Introduction
Human immunodeficiency virus-positive/acquired immunodeficiency syndrome (HIV/AIDS) and malnutrition form a deadly duo with each one fuelling the other. Malnutrition increases susceptibility to infection by causing immune dysfunction in manifold ways. The depressed immune status can amplify HIV replication and accelerate the progression of HIV disease to AIDS stages(Reference Oumer, Kubsa and Mekonnen1,Reference Poda, Hsu and Chao2) . Untreated or advanced HIV/AIDS is again associated with a compromised immune status that makes patients susceptible to lethal opportunistic infections. Rationally, the HIV disintegrates the primary physical defense barriers of collagen levels in skin cells and alters intracellular neovascularisation(Reference Taye, Shiferaw and Enquselassie3,Reference Jobiba, Andrew and Bandac4) . Evidence suggested that inadequate calorie intake due to elevated levels of proinflammatory cytokines like interleukin-1(IL-1), interleukin-6 (IL-6) and tumour necrosis factor-α (TNF-α) can cause anorexia and oxidatively stressed derangements of metabolites(Reference Taye, Shiferaw and Enquselassie3,Reference Kebede5,Reference Willumsen6) .
The clinical presentation of severe acute malnutrition (SAM) includes severe visible wasting (marasmus), nutritional oedema (kwashiorkor) or marasmus–kwashiorkor. However, kwashiorkor and marasmus among HIV-infected children have a higher risk of mortality(Reference Chanie, Legas and Zewude7,Reference Kebede, Kebede and Negese8) . In sub-Saharan Africa, the epidemiology of SAM has increased among children requiring hospitalisation composed of those who are HIV infected, with case fatality rates reaching as high as 20–50 %(Reference Tiyou, Belachew and Alemseged9–Reference Alebel, Demant and Petrucka11). HIV-infected children with SAM are more likely to present with other comorbidities and complications (e.g. TB, severe anaemia, persistent diarrhoea or villous atrophy and disaccharide intolerance), which contribute to loss of appetite, poor oral intake and continual lean body mass(Reference Chanie, Legas and Zewude7,Reference Fergusson and Philippa12) .
Globally, in 2019, 1⋅7 million children are infected with the HIV; in the same year, 44 229 children in Ethiopia were infected with HIV, and 2055 died because of AIDS(Reference Kebede5,Reference Chanie, Legas and Zewude7) and case fatality rates among children with SAM are often as high as 20–30 %(Reference Schofield and Ashworth13). The greatest burden of malnutrition is seen in the peak age group of 2–5 years(Reference Oumer, Kubsa and Mekonnen1,Reference Gebremichael, Hadush and Kebede14) . HIV increased poor nutrition as a result of poor food intake and fuelled nutrient usage from the body increases the incidence of lethal opportunistic infection for further hospitalisation(Reference John, Diala and Adah15). A systematic review and meta-analysis on children with SAM in sub-Saharan Africa revealed that children with HIV infection were more likely to die than those not infected with HIV (30⋅4 v. 8⋅4 %)(Reference Willumsen6,Reference Marotta, Abate and Aragie10,Reference Alebel, Demant and Petrucka11) . When weight loss was >10 % below the baseline weight, the relative risk of death increased nearly six-fold(Reference Oumer, Kubsa and Mekonnen1,Reference Kebede5,Reference Hecht, Weber and Grote16) . In addition, even relatively small losses in weight (5 %) are associated with a decrease in the survival rate of HIV-infected children.
