Determinants of anemia among HIV positive children on Highly Active Antiretroviral Therapy (HAART) attending hospitals of North Wollo Zone, Amhara Region, Ethiopia, 2019: A case-control study

Introduction Anemia is one of the most commonly observed hematological abnormalities and an independent poor prognostic marker of HIV disease. The rate of progression and mortality in this subgroup of patients is high compared to non-anemic patients. WHO estimates that over 2 billion people are anemic worldwide and young children bear the world's highest prevalence rate of anemia. In Ethiopia, there is limited information about the determinant factors associated with anemia among HIV positive children. Objective The main purpose of this study was to assess the determinants of anemia among children on highly active anti-retroviral therapy attending hospitals of North Wollo Zone, Amhara Region, Ethiopia. Methods A case-control study was conducted on 350 HIV infected children on HAART attending Hospitals of North Wollo Zone, from February 1 to March 30, 2019. The study participants were selected with consecutive sampling technique. An adapted, interviewer-administered, and pre-tested questionnaire and chart review were employed to collect the data. Besides, blood and stool sample was investigated to determine hematologic indices and malaria and to investigate intestinal parasites respectively. Data were analyzed by using the SPSS version 24 statistical software and bivariate and multivariate logistic regression was used to identify predictors. Results A total of 350 HIV positive children (117 cases and 234 controls) were included in this study with an overall response rate of 100%.On multivariate analysis, variables which have spastically signicance association with anemia were; had Amebiasis (AOR = 7.29, 1.22 - 43.56), had history of opportunistic infections (AOR = 9.63, 1.94 - 47.85), had malaria infection (malaria pf) (AOR = 4.37, 1.16 - 16.42),eating non-diversied (AOR = 10.39, 2.25 - 48.0), WGT –Age Z score value between -2_-3 (AOR = 9.80, 2.46 - 39.14),level of adherence (AOR = 2.31, 1.92, 7.77) and being from rural area(AOR = 8.8, 2.07 - 37.79). Conclusion In this study having parasitic infections, history of opportunistic This study revealed that 36.6% of children on HAART, 62 (17.7 %) of cases and 66 (18.9%) controls had a history of severe acute malnutrition. Similarly, 86 (24.6%) of cases and 114 (32.6%) controls had a history of opportunistic infections. The proportion of malaria infection was around two times higher among cases, 58 (16.6%) than controls, 31 (8.9%). Besides, the proportion of having low CD4-count (<350) was approximately 6 folds higher among cases 56 (16.0%) compared to controls, 10 (2.9%). Regarding HAART, the proportion of poor adherence was around three times higher among cases 16(4.6%) compared to controls 5 (1.4%) (Table–3). Bivariate logistic regression analysis was done for different predictor variables. Intestinal parasite, WGT-AGE Z score, history of oro-pharyngeal disease, history of opportunistic infections, history of chronic diarrhea, history of Missing Follow up, HAART duration, Level of Adherence to HAART, on Cotrimoxazole, Viral Load, Malaria history, Highest WHO staging, Care giver’s residence, Orphanhood of the child, Wealth Index, household food security, Food diversity, Feeding frequency, Counseling on child feeding, were signicantly associated with the HGB level of respondents. All variables with 𝑃 value less than 0.25 were tted into the backward stepwise multivariate logistic regression model. But, on multivariate analysis, having parasitic infections, history of opportunistic infections, being malnourished, poor adherence to ART, caregivers living in the rural area and eating non-diversied foods were signicantly associated with anemia among HIV positive children on HAART.


