Incidence and Predictors of Severe Adverse Drug Reactions among Patients on Antiretroviral Drugs in Harari Regional State, Eastern Ethiopia

Background The introduction of combination antiretroviral therapy improves the quality and longevity of people living with HIV/AIDS. However, adverse drug reactions associated with antiretroviral therapy compromise the resulting benefits and have been reported differently worldwide, including Ethiopia. Severe adverse drug reactions are one of the major public health concerns for the reason that they can potentially impede the benefit of antiretroviral therapy and put the patient's survival at risk. Despite many successes achieved with the introduction of the combined antiretroviral therapy, the majority of the patients on antiretroviral therapy experience adverse drug reactions associated with the drugs. Consequently, little is known about the problem in the current study area. This is, therefore, to study incidence and predictors of severe adverse drug reactions among patients on antiretroviral drugs in the Harari region, Eastern Ethiopia. The aim of this study was to assess the incidence and predictors of severe adverse drug reactions among patients on antiretroviral therapy from February 25, 2022, to March 25, 2022, in the Harari region, Eastern Ethiopia. Methods A hospital-based retrospective cohort study was conducted among 449 randomly selected medical records of people living with HIV on first-line antiretroviral therapy. Collected data were entered into EpiData version 3.1 and exported to STATA version 15 for analysis. Kaplan–Meier survival curve with log-rank test was used to compare survival curves for categorical independent variables. A p value ≤0.05 was declared as significant, and an adjusted hazard ratio was used to report the effect size using the multivariate Cox proportional hazard model. Result The overall incidence density of the severe adverse reactions was 7.22 per 1000 months (95% CI: 5.5, 9.6). After adjusting for all potential confounders using multivariable Cox proportional hazard ratio, advanced clinical diseases (AHR = 3.44; 95% CI: 1.54, 7.65), HIV/tuberculosis confections (AHR = 2.38; 95% CI: 1.23, 4.62), and being female (AHR = 3.12; 95% CI: 1.57, 6.18) were significantly associated with the experience of severe adverse drug reactions. Conclusion In this study, the incidence of severe adverse reactions was consistent with the previous studies, and advanced World Health Organization (WHO) clinical stage, HIV/TB confection, and being female were the independent predictors of the severe adverse drug reactions.


Introduction
Human immunodefciency virus (HIV) infected more than 79 million people and killed more than 36 million lives since the start of its epidemics.It remains one of the most serious global health threats of our time [1,2].Currently, 37.7 million people are living with HIV (PLWH) globally.In Ethiopia, 620,000 people were living with HIV of which 78% were on ART at the end of 2020 with an estimated prevalence of 0.9% nationally [3].
Te introduction of antiretroviral therapy signifcantly modifed the natural history of HIV infection and changed it from an end-of-live event to a manageable chronic disease [4].However, alongside of all its advantages, challenges related to its associated adverse drug reactions (ADRs) are threatening the progress still, which may range from mild to serious life-threatening ADRs with both short-and longterm efects.Although the most recently approved drugs generally show an improved safety profle than frstgeneration antiretroviral, drug-related ADRs continue to occur.ADRs are one of the major constraints of the combined antiretroviral therapy (cART) among patients receiving cART, and it is a very common complication of the cART therapy and a major reason for defaulting during the treatment [5].
An adverse drug reaction is a harmful event or unpleasant reaction, attributed to an intervention related to the use of a medicinal product, which predicts hazards from future administration and warrants prevention or specifc treatment, or alteration of the dosage regimen, or withdrawal of the product [6].Te ADRs in resourcelimited settings may diverge from those developed countries because of diferent reasons such as the high prevalence of TB, reliance on traditional medicines, and malnutrition and poverty, and also patients come with advanced stages of the disease [7][8][9][10].Te adverse events/ toxicity associated with cART drugs are more likely to lead to the drug switch among the patients on the treatment, negative impact on the public health system, and poor adherence to the treatment which may result in negative clinical outcomes [11,12].
Adverse drug reaction surveillance (drug safety surveillance) was established in Ethiopia in 2002 under the Food, Medicine and Health Care Administration and Control Authority currently named as Food and Drug Authority when the country became a member of the WHO program for international drug monitoring.Te federal regulatory authority is responsible and empowered with monitoring and ensuring the postmarket drug safety, enabling all issues with ADR.Despite this, the organization collected and documented only a few reports from the health facilities due to diferent reasons.As the case report analysis conducted showed there is a signifcant decrease in ADR reporting from 2016 to 2018 which does not show the absence of ADR, majority of the healthcare professionals were not aware of the existence of the spontaneous ADR reporting system which shows a low response and lack of knowledge at ground level [13][14][15][16][17].
To overcome the adverse drug reactions associated with antiretroviral therapy (ART), diferent intervention was undertaken of which Stavudine (d4T) which was one of the frst-line ART drugs was recommended by WHO to phase out from the frst-line regimens due to its well-known longterm mitochondrial toxicity [18,19].Ethiopia also amended its guidelines to initiate newly diagnosed patients on non-d4T regimens starting from 2012 and other patients gradually as recommended by WHO [20].Several previous studies on adverse reactions of ART were focused on the incidence and predictors of all forms of adverse reactions, and few assessed the severe forms of the adverse reactions of ART drugs in Ethiopia, particularly in the current study area.Tus, the current study aimed to assess the incidence and its predictors of severe adverse reactions among patients with advanced WHO clinical stage as exposed and those with nonadvanced WHO clinical stage as unexposed at baseline in Harari, Eastern Ethiopia, from January 1, 2018, to December 2020.