According to the Health and Health-Related Indicators (HHRIs) of 2014, SAM was the third leading cause of mortality (accounting for 8⋅1 % of deaths) and 20 % of hospital admissions for children in Ethiopia(Reference Wagnew, Tesgera and Mekonnen17–Reference Abate, Aragie and Tesfaw19) and mortality rate in HIV children undergoing SAM treatment is found to be 5⋅4 per 100 child-years in northwest Ethiopia(Reference Kebede20). A major contribution to the excess mortality due to malnutrition in HIV-positive children is the inaccessibility of timely and appropriate medical services, which are plentiful and vary at regional and district levels(Reference Rose, Hall and Martinez-Alier21). Financial fees for accessory transport and treatment cost, which are frequent barriers to paediatric healthcare services in many developing countries including Ethiopia(Reference Hecht, Weber and Grote16,Reference Rose, Hall and Martinez-Alier21) . Even though SAM is a leading cause of hospitalisation and mortality for HIV-infected children, there is no prior evidence in Ethiopia on when to develop SAM and associated risk factors for seropositive children after antiretroviral therapy (ART) is initiated. Hence, the present study aimed to estimate the half-life time to develop SAM in HIV-infected children treated with ART care in selected health facilities of the Metekel Zone including Assosa Hospital, Northwest Ethiopia.
Methods
Study setting and time
The present study was conducted in the Metekel Zone at a selected health facility from 1 January 2011 to 30 December 2021. The Metekel Zone is one of the three administrative zones, located in the Benishangul Gumuz region. The region has two referral hospitals, three primary hospitals and more than thirty-two health centres. The present study was conducted in two health centres and two referral hospitals. The main reason including only four health institutions is earlier endorsement and initiation of ART care for more than 2968 catchment adult and children population received ART care(Reference Bizuneh, Jara and Abate22).
Study design
A facility-based retrospective cohort study was conducted.
Source population
The source population was the entire HIV-positive children who had enrolled in HIV/AIDS chronic care in selected health facility in northwest Ethiopia.
Study population
All HIV-positive children enrolled and started ART in selected health facility from 1 January 2011 to 30 December 2021.
Inclusion criteria
All HIV-positive children (≤15 years of age) were enrolled for HIV/AIDS chronic care without SAM at baseline from 1 January 2011 to 30 December 2021 in selected health facility.
Exclusion criteria
Children at baseline registered for ART and whose ART follow-up care did not contain variables including age of children and date of ART initiation were excluded.
Study population
All HIV-positive children who started taking ART between 1 January 2011 to 30 December 2021, at two health centres and two referral hospitals qualified as study participants and were eligible for the final analysis.
Sample size determination
The sample size was calculated by using log-rank survival data analysis of the double population proportions formula by using the following assumption, 95 % confidence level, 80 % optimum statistical power and taking one error of 5 %. Considering a study that was conducted in the same place in northwest Ethiopia and taking rural residents as a predictor variable for the exposed group of seropositive children denoted by q1 (0⋅52) and urban residents not exposed group denoted by q0 (0⋅48)(Reference Kebede5). After adding 10 % incomplete medical records, the final sample size was found to be 630. Nevertheless, since children ≤15 years enrolled for ART care in two hospitals and health centres from 1 January 2011 to 30 December 2021 were found to be 732. Hence, all study participants were included, and all study participants without a sampling procedure due to increase statistical power.
Finally, since children under the age of 15 were enrolled for ART care in two hospitals and health centres from 1 January 2011 to 30 December 2021, and after we had retrieved seropositive children's ART follow-up cards and screened for eligibility requirements (11 files are removed), we included 721 individual files without any sampling procedure to increase statistical power.
Dependent variable
Incidence of severe acute malnutrition.
Independent variables
Socio-demographic characteristics of the children (age, sex, residence, family size, caregiver, parental status), baseline clinical and laboratory factors, characteristics like ART regimen, functional status, developmental status, nutritional status, and opportunistic infections, and follow-up factors (level of ART adherence, CPT, IPT, TB contact history and vaccination status).
Operational definitions
Events: Children with under-fifteen HIV-positive children who had an admission history of SAM (MUAC <11⋅5 cm, or weight-for-height/length Z-score <−3, or bilateral pitting oedema of other causes excluded with or without severe wasting Z-score <−3) were an event for this study. Censored: If the child had lost follow-up or transferred out to another service before developing SAM, or if the child was free from SAM until the end of our data collection day.