Background
HIV is still a global problem causing extensive morbidity and mortality worldwide, especially in sub-Saharan Africa, where more than 70% of all HIV cases occur. (1)(2)(3)(4). Hematological abnormalities are a common occurrence in individuals diagnosed with human immunode ciency virus (HIV) (5)(6)(7)(8)(9), of these hematological abnormalities anemia is the most common in HIV-infected children that have a signi cant impact on the quality of life and clinical outcomes. (7)(8)(9) According to World Health Organization (WHO) de nitions, anemia is de ned as hemoglobin level less than 11 g/dl for children <5 years old, <11.5 g/dl for children 5-11.9 years old and <12 g/dl for children 12-14.9 years old after altitude adjustment (10). Anemia is classi ed as mild degree (Hgb 10.0-10.9 g/dl), moderate (Hgb 7.0-9.9 g/dl), severe (Hgb 4.0-7.0 g/dl), and very severe Hgb less than 4.0 g/dl. It can also be classi ed based on the hematocrit (PCV) percent. Packed cell volume (PCV) of less than 33.0 % is regarded as anemia by the World Health Organization. (10) The risk factors of anemia are multiple and vary across geographical areas. In HIV-infected patients, anemia may be caused by nutrient de ciencies (iron, folic acid, and vitamin B12), sickle cell disease, HIV/AIDS itself, malaria, hookworm, and other infections. Iron de ciency anemia is the leading(50%) cause of childhood anemia in developing countries (8,9,11) In addition to the direct effect of HIV, highly active antiretroviral therapy (HAART) also causes anemia. Although HAART has the capability of reducing the incidence of anemia [17] by suppressing viral replication and increases CD4 cell count, anemia remains a common problem even for patients treated with antiretroviral agents. Anemia due to drugs, such as cotrimoxazole, pentamidine, foscarnet, and zidovudine (AZT) often re ects reticuloendothelial iron block. (12) There are several factors believed to contribute to the pathophysiology of anemia observed in HIV patients. First, many of the opportunistic infections or malignancies to which HIV patients are susceptible can lead to anemia (13).
As with anemia in the HIV population, micronutrient de ciencies can play a role in contributing to anemia in HIV patients (13). Finally, many of the most common ART drugs included in the standard also harm hematopoiesis and can thus contribute to anemia in HIV. The molecular mimicry between erythropoietin (EPO) and the HIV-1 p17 protein can lead to circulating auto-antibodies against endogenous EPO in some HIV patients (4,13).
Anemia is a common co-morbid condition among HIV infected children and has a profound impact on disease progression and has been noted as a signi cant predictor of decreased survival time and death (14,15). Especially children are more prone to the consequences of anemia due to high iron requirements, low intake of iron from foods, and frequent episodes of infection (10).
Anemia is one of the global public health problems that affect more than one-third of the world population. It has been strongly associated with poor growth and development, limited psychomotor development, and poor long-term performance in cognitive, social, and emotional functioning in children. It is also associated with impaired mental, physical and language development; and poor coordination, scholastic achievement and immune function (16).
Also, the outcomes of anemia can be considered from a variety of perspectives, including stunt growth, increase morbidity from infectious diseases, detrimental impacts on economic and social development (2,17).
According to World Health Organization (WHO) 2015 report, about 43% of under-ve children were anemic, with regional variations of 62.3% in African, 53.8% in South-East Asia and 21.9% in Western Paci c Region (8,13).
Anemia is a common childhood health problem in Africa, and its prevalence among children under the age of ve years is estimated at 62%, which is above the cut off points (40%). It is a leading cause of pediatric mortality and impaired development and is highly prevalent in young children in sub-Saharan Africa (9,18).
In Ethiopia As many as six in ten under-ve children are anemic (19). As per EDHS 2016, Overall, more than half of children 6-59 months (56 percent) suffered from some degree of anemia: 25 percent were mildly anemic, 28 percent were moderately anemic, and 3 percent were severely anemic (11). Despite causing such high levels of burden, the public health community does not pay anemia the attention it deserves.
The government of Ethiopia has been working to reduce childhood anemia. Accordingly, it endorsed the national nutrition program, bimanual school deworming, developed micronutrient de ciency prevention and control guideline and implemented micronutrient forti cation. But studies from different corners of the country have shown that childhood anemia is still a major public health problem in this country (20).
Most of the studies on the HIV-anemia interplay were done in developed countries and mainly included adults.
Determinants of anemia among HIV/AIDs positive children on HAART has not been very well understood in resource-poor settings like Ethiopia, although there are few crossectional studies conducted at different parts of the country. However, knowing the determinants of anemia among HIV/AIDS positive children on HAART through the case-control study will help to screen and manage anemia according to its cause. Therefore this study intends to ll these gaps by identifying the determinant factors of anemia among HIV positive children on HAART.