Study Design, Setting
, and Period.Tis study was conducted in public hospitals of Harari Regional National State, Eastern Ethiopia.Harari region is located in the Eastern part of Ethiopia at a distance of 526 km away from the capital city Addis Ababa/Finfnnee.Te region has a total population of 276,424.Te availability of health facility coverage in the region is more than 100%, having 4 hospitals (1 private, 2 public, and 1 governmental hospital), 8 health centers, and 2 not-for-proft nongovernmental organization (NGO) health facilities (HFs) (one fstula center and sexual reproductive health (SRH) center/clinic).Currently, there are a total of 9 ART sites (4 hospitals, 4 urban health centers, and 1NGO clinic) with a total of 4,224 HIV patients on ART in the region.Randomly, selected public health hospitals were the study setup [21].Te data extraction was conducted from February 25 to March 25, 2022.

Study Population and Eligibility
Criteria.Te study populations were PLWH patients attending ART clinics in Harari region hospitals.Te study cohort assigned the study participants into the exposed group (those with advanced WHO clinical stage (III & IV) at baseline) and the unexposed group (those with unadvanced WHO clinical stage (I & II) at baseline).Te eligibility criteria were all adult (age 15 and above) HIV patients who were on ART and enrolled from January 2018 to December 31, 2020, at the study setups and on follow-up with at least one follow-up/repeated visit and those with complete information about severe ADRs.

Sampling Technique, Procedure, and Sample Simple
Calculation.Te sample size of the study was determined for survival data, Cox proportional regression model by using Epi Info 7.0 by taking the percentage of exposed group and unexposed group who experienced severe adverse drug reactions 19% and 9%, respectively, from previous study conducted in Debre Markos referral hospital [22], with the consideration of the following assumption margin of error 5% and 95% confdence interval, power 80%, and by adding 10% of the loss to follow-up which gives 150 of exposed group and 299 of unexposed group and total sample size of 449 with a ratio of exposed to unexposed group 1 : 2. Te study settings were selected randomly, and lists of all eligible participants were obtained from ART clinics in the selected facilities and were used as the sampling frame.Te patient's ART unique identifcation number (ID) was used to select patients' medical record using computer-generated simple random sampling techniques.

Data Collection and Quality Control.
A structured data extraction form was developed from ART follow-up charts and relevant literature and used [22][23][24][25].Tree clinical nurses were deployed for data collection.One BSC nurse who received a training on ADR in ART was assigned to supervise and coordinate the data collection processes.A two-day training on the objective of the study, data collection tool, and extraction technique was given by the authors.Te supervisor followed the day-to-day data collection activities to make sure that the collected data are complete, consistent, and accurate.