Adherence to ART: ART was classified based on the percentage of drug dosage calculated from the total monthly doses of ART drugs (Good > 95 %, fair 85–94 % and poor < 85 %)(Reference Getahun, Teshome and Fenta18).
Anaemia: It was defined as having a haemoglobin level ≤10 mg/dl(Reference Muenchhoff, Healy and Singh23).
CD4 count: CD4 levels below the threshold level were classified based on the child's age (i.e. infants CD4 1500/mm3, 12–35 months 750/mm3, 36–59 months 350/mm3 and 5 years 200/mm3)(24).
Time to SAM: Newly diagnosed of SAM in HIV-positive children after ART initiated; Severe acute malnutrition: which is defined by the WHO as a weight-for-height Z-score of less than −3 Z-curve, or a mid-upper arm circumference of less than (MUAC) < 11⋅5 cm in a child aged 6 months to 5 years, or the presence of bilateral oedema with failed appetite test(Reference Kebede5).
CD4 count: CD4 levels below the threshold level were classified based on the child's age (i.e. infant CD4 1500/mm3, 12–35 months 750/mm3, 36–59 months 350/mm3 and 5 years 200/mm3)(Reference Molla, Kebede and Kebede25).
Incomplete data: Children at baseline registration for ART, whose ART follow-up care did not contain variables including age of children, and ART initiation date.
Data collection instruments and quality control
Standard and pretested data extraction tools were used to extract the required information from the case notes both for new and readmitted cases. Before the actual data collection, the prepared checklist of variables was pretested in thirty-seven case notes of HIV-infected children at Jawi Primary Hospital. Two-day training was given for two diploma nurse data collectors and for a degree public health officer about the objective of study outcome and importance of maintaining data confidentiality.
Data processing and analysis
After coding, data were entered into Epi-Data version 4.2, and then exported to STATA (se)/14 for further analysis. The WHO Anthro-Plus-Version 1.04 and ENA for Nutrition Smart Software were used to generate the Z-score (WAZ, HAZ and WHZ/BAZ) to define the nutritional status of HIV-positive children. The incidence rate of SAM is calculated using the total number of people per year (PPY) individual contribution to follow-up as a denominator. Descriptive nonparametric statistical tests such as the Kaplan–Meier plot were used to estimate the median SAM-free survival time. Assessing whether a real statistically significant survival difference between the two groups was tested by using the log-rank test.
The necessary Cox-proportional hazard model assumption was checked using a graphical diagnostic based on the scaled Schoenfeld residuals (log–log survival plot) and statistical tests (using the global test estimations). The multivariable Cox-proportional hazard regression for this study was written during consecutive follow upon HIV-infected children individuals under observation experiencing the event (SAM) in a period centred on that point in time. The covariates of the multivariate analysis were selected using the enter method. Variables with P-value <0⋅2 in the bi-variable Cox regression analysis were included in the multivariable Cox regression model to determine the factors associated with SAM incidence. Deviance Information Criteria (DIC), Akaike's Information Criterion (AIC) and Bayesian Information Criterion (BIC) are used to compare the candidate Cox-proportional hazard to the checked final model fitted. The model with P-value less than 0⋅05 will be selected as the final model of the analysis and checked using Nelson-Aalen and Cox-Snell residual tests (Fig. 6).
Results
Socio-demographic characteristics of HIV-infected children
Overall, 732 children living with HIV files were reviewed for this study. However, eleven individual charts (1⋅6 %) were excluded due to incompleteness. The majority 384 (53⋅26 %) of the respondents were female in gender, and 510 (70⋅74 %) of those were urban inhabitants. The overall mean (±sd) age of participant children was found to be 9⋅83 (±3⋅3) years. A large proportion of 381 (52⋅84 %) children had lived with parents (Table 1).
