resources, must assume additional costs for care not covered by the various programs and partners involved in the fight against TB and HIV [13,19,20]. This situation constitutes a major obstacle to the achievement of the Sustainable Development Goals, in particular objective 3.3, which aims at ending the epidemics of TB, AIDS, malaria and neglected tropical diseases by 2030 [6]. The WHO recommends systematically screening PLHIV for TB at each medical visit and prescribing Tuberculosis Preventive Treatment (TPT) to all new patients whose TB screening test is negative [3,7,21-23]. Numerous studies carried out around the world have demonstrated the effectiveness of TPT. Thanks to this prophylaxis, the incidence of TB can be reduced by up to 40%, while its mortality rate can be reduced by approximately 50% [4,5,24,25].

Despite WHO guidelines, countries with high TB incidence often experience serious difficulties in implementing these measures on the ground. Stock-outs of TPT and provider concerns about side effects are among the major barriers to effective implementation of this prophylaxis [25]. In addition to systematic TB screening and TPT prescription, the DRC has established a policy called “One Stop Shop” [26]. This strategy aims at adopting an integrated approach to simultaneously manage HIV and TB within the same health care facility [27]. Notwithstanding the various strategies put in place, the DRC continues to record new cases of TB among patients on antiretroviral drugs (ARVs) each year. In 2019, the country reported an estimated TB incidence rate among PLHIV of 34 per 100,000 population [28]. According to the National HIV/AIDS Control Program, the detection rate of TB/HIV co-infection increased from 27% in 2019 to 32% in 2021 [29,30]. Kinshasa remains the DRC province with the highest HIV and tuberculosis prevalence rates [14,28]. This study aimed to identify the factors associated with active TB among PLHIV on ARVs at Luyindu Hospital Center (HC), in the Binza Ozone Health Zone (HZ) of the city of Kinshasa, from January 1, 2021 to December 31, 2022.

Methods     

Study site and period: the study was conducted at the Luyindu HC, located in the Binza Ozone HZ, Kinshasa, from January 1^st, 2021, to December 31^st, 2022. Luyindu HC, supported by Médecins Sans Frontières (MSF), served as a referral hospital for six other healthcare facilities also supported by MSF during this period.

Justification for the selection of Luyindu Hospital: Luyindu Hospital was chosen because of its technical capabilities, allowing it to manage patients, including those at an advanced stage of HIV infection. During this study, it was observed that, in addition to the patients routinely followed at Luyindu, a significant number came from other MSF partner facilities due to the severity of their condition. These patients were referred to Luyindu for hospitalization or specific exams. They were sent to Luyindu rather than to other MSF-supported referral hospitals, depending on their place of residence. Once the care for which they were referred had been completed, they were referred back to their original healthcare centers to continue their regular follow-up.

Role of médecins sans frontières in patient care: médecins sans frontières support for Luyindu Hospital included the renovation of the emergency room and hospital wards, provision of an X-ray machine, laboratory supplies, and pharmaceuticals (excluding ARVs). Médecins sans frontières also provided training for healthcare providers involved in the care and follow-up of PLHIV, covered operational costs at the hospital, and paid a monthly allowance to all healthcare providers responsible for the care of PLHIV.

Type of study: from August to October 2023, a matched case-control study, based on data from medical records and interviews of PLHIV on ARVs, was conducted at the Luyindu HC.

Study population: the study population consisted of PLHIV aged 15 years or older who did not have TB at ART initiation. A total of 1,380 PLHIV were followed up at Luyindu HC between 2021 and 2022, mostly from seven partner care facilities in the Binza Ozone, Lingwala, Bumbu, and Kintambo HZs.

Participant inclusion and exclusion: from the registers of PLHIV, patients aged 15 years and over without TB at the start of ART were identified and listed. Cases included were those of the patients who consented to participate in the study and who subsequently developed TB after initiation of ART during the study period. Controls were patients who were free of TB during the same period and were matched to cases by sex and age (± 3 years). They also consented to participate in the study. Exclusions for cases and controls were based on criteria such as lack of informed consent, lack of documentation of TB status at baseline, or loss to follow-up of the patient.

