Eligibility criteria

We will include any randomised-controlled trial (RCT) that compares the outcomes of antibiotic therapy, ERAT and appendectomy in the management of patients with UAA.

UAA will be confirmed via CT, defined as an appendix with a diameter greater than 6 mm, accompanied by wall thickening, periappendiceal oedema and/or a small fluid collection, without evidence of perforation, abscess or appendicoliths.

Categorisation of studies

To facilitate decision-making and improve clarity, we will categorise intervention arms into antibiotic therapy, ERAT, open appendectomy and laparoscopic appendectomy. This classification system allows for meaningful comparisons and more effective guidance in clinical decision-making.

Data sources and search strategy

We will conduct comprehensive searches in PubMed, Web of Science, Embase, the China National Knowledge Infrastructure and the Cochrane Library to identify relevant RCTs. The search will cover the period from the inception of these databases until September 2024. To ensure a thorough search, we will use a combination of Medical Subject Headings and free-text terms. These terms will fall into three main categories: clinical conditions (eg, acute appendicitis, UAA), interventions (eg, antibiotic therapy, ERAT, appendectomy) and study design (eg, RCTs, controlled trials). Search terms will be customised for each database. Additionally, reference lists of included studies will be manually searched to identify any eligible studies that may have been missed during the initial database search. The detailed search strategies for all databases are provided in the online supplemental file.

Study selection

The search results from each database will be imported into EndNote X9. After removing duplicates, two independent reviewers (XW and XL) will screen the titles and abstracts of the identified studies to exclude irrelevant articles. The remaining studies will undergo a full-text review based on the predefined inclusion criteria. Any disagreements between reviewers regarding study selection will be resolved through discussion. Detailed records will be kept for excluded studies, including the reasons for exclusion. Figure 1 presents the PRISMA flowchart, which visually outlines the study selection process and provides an overview of the screening procedures.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Selection of studies for inclusion.

Data extraction

Two independent reviewers (LL and YL) will use a predesigned form to extract relevant data from the included studies. Extracted data will include details about the publication (eg, first author, year of publication, country), population characteristics (eg, age, sex, sample size), interventions and outcomes of interest. If any data are missing, we will attempt to contact the original study authors to obtain the necessary information. Any discrepancies or disagreements during data extraction will be resolved through discussion to reach consensus.

Outcome measurements

The primary outcome in this study will be the treatment success rate, defined as the resolution of abdominal pain, absence of complications and improvement in inflammatory markers within 1 month after hospital discharge. Secondary outcomes will include major complications, hospital costs, length of hospital stay and recurrence of appendicitis. Major complications are defined as appendiceal perforation, persistent abdominal pain, surgical site infections, incisional hernias and fever. Recurrent appendicitis refers to recurrence within 1 year after treatment. We may be unable to conduct a formal meta-analysis on hospital costs due to variations in currency and cost analyses across trials.

Risk-of-bias assessment

We will use the Cochrane Risk of Bias 2 tool to assess the potential biases in the included studies.13 This tool examines possible biases related to the randomisation process, deviations from intended interventions, missing outcome data, outcome measurements and selection of reported results. Two independent reviewers (XW and XL) will assess the risk of bias. Based on the responses, studies will be categorised as having a ‘low risk of bias’, ‘some concerns’ or a ‘high risk of bias’. Any disagreements between reviewers will be resolved through discussion. A thorough assessment of bias is critical to ensuring the reliability and validity of the included studies, thereby enhancing the credibility of the evidence in the analysis.

Data synthesis

The characteristics of the included trials will be synthesised and presented in tabular form, providing a comprehensive overview of key elements such as participant demographics, study design, sample size and interventions. To compare the efficacy of different treatments for UAA, we will conduct a Bayesian NMA using the ‘netmeta’ package within the R software framework. This package offers the tools and algorithms needed for a robust NMA, allowing us to synthesise evidence from diverse sources. Our analysis will include both direct and indirect comparisons of the interventions, giving a nuanced understanding of their relative effectiveness.

The results will be visually represented in a network diagram, showing the relationships between different interventions and their outcomes. Each node will represent an intervention, with node size reflecting the number of patients receiving that treatment. Pairwise comparisons will be shown as edges between nodes, with the thickness of the edges corresponding to the weight of each comparison.

We will also generate a contribution matrix to assess the impact of individual comparisons and the influence of both direct and indirect evidence on the overall summary effects. This matrix will provide insights into the contributions of each comparison and the strength of evidence derived from both types of sources. If quantitative synthesis is not feasible or appropriate, we will conduct a narrative synthesis, offering a descriptive summary and interpretation of the findings to facilitate a qualitative understanding of the evidence from the trials.

Assessment of transitivity and meta-biases

Based on our initial findings, we anticipate that the interventions for UAA will meet the transitivity assumption, which ensures compatibility for joint randomisation. Consequently, comparisons between interventions within the network will be informed by a combination of direct evidence (eg, pairwise RCTs) and indirect evidence (eg, inferring the effect of B–C from comparisons of A–B and A–C). We will use both direct and indirect evidence, or a combination of the two, to guide our analysis and assess the relative effectiveness of the interventions.

Network meta-analysis

Assuming homogeneity in the distribution of effect modifiers across studies, we will conduct a frequentist NMA, accounting for the proposed closed network geometry. This approach will use all available evidence within the network to compute pairwise effect sizes, evaluating relative treatment effects. When direct comparisons between two interventions are unavailable, indirect comparisons will be made using a common comparator. To rank the mixed effect sizes (combining both direct and indirect evidence) and present the 95% CIs for all treatment combinations, we will use graphical tools such as network forest plots, interval plots and league tables. These visual aids will offer a comprehensive overview of treatment effects and will assist in interpreting and comparing the interventions within the network.

Detection of heterogeneity

Our meta-analysis aims to combine the findings of studies that show sufficient homogeneity in terms of participants, interventions and outcomes. To assess the level of heterogeneity across studies, we will use the χ² test and the inconsistency index (I²).14 15 If the I² value exceeds 50%, indicating substantial heterogeneity, we will use a random-effects model to account for variability among the studies. If the I² value is 50% or lower, indicating low-to-moderate heterogeneity, we will use a fixed-effects model to combine effect sizes.

Additional analyses

To explore potential effect modifiers and their impact on the primary outcome, we will conduct network meta-regression using a random-effects model, provided sufficient data and information are available across the included studies. Potential effect modifiers may include the average age of participants, sex distribution, year of publication and study quality. Subgroup analyses will be performed based on factors such as age, sex, region, study quality and publication date to identify potential sources of heterogeneity and assess their influence on the primary outcome. To ensure the robustness of our findings, we will conduct sensitivity analyses by systematically excluding one study at a time. This approach will help evaluate the impact of individual studies on the overall results, providing insights into the stability and reliability of our findings. Potential publication bias will be assessed using a comparison-adjusted funnel plot,16 which will help detect any asymmetry in study effects, indicating the presence of publication bias.

Credibility of the evidence

We will use the Confidence in Network Meta-Analysis approach to assess the credibility of the evidence derived from our NMA.17 Two independent reviewers (XW and XL) will evaluate six domains: within-study bias, cross-study bias, indirectness, imprecision, heterogeneity and incoherence. Any disagreements between reviewers will be resolved through discussion. The level of confidence in our findings will be classified as ‘high’, ‘moderate’, ‘low’ or ‘very low’, based on the quality and reliability of the evidence from the NMA.