Coverage of isoniazid preventive therapy among people living with HIV; A retrospective cohort study in Tanzania (2012-2016)

Background: Isoniazid Preventive Therapy (IPT) is a proven public health tool to reduce Tuberculosis (TB) among people living with HIV. However, its implementation in most countries is suboptimal. This retrospective study was conducted to determine the coverage and factors associated with IPT initiation to inform program scale up and improve quality of service in Tanzania and elsewhere. Methods: Retrospective cohort study design involving HIV clinics in Dar es Salaam, Iringa and Njombe regions from January 2012 to December 2016. Results: A total of 171,743 PLHIV were in the cohort from 2012 to 2016. Of these, 166,709 were involved in the analysis. Of those analyzed, 23,970 (14.38%) were ever initiated on IPT. Female sex (aOR = 1.72, 95% CI: 1.13, P < 0.001), obesity (aOR = 1.29, 95% CI:1.20 (cid:1) 1.39, P < 0.001), WHO clinical stage II (aOR = 1.48, 95% CI: 1.42 (cid:1) 1.55, P < 0.001), enrolment in hospitals (aOR = 1.98, 95% CI: 1.89 (cid:1) 2.06, P < 0.001), enrolment in were with increased IPT uptake. Being on ART status associated with decreased IPT initiation. Conclusions: Our study documented low IPT initiation in the study area as well as documented factors which enable IPT initiation and those which impair IPT initiation. Strategies are needed to work on barriers and sustain enabling factors to improve IPT coverage.


Background
Tuberculosis (TB) among people living with HIV (PLHIV) is still a public health concern (WHO, 2018) causing increased morbidity and mortality (Dravid et al., 2019). TB disease among PLHIV threatens the benefits gained through other interventions such as antiretroviral Therapy (ART) (Diwakar et al., 2020) and such interferences may affect achieving Sustainable Development Goals and Joint United Nations Programme on HIV/AIDS goals of ending AIDS epidemics by 2030 (UNAIDS, 2014). Worldwide, in 2018, 9% of individuals diagnosed with TB disease were HIV positive. Fifty-one percent, of those with HIV had TB disease with 84% of these being on ART. There were 1.3 million, among people diagnosed with TB and 300,000 (23%) were HIV positive (WHO, 2018). Tanzania was ranked among 30 countries in the World with high TB burden in 2018. TBHIV co-infection among confirmed TB patients was 28%, and TBHIV mortality rate due to TBHIV co-infection was 29 per 100,000 population which accounted for up to 73% of all deaths among TB patients (UNAIDS, 2019;Mollel et al., 2019).
Isoniazid Preventive Therapy (IPT) which entails use of an anti-TB Drug called Isoniazid (INH) for at least 6 months, treats latent TB infection thus preventing the development of active TB disease in high risk population like PLHIV (Getahun et al., 2010). IPT is a proven public health intervention to reduce TB disease among PLHIV in real world settings. The effect is even more when IPT is used in combination with ART (Yirdaw et al., 2019;Golub et al., 2011;Shayo et al., 2017). Following this, the World Health Organization (WHO) recommends that IPT is part of comprehensive HIV care and treatment package (WHO, 1998).
Despite available evidence on the benefits of IPT on PLHIV, use of this effective preventive tool is suboptimal (Lester et al., 2010;Grant et al., 2005;Ayele et al., 2015;Makoni et al., 2015). Suboptimal implementation of IPT is equally present even in parts of the world hard-hit by both HIV and TB such as Sub-Saharan Africa (Wilkinson and Davies, 1997) where service is needed most. In 2017, IPT among PLHIV Worldwide ranged from 1% in Eswatini to 53% in South Africa (WHO, 2018). Reasons for low IPT coverage range from health provider factors, to client factors, to general health system factors (Lester et al., 2010;Mindachew et al., 2014).
Tanzania has been implementing the use of IPT among PLHIV in HIV care and treatment clinics to prevent development of TB diseases since 2011 (Sabasaba et al., 2019). Here we describe characteristics of PLHIV who were screened TB negative (prerequisite for IPT initiation), determine characteristics of PLHIV who initiated IPT as well as determine factors associated with IPT initiation as IPT has been recently integrated in HIV care and treatment clinics in Tanzania. The findings from this study will inform IPT program scale up in Tanzania and elsewhere.

