Abstract Book

S32

ESTRO 37

del Sacro Cuore – Rome- Italy, Gemelli ART- Department of Radiation Oncology, Rome- Italy, Italy 2 Università Cattolica del Sacro Cuore, Radiation Oncology - Gemelli ART, Rome, Italy 3 KBO Labs- Gemelli ART- Department of Radiation Oncology- Fondazione Policlinico A. Gemelli- Università Cattolica del Sacro Cuore – Rome- Italy, KBO Labs- Gemelli ART- Department of Radiation Oncology, Rome, Italy 4 Gemelli ART- Department of Radiation Oncology- Fondazione Policlinico A. Gemelli- Università Cattolica del Sacro Cuore – Rome- Italy, Gemelli ART- Department of Radiation Oncology, Rome, Italy Purpose or Objective Aim is to exploit process mining (PM) techniques to develop a system able to detect, compare and verify adherence of clinical activities to Guidelines (GL) and good clinical practice (GCP) by reading information contained in existing Electronic Health Records (EHR), without additional data management burden. Material and Methods A regional consensus-based clinical GL (fig.1) concerning the clinical management of patients affected by locally advanced rectal cancer (LARC) was translated and implemented as a computer-interpretable GL. Subsequently, EHR data regarding diagnostic and therapeutic pathways of 485 patients affected by LARC were automatically read as a valid input, called event- log, by the PM software. In this event-log, each patient had a series of events and event’s attributes. The occurrence of an event can trigger logical conditions allowing the patient to move from a state of the GL to another one. Adherence analysis to the clinical GL was performed considering the ratio between the number of patients in pre-defined completion states and patients that entered the GL. Moreover, a focused analysis to verify the major dropout condition from GL and a conformity analysis of the waiting time from treatment to surgery in the different branches of the GL were performed . Results The computer-interpretable GL and the flow of patients within the GL is shown in Fig. 2; among the total of 485 patients, 473 patients (97.5%) entered the GL. Of the 473 patients, 354 (75%) reached one of the completion states. By comparing the flows of patients among the three different branches of the GL, it emerged that 54 out of 55 patients (98%) completed the pathA, 270 out of 380 (71%) completed the pathB, and 30 out of 38 (79%) completed the pathC. The single highest dropout rate was found in the condition on concurrent chemotherapy (cCT) agent in pathB, with a value of 74.4% of dropouts (from 371 to 276). Among these 95 patients, 32 (33.7%) dropped out because of missing data about the cCT regimen used, 63 patients dropped out because they underwent to another chemotherapy (not 5-Fluorouracil, not Capecitabine). The waiting time distributions from treatment end to surgery in the 3 paths of the GL didn’t show, as expected, any statistically significant difference (p-value 0.85 for pathA-pathB, 0.71 for pathA-pathC, 0.41 for pathB-pathC).

Conclusion This study shows the potential that a PM approach has in managing patient processes and workflows within a radiation oncology department. It was possible to check and measure the difference between the actual process and the recommended one, exploiting, without additional data management cost, the data already stored in EHR. The level of non-adherence can be then used as a metric of conformance to improve either the actual process execution, the data storage system, or the guideline itself depending on the causes of such deviations from the GL or from the GCP. OC-0070 Image guided brachytherapy for primary vaginal cancer: results of an international multicenter study H. Westerveld 1 , R. Nout 2 , M. Schmid 3 , C. Chargari 4 , N. Nesvacil 3 , R. Mazeron 4 , L. Fokdal 5 1 Academic Medical Center, Radiotherapy, Amsterdam, The Netherlands 2 Leiden University Medical Center, Radiotherapy, Leiden, The Netherlands 3 Comprehensive Cancer Center- Medical University of Vienna, Radiotherapy, Vienna, Austria 4 Gustave Roussy Cancer Campus, Radiotherapy, Villejuif, France 5 Aarhus University Hospital, Oncology, Aarhus, Denmark Purpose or Objective Definitive (chemo)radiotherapy, including brachytherapy, is the mainstay of treatment for primary vaginal cancer. Because vaginal cancer is very rare, most studies have included few patients over a long time span and most are from the radiographic era, prior to the use of CT or MRI for brachytherapy treatment planning. The purpose of this study was to assess the outcome of patients who had CT or MRI based image guided adaptive brachytherapy (IGABT) for primary vaginal cancer in a multicenter setting, prior to the development of recommendations for a common target concept. Material and Methods Patients were eligible for inclusion with: a histological confirmed vaginal cancer; FIGO stage I-IVA; MRI at diagnosis; CT or MRI based IGABT (PDR or HDR) with or without chemotherapy. Patients’ charts were retrospectively reviewed to obtain patient, tumor, and treatment characteristics. Late morbidity was scored according to the CTCAE 3.0 scale. The prescription dose, the clinical target volume (CTV) to which this dose was prescribed and the fractionation schedule were according to the local guidelines of each institute. Additionally, organs at risk (OAR) were delineated. Survival rates were calculated using the Kaplan-Meier method and log rank test. Results A total of 148 patients were included from five participating centers. All but three patients, received pelvic EBRT (45-50,4 Gy) and 94 patients (63,5%) received Proffered Papers: BT 1: Gynaecological brachytherapy