85 Validation of a multicenter risk model for implant loss following implant-based reconstruction with permanent implant because of flap problems, infection, skin necrosis, hematoma or seroma. The following indications for explantation or revision were not considered as implant loss due to a wound healing related complication: dissatisfaction with size, asymmetry, breast pain, autoimmune syndrome induced by adjuvants (ASIA), suspected anaplastic large cell lymphoma (ALCL), newly diagnosed breast cancer, device malposition, scarring, capsular contracture or device rupture. The following incision sites available in the DBIR were considered as nipple preserving procedures: mastectomy scar (nipple sparing), inframammary, periareolar and axillary incisions. Mastectomy scar (general) was interpreted as not nipple sparing. Statistical analysis Categorical variables were depicted as frequencies with percentages and continuous variables are presented as mean with standard deviations (SD) or median with interquartile range (IQR) based on the distribution. Differences in baseline characteristics between groups were tested with unpaired T test, MannWhitney U test or chi-square tests. To assess the validity of the local risk model, the beta regression coefficients listed in Table 1 were used to calculate the predicted probability of implant loss in the validation cohort. For each predicted probability group the observed probability, with corresponding SD, was calculated. This was visualized in a calibration plot, with predicted probability on the y-axis and observed probability on the x-axis. Finally, univariate logistic regression was performed to determine the association between risk factors and implant loss in the current cohort, providing odds ratios (OR) with 95% confidence intervals (CI) and P-values. IBM SPSS statistics (version 26) was used for statistical analysis, and a P-value <0.05 was considered statistically significant. RESULTS Validation study population A total of 9373 implant-based breast reconstructions were registered in the DBIR between inception and January 2021; 6194 reconstructions were registered during the study period. After exclusion of patients in whom autologous adjunctive procedures were used, 5699 reconstructions remained. To validate the previously described risk model, 1930 patients were excluded because data of one or more risk factors was missing, resulting in a total of 3769 reconstructions. The mean age in this cohort of 3769 reconstructions was 48.8 ± 11.3 years with a mean BMI of 24.7 ± 4.2 kg/m2. Patient selection and distribution are visualized in a flowchart in Figure 1. The baseline characteristics of the validation study population were compared to the baseline characteristics of the previous multicenter risk model population (Table 2). 6
RkJQdWJsaXNoZXIy MTk4NDMw