Sample size: the sample size was determined using the Open Epi software version 3.01 with the following parameters: a two-sided significance level of 5%, a power of 80%, a case-to-control ratio of 1:1, a proportion of controls with a CD4 count < 200 cells/mm³ of 41.9% and a minimum detectable odds ratio of 2.02 as found in a study conducted by Nugus and colleagues in 2020 on the determinants of active TB among adult PLHIV on ARVs in Ethiopia [5]. For the sample size calculation, a CD4 count < 200 cells/mm³ was used as the exposure status for the occurrence of active TB among PLHIV after ART initiation, as reported in several articles on risk factors for active TB [3, 9, 31]. The Fleiss method with continuity correction was used to calculate the sample size. To increase the power of this study, the sample size calculated at 280 was raised to 308, i.e., 154 cases and 154 controls.

Sampling technique: a record was drawn up of all PLHIV on ARVs received from January 1, 2021 to December 31, 2022. From this record, we identified two lists: the first list of patients who developed TB and the second list of those who were free of TB. Then, separately, the cases and controls were selected using a simple random sampling technique. Under this procedure, we used numbers generated by the RAND function within limits in Excel until reaching the size of 154 cases and 154 controls. However, if some of these cases or controls were found not to meet the inclusion criteria, they were excluded and replaced by other subjects until the required maximum number of 154 subjects was reached for both lists.

Operational definitions of study variables

HIV/TB co-infection status: new case of TB after initiation of ART and confirmed by positive acid-alcohol-resistant bacilli microscopy, positive Xpert Mycobacterium/Rifampicin test, culture, or based on the clinical decision rendered by an expert clinician.

History of tuberculosis: patient who suffered from tuberculosis before the diagnosis of HIV infection and was declared TB-free at the time of HIV diagnosis. This includes patients who had previously been treated and cured of TB prior to their enrollment for ART.

CD4: the CD4 count, measured at admission, is categorized into two groups: i) ≥ 200 cells/mm³, and ii) > 200 cells/mm³. Viral load (VL): HIV viral load, measured at admission, is categorized into two groups: i)< 1,000 copies/ml, and ii) ≥ 1,000 copies/ml.

HIV-TB delay: time interval in years between the date of diagnosis of HIV infection and the date of confirmation of diagnosis of active TB.

Body mass index (BMI) at hospital admission: ratio of weight in kilograms to the square of height in meters (Kg/m²). The variable is summarized in two groups: 1. BMI < 18.5 Kg/m² underweight (undernutrition), 2≥ 18.5 Kg/m² [32]. Body mass index was calculated based on weight and height data recorded at hospital admission during the study period and retrieved from medical records.

Index of economic well-being: composite score of household’s standard of living, obtained by assigning weights to possessions, and generated from a principal components analysis. Classified in quintiles: i) lowest; ii) second quintile; iii) middle quintile; iv) fourth quintile and v) highest [33].

Overcrowding index: ratio of the number of adults (15 years and over) in a house compared to the number of rooms. It includes 3 categories: < 1 (normal crowding), 1-2 (moderate crowding) and > 2 (severe crowding) [32].

Antiretroviral treatment adherence: level of adherence to ART assessed using the 8-question Morisky scale, summarized into 3 categories: Good adherence (≥8), Average adherence (6 to 7) and Poor adherence (< 6) [34].

Lost to follow-up: patient absent for at least 90 days for no obvious reason.

Tuberculosis preventive treatment: regular and complete intake of isoniazide (INH) or the INH-rifapentine combination before the study period.

Data gathering: the monitoring registers and medical records of PLHIV on ARVs served as sources for collecting information. Data were collected using Android phones equipped with the KoboCollect app and a structured interviewer-administered questionnaire. The selected PLHIV were contacted and interviewed by peer educators. A unique code was assigned to each participant by the principal investigator to ensure the confidentiality of the patients' identities. Subsequently, investigators reviewed the medical records to complete the collected information.