Table 1. Socio-demographic characteristics of children living with HIV in the Metekel Zone at selected health facilities in northwest Ethiopia from 1 January to 30 December 2021 (n 721)
Baseline clinical and comorbidity characteristics
More than half 487 (67⋅47 %) of the participant children had PMTC follow-up during pregnancy and almost all 467 (94⋅47 %) of the caregivers in their dyads had nutritional counselling at the baseline. Nevertheless, 86 (11⋅93 %), 139 (19⋅2 %) and 70 (9⋅9 %) of those children developed severe wasting, stunting and being underweight, respectively. The majority (57⋅3 %) of the participant children had a CD4 count above the threshold, but nearly two in five 287 (39⋅8 %) of those were in CLINICAL stages III&IV. Nearly half 302 (41⋅6 %) of the participant children did not receive CPT, 270 (37⋅4 %) did not receive IPT, 510 (70⋅7) had no ART regimen change and 463 (64⋅2 %) had no opportunistic infection. At the baseline, the mean (±sd) weight and MUAC of participant children were found to be 11⋅3 (±23⋅6) kg and 10⋅8 (±1⋅4 cm), respectively (Table 2).
Table 2. Baseline clinical and immunological characteristics of seropositive children in selected health facilities in northwest Ethiopia from 1 January to 30 December 2021 (n 721)
Time to develop SAM in HIV-infected children
A nine-year retrospective cohort study of 721 participant children yielded 21 685 months. The median time to develop SAM was found to be 30⋅3 (IQR ± 13⋅4) months with the crude incidence of SAM at 14⋅2 %. During the follow-up period, about 612 (84⋅88 %) children were on follow-up, while 39 (5⋅4 %) died, 58 (8⋅04 %) lost from follow-up and 12 (1⋅67 %) were transferred out. The median survival probability for participant children was found to be 89⋅76 % (86⋅74; 92⋅12). The overall incidence density rate of SAM was 5⋅64 per 100 child-year observations (95 % CI 4⋅68, 6⋅94).
Fig. 1. Types of ART regimen for children receiving ART at selected health facilities in northwest Ethiopia from 1 January to 30 December 2021 (n 721).
Fig. 2. Types of opportunistic infections for children receiving antiretroviral therapy at selected health facilities in northwest Ethiopia from 1 January to 30 December 2021 (n 721).
Kaplan–Meier survival of SAM
There was a significant survival difference in the time of developing SAM among the categorical variables like below threshold CD4 count disclosed HIV status, and levels of haemoglobin at the baseline had a significant survival difference (Fig. 3).
Fig. 3. Overall hazard-estimate of attrition for under-five children after started inpatient treatment for complicated SAM at Pawe general hospital in northwest Ethiopia from 1 January to 30 December 2021 (n 721).
Accordingly, there was a significant survival difference in the incidence of SAM between children who had haemoglobin ≤10 gm/dl as compared with those who had haemoglobin levels > 10 gm/dl and evidence from the log-rank test (Chi2(1) = 52⋅1, P = 0⋅001) (Fig. 4).
Fig. 4. Kaplan–Meier survival curve of children living with HIV stratified by levels of haemoglobin who enrolled for ART in northwest Ethiopia from 1 January to 30 December 2021 (n 721).
Furthermore, there was also a significant survival difference for children who had WHO clinical stages III&IV at the baseline as compared with those who had children WHO clinical stages I&II; which is evidenced by the log-rank test (Chi2(1) = 77⋅1, P = 0⋅001) (Fig. 5).
Fig. 5. Kaplan–Meier survival curve of children living with HIV stratified by WHO clinical stages in enrolled for ART care in northwest Ethiopia from 1 January to 30 December 2021 (n 721).
