109 Risk prediction of implant loss following implant-based breast reconstruction to the original model and good agreement was observed in the calibration plot. Consequently, the risk prediction model for TE reconstructions can be used for this subgroup of patients. The results of the DTI model showed a decreased mean ratio between the observed and predicted probability compared with the original and TE model. Therefore, based on the data used in this study, it is suggested that this DTI model is not superior to the original risk prediction model and could not be accurately validated. According to the literature, the incidence of implant loss varies between 1.8 and 16.9%.(2-7) With 6.7%, the incidence of implant loss in this study lies well within this range. Even though the accuracy of DBIR data is annually published, with a completeness of 93% or more for most variables,(9) an underestimation of the actual implant loss rate due to possible underreporting of explantations could still be present. The incidence is also depending on the definition of implant loss and not all studies have used the same definition. In the current study, implant loss was defined as the necessity of explantation or replacement (with the same or other implant or autologous tissue) due to postoperative complications related to wound healing problems. If the definition would be expanded to implant loss due to any reason (i.e. device rupture, capsular contracture, pain, malposition of the prosthesis), the incidence is expected to increase. The four risk factors included in the prediction model were BMI, active smoking status, previous radiotherapy and prepectoral placement. Previous studies have determined patient characteristics and comorbidities affecting the risk of complications after implant-based reconstructions: smoking and obesity are wellknown risk factors. In literature, reported cutoff points for obesity are a BMI ≥ 25 or 30 kg/m2.(13-18) However, it is more accurate to address BMI as a continuous variable, as some information will be lost when converting continuous to binary data. Previous radiotherapy has been described as risk factors for complications as well,(4) unlike prepectoral implant placement. Two previous meta-analysis have shown that either a prepectoral or subpectoral implant position is not associated with the occurrence of complications.(19, 20) This illustrates that most of the risk factors from the risk prediction model in the current study all have been previously described as risk factors for complications or implant loss after implant-based reconstruction. Furthermore, in this study, neither DTI or two-stage reconstructions were significant risk factors. Previous studies comparing DTI with two-stage reconstructions showed contradictory outcomes in complication rates.(21-24) A large study of Singh et al. reported similar results after one-stage and two-stage reconstructions 7
RkJQdWJsaXNoZXIy MTk4NDMw