Statistical data analysis and processing: data from registers, medical records, and interviews of PLHIV were exported from the KoboCollect application to MS Excel, and then imported into SPSS 25 software for analysis. Categorical variables are presented by absolute and relative frequency. For quantitative variables, the normality of their distribution was tested using the Kolmogorov-Smirnov test at a 95% confidence level. The mean and standard deviation were used when the distribution was normal, while the median and interquartile range were used for skewed distributions. The Chi-square test was used to compare proportions. Multivariate logistic regression with the backward conditional method was then performed by including all variables with a p-value ≤ 0.020 from the bivariate analysis. This regression was used to identify factors independently associated with HIV/TB co-infection, calculating odds ratios (OR) and their 95% confidence intervals (CI). Model fit was assessed using the Hosmer-Lemeshow test and the goodness-of-fit test. Collinearity was evaluated using a correlation matrix and variance inflation factor (VIF) values on SPSS, with all VIF values being less than 2, indicating negligible collinearity among the explanatory variables.

Ethical considerations ethical authorization was obtained from the Ethics Committee of the Kinshasa School of Public Health, University of Kinshasa, under Approval number: ESP/CE/O54/2024. Authorization was also obtained from the head of the Kinshasa Provincial Health Division and the administrative body of Luyindu HC to access patient medical records. Written informed consent was obtained from all study participants. The confidentiality of the participants was ensured by not disclosing any private information that could lead to the precise identification of the patients.

Results     

During the two years of the study, Luyindu HC treated a total of 1,380 PLHIV, including its own cohort and patients from other MSF partner establishments. Of this total, 407 had developed active TB, a proportion of 29.5%. Three hundred and eight patients from this group were recruited for the study, half of whom presented with TB/HIV co-infection (Figure 1).

Sociodemographic characteristics of cases and controls: the majority of participants were aged between 15 and 50, with a slight female predominance (sex ratio of 1.4 female/male). The median age was 46 (Interquartile range [IQR]: 36 - 56), with a minimum age of 15 and a maximum age of 74. In terms of education, most participants had attended school. Concerning marital status, most participants lived as a couple. The majority reported being affiliated with the Christian religion. More than half of the participants lived in conditions of moderate overcrowding. No significant differences were observed between cases and controls in terms of socioeconomic status (Table 1). Over 70% of participants used charcoal as their main cooking fuel. More than 80% of participants lived in houses with finished floor materials (cement or tiles). The majority of participants had access to drinking water through a tap in their home. Over 80% of respondents had roofs made of manufactured materials (sheet metal, ceiling tiles, or cement) and walls made of cement blocks (Table 1).

Clinical and biological characteristics of cases and controls: tuberculosis was primarily diagnosed using the Lipoarabinomannan Tuberculosis test (TB-LAM) (84.4%). The majority of TB cases (94.8%) occurred in patients after more than 12 months of ART. Tuberculosis cases occurring before the 3^rd month of ART were rare (0.6%), increasing slightly to 1.3% between the 3^rd and 6^thmonths, and then to 3.2% between the 6^th and 12^th months after the start of treatment. More than half the cases, compared with controls, had a history of TB. A greater number of cases had a lower-than-normal BMI compared to controls. The majority of participants in both groups had not received TPT. Cases were less adherent to ART than controls. Most participants remained on first-line ART. A lower percentage of cases compared to controls did not receive cotrimoxazole prophylaxis (26.6% vs. 14.9%, respectively). The difference between cases and controls was statistically significant (p = 0.008) (Table 2). The majority of cases had hemoglobin levels < 10 g/dl, unlike the controls. A higher percentage of cases than controls had blood glucose levels > 126 mg/dl. Viral load was detectable (≥ 1,000 copies/ml) in more than half of the cases, compared to less than a fifth of the controls. A higher proportion of co-infected cases had a CD4 count ≤ 200 cells/mm³ compared to controls (Table 2). There were no significant differences between participants with hemoglobin levels < 10 g/dl and those with levels ≥ 10 g/dl. The majority of participants had blood glucose levels < 126 mg/dl. Viral load was detectable (≥ 1,000 copies/ml) in more than half of the cases, compared to less than a fifth of the controls. A higher proportion of co-infected cases had a CD4 count ≤ 200 cells/mm³ compared to controls (Table 2).

Profile of participant comorbidities and behaviors: participants had various comorbidities, including high blood pressure (18.5%), bronchial asthma (7.1%), diabetes mellitus (2.9%), kidney disease (1.9%), and heart disease (1.3%). Less than 10% of participants used substances such as tobacco or cannabis. Approximately 50% of both cases and controls consumed alcohol. More than 90% of participants had never been incarcerated (Table 1).