Study design and setting
We described a secondary data analysis of a retrospective cohort study using the STROBE checklist (Cuschieri, 2019). Anonymized data from all PLHIV enrolled in 315 HIV care and treatment clinics (CTCs) in Dar es Salaam, Iringa and Njombe regions over the period from January 2012 to December 2016. Iringa and Njombe regions have higher HIV prevalence, above the national average (URT, 2018) and Dar es Salaam to has the highest TB case notifications in the country (Said et al., 2017).

Study population
This study used retrospective data collected from PLHIV enrolled in HIV care and treatment clinics in the three regions enrolled in the specified study period. PLHIV who had TB disease diagnosis before CTC enrolment and those who attended clinics in health facilities which never implemented IPT throughout the study period were excluded from the analysis.

Data collection
In Tanzania TB management is integrated in HIV care and treatment services to reduce TB among PLHIV. Information regarding the cohort is detailed elsewhere (Maokola et al., 2020). In summary, three interventions to reduce TB are implemented in HIV care and treatment clinics; these are Intensified TB case finding (ICF), IPT and TB infection prevention and control. During every clinic visit PLHIV are screened for active TB using symptoms/signs as part of ICF. Those who are found with at least one symptom or sign of TB undergo further laboratory and radiological investigations to diagnose TB disease. Those who are found to have no symptoms/signs of active TB are offered IPT for at least 6 months provided other criteria are fulfilled. TB infection and prevention measures are also implemented in all HIV care and treatment clinics to reduce TB infection transmission.
CTCs have patient-level data collection containing data on every visit made by PLHIV. For up to 80% of clinics, data including those related to TB are electronically captured at health facility level into an electronic database called the CTC2 database. Data from the CTC2 database is transferred to the national level database (CTC3) in real time, where demographic and clinical information of each PLHIV identified only by their unique identification CTC number is available for analysis.
Information for PLHIV enrolled in the three study regions were extracted from the CTC3 database. The extracted database had the following variables: The extracted database contained study participant unique identification, information regarding demographic data of the study participants, characteristics of the health facilities, clinical information of the study participants including ART information and TB management. The data were cleaned by checking missing values and abnormal entries. STATA version 14 was used for data analysis. Demographic and clinical variables were used as independent variables and IPT status (received/not received) was the dependent variable. PLHIV with missing data for covariates were excluded from the analyses on a case by case basis, giving a complete case analysis. Missing data were most frequent for clinical characteristics which are not routinely measured in CTC, although these are available at baseline (enrolment into the clinic). As less than 1% of PLHIV were missing these clinical characteristics, the analysis was conducted based on complete case analysis.
Descriptive analysis was used to describe baseline characteristics of study participants and uptake of IPT. Multivariate logistic regression was used to determine factors associated with IPT initiation status involving only exposure variables which were statistically significant during univariate analysis. Multilevel analysis was done to account for cluster effects at the health facility. A statistical significance level of 5% was used.

Baseline characteristics of study participants
A total 171,743 PLHIV were enrolled in CTCs in the 3 regions during the study period from 2012 to 2016. Four (UNAIDS, 2014) percent of study participants were excluded from the analysis as they either had TB before CTC enrolment or they were in clinics which never implemented IPT throughout the study period. Hence, 166,709 were included in the final analysis; of these 23,970 were initiated on IPT (Figure 1). Most of the study participants were females (68.57%), aged 25À49 (71.98%), with walking functional status (95.95%), not on ART (64.79%), with normal nutritional status (92.01%) and normal BMI (56.33%). Study participants were also more likely to be WHO clinical stage I (36.58%), enrolled in 2014 (21.20%), from dispensary level (37.71%), with good adherence to ARV (98.54%), enrolled in Dar es Salaam (72.12%) and from public health facilities (72.82%) ( Table 1).

