91 Validation of a multicenter risk model for implant loss following implant-based reconstruction for surgery, permanent implant volume and tissue expander volume, the rate of nipple sparing reconstructions was significantly lower in the validation cohort compared to the multicenter cohort (29.9% vs 40.5% respectively). Furthermore, the incidence of obesity was significantly lower in the validation cohort (10.6% vs 15.8%). The other risk factors were not significantly different between the two cohorts. A nipple sparing procedure and a DTI approach were not significantly associated to implant loss in the current validation cohort. Since these factors represented half of the risk model, it is understandable that the risk model was not accurate in the current validation cohort. It could be hypothesized that the risk of implant loss increases in a nipple sparing procedure, as wound problems or necrosis seem to be most common in the nipple area. However, to date, a nipple sparing procedure has not been described as a risk factor for implant loss, thereby confirming the results of this validation cohort. In addition, a DTI approach is a frequently described risk factor for implant loss,3 but this was not observed in the current validation cohort. However, the literature is contradictory on this topic, and critical patient selection, for instance by judgment of mastectomy flap tissue quality, is an important component.16-18 Although the current study contained a large sample size with data of a nationwide population, this database study has certain limitations. First of all, the accuracy of all DBIR data could not be confirmed due to its anonymized nature and privacy regulations. Another limitation is the restriction to the data collected in the database. One of the risk factors in the multicenter risk model was a nipple sparing procedure, which was not a direct variable in the DBIR database. However, this factor could be indirectly derived from the variable ‘incision site’. The same applied to the definition of implant loss, which was created based on the available data in the DBIR database. However, the accuracy of these definitions could not be confirmed due to privacy regulations within the anonymized data. Finally, the registration of explantations might be an underestimation of the clinical practice due to under registration. In conclusion, the observed incidence of implant loss in the validation cohort was 8.1% and does increase if the number of risk factors accumulates. However, the predicted probability of implant loss based on the multicenter risk model did not match the observed probability in the current nationwide cohort, indicating that the multicenter risk model is not accurate in Dutch practice. In the future, attempts will be made to improve the risk model and provide a validated tool for the risk assessment of implant loss. This could lead to improved pre-operative information for patients and the ultimate goal to decrease the risk of implant loss by optimizing the surgical strategy in a personalized fashion. 6
RkJQdWJsaXNoZXIy MTk4NDMw