General methods
This protocol has been registered with the PROSPEROinternational prospective register of systematic reviews(registration number CRD42020153726) and will be reported adheringto the Preferred Reporting Items for Systematic Review andMeta-Analysis Protocols (PRISMA-P) 2015 statement (20). ThePRISMA-P checklist for this study is included as an additional file(supplementary file 1).
The final review will be reported following the PRISMA statement(21) and the Meta-Analysis of Observational Studies in Epidemiology(MOOSE) guidelines (22).
Study eligibility criteria
Studies will be selected according to participants, condition oroutcome(s) of interest, and study design.
Study designs
We will include randomized controlled trials (RCT) andcontrolled observational studies assessing the management ofSeymour fractures. We expect the majority of studies to beretrospective or prospective observational studies (cohort orcase-control) with a comparative group. Due to the anticipatedpaucity of comparative studies, we will also includenon-comparative studies e.g. case series, with intention of poolingoutcomes with single-arm data from comparative studies into ameta-analysis of proportions. Case series of less than 3 patients,case reports, study protocols, animal studies and review articleswill be excluded.
Participants
We will include studies examining the management of Seymourfractures in children and adolescents, where these are persons agedunder 18 years with open physeal plates. Studies reporting adults(aged over 18) or any persons with fused epiphyseal plates will beexcluded, as will patients without radiological confirmation of ajuxta-epiphyseal fracture. Closed injuries will also beexcluded.
Interventions and Comparators
We will broadly group patients based on two treatmentmodalities, namely debridement and prophylactic antibiotics and thetiming of these interventions. Debridement will be defined asformal washout, debridement of soft tissues and any form ofsplinting or fixation. This may take place in the emergencydepartment, clinic room or operating theatre.
Antibiotics may be in the oral or intravenous form and must beprescribed from time of presentation (<24h).
To establish the isolated and combined importance of earlydebridement and prophylactic antibiotics, we will compare thefollowing:
Early, late or no formal debridement +/- reduction and fixationas needed;
Early vs late or no antibiotics.
Early antibiotics will be defined as prophylactic antibioticsadministered from the time of presentation, as long as presentationwas within 24 hours of injury.
Early debridement eludes to debridement within 48 hours ofinjury.
In addition, these will be compared in combination as ‘complete’vs ‘incomplete’ treatment, where complete treatment is defined as acombination of early antibiotics and debridement +/- fracturereduction and fixation if appropriate. Incomplete treatment will bedefined as either a lack of debridement +/- reduction and splintingor fixation or a lack of early prophylactic antibiotics but thepresence of the other. No treatment pertains to a lack of bothcomponents.
For those who underwent debridement, a subgroup comparison willbe conducted, examining emergency department vs operating theatremanagement. Those patients who were delayed presenting to medicalcare will also be analysed as a separate group. If reported, usualcare such as pain relief, anaesthesia and immobilisation techniquewill be examined in addition.
Outcome measures
The primary outcomes will be the incidence of soft tissue orbony infection. Soft tissue infection is defined as those withcharacteristics signs of skin and subcutaneous tissue infection(erythema, warmth, purulence). Bony infection (osteomyelitis) isdefined as those that had signs of infection combined withradiographic evidence of focal bony lysis or cortical loss or aperiosteal reaction.
Secondary outcomes will include other adverse events suchmal-union, non-union, need for re-operation, physeal disturbance,nail dystrophy/atrophy (all as defined by the study inquestion).
Mal-union, non-union will be assessed up to one year; nailgrowth and physeal disturbance will be assessed with a minimumfollow up of 3-months post injury.
Setting
Studies performed in the hospital and emergency departmentsetting will be included. Studies performed in a primary caresetting will be excluded.
Language
No limitations will be imposed on language.
Information sources
The primary source of literature will be a structured search ofthe following major electronic databases from inception to April2020: MEDLINE (Ovid) SP; EMBASE (Ovid SP); CINAHL (Cumulative Indexto Nursing and Allied Health Literature) and the Cochrane Library(Cochrane Database of Systematic Reviews, Cochrane Central Registerof Controlled Trials [CENTRAL], Cochrane Methodology Register), incollaboration with a medical research librarian. PROSPERO will besearched for ongoing or recently completed systematic reviews.