Predictors for time to develop SAM in HIV-infected children
During bi-variable Cox regression analysis, variables were checked whether they were factors associated with time to develop SAM at P-value <0⋅2 for a candidate transferee of multivariable Cox regression. After adjusting certain confounding, eleven variables were fitted to build the final model with three independent factors associated with the incidence of SAM. The risk of SAM among children with CD4 counts below the threshold level was 2⋅6 increased as compared with those child having CD4 counts above the threshold [AHR 2⋅6 (95 % CI 1⋅2, 2⋅9, P = 0⋅01)]. Likewise, the risks of SAM for HIV-positive children with disclosed their HIV status were 1⋅9 times the increased risk of developing SAM as compared with those children not disclosed HIV status [AHR 1⋅9 (95 % CI 1⋅4, 3⋅39, P = 0⋅03)]. Moreover, the risk of developing SAM for baseline levels of haemoglobin ≤10 mg/dl is nearly two [AHR 1⋅8 (95 % CI 1⋅2, 2⋅9, P = 0⋅03)] times increased as compared with the counter group having >10 mg/dl at the baseline (Table 3).
Table 3. Bi-variable and multivariate Cox regression analysis for time to develop SAM among HIV-infected children in northwest Ethiopia from 1 January to 30 December 2021 (n 721)
Overall, the model adequacy test
The overall multivariable Cox regression test of model adequacy for SAM incidence in HIV-positive children after ART started and it indicates that the line is on the straight origin (Fig. 6).
Fig. 6. Nelson-Aalen and Cox-Snell residual tests for the final model adequacy test.
Discussion
At the end of the study period, 103 (14⋅2 %) participants’ children developed SAM with maximum and minimum time to develop between 3 and 98 months. This finding is lower than previously reported studies done in south Gondar hospital (26⋅72 %)(Reference Chanie, Legas and Zewude7), East Africa (24⋅65 %)(Reference Marotta, Abate and Aragie10), sub-Saharan Africa (24⋅5 %)(Reference Boettiger, Aurpibul and Hudaya26), West Africa (26 %)(Reference Jesson, Masson and Adonon27), Malawi (17⋅5 %)(Reference Jobiba, Andrew and Bandac4) and Burkina Faso (63 %)(Reference Poda, Hsu and Chao2). This might be due to the difference in quality care, cut-off point for SAM diagnosis and ability of healthcare providers to screen in different health institutions(24). Likewise, the overall incidence density rate in our study was found to be 5⋅64 per 100 PPY (95 % CI 4⋅68, 6⋅94) and is comparable with previously reported studies in 5⋅4 per 100 PPY at Debre Tabor(Reference Chanie, Legas and Zewude7) and 5⋅42 per 100 PPY in Pawe hospitals(Reference Kebede5). This might be due to the difference in socio-demographic characteristics, study population and study design(Reference Chanie, Legas and Zewude7). In contrast to the 126-month finding in the Debre Tabor Hospital, the median time to develop SAM in our report was found to be 30⋅3 months. This could be explained by the quality of treatment and screening capabilities of the healthcare workers in the two institutions. Given that Debre Tabor Hospital is a specialised university hospital, our study setting consists of health centres and general hospitals(Reference Chanie, Legas and Zewude7,Reference Molla, Kebede and Kebede25) .