General Objective
To assess determinant factors of anemia among HIV Positive Children on Highly Active Antiretroviral Therapy (HAART) attending hospitals of North Wollo Zone, Amhara, Ethiopia, 2019.

Speci c objectives
To identify socio-demographic factors associated with anemia among HIV Positive Children on Highly Active Antiretroviral Therapy.
To determine nutritional related factors associated with anemia among HIV Positive Children on Highly Active Antiretroviral Therapy.
To identify drug-related factors associated with anemia among HIV Positive Children on Highly Active Antiretroviral Therapy.
To identify disease-related factors associated with anemia among HIV Positive Children on Highly Active Antiretroviral Therapy.

Study Area
This study was carried out in North Wollo Zone Hospitals, Amhara region, Ethiopia.

Study Design and Period
A health facility-based case-control study was conducted from February 1 to March 30, 2019.

Population Source Population
All HIV Positive Children on Highly Active Antiretroviral Therapy (HAART) attending hospitals of North Wollo Zone, Amhara, Ethiopia.

Study Population
All HIV Positive Children on Highly Active Antiretroviral Therapy (HAART) attending selected hospitals of North Wollo Zone, Amhara, Ethiopia.

Sample Size Determination
The sample size is determined using a double population proportion difference formula by using major determinant variables such as iron de ciency, in ammation, underweight, malaria Pf, albumin de ciency, and ART regimen. Since it gives the maximum sample size, iron de ciency was considered as an independent predictor (exposure) variable ( Table 1). From that study, the proportion of children among cases with iron de ciency was 26.9% and the proportion of children among controls with iron de ciency was 8.3%. (30). Using Epi-Info version7, one to two ratio (1:2) of cases to controls and with a power of 90%, the total sample size becomes 350 with 117 cases and 234 controls.

Exclusion criteria for controls
Children with an incomplete chart, children passed-away on arrival, children admitted without their mothers.

De nitions of terms and operational de nitions
Case (anemia) -HIV Positive Children on Highly Active Antiretroviral Therapy (HAART) whose Hg level is less 11 g/dl for children <5 years old, <11.5 g/dl for children 5-11.9 years old and <12 g/dl for children 12-14.9 years old or may not be diagnosed with other medical illness (31).
Control (non anemic)-HIV Positive Children on Highly Active Antiretroviral Therapy (HAART) whose Hg level is more than or equal to 11 g/dl for children <5 years old, 11.5 g/dl for children 5-11.9 years old and 12 g/dl for children 12-14.9 years old age and sex-matched. Nutritional assessment: Dietary Diversity Score (DDS) was calculated from a single 24-hour dietary recall data. Additionally, study participants were screened for nutritional status using anthropometric measurements.
Chart review was done to identify hemoglobin concentration (Hgb), blood history of malaria and Viral load.
Stool specimen collection and examination: A stool sample was collected from each study participants using clean, widemouthed and leak-proof stool cup by laboratory professionals. Then, stool wets mount was prepared using saline and/or iodine and examined microscopically for identi cation of intestinal helminths.

Data quality control
A pretest was conducted on 5% (17) of the sample size among children on HAART attending Dessie referral hospital which was not included in the study settings and was selected by lottery method among the hospitals not included in the study settings. The pretest study was conducted 2 weeks before the actual data collection period. The study participants who ful ll the inclusion criteria were selected by simple random sampling technique. The result of the pretest was discussed and all the necessary amendments were made on, the instructions, contents, order and grammatical issues based on the results of the pretest.
Three-day intensive training was given for data collectors and supervisors on the overall data collection procedure to minimize systematic errors. A clear explanation of the purpose and objective of the study was provided for the respondents at the beginning of the interview. On spot-checking and correction was made for incomplete questionnaires by supervisors. Laboratory test quality was assured by giving training for laboratory professionals by MSc laboratory professionals using the standard operating procedure (SOP) and regular monitoring of reagents for manufacturing, expiry date, and proper storage. The sample was processed immediately after collecting from the study participants to minimize errors and the overall data collection process was monitored by investigators.

