Jordy van Sambeeck

Trochleoplasty procedures show complication rates similar to other patellar stabilizing procedures 7 105 loss of up to 20° ROM not requiring surgical treatment, increase in PF OA to grade 2 or 3 according to Iwano classification, superficial wound infection, anesthetic complications. Major complications included redislocation of the patella, return to OR due to increase in pain or recurrence of instability or any other cause, reduced ROM requiring arthrolysis, hardware removal because of pain or crepitus, progression to grade 4 PF OA, venous thrombotic event. Residual pain, swelling or crepitus not leading to OR were considered outcomes of the procedure and not complications. Data collection and analysis Data were extracted from the included articles by two reviewers (JvS, SvdG) and included: study ID, number of patients, number of knees, type of trochlear dysplasia, duration of symptoms, indication for surgery, mean patient age at surgery, patient sex, previous surgery on the involved knee, type of trochleoplasty performed, additional procedures performed, type and rate of complications and (if mentioned) time when complication occurred, length of follow-up and patients lost to follow-up. In studies that reported only percentages of complications and no absolute numbers, absolute numbers of complications were calculated based on the number of patients or surgical procedures reported. Subsequently a meta- analysis was performed whenever three or more studies per surgical technique that reported on a type of complication could be included. Despite anticipated heterogeneity, the individual study proportions were pooled. Pooled estimates of proportions with their corresponding 95% confidence intervals were calculated using Freeman-Tukey double arcsine transformation within a random effects model framework. Heterogeneity of combined study results was assessed by I 2 , and its connected chi-square test for heterogeneity, and the corresponding 95% confidence intervals were calculated. Restricted maximum likelihood was used to estimate the heterogeneity variance. Statistical analyses were performed using R version 3.4.0 (R Foundation for Statistical Computing, Vienna, Austria) with package ‘meta’. Quality assessment Quality assessment was not performed as the included articles were retrospective or prospective single-arm cohort studies and no validated scores for the methodological quality of these type of studies are available. Results The search strategy retrieved 1,848 unique records. Subsequent selection procedure resulted in 55 eligible articles of which 20 studies could be included in this systematic review (Fig. 1).