108 Chapter 7 OR: 4.458, CI: 2.563-7.754), previous radiotherapy (β: 0.989, OR: 2.689, CI: 1.0396.959) and increase in implant volume per 100 cubic centimeters (β: 0.290, OR: 1.336, CI: 1.117-1.599). The model was applied to the validation cohort and divided into quintiles, containing 41 to 46 subjects. The ratio between the observed and predicted probability ranged from 0.256 to 1.003, with a mean ratio of 0.676. The calibration plot is shown in Figure 4. The reference line lies within the CIs of three out of five quintiles. Figure 4. Validation of the DTI risk prediction model using a calibration plot DISCUSSION In this study an internally validated risk prediction model for implant loss following DTI or two-stage implant-based breast reconstruction was created using nationwide population-based data of the DBIR database. Four risk factors were included in the model (BMI, active smoking status, previous radiotherapy and prepectoral placement). The calibration plot showed good agreement, indicating the model may be extrapolated to the Dutch reconstructive population at large. Alternative risk prediction models were created, in which a subgroup of TEs and DTI reconstructions was analyzed separately. Results showed that in the TE model (including: BMI, active smoking status and prepectoral placement) the mean ratio between the observed and predicted probability was slightly better compared
RkJQdWJsaXNoZXIy MTk4NDMw