Data analysis and presentation
The lled questionnaires were checked for completeness, coded and entered into Epi Data version 4.2 and then exported to SPSS version 24 for further analysis. Descriptive statistics were computed and the result was reported using frequencies and percentages. Bivariate logistic regression was performed and variables with p < 0.25 were transported to multivariable logistic regression. Finally, variables with P-value < 0.05 in the multivariable logistic regression model were taken as statistically signi cant and adjusted odds ratio with its 95% con dence interval was considered to see the strength of association between the exposure variables and the outcome of interest.

Ethical consideration
Ethical clearance was obtained from the Woldia University research ethics committee and it approves the whole part of the research including the method. Permission and support letter was sought and obtained from the North Wollo Zone Health O ce. Then, o cials at different levels in the hospitals were communicated through letters. The responsible bodies at each hospital were told about the purpose of the study. Written informed consent was obtained from parents of every study participants after a detailed explanation on the purpose and bene t of the study right before data collection and assent was obtained from the study participants and the ethics committee approved this procedure. The respondents were noti ed that they have the right to refuse or terminate at any point during data collection. All data collectors and supervisors were instructed on how to comply with strict con dentiality practices for all participants both during and after data collection. The identity of the participants was kept anonymous.

Dissemination of the result
The results of the study were disseminated to Woldia University, faculty of health science, research directorate o ce. Furthermore; the ndings and recommendations of the study will also be shared in a written document to North Wollo Zone Health O ce and Hospitals in North Wollo Zone. After public defense and incorporation of comments, the result will also be disseminated through presentations on research conferences, workshops, and symposiums. Finally, the attempt will be made to publish in a reputable peer-reviewed journal. The study participants will remain anonymous during and after publication. The owners of the research are the researchers and Woldia University

Results
Socio-demographic characteristics of the respondents A total of 350 HIV positive children (117 cases and 234 controls) were included in this study with an overall response rate of 100%. One hundred eighty-two 182(52%) of them were females. The mean and the median age of the child were 8.91 and 3.14 years respectively. Most of the participants were from urban areas (  Care giver's residence, Orphanhood of the child, Wealth Index, household food security, Food diversity, Feeding frequency, Counseling on child feeding, were signi cantly associated with the HGB level of respondents. All variables with value less than 0.25 were tted into the backward stepwise multivariate logistic regression model. But, on multivariate analysis, having parasitic infections, history of opportunistic infections, being malnourished, poor adherence to ART, caregivers living in the rural area and eating non-diversi ed foods were signi cantly associated with anemia among HIV positive children on HAART. The odds of being anemic among HIV positive children on HAART who had Amebiasis was around 7 times higher than Eating non-diversi ed foods was found to be signi cantly associated with anemia among children on HAART. The odds of being anemic among HIV positive children on HAART, who eat non-diversi ed foods was about 10 folds higher than those who eat diversi ed foods (AOR = 10.39, 2.25 -48.0). At the same time, the odds of children on HAART with WGT -Age Z score value between -2_-3 was also about 10 times higher to be anemic than those with WGT -Age Z score value between 1_-2 (AOR = 9.80, 2.46 -39.14).
The odds of children on HAART who have poor adherence to ART drugs were two times higher compared to those who have good adherence (AOR = 2.31, 1.92, 7.77). Besides, caregivers residence was also signi cantly associated with anemia. The odds of being anemic among children on HAART whose caregiver living in the rural area was approximately 9 folds higher than those whose caregiver living in an urban area (AOR = 8.8, 2.07 -37.79) ( Table 4).

