Gabapentin for phantom limb pain after amputation in pediatric oncology: a systematic review protocol

The aim of this review is to systematically appraise the evidence of gabapentin for PLP after amputation in pediatric oncology.

This study sought to answer the following questions:

We will systematically retrieve electronic databases (Cochrane Library, MEDLINE, EMBASE, Web of Science, CINAHL, PsychINFO, Scopus, WANGFANG, and Chinese Biomedical Literature Database) from the inception to the present without restrictions to publication status and language. The search strategy is built using keywords including “neoplasms,” “pain,” “cancer pain,” “phantom limb,” “pain intensity,” “neuropathic,” “chronic pain,” “gabapentin,” “anti-epileptic drug,” “anticonvulsant,” “neurontin,” “gralise,” “GABA analogs,” “random,” “allocation,” “placebo,” “sham,” “blind,” “control trial,” “case-control,” “case control,” “case-comparison,” “case-referent,” “clinical trial,” “observational study,” “study,” and “trial.” A detailed search strategy for Cochrane Library is presented in Table 1. Similar search strategies for other electronic databases will be adapted and applied. Translations will be performed when necessary in Chinese databases. At the same time, we will examine unpublished and ongoing work in clinical trial registry, conference proceedings, and reference lists of eligible studies. Two independent reviewers (X-R and B-JH) will carry out the whole process of systematic searches. Any disagreements will be solved by discussion with the help of a third reviewer (Y-LH). A consistent decision will be reached after discussion.

Records retrieved from the literature sources will be imported to a reference management, and duplicates will be eliminated. Two independent authors (J-S and Z-MM) will carry out a broad scan of study titles and abstracts; and unrelated studies will be removed. Then, full-paper of potential eligible studies will be checked to make sure whether they fulfill all eligibility criteria. Any divergences between two authors will be solved by consulting a third author (X-R), and a final decision will be reached after the discussion. Reasons for all excluded studies at different stages will be recorded. The results of study selection process will be summarized in a PRISMA flowchart.

Data from selected RCTs and CCSs will be transferred from their original presentation to a standard form with each included study receiving a reference code. If necessary, we will also extract indirect data from figures and charts.

For all included RCTs and CCSs, two authors (J-S and Z-MM) will independently obtain the data from eligible trials according to the predefined data extraction sheet developed specifically for this study. Any opposite views regarding the data extraction will be resolved by discussion with the help of another author (X-R), and we will make a final consistent decision. The extracted information consists of study characteristics (such as country, title, language, publication time, and funding source), patient characteristics (such as age, gender, and diagnostic criteria), study design (such as randomization details, blind, and lost to follow-up), intervention and control details (such as treatment types, duration, and number and length of sessions), and outcomes, safety, and other related information (such as confounding factors).

The study quality of each eligible study will be examined by two independent authors (X-R and B-JH) using Cochrane Collaboration’s Risk of Bias Tool for RCTs [36] and Newcastle-Ottawa Scale for CCSs [37], with predetermined criteria. RCTs will be assessed on seven aspects, and each one is further rated as high, unclear, or low risk of bias [36]. CCSs will be appraised on three broad perspectives with eight specific items [37]. Any doubt between two authors will be answered with the help of a third author (Y-LH) through discussion, and a final consistent decision will be made.

We will contact corresponding authors of primary studies to obtain any missing or insufficient or unclear data by email or fax. If we can not achieve those data, we will analyze available data using Intention-To-Treat approach and will discuss its potential affects in the manuscript.

If there is a minimum of 10 trials in any meta-analysis, we will examine reporting bias using a funnel plot [38], and symmetry of the funnel plot will be performed using Egger’s regression test [39].

We will use RevMan 5.3 software to synthesize and analyze all outcome data. We plan to carry out separate analysis based on types of study, including RCTs and CCSs. We will calculate the treatment effect of dichotomous data using risk ratio and 95% confidence intervals (CIs) and that of continuous data using mean difference (MD) or standardized MD and 95% CIs. We will examine heterogeneity using I² statistic, and we will undertake statistical pooling on groups of trials which are considered to be sufficiently similar [40, 41]. Where heterogeneity is low or minor (I² ≤ 25%), we will utilize a fixed-effect model to pool the data; if heterogeneity is moderate (25% < I² ≤ 75%), we will apply a random-effect model to synthesize the data, and if heterogeneity is obvious (I² > 75%), we will not pool the data [40]. Meta-analysis will be carried out based on the sufficient homogeneity regarding on participant characteristics, types of intervention and outcome, and comparability between methods and ability to aggregate data. A narrative synthesis of eligible trials will be performed if the extracted data is too diverse to fulfill the threshold for meta-analytic approach. We will build a “summary of findings” table for the outcomes, and we will appraise evidence quality of primary outcome using Grading of Recommendations Assessment, Development and Evaluation [42, 43], which covers five aspects of risk of bias, imprecision, consistency of effect, indirectness, and publication bias.

We will carry out subgroup analysis to test the sources of significant heterogeneity based on the following:

We will investigate the sensitivity analysis to test the stability and robustness of study findings based on the sample size of included trials, and study quality.

We will publish this study on a peer-reviewed journal or a conference meeting.