Protocol and registration
The protocol of this systematic review and meta-analysis was found in the PROSPERO which can be accessed in the web address of https://www.crd.york.ac.uk/PROSPERO/#myprospero. Moreover, the protocol has been registered in the international prospective systematic review and meta-analysis (PROSPERO) with a registration number of CRD42018110570. For reporting of the findings, the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guideline has been utilized (Additional file 1).
Study area: All included studies have been conducted in Ethiopia.
Types of studies: In this systematic review and meta-analysis observational studies including case-control, cross sectional and cohort studies that report childhood TB treatment outcome and its associated factors were included. However, we did get case-control studies.
Type of participants: Children less than 15 years those who had tuberculosis and receiving tuberculosis treatment had been enrolled in the review.
Types of intervention: In this review, we compared the TB treatment success rate of HIV Co- infection compared to the counterparts.
Types of outcome measures: In this study, TB treatment success rate defined as both treatment completed and cured cases.
Language: Articles reported in English language were included
We excluded papers that were not fully accessible, after at least two email contact with the primary author of the paper. Moreover, Editorials, trials, conference papers, qualitative studies, and reviews were excluded.
Both published and unpublished research reports that revealed the childhood TB treatment outcomes and/or its association with HIV co-infection in Ethiopia were included. We searched all available articles using PubMed, Google scholar and a web of science. Additionally, reference lists of included studies and institutional research repositories were explored. Searching was limited to studies conducted in Ethiopia and published in English language.
We used the following searching terms to search all databases: “tuberculosis treatment”, “childhood tuberculosis treatment”, “childhood tuberculosis treatment outcome”, “tuberculosis”, “cured”, “completed”, “treatment completed”, “ relapse”, “ treatment failure”, “died”, “ successful TB treatment outcome”, “ unsuccessful TB treatment outcome”, “childhood tuberculosis treatment”, “outcome of tuberculosis treatment”, “factors”, “risk factors”, “ associated factors”, “predictors”, “ HIV co infection”, and “Ethiopia”. The searching string was developed using the Boolean operators ‘AND’ and ‘OR’. PubMed searching was done using this searching strategy (Additional file 2). The last date of PubMed searching was on August, 23, 2019.
At the beginning, we retrieved all available studies from electronic databases. Those identified studies were imported to the Endnote citation manager and duplicates removed carefully. Second, two independent authors (GMB and CAW) screened and assessed the title and abstracts of studies that imported in the endnote followed by reviewing of the full texts. Discrepancies among reviewers have been solved through discussion and consensus. In addition, disagreements were solved by communicating the authors of the primary studies.
Two authors (GMB and CAW) assessed the quality of included studies independently. The Joana Briggs Institute (JBI) critical appraisal tool has been utilized to critically appraise the quality to the studies. JBI critical appraisal checklist for cohort and cross-sectional studies was employed (Additional 3). Discrepancies between the authors have been solved through discussion, repeating the procedure and reaching upon the consensus.
Data collection process and extraction
After we have developed a data extraction sheet, the first author (GMB) had extracted the following data from the included studies, first of author of the study, study area, region, population, design, sample size, proportion of treatment outcome including cured, treatment completed, treatment failure, died, and relapse, odds ratio of HIV co infection, proportion of treatment success rate, log p, and SE log p and the second author (CAW) checked the extracted data. Any disagreement between authors has been solved by discussion.
Treatment completed: A patient with TB who completed treatment without evidence of failure, but with no record of sputum smear or culture results, in the last month of treatment.
Cured: A patient with bacteriologically confirmed pulmonary TB at the beginning of treatment who was smear or culture-negative in the last month of treatment
Treatment failure: A TB patient whose sputum smear or culture is positive at 5 month or later during treatment.
Died: A TB patient who died from any cause during treatment.
Successful treatment outcome: If PTB patients were cured (i.e., negative smear microscopy at the end of treatment and on at least one previous follow-up test) or completed treatment with resolution of symptoms.
Unsuccessful TB treatment outcome: If treatment of PTB/EPTB/ patients resulted in treatment failure (i.e., remaining smear-positive after 5 months of treatment), default (i.e., patients who interrupted their treatment for two consecutive months or more after registration), or death.
Synthesis of results
Before meta- analysis of the research report, the proportion of treatment outcome and OR of HIV co infection had been transformed to logarism in excel spreadsheet. Then, the excel data exported to STATA version 11 for further analysis. To determine the pooled effect of TB treatment outcome a weighted inverse random effect model was employed. The descriptive data were presented using a table to describe the characteristics of each primary study. Besides, the point prevalence of each study as well as the overall prevalence were described using a forest plot graph. The forest plot was interpreted as follows: the horizontal line shows the 95% CI and the black box represents the Wight of each study. Moreover, an explanatory data analysis using I2 test was conducted to assess the random variations between each primary study. In this study, heterogeneity was interpreted as an I2 value = 0% no heterogeneity, 25% = low, 50% = moderate, and 75% = high (11). Based on the above testes, the primary studies included in this meta-analysis exhibited a significant random variation (I2 = with Eggers regression test p-value < 0.001), which forced us to use a random effects meta-analysis model to compute pooled effect. Publication bias was assessed by funnel plot and Egger’s test. Statistically significant publication bias was declared at p-value less than 0.05.