Database search results
We retrieved 291 relevant articles from PubMed, Google Scholar, Science direct, Scopus, and other sources. Of these initial articles, there were about 236 non-duplicated articles. From the remaining articles, 204 articles were excluded after review of their titles and abstracts. Then, 32 potentially full text articles were assessed for eligibility based on the pre-set criteria and 22 articles were further excluded due to different reasons. Lastly, 10 articles met the eligibility criteria and included in the final meta-analysis to determine the prevalence and associated factors of treatment failure (Figure 1).
General descriptions of included studies
As described in Table1, the 10 included studies were both cohort and cross-sectional study design, and done from 2003 to 2021. Among the included studies, five studies were conducted in the Amhara region (19, 20, 25-27), whereas two in Oromia (21, 24), two in Addis Ababa (23, 28), and one in SNNP (22). In this meta-analysis study, 4,572 participants were involved to determine the pooled prevalence of treatment failure. The sample size of the studies fluctuated between 96 (21) and 1,186 (23). Regarding prevalence, the lowest prevalence (3.1%) of treatment failure was reported from a study done in Southern Nations, Nationalities, and Peoples Region (SNNPR) (22), whereas the highest prevalence (18.8%) was reported in study done at Fiche and Kuyu hospitals in Oromia Region (24). Concerning quality of the included studies, two studies (20, 26) were assessed based on JBI check list for cross-sectional studies, and the remaining eight studies (19, 21-25, 27, 28) based on JBI check list for cohort studies. According to this quality assessment criterion, all studies were included (Table 1).
Meta-analysis of treatment failure
Publication bias
Publication bias was checked using the Begg’s test which showed no statistically significant publication bias with p-value of 0.93. We also performed publication bias assessment for overall treatment failure using funnel plot (Figure 2).
Prevalence of HIV/AIDS treatment failure
The pooled prevalence of treatment failure among HIV-positive children in Ethiopia was found to be 12.34 (95%CI: 8.59, 16.10) (Figure 3). Based on clinical definition the pooled prevalence of treatment failure was 27.65 (95% CI: 1.23, 54.06) (Figure 4). Concerning immunological failure, the pooled prevalence was 17.97 (95% CI: 10.13, 25.82) (Figure 5). Furthermore, the pooled prevalence of virological treatment failure was 23.93 (95% CI: 3.21, 44.65) (Figure 6).
Heterogeneity and Sensitivity analysis
Heterogeneity test (I2) was 94.7%, p < 0.001 which shows that there is significant variation across the included original studies. Then, we calculated a random effect meta-analysis model to estimate the pooled prevalence of treatment failure. In the sensitivity analysis, there is no study away from the lower and upper limit of confidence interval.
Subgroup analysis
In this review, a multiple comparison was executed in the prevalence of treatment failure among HIV positive children using different factors. Thus, subgroup analysis was employed based on regions, study design, and sample size. Higher prevalence of treatment failure was observed in Addis Ababa with a prevalence of 15.92 (95% CI: 11.72, 20.12) followed by Oromia regions at 14.47 (95% CI: 9.79, 19.14) (Figure 7). On the other hand, the prevalence of treatment failure was lower in studies having a sample of size 628, 3.03 (95%CI: 1.69, 4.37) compared to those having a sample size 391, 18.41 (95% CI: 14.57, 22.26) (Figure 8). Furthermore, we performed subgroup analysis based on study design. Accordingly, the prevalence of treatment failure in cross-sectional study was 13.43 (95% CI: 10.76, 16.11) and in cohort study 12.14(95% CI: 7.83, 16.44) (Figure 9).
Associated factors of HIV/AIDS treatment failure
In this meta-analysis, the associated factors were categorized in to two thematic areas. These were: 1. Socio-demographic 2. Clinical and drug related factors.
Socio-demographic factors
Based on the finding of a study (20), male children were more likely (AOR= 3.15, 95% CI: 1.18, 8.39) to develop HIV treatment failure. From a single study, HIV positive children have not both parents as primary caretakers (AOR= 2.72, 95% CI: 1.05, 7.06], and negative serologic status of their caretakers (AOR= 2.69, 95% CI: 1.03, 7.03) were at high risk of treatment failure (28). Age of children below 5 years (AOR = 2.4, 95% CI: 1.0–5.7) (26) was also reported as a contributing factor for the occurrence of HIV treatment failure as compared to their counterparts. Likewise, HIV positive children with age below 3 years were at high risk (AHR=1.85, 95% CI: 1.24, 2.76) to develop treatment failure as compared to those with age between 5 and 15 years (23). On the other hand, the age of HIV positive children between 6 and 9 years was protective (AOR=0.26; 95% CI: 0.09, 0.72) from the occurrence of treatment failure compared to age between 10 and 15 years (24).
Clinical and drug related factors
The pooled odds ratio of this meta-analysis revealed that HIV positive children with history of opportunistic infections were 2.6 times more likely (AOR=2.64; 95% CI: 2.19, 4.31) to develop treatment failure as compared to their counterparts. The estimated pooled effects of advanced WHO clinical stage III/IV on HIV treatment failure was (AOR=1.66; 95% CI: 1.24, 3.21) as compared to those children categorized on WHO clinical stage I/II. Report from a single study showed that children who interrupt and restart their HIV treatment were positively (AOR=2.21, 95% CI: 1.09, 4.54) associated with treatment failure (26). HIV positive children who had height for age below or on 3rd percentile were at high risk (AHR=3.3, 95% CI: 1.0, 10.6) of treatment failure (23). Additionally, children who was not disclosed their HIV status were also at high risk (AHR= 4.4, 95% CI: 1.8, 11.3) to develop treatment failure (25). Original regimen change (AOR=9.22, 95% CI: 3.36, 25.03) (20) and ART drug substitution (AHR=1.7, 95% CI: 1.1, 2.7) (23) were reported significantly associated factors HIV treatment failure. The result of this meta-analysis revealed that adherence level was positively associated with HIV treatment failure. From this finding, the estimated pooled effect of poor ART adherence level on childhood HIV treatment failure was (AOR=2.53; 95% CI: 2.03, 4.97) as compared to good adherence level counterparts.