Bivariate and multivariate analysis: in the bivariate analysis, variables with a p-value ≤ 0.20 were considered for inclusion in the multivariate model to control for potential confounding factors. These variables included: economic well-being index "fourth quintile, history of tuberculosis, absence of cotrimoxazole prophylaxis, absence of TPT, poor adherence to ART, low BMI, low hemoglobin level, low CD4 count, and high viral load". After adjusting for these potential confounders in the logistic regression model, several variables remained significantly associated with the occurrence of active TB among PLHIV on ARVs. Detailed results, including crude odds ratios (cOR), adjusted odds ratios (aOR), and confidence intervals, are presented in Table 3. In the logistic regression model, a history of TB was associated with an almost six-fold increased risk of developing active TB. Poor ART adherence quadrupled the risk of TB. Both the absence of TPT and the absence of cotrimoxazole prophylaxis doubled the risk of TB. A low BMI and high viral load also doubled the risk of TB. Additionally, a low CD4 count quadrupled the risk (Table 3).

Discussion     

This study aimed at identifying the factors associated with active TB among PLHIV on ARVs at Luyindu HC, in the Binza Ozone Health Zone. Tuberculosis incidence was low (0.6%) before 3 months of ART, but increased to 94.8% after 12 months of ART. This variation from a previous study, which reported a higher initial incidence (19%) and an incidence of 60.4% after 12 months [4], could be explained by more rigorous screening in our study, reducing the risk of immune reconstitution syndrome and reactivation of latent TB. It is crucial to strengthen screening efforts and ensure regular follow-up. Patients with a CD4+ count ≤ 200 cells/mm³ have an increased risk of developing active TB, as observed in other studies [3, 5,31,35,36]. Similarly, an HIV viral load ≥ 1000 copies/ml increases this risk, in line with observations from previous studies [20]. High viremia aggravates immunodeficiency, which in turn increases susceptibility to TB. Severe immunodepression not only favors the reactivation of latent TB but also reduces the body's ability to control infection. Regular assessment of immune status, ART optimization, and structured follow-ups are crucial to mitigate TB risk. Early detection and immediate ART initiation remain key strategies, reinforcing the need to strengthen the 'Test and Treat' approach for prompt patient management [37,38]. In addition to optimizing treatment, healthcare providers should ensure adequate nutritional support and organize regular clinical visits to enhance patient outcomes.

A BMI < 18.5 Kg/m² increases the risk of TB compared with a BMI ≥ 18.5 Kg/m² in agreement with other observations [17,18]. Malnutrition affects cell-mediated immunity, the main defender against TB, making patients more vulnerable. It is essential to strengthen the integration of nutrition into the care of PLHIV to improve clinical outcomes and quality of life. A significant association was found between a history of TB and an increased risk of recurrence, in agreement with studies conducted in Ethiopia [3,5,39]. Reactivation of latent infection is often favored by a drop in immunity due to HIV [3,5,39]. It is imperative to exclude active TB before starting TPT to avoid treatment failures and resistance associated with prescribing INH monotherapy. Rapid initiation of ART as soon as HIV is diagnosed should be a top priority for care providers. Failure to take TPT is linked to the onset of TB among PLHIV, corroborating previous studies [16,17,40]. The combination of ART and TPT significantly reduces the risk of TB [23]. Adequate funding of TB/VIH programs, reinforcement of the “one stop shop” strategy, and training of healthcare professionals are essential. Routine TB screening and education on the benefits of TPT need to be reinforced. Community involvement is crucial to reduce stigma, which can inhibit access to care and affect adherence.

In our study, socioeconomic status was not significantly associated with TB occurrence in PLHIV, unlike other studies that regularly identify low socioeconomic status as an important risk factor [4,6]. This lack of association may be due to the use of the Economic Welfare Index to assess socioeconomic status and less pronounced socioeconomic disparities among participants in our sample. Additionally, although malnutrition is a possible confounder in other studies [41]. it was not identified as a confounder in our analysis. This observation suggests that malnutrition does not explain the lack of a significant association. To better understand these discrepancies, it would be useful to review the methodology for assessing socioeconomic status using methods more aligned with those of previous studies. The absence of a significant association between overcrowding index and active TB in our study is consistent with the findings of Mohamed [32]. Although overcrowding can theoretically increase the risk of transmission [42], other factors such as access to care, adherence to treatments, and hygiene are more decisive. It is pertinent to focus on these key factors for prevention and treatment, while continuing to monitor housing conditions.

