In the initial search, we found a total of 1847 records from the electronic search database of Midline/PubMed, Science direct, Hinari, Google, and Google scholar. After removing duplication 1028 records remained. After reviewing their titles and abstracts, we were excluded 992 records due to these articles were unrelated to our objective. Then after assessing 36 full articles 10 articles were further excluded due to differences in the study population and unspecified outcome. Finally, 26 studies were included in this systematic review and meta-analysis (figure 1).
Characteristics of the included articles
This meta-analysis included 26 different studies covering a total of 13,212 children aged 0 to 18 years. The studies were conducted from 2010 to 2020 among more than 14 sub-Saharan Africa countries [7-9, 19, 28-49] (Table 1). All of the studies included in this review were observational studies conducted in a health facility with the sample size ranging from 28 to 3195 participants as reported from a study in South Africa  and West Africa  respectively. All included studies were used WAZ, WHZ, and HAZ below -2 Z- score (WHO standard) to diagnose under-weight, wasting, and stunting respectively.
The highest prevalence of stunting was reported from a study in Cameron 77.0% , and the least was from a study in Mail 20.0% . Similarly, the highest prevalence of underweight was reported from a study in Nigeria 58.6% , and the minimum was from a study in Tanzania 6.8% . The highest (52.0%) and the least (5.8%) prevalence of wasting were also reported from studies conducted in Senegalese  and Ethiopia  respectively.
To estimate the prevalence of stunting, 26 studies were included in the analysis; the overall pooled prevalence of stunting was 51.4% (95% CI: 46.3, 56.4) (Figure 2). Similarly to estimate the prevalence of underweight, 24 studies were included in the analysis, the overall pooled prevalence of under-weight was 39.0% (95% CI: 33.7, 44.3) (figure 3). Twenty-five studies were also included in the analysis to estimate the prevalence of wasting; the overall pooled prevalence of wasting was 24.5% (95% CI: 19.2, 29.8) (figure 4). High heterogeneity was observed between studies on the prevalence estimate of stunting as evidenced by (I2= 96.9% and p=000). The heterogeneity of the prevalence estimates on underweight was also high (I2 = 97.0% and p=0.0000). Similarly, there was also high heterogeneity of the prevalence estimate on wasting as evidenced by I2= 97.1% and P-value=0.000). Publication bias was checked using the Eggers test and its result showed that there is no significant publication bias as evidenced by p = 0.590, 0.206, and 0.197 for stunting, under-weight, and wasting respectively. We also observed the symmetrical distribution of the funnel plot indicating the absence of publication bias (figure 5& figure 6).
We also done subgroup analysis by the country having more than two studies conducted. According to the result, the pooled prevalence estimate of stunting was highest in Cameroon, 65.6% (95% CI: 52.8, 78.3), I2 = 86.7% and the least was in Nigeria, 46.0% (95% CI: 32.9, 59.0), I2= 89.2. The highest pooled prevalence estimate of under-weight in sub-group analysis was in West Africa, 55.5% (95% CI: 53.8, 57.2), I2 = 0.0% and the least was in Tanzania, 24.1% (95% CI: 10.9, 37.3), I2 = 97.9 %. Similarly, the highest pooled prevalence of wasting in sub-group analysis was in West Africa studies, 39.5% (95% CI: 37.8, 41.2), I2 = 0.0% and lowest was in Tanzania, 15.1% (95% CI: 0.52, 29.62), I2 =98.3%.
Factors associated with under-nutrition of HIV infected children
Factors associated with stunting
During the review of primary articles, we have identified numerous factors associated with stunting in the primary study. Variables reported as a significant association with stunting in at least three primary studies were included in this metal analysis. Accordingly, household food insecurity was found to have a significant association with stunting among HIV infected children.
Household food insecurity:
Household food insecurity was reported as factors associated with stunting among three primary studies included in this review [7, 32, 45]. A total of 1180 children were included to analyze the association between household food insecurity and stunting among HIV infected children. The pooled odds ratio showed that children from food-insecure households were 5.50 times more likely to have stunting as compared with their counterparts [OR= 5.50 (95% CI 3.36, 8.98] (Figure. 7).
Factors associated with under-weight
To identify factors associated with stunting, we reviewed more than 7 primary studies and identified numerous factors for the occurrence of under-weight in HIV infected children in the primary study. Variables reported as a significant association with under-weight in three primary studies were included in this meta-analysis. Accordingly, low family income, feeding frequency, and caretakers attending dietary counseling were significantly associated with the under-weight.
Family economic status
Family economic status was identified as a factor associated with underweight among three primary articles included in this review [28, 30, 45]. A total of 940 participants were included to analyze the association between monthly family income and under-weight among HIV infected children. The odds of developing under-weight among HIV infected children who have low family economic were 6.28 times more likely to be underweight as compared with their counterparts [OR= 5.25 (95% CI: 2.52, 10.92)] (Figure. 8).
Feeding frequency was identified as a factor associated with under-weight among HIV infected children in three primary studies included in the meta-analysis [7, 28, 32] with a total of 1, 381 study participants. The odds of under-weight among HIV infected children who feed 4 times and more per 24 hours were 67.8 % less likely to become under-weight than children feeding less than 4 times per 24 hours [OR= 0.32 (95% CI 0.172, 0.605)] (figure 9).
Child caretaker dietary counseling
Caretaker dietary counseling was identified as a factor for under-weight HIV infected children in two primary studies [28, 30] with a total of 721 study participants. The pooled odds ratio showed that the risk of underweight among children whose caregivers taken dietary counseling sessions was significantly lower (64.3%) as compared to their counterpart [OR = 0.367 (95% CI: 0.182, 0.739)] (figure 9).
Factors associated with wasting
In this review, we have found numerous factors associated with wasting reported from different primary studies. No variables were identified as factors for wasting in at least three primary studies. Therefore meta-analysis to identify the associated factors for wasting was not done, but some factors reported as significant association with wasting in at least two primary studies were summarized through the table as below (Table 2).