The secondary source of potentially relevant material will be asearch of the grey or difficult to locate literature, includingOpen Grey and dissertation databases (e.g. Open Access Theses andDissertations). We will hand-search and screen the reference listsof included studies, relevant reviews, national clinical practiceguidelines or other relevant documents to identify cited articlesnot already in our list of included studies. Content experts andauthors who are prolific in the field will be contacted. Theliterature searches will be designed and conducted by the reviewteam which includes two experienced health informationspecialists.
Search strategy
The search strategy used will include a range of text words aswell as Medical Subject Headings (MeSH) terms related to 'Seymourfractures' and 'juxta-epiphyseal fractures.' The draft searchstrategy for MEDLINE is presented in supplementary file 2. Thesesearch terms will be adapted for use with other bibliographicdatabases.
No restrictions will be placed on the timing of publication. Thesearch will be performed in English and translations will be soughtfor articles published in other languages. No restriction will beplaced on publication status (i.e. unpublished studies will beincluded).
Selection of studies
Once the text and MeSH searches have been combined, duplicateswill be removed using EndNote (Clarivate Analytics, Boston, MA,USA). Citations will also be managed using this software.
The collated reference list of studies meeting the inclusioncriteria will be searched to identify additional relevant studies.Two independent researchers (A.K. and G.N.) will screen titles andabstracts for eligibility against a pre-defined list of inclusionand exclusion criteria. This process will be carried out usingRayyan (23), a bespoke web and mobile app for systematic reviews.At this stage, any reference deemed eligible for inclusion byeither reviewer will be included. Two reviewers (A.K. and G.N.)will then screen the full text of potentially relevant articles foreligibility. Reasons for exclusion will be recorded whereapplicable.
Where disparity occurs between references, consensus will besought, and all remaining articles will be read in full before adecision on inclusion is made. If disagreements remain between thescreening authors, the texts will be screened by a third author(L.C.).
The bibliography of the final included studies will be screenedto check for additional publications that may be relevant. Thesearch results, including abstracts, full-text articles, and recordof the reviewer’s decisions will be recorded first in Rayyan (23)and then in a pre-defined data collection sheet in Microsoft Excel(Microsoft Corporation, 2018).
Data extraction and management
Two reviewers (A.K. and G.N.) will collect data independentlyand in duplicate using a pre-defined electronic data extractionform. The data collection process will be in keeping with theCochrane Handbook of Systematic Reviews of interventions (24).
The following data will be extracted: first authors; year ofpublication; study design; inclusion & exclusion criteria; numberof patients; method of diagnosis; age; sex; relevant medicalhistory; mechanism; time since injury; digit involved; type ofintervention, (debridement, fixation, anaesthesia); duration ofintervention; specialty performing intervention; location ofintervention (theatre or A&E); antibiotic regimen; time tofirst antibiotic; analgesic regimen and primary and secondaryoutcomes.
If authors report on adult patients, these will not be includedin the analysis, if the data is clearly distinguishable.
In addition, the statistical analysis models and outcomemeasures used will be noted. Divergences will be resolved byconsensus or with a third reviewer (L.C.) if needed.
Dealing with missing data
Where relevant, study authors will be contacted if data relevantto the systematic review are missing in the study report. Whereauthors fail to reply after first contact or after one reminder,the missing data will be acknowledged, and we will proceed with theanalyses.
Assessment of risk of bias of includedstudies
We expect that most included studies will be observationalrather than randomised studies, of which some will be uncontrolled(e.g. case series of one surgeon’s outcomes). Each study designwill be assessed using a relevant tool.
If there are any randomised trials, they will be assessed usingthe Cochrane Risk of Bias 2 (RoB 2)(25) tool, which focuses onaspects of trial design, conduct and reporting.
Non-randomised comparative studies (e.g. cohort and case controlstudies) will be assessed using ROBINS-I tool (26), which holdsstudies to standard against a hypothetical pragmatic randomisedtrial. It covers seven potential domains of bias.
