A systematic search of the Ovid database (including MEDLINE, PreMEDLINE, EBM Reviews, Cochrane Database, Cancerlit, CINAHL, and EMBASE) was performed on December 4, 2016 to identify studies reporting clinical and sport-specific outcome after apophyseal fractures of the pelvis. This study was performed in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis) guidelines  The search strategy comprised the Boolean operators (AND; OR) that combined the following terms in the title and abstract fields: apophysis, apophyseal, fracture, injury, avulsion, avulsion*, pelvis and pelvic.
No limits were set on the date of publication. The inclusion criteria were specified in advance: Clinical studies of all kind and all levels of evidence published in English or German language, reporting clinical outcome after operative or conservative treatment of apophyseal avulsion fractures of the pelvis in the adolescent athlete, published online or in print in a peer-reviewed journal, including the results of at least 5 or more patients (till the age of 18 years, both sexes). Exclusion criteria were as follows: None-English or None-German-language studies, articles that were off-topic, study collectives with concomitant injuries (as pelvic ring fractures) or procedures, in-vitro or animal studies, radiologic or diagnostic studies, epidemiologic studies, and other types of articles such as technical notes, or narrative reviews.
Two reviewers (HE and LH) independently screened all articles for relevance by title and abstract according to the defined in- and exclusion criteria. If no abstract was available, the full-text article was obtained to assess the relevance of the study. The full-text of all articles not excluded during the initial screening process was obtained and reviewed by the same two reviewers for possible inclusion in the study. Any disagreement on article eligibility was resolved through discussion. To ensure that all studies were recorded, references within included studies and all review studies were cross-referenced for inclusion if missed by the initial search.
All studies that met the inclusion criteria were abstracted regarding study characteristics, patient demographics, location and cause of injury, treatment protocols, and surgical technique. Data was collected by one reviewer (HE) in an Excel extraction form and verified by a second reviewer (LH). Any disagreement that arose was resolved by consensus between both reviewers. Study characteristics of interest included author names, year of publication, number of patients at final follow-up, length of follow-up, and the quality assessment by the Coleman Methology Score. Patient demographics included sex and the mean age of the patients. Outcomes of interest were the clinical outcome at follow-up, the complication rate (in particular non-union, heterotrophic ossifications, neurological sequelae, wound infections, persistent pain or functional restrictions like limited range of motion) and the proportion of patients returning to sports. Excellent outcome was specified as a reported excellent outcome and the absence of complications at follow-up. Return to sport was defined as the rate of patients returning to sports to the preinjury sport-level at follow-up.
The methodological quality of the included studies was evaluated with the modified Coleman Methodology Score. This score is widely used in systematic reviews and meta-analyses to assess the methodology of clinical studies by using 10 specific quantitative and qualitative criteria: study size, mean follow-up, number of surgical procedures, type of study, diagnostic certainty, description of surgical procedure, postoperative rehabilitation, outcome measures, outcome assessment, and selection process [14–16]. The final score ranges from 0 to 100, with a score of 100 indicating the highest study quality . If the score is greater than 85, the study is designated as excellent, between 70 and 84 good, from 50 to 69 moderate, and below 50 poor . In addition, all included studies were assessed according to the level of evidence defined by the Oxford Centre for Evidence Based Medicine .
The data was analyzed using established statistical software (Excel Version 14.3.9, IBM SPSS Statistics for Macintosh version 22.0 and R-Project Version 3.2.4, package ‘metaprop’, by G. Schwarzer, University of Freiburg). Descriptive statistics and calculation of frequency weighted means and standard deviations were used to report study characteristics, patient demographics, and outcomes. Pairwise comparisons were evaluated using the chi-squared test. A p-value of <0.05 was considered significant for all comparisons. Meta-analyses were performed to calculate the overall pooled rate of clinical outcome, complications, and return to sports after avulsion fractures of the pelvis. The random effects model was used for the meta-analysis. The result of the meta-analysis was presented as a forrest plot and heterogeneity was calculated according to the method of Higgins et al. and is expressed as I2. The value of I2 ranges from 0% expressing complete consistency to 100% expressing complete inconsistency of the data .