Prophylactic Replacement of Voice Prosthesis | 177 7 model, the DL70 was chosen as the number of days at which the Kaplan-Meier curve for device lifetime in the different type of groups of patients crosses the 70% boundary. Several models were considered: one DL70 for all patients, separate DL70s for used type of VPs (VPG1 and VPG2), and separate DL70s for indication TL (PG1 and PG2). All models assume that problem solving VPs will never be prophylactically replaced. Figure 1. Flowchart of included patients and subgroups Abbreviations: VP=voice prosthesis, TL=total laryngectomy For evaluation of the prediction models, we counted how many of the VPs received by the patients would be prophylactically replaced and how many would (still) have been replaced because of leakage if the various prediction models had been used at the time. VPs with a device lifetime shorter than DL70 are considered ‘replaced because of leakage.’ VPs with a device lifetime at DL70 or higher are considered ‘prophylactically replaced.’ Time in situ of additional VPs (that would be used in case of PVPR) was estimated from known history of VP use per year. Thus, for each patient the number of needed VPs per year under the various prediction models were calculated. In the evaluation of the model that uses different DL70s for VPG1 and VPG2 we assume that all patients start in VPG1 and switch to VPG2 as soon as they have their first problem solving VP. The data is displayed in medians and ranges. Individual adaptive simulations As there is a high device lifetime variability, an individual adaptive model per patient was made. This model uses the device lifetime of the first three used VPs to calculate the replacement moment for the coming VP, as the DL70 based on these three VPs.
RkJQdWJsaXNoZXIy MTk4NDMw