Regarding predictors of SAM, the risk of SAM among children with a baseline CD4 count below the threshold level was three times increased than that of children with a CD4 count above the threshold levels [AHR 2⋅6 (95 % CI 1⋅2, 2⋅9)]. This is comparable with findings reported in Pawe Hospital(Reference Kebede5), Debre Tabor Hospital(Reference Chanie, Legas and Zewude7), Burkina Faso(Reference Léon, Savadogo and Donnen28), Nigeria(Reference Chiaha and Okechukwu29), Malawi(Reference Jobiba, Andrew and Bandac4) and Asia Pediatric HIV Observational Database(Reference Boettiger, Aurpibul and Hudaya26). This might be because children who have low CD4 count can be exposed to chronic diarrhoea, tuberculosis and lethal opportunistic infections, which ought to cause a significant imbalance in nutritional demand and individual intake both in quantitative (number of kilocalories/day) and qualitative (vitamin and minerals, etc.) deficiencies. What is more, the proinflammatory cytokines like interleukin-1 (IL-1), interleukin-6 (IL-6) and tumour necrosis factor-α (TNF-α) for a long time cause anoxia and lack of interest in feeding(Reference Taye, Shiferaw and Enquselassie3,Reference Kebede5,Reference Willumsen6) , this might expose for metabolic derangement due to oxidative stress and hasten viral replication in progressions of AIDS disease(Reference Hecht, Weber and Grote16,Reference Moolasart, Chottanapund and Ausavapipit30) .
Furthermore, compared to their not disclosed HIV status, children that disclosed they were HIV positive had a nearly two-fold higher risk of developing SAM [AHR 1⋅9 (95 % CI 1⋅4, 3⋅39)]. This might be due to the effect of disclosing HIV status for children ≤15 years. Disclosure of a child's HIV status to the child has value in terms of positive health outcomes such as better adherence and slower disease progression. However, it has negative health consequences on increased psychiatric hospitalisation(Reference Odiachi31), and prolonged loss of appetites with a marked weight reduction loss(24), all of these further push children towards nutritional impoverishment.
Consistent with previous findings at worksite hospital(Reference Oumer, Kubsa and Mekonnen1), south wall Hospital(Reference Marie, Weldemariam and Dagnew32), Pawe Hospital(Reference Molla, Kebede and Kebede25), Deber Tabor Hospital(Reference Chanie, Legas and Zewude7), Western Kenya(Reference Willumsen6) and Burkina Faso(Reference Poda, Hsu and Chao2); the risk of developing SAM after ART for seropositive children with Hgb ≤10 mg/dl were nearly two times increased as compared with counter group [AHR 1⋅8 (95 % CI 1⋅16, 2⋅9)]. A consistent report on earlier research suggested that HAART regimens containing zidovudine (ZDV) are a substantial contributor to severe anaemia, particularly in children with HIV(Reference Beletew, Mengesha and Ahmed33). Reduced synthesis of red blood cells (RBCs) and increased RBC oxidation are two ways this is shown with this HAART regimen and malnutrition will therefore inevitably increase along with the phase of HIV disease.
Limitations
The present study had inherent limitations resulting from its retrospective study design; however, missing significant variables like caregivers’ household economic assets and lack the educational status in the recoded files might bias final interpretations.
Conclusion
Significant predictors of acute malnutrition were having a CD4 count below the threshold, children who had previously reported their HIV status, and having haemoglobin <10 mg/dl. To ensure better health outcomes, healthcare practitioners should improve earlier nutritional screening and consistent counselling at each session of care.
Acknowledgements
We would like also to thank the data collectors and supervisors for their unreserved cooperation during data collection.
F. K.: Conceptualisation, Data curation, Formal analysis, Funding acquisition, Methodology and Investigation, Project administration, Resources, Software and Writing – review and editing; T. K.: Funding acquisition, Methodology and Investigation, Supervision, Validation, Visualisation, Writing the original draft.
We have not received any financial support for this paper preparation.
The authors declare that there is no competing interest.
All methods were performed following the relevant guidelines and regulations of the Helsinki Declaration. The Ethical Review Board of Woldia University, College of Health Sciences, Research and Community Service, Technology Transformation, and University-Industry Linkage office Ethically Cleared With Refill Number (RCS, TT and UIL PH;0015/2015) with issue date on June 09/2022 E.C. Furthermore, offices of RCS, TT and UIL PH has waived the patient written consent because the authors had no physical contact with either of the caregiver or case children and the data were collected from medical charts alone from files recorded retrospectively.
There is no consent of the study for this publication.
Open access