Our results show that the absence of cotrimoxazole prophylaxis is associated with an increased risk of developing tuberculosis among PLHIV receiving antiretroviral therapy. This observation aligns with several studies, particularly those conducted in the Swiss HIV Cohort Study and the TREAT Asia HIV Observational Database, which reported a significant reduction in TB risk among patients receiving cotrimoxazole [43,44]. The protective effect of cotrimoxazole against tuberculosis may be explained by two main mechanisms: a direct antimycobacterial effect, suggested by in vitro studies showing sulfamethoxazole activity against Mycobacterium tuberculosis, and an indirect effect through the prevention of other opportunistic infections. By reducing the occurrence of bacterial, parasitic, and fungal infections, cotrimoxazole helps limit immune deterioration in PLHIV. Indeed, recurrent opportunistic infections worsen systemic inflammation and immune stress, creating a favorable environment for the reactivation of latent TB or the acquisition of a new infection.

Study limitations: as a matched case-control study, it is subject to potential biases such as information, recall, and selection, which may restrict the generalizability of the results to the general population. Additionally, data on TPT uptake were obtained from patient self-reports due to the unavailability of this information in medical records. This introduces a potential bias regarding the reliability of the data, which may impact the accuracy of the TPT uptake information. One of the main limitations is the temporal ambiguity of certain factors associated with tuberculosis, such as low BMI, anemia, or hyperglycemia, which may be consequences rather than causes of the disease.

Study strengths: the rigorous study design allows effective control of bias and provides valuable information on risk factors for TB among PLHIV on ART. The scarcity of similar studies in the DRC reinforces the value of our work, underlining the importance of an integrated approach to influencing health policies and programs to combat HIV/TB co-infection in the DRC.

Conclusion     

To reduce the prevalence of active TB among PLHIV on ART in the DRC, optimizing treatment adherence, promoting TPT, and providing nutritional support must be key priorities. Expanded coverage of TPT and rigorous clinical follow-up are crucial to strengthening the effectiveness of care. Longitudinal research is needed to evaluate specific interventions, and interdisciplinary collaboration between researchers, policy-makers and healthcare providers will help refine public health policies.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Guylain Katayi Ngoy: conceived and developed the design of this study, with the approval of the Director. Aimée Mampasi Lulebo: the data collection tool, developed by Guylain Katayi Ngoy on the basis of the literature review and the objectives of the study, was validated by the Director. The manuscript was successively reviewed by Aimée Mampasi Lulebo, Nadine Ngongo Mayasi, Ngoyi Kashiba Zacharie Bukonda, Benito Maykondo Kazenza and Jean Claude Maleshila Mikobi. Statistical analysis of the data was carried out by Guylain Katayi Ngoy in collaboration with Jean Claude Maleshila Mikobi. All the authors have read and agreed to the final manuscript.

Acknowledgments     

We would like to thank all the faculty and the ethics committee of the Kinshasa School of Public Health for their ethical authorization, which enabled us to conduct this study in accordance with established standards. We would also like to express our gratitude to the authorities of the Kinshasa Provincial Division for Health and the Binza Ozone HZ, as well as to the authorities and staff of the Luyindu Hospital and the investigators, whose assistance was essential for data collection.

Tables and figures     

Table 1: sociodemographic characteristics of participants living with HIV, recruited from the Luyindu Hospital Center (Kinshasa, Democratic Republic of Congo), from January 2021 to December 2022 (N=308)

Table 2: clinical and behavioral characteristics of participants living with HIV, recruited from the Luyindu Hospital Center (Binza Ozone Health, Kinshasa, Democratic Republic of Congo), from January 2021 to December 2022 (N=308)

Table 3: factors associated with active TB among participants living with HIV, on antiretroviral drugs at Luyindu Hospital Center (Binza Ozone Health zone, Kinshasa, Democratic Republic of Congo) from January 2021 to December 2022 (N=308)

Figure 1: flow diagram of participants living with HIV, recruited from the Luyindu Hospital Center (Binza Ozone Health zone, Kinshasa, Democratic Republic of Congo), from January 2021 to December 2022

References