Uncontrolled studies (e.g. case series) will be assessed using atool which has been specifically for this purpose (27). It isformed from an adaptation of previous criteria from Pierson (28),Bradford Hills (29) and Newcastle Ottawa scale (30) modificationswhich assesses bias in four domains: selection, ascertainment,causality and reporting.
Disagreements between the review authors will be resolved byconsensus or with a third reviewer where necessary. A narrativesummary of the risk of bias of the included studies will beperformed, which will be supported by a figure and table showingthe results of the critical appraisal. The results of the risk ofbias tool will be used in a sensitivity analysis to ensure studiesjudged to be at ‘high’ risk of bias do not affect the robustness ofour results in any subsequent meta-analysis.
Data analysis and synthesis
To answer the review question of determining the optimalmanagement of Seymour fractures, the data from each paper will beused to build evidence tables providing an overall description ofincluded studies. The tables will contain data including studycharacteristics, context, population, outcomes and findings foreach included study. This will be accompanied by a narrativesynthesis of the data.
Clinical and methodological heterogeneity will be assessedacross each study in early vs late debridement, early vs lateantibiotics and complete vs incomplete treatment groups (31). Thiswill determine whether it may be feasible to perform ameta-analysis. If possible we will perform a random effectsmeta-analysis (31). We will present the results as a pooledestimate for each of the primary outcomes comparing early vs latedebridement, early vs late antibiotics and complete vs incompletemanagement as relative risk and 95% confidence intervals. Theresults of this will be presented in a forest plot. If feasible andappropriate, studies that do not contribute comparative data willbe pooled with single-arm data from comparative studies todetermine the overall incidence in the treatment group of relevanceby meta-analysis of proportions. A forest plot may be produced toshow the pooled effect of findings.
Heterogeneity will be assessed visually by examining the overlapof confidence intervals in the forest plot. We will quantifystatistical heterogeneity by estimating the variance betweenstudies using the I2 statistic which examines thevariance between studies to produce a percentage score of between 0and 100% which will be interpreted as per the Cochrane handbook(24). Tau squared and chi-squared tests will also be appliedwhere
a P value of <0.05 is considered statisticallysignificant for heterogeneity.(31)
A summary of findings table will be created for the primaryoutcome measure. We will rate the overall quality of evidence ofthese outcomes using the Grading of Recommendations, Assessment,Development and Evaluation (GRADE) Working Group methodology (32).Each critical outcome’s quality of evidence is rated, taking intoconsideration five defined criteria (risk of bias and limitationsof design, consistency of analysed studies and their results,directness, precision, and publication bias) that may lead tograding down, and three criteria (large effect, dose-response, andopposing bias and confounders) that may lead to grading up (33,34).
Additional analyses
If it is not possible to combine the data in the above manner,then we will determine the crude incidence estimates of infection(number of infections/sample size) along with the 95% confidenceintervals associated with timing of debridement and prophylacticantibiotics for each study using a meta-analysis ofproportions.
If sufficient studies are identified and data points areavailable, potential sources of heterogeneity will be investigatedfurther by subgroup analyses according to baseline characteristicsand methodological covariates. We plan to conduct analyses bygender (male vs female), age (children vs adolescents) and risk ofbias (e.g., high vs moderate/low risk of bias). A subgroup analysiswill be carried out of patients who were delayed presenting tohospital, as a meta-analysis of proportion to ascertain theincidence of infection in this cohort. For all patients whounderwent debridement, a subgroup comparison will be conducted,examining emergency department vs operating theatre management.
Meta-bias
Publication bias will be investigated, and a funnel plotwill be generated for each meta-analysis containing 10 or morestudies. Small study effects (or "publication bias" across studies)will be assessed by inspection of the funnel plots for asymmetryand with Egger's test (35) where appropriate, with the resultsconsidered to indicate potential small study effects when P valuesare < 0.10. Depending on the number of included studies in thereview, we will undertake a sensitivity analysis to ensure therobustness of our results. We anticipate that the systematic reviewwill identify studies judged to be at serious risk of bias and wewill perform a sensitivity analysis where these are excluded.