Data Processing and Statistical
Analysis.Data were entered using EpiData version 3.1 and exported to STATA for cleaning and analysis.All randomly selected study participants were followed for a maximum of 36 months (from January 2018 to December 31, 2020), and a total of 449 study participants were included in the analysis.Patients who lost to follow-up and transferred out after at least repeated follow-up and died before experiencing the events were considered as censored during analysis.All patients were followed until the occurrence of an event of interest (severe adverse drug reaction) or censored.Descriptive statistics such as frequency, percentage, mean, and standard deviation (SD) were calculated.Kaplan-Meier (KM) survival curve with log-rank test was used to compare survival curves for categorical independent variables.Te median and mean survival time were estimated using the KM method.To predict the association between independent and outcome/ dependent variables, the Cox proportional hazard regression model was used.95% CI of the hazard ratio was calculated, and the variable with p value ≤ 0.05 in the Cox proportional hazards model was considered and declared as signifcant.
For the evaluation of the proportional hazard assumption, log-log survival curves and Schoenfeld's goodness-of-test approach were used.A life   1).

Behavioral Characteristics.
Of a total of 449 (150 exposed and 299 unexposed), 67.26% disclosed their serostatus to family members or sexual partners and 30.8% were not disclosed yet.

Clinical Characteristics.
From a total of 449 (150 exposed and 299) HIV-positive patients on ART follow-up, 80.62% had a CD4 count of more than or equal to two hundred cells/mm 3 and more than one-third of the patients had normal BMI during ART initiation.Most of the HIVpositive patients at ART initiation time were WHO clinical stage I (39.2%), around 15% of the patients experienced HIV/TB infection at baseline, one-third of the patients were on 1e (TDF-tenofovir + 3TC-lamivudine + EFV-efavirenz) regiment at baseline, and more than half of the patients experienced regimen change.More than three-ffth of 61% of the participants were on working functional status during the ART initiation, and 80% of the patients had good Canadian Journal of Infectious Diseases and Medical Microbiology adherence to ART.More than one-third of the patient's nutritional status was normal at the time of ART initiation (Table 2).

Survival Probability.
Due to the smaller events that occurred in the cohort, the median survival time was not computable.About 40 (81.63%)severe adverse drug reactions occurred in the frst year of ART enrollment, of which 29 (59.1%)occurred in the frst six months of ART enrollment.Te cumulative survival probability of developing severe ADRs of the HIV-positive patients on ART at six, twelve, twenty-four, and thirty-six months was 0.82, 0.57, 0.23, and 0.036, respectively (Table 3).Te overall survival probability of the patients through the treatment time decreased as the time increased (Figure 1).
Te survival probability among the unexposed group is high compared to the exposed group throughout, and the log-rank test indicates a statistically signifcant survival probability between the two groups (Figure 2).
Te survival probability of PLWH who did not experience TB/HIV confection was high when compared to those who experienced it, and the log-rank test indicates a statistically signifcant diference in survival between the two groups (Figure 3).
Te survival probability of males was high when compared to females, and the log-rank test indicates a statistically signifcant diference in survival between the two groups of PLWH (Figure 4).

Predictors of the Severe ADRs.
After proportional hazard assumptions were checked for each covariate, all covariates with a p value of ≤0.25 in bivariate analysis (health facility, sex, baseline BMI, WHO clinical stage, HIV/TB confection, experiencing other OIs than TB, and taking other medication at baseline) were considered as a candidate for multivariate Cox proportional hazard regression analysis.In the fnal model, three variables, gender, WHO clinical stage, and HIV/TB coinfection, were signifcantly associated with severe ADRs after model ftness was checked with Schoenfeld residual test.Female HIV-positive patients on ART follow-up were about three times (AHR � 3.14 95% CI: 1.59, 6.22) more likely to experience severe ADRs compared to male patients.Te risk of experiencing severe ADRs among patients presented with

Discussion
A total of 449 patients with HIV/AIDS who started ART between January 2018 and December 2020 were followed up and included in the analysis, of which 49 patients experienced severe ADRs.In this study, we found that 10.91% of patients enrolled in the study experienced severe ADRs over a period of 6,782 person-month observations.Being female, advanced clinical diseases and TB/HIV confection were signifcantly associated with the time to develop severe adverse drug reactions among PLHIV on ART.
On the other hand, the fndings of this study were lower than the study conducted in India, 23.1% [30], Nigeria, 16.8% [31], Northern and Southern Ethiopia, 3.2/100 person month [32], and 22.9% [33], respectively.Tis variability may be due to diferences in the follow-up period, which may afect the occurrence of adverse events in HIV patients on ART.In our case, the follow-up period was three years, while the follow-up period was six months in a study conducted in India [30] and 12 months in a prospective study conducted in Ethiopia [32] as ART-associated adverse drug reactions during the frst six months of treatment are high/common [25,[34][35][36][37].In addition, data collection methods or study design also may infuence the incidence of the ADR.Furthermore, the diference in ADR defnition among those studies is a possible reason.For instance, studies in India and in Southern Ethiopia defned ADRs as the occurrence of undesirable efects registered on patient follow-up charts such as nausea, diarrhea, anemia, abdominal pain, skin rash, and anemia, but in our case, severe ADRs were defned as serious/life-threatening reactions which consist of grade 3 and 4 of WHO ADR classifcation [18].Another possible reason is that the sample size of the present study is relatively larger than that of the study conducted in India, Northern Ethiopia, and Southern Ethiopia.
In this study, the incidence of severe ADRs is relatively higher than in the two studies conducted in the Northern parts of Ethiopia Tigray (3.6 per 100 person year) [24] and Amhara region (3 per 100 person-years) [22].Tis difference may relate to the diference in follow-up time in which the follow-up time in our case is lower than that of  the two studies.As the follow-up time increases, the denominator increases which implies a decrease in incidence and the ADRs are likely to occur in the early treatment life [34,35].
In the current study, the exposed group, those who presented with the advanced clinical disease at ART initiation time (AHR � 3.29; 95% CI: 1.47, 7.39), is three times more likely to experience severe ADRs compared to those unexposed.Tis fnding is supported by an observational study conducted in Nigeria [11], Debre Markos, Northern Ethiopia, Arbaminch, Southern Ethiopia, Kafa Zone, and Jimma, South West Ethiopia [22,23,29,36].Patients with advanced disease are more likely to have various comorbidities that can lead to multiple medications being taken at the same time.Patients on multiple medications or polypharmacy are more likely to experience serious side efects due to overlapping toxicities.Patients with advanced disease may also have altered pharmacodynamics and pharmacokinetic mechanisms due to drug reactions that predispose patients to adverse events [37][38][39][40].To address new HIV infections and early ART initiation, Ethiopia initiated casebased surveillance integrated with the testing of index cases and added to notifable diseases.However, the available information has not yet been well analyzed, reported or presented to the scientifc community.Tis surveillance system uses the hybrid model consisting of paper-based and electronic reporting in which there was nonalignment between them.Behavioral and technical factors such as poor documentation practice, skill gap, and lack of experience which result in poor data quality and poor information utilization for decision-making were the other challenges for the surveillance system in Ethiopia.Moreover, the challenges to the surveillance of low-and middle-income countries including Ethiopia are limited resources, both fnancial and capacitated human resources [41][42][43][44][45].
Patients experiencing HIV/TB confection at ART initiation were about two times more likely to experience severe ADRs (AHR � 2.35, 95%: 1.22, 4.54) when compared to those who did not experience them.Tis fnding is similar to study fndings in seven teaching hospitals in Ethiopia and southern Ethiopia [25,33].Tis could be explained by the overlapping toxicities of anti-TB and ART drugs, which have been well documented, and even some literature delay the initiation of ART during the intensive phase of the TB treatment course.Tere is also the documented drug-to-drug reaction of the anti-TB and ART drugs [39,[46][47][48][49].
In this study, female patients were three times (AHR � 3.06; 95% CI: 1.54, 6.06) at higher risk of experiencing severe ADRs at any time compared to males.Tis fnding is consistent with the study conducted in India [50], Nigeria [31], Ghana [51], Arbamanchi, and southern Ethiopia [29].Possible explanations may include physiological diferences in women having a lower lean body mass than men, reduced hepatic clearance in women, and differences in cytochrome P450 (CYP) enzyme activity and metabolism at diferent rates compared to men.Another possible explanation is a higher concentration of the drug in the blood and a longer elimination time in women than in men.In addition, women may use birth control and hormone supplements during their reproductive years, which increases the number of medications that can increase the likelihood of ADRs [37,39,52,53].
Being a retrospective study has its limitations since it is based on secondary data.Te event documentation may be diferent depending on the service provider's documentation practice and data utilization culture which leads to underestimation of the event of interest.Te other limitation of this study was that the side efects of the ART were not included and reported.Canadian Journal of Infectious Diseases and Medical Microbiology

Conclusion and Recommendation
In this study, the incidence of severe ADRs was relatively consistent with other studies conducted in diferent parts of the developing and developed countries with a considerable burden.Patients with advanced WHO clinical disease, HIV/TB coinfection, and female patients on the ART initiation should get due attention and close follow-up to minimize the preventable severe ADRs.Routine screening of patients for signs and symptoms of severe ADR is very crucial to prevent timely response and management of it.Additionally, we recommend a prospective study to measure variables more precisely that cannot be addressed through a retrospective study.

Figure 2 :
Figure 2: Kaplan-Meier survival estimate curve for the time to develop severe ADRs among PLWHA on ART by their exposure status in Harari, Eastern Ethiopia.

Figure 3 :Figure 4 :Figure 1 :
Figure 3: Kaplan-Meier survival estimate curve for the time to develop severe ADRs among PLWHA on ART by HIV/TB coinfection in Harari, Eastern Ethiopia.
, and baseline nutritional status.Te study participants were categorized as exposed and unexposed groups based on the baseline WHO clinical stage of HIV/ AIDS.Patients with WHO clinical stages III and IV and I and II were classifed as exposed and unexposed groups, respectively.
Ethical clearance was obtained from the Institutional Health Research Ethics Review Committee (IHRERC) with ref. no.IHRERC/026/2022 of the College of Health and Medical Science, Haramaya University Health and Medical Science College.An ethics approval letter from IHRERC was delivered to study facilities in the Hiwot Fana Comprehensive Specialized University Hospital and JugolHospital.Final approval for data extraction was obtained from study facilities.Informed, voluntary, written, and signed consents were obtained from each head of the respective health facility before data collection was started.As the study involves retrospective data, there was no direct harm to the patients.To ensure the confdentiality of the patients' records, data extracted from the patient follow-up charts were deidentifed and stored in a secured computer which was only accessible to authors and supervisors.
table was used to calculate the cumulative survival probability of the patients.Adjusted hazard ratio was used to report the efect size.Te Schoenfeld residuals and global test were used to check the goodness-of-model ftness.2.7.Ethical Statement.

Table 1 :
Sociodemographic characteristics of the patients on ARTfollow-up in Harari, Eastern Ethiopia, and February 25 to March 25, 2022.
advanced WHO clinical stage (exposed) at baseline was about three times (AHR � 2.96; 95% CI: 1.33, 6.58) as compared to unexposed patients on ART.Patients with TB confections at baseline were double (AHR � 2.11; 95% CI: 1.09, 4.06) times more likely to experience severe ADRs as compared to a patient with no TB infection at baseline (Table4).

Table 3 :
Life table for severe adverse drug reactions among HIV-positive patients on ART in Harari, Eastern Ethiopia, February 25 to March 25, 2022.

Table 4 :
Multivariate Cox regression analysis of the associations of the predictor variables and time to develop ADRs, February 25 to March 25, 2022.p value less than 0.05 indicates statistical signifcance in multivariable Cox proportional hazard, and 1 indicates a reference value. *