Discussion
The study registered low uptake of IPT among PLHIV. Female sex, ART, obesity, WHO clinical stage, enrolment in Njombe region, and being in public hospitals were the determinants of IPT initiation.
Only about 80% of the HIV clinics in the country had an electronic HIV database and hence were capable of contributing data to the national database. Health facilities at low levels did not have electronic databases and were unable to contribute data. Secondly, the analysis included 3 regions out of 26 regions. These regions were purposively chosen as they had the highest HIV and TB prevalence. They may therefore not represent a true picture of other regions which were not involved in the analysis.
Our study registered lower IPT initiation compared to other studies which were also conducted in routine settings in Zimbabwe, Ethiopia and Nepal (Makoni et al., 2015;Nyathi et al., 2019;Yirdaw et al., 2019;Teklay et al., 2016;Wesen and Mitike, 2012;Dhungana et al., 2019). These studies documented IPT initiation ranging from 20% to 54%. The observed low IPT initiation in the current study needs further research. Nevertheless, most of the studies, despite being conducted in routine settings like our study, involved small numbers of health facilities. None of the above studies involved more than 20 high level health facilities. Our study involved a total of 315 in the 3 regions containing a mix of high and low level health facilities. Shortage of INH, fear of INH resistance, lack of confidence in ruling out TB disease, lack of commitment among providers and hesitancy in accepting IPT among clients were among the barriers for IPT implementation documented in several other studies (Nyathi et al., 2019;Teklay et al., 2016;Wesen Denegetu and Lovely Dolamo, 2014). Improvement in recording and reporting in order to identify PLHIV who are eligible for IPT, health education to curb fear of Table 3 Factors associated with IPT initiation (N = 166,709).

Covariate
Univariate analysis Multivariate analysis cOR, 95% CI, P-value aOR, 95% CI, P-value adherence and side effects among PLHIV, interventions to reduce staff workload and avoiding INH stockout are the interventions recommended to improve IPT coverage (Teklay et al., 2016;Ogunsola et al., 2019;Legese et al., 2020) which could be applied in our setting as well. Consistent with our study, other studies also documented a high IPT initiation among females, ART recipients, PLHIV enrolled in public hospitals and PLHIV with relatively good health (WHO clinical stage II) (Nyathi et al., 2019;Trinh et al., 2016;Van Ginderdeuren et al., 2019;Ayele et al., 2017). High uptake among females could be due to more health information for women during medical encounters such as maternal and child health services (Wesen Denegetu and Lovely Dolamo, 2014). ART experience and recipients' good health may increase confidence for IPT initiation among health care providers.
In our study determinants of IPT initiation were female sex, ART, obesity and WHO clinical stage II. Other IPT initiation determinants were enrolment in the Njombe region and in public hospitals. As stated above, females have a higher threshold embracing health services due to their health-seeking behavior (Thompson et al., 2016) as well as their increased opportunity to encounter health care providers for reproductive health (Wesen Denegetu and Lovely Dolamo, 2014). Other studies also found an association between female sex and IPT uptake (Adeniyi et al., 2019). We also found IPT initiation to be associated with health individuals (WHO clinical stage II). Our finding is in agreement with what was found in Kenya (Nyathi et al., 2019). The reason for avoiding IPT in severely ill PLHIV can be attributed to fear of clinicians due to difficulty in ruling out TB disease in this population as speculated in the Kenyan study (Nyathi et al., 2019). High IPT initiation in hospitals and public health facilities could be explained by the fact that IPT was rolled out gradually starting with Hospitals. Higher initiation in public versus in private health facilities could be due to a slower pace of private health facilities in implementing new policies in comparison with public owned health facilities (Buregyeya et al., 2017). Higher IPT initiation in Iringa and Dar es Salaam in comparison with Njombe explains dynamics in IPT implementation, which also differs from health facility to health facility, as reported in another study showing different implementation levels in different places (Dhungana et al., 2019;Omondi et al., 2018;Ngugi et al., 2020).

Conclusion
Our study, which used routine HIV data from 315 clinics, documented low IPT initiation proportion among those who were enrolled in HIV care and eligible for IPT in Dar es Salaam, Iringa and Njombe during the study period. Variations in IPT initiation among regions signal different dynamics affecting IPT uptake in different regions and hence require customized approaches in quality improvement. Implementation research is needed to understand health systems as well as cultural barriers in the uptake of IPT intervention and design quality improvement initiative accordingly.

Declarations
This article is an original research which is one of the required number of published articles for PhD training of the first author. We declare that this article has not been submitted to any other journal for publication.

Ethics approval and consent to participate
The study used routine HIV data from Tanzania. Hence, the study did not have contact with human subjects. Approval for conducting the study was obtained from Kilimanjaro Christian Medical University College in Moshi Tanzania. Permission to use routine HIV data was obtained from Ministry of Health, Community Development, Gender, Elderly and Children (MoHCDGEC) Tanzania by signing Data Transfer Agreement (DTA).

Consent for publication
Authors of the study collaboratively give permission to the publisher to publish the study findings as presented by the authors.

Availability of data and materials
Data and other materials used for the study belong to the MoHCDGEC Tanzania. Permission to access data and other materials used for the study can be obtained from the Permanent Secretary MoHCDGE in Tanzania.