Discussion
Anemia is a common co-morbid condition among HIV infected children and has a profound impact on disease progression and has been noted as a signi cant predictor of decreased survival time and death (14,15). Especially children are more prone to the consequences of anemia due to high iron requirements, low intake of iron from foods, and frequent episodes of infection (10).
In Ethiopia, there is limited information about the determinant factors associated with anemia among HIV positive children. Hence, this study aimed to determine the determinant factors of anemia among HIV infected children on HAART.
This case-control study included socio-demographic variables like residence, occupation of mothers, household wealth index status, and disease-related variables like history malaria and opportunistic infections. It also included drug-related and nutrition-related variables like HAART duration, HAART regimen, dietary diversity score and nutritional status of children.
In this study, having parasitic infections, history of opportunistic infections, being malnourished, poor adherence to ART, caregivers living in a rural area and eating non-diversi ed foods were signi cantly associated with anemia among HIV positive children on HAART (Table-4).
This study result revealed that the odds of being anemic among HIV positive children on HAART who had Amebiasis was around 7 times higher than those who had no Amebiasis. This might be because children with Amebiasis might cause dysentery which will result in GI blood loss() Similarly, the odds of being anemic among children on HAART who had malaria infection (malaria pf) was also around 4 folds higher than those who have no malaria infection. This nding is in line with a study conducted at Zewditu Memorial Hospital, Ethiopia in which intestinal parasites and malaria were independently associated with the odds of being anemic (20). It is also consistent with research in south-central Ethiopia in that the risk of anemia was high (adjusted Hazard Ratio = 10) among children with malaria (25). The possible justi cation for this might be due to falcifarum malaria can cause hemolysis which in turn results in hemolytic anemia.
Similarly, research from Addis Ababa, Ethiopia revealed having intestinal parasitic infections (AOR = 2.7, 95% CI, 1.1-7.2) were identi ed as signi cant predictors of anemia (26). This might be because some parasitic infections including malaria are the major causes of blood loss and destruction of red blood cells.
The odds of children on HAART who had a history of opportunistic infections was about 10 times higher to be anemic compared to those who had no history of opportunistic infections. This is consistent with a study at Zewditu Memorial Hospital in which TB-HIV co-infection was independently associated with the odds of being anemic (20). This could be because children who have been attacked with the opportunistic infection will have a loss of appetite and more likely to be malnourished, which is a signi cant predictor of anemia.
Eating non-diversi ed foods was found to be signi cantly associated with anemia among children on HAART. The odds of being anemic among HIV positive children on HAART, who eat non-diversi ed foods was about 10 folds higher than those who eat diversi ed foods. This is consistent with a research nding from Waghimra, Northeast Ethiopia which revealed poor dietary diversity being signi cantly associated with anemia (29). This is because when children got diversi ed food they will have access to iron-rich foods which will prevent iron de ciency anemia.
The odds of children on HAART with WGT -Age Z score value between -2_-3 was also about 10 times higher to be anemic than those with WGT -Age Z score value between -1_-2. This is in line with ndings of other studies in which stunting and vitamin A de ciency were identi ed as an independent risk factor for anemia (21,29). This might be because malnourished children will have associated iron de ciency which is the most common cause of anemia.
The odds of children on HAART who have poor adherence to ART drugs were two times higher compared to those who have good adherence (AOR = 2.31, 1.92, 7.77). The possible justi cation for this might be those children with poor adherence are at higher risk of developing opportunistic infections which intern results in anemia.
Besides, caregivers residence was also signi cantly associated with anemia. The odds of being anemic among children on HAART whose caregiver living in a rural area was approximately 9 folds higher than those whose caregiver living in an urban area. This is in line with the 2016 EDHS report in which children in rural areas were more likely to be anemic than those in urban areas (57 percent and 47 percent respectively). This might be because children family from urban resident have better consumptions of diversi ed foods; in addition to this, the prevalence of intestinal parasites is low in urban.
Abbreviations And Acronyms communicated through letters. The responsible bodies at each hospital were told about the purpose of the study. Written informed consent was obtained from parents of every study participants after a detailed explanation on the purpose and bene t of the study right before data collection and assent was obtained from the study participants. The respondents were noti ed that they have the right to refuse or terminate at any point during data collection. All data collectors and supervisors were instructed on how to comply with strict con dentiality practices for all participants both during and after data collection. The identity of the participants was kept anonymous.

Consent for publication
Note applicable.Availability of data and materialsData is available and it can be accessed from the corresponding author when asked with a reasonable inquiry.