WITHDRAWN: Treatment Course Algorithm Developed from a Claims Database: Example From Patients With Locally Advanced or Metastatic Urothelial Cancer in France
Speaker(s)
ABSTRACT WITHDRAWN
OBJECTIVES:
Urothