Margriet Kwint

Real world evidence to audit NTCP-models for acute esophagus toxicity 133 7 risk were; esophagus: Dmax ≤66 Gy and V(50Gy) ≤50% (EQD2 10 ), mean lung-dose ≤20 Gy (EQD2 3 ), spinal cord ≤52 Gy (EQD2 2 ), total heart ≤40 Gy and ⅔ of the heart ≤50 Gy and ⅓ of the heart ≤66 Gy (EQD2 3 ). Equally spaced, 7-field IMRT-plans were calculated and optimized using 10 and/or 6 MV photons. Optimization was done using direct machine parameter optimization in Pinnacle version 9.0, (Philips, Best, The Netherlands). The prescription dose was specified at a representative point in the PTV. The dose inhomogeneity within the PTV was >90% and <115%. Electronic toxicity registration method From December 2012, a prospective, electronic toxicity (grade ≥2 toxicities (CTCAE v4.0)) registration was implemented within our department. Registration of toxicity is performed by the treating physician during each patient consultation. Simultaneously, a data management infrastructure was built to merge the toxicity data to patient characteristics and treatment parameters. AET was scored using the Common Toxicity Criteria 4.0 from start of treatment, till 3 months after. Toxicity was scored at baseline and weekly during treatment, until 3 weeks after treatment by the treating physician. Thereafter the patients were followed with 3 monthly intervals or more frequently if indicated. The toxicity was scored in the electronic patient file (Chipsoft, Amsterdam, The Netherlands). The physician first indicated whether or not any toxicity grade ≥2 was present. In case no or grade 1 toxicity was present, ‘no toxicity’ was scored to distinguish between missing data and an explicit registration of ‘no toxicity’. When there was grade ≥2 toxicity, the type and grade of toxicity was registered. The highest AET grade was used for this analysis. The data from the first year of the toxicity registration (December 2012 until December 2013) was not used for this study. This first year was used to illustrate the learning curve of the registration of the physicians. To assess the validity and completeness of the electronic registration, a sample test of N=77 (35%) was performed to check the accuracy of the registration. This was done separately by two individuals, by retrospectively reviewing the patients file and compare this to the electronic toxicity registration. For this sample test, a cross table was made and sensitivity and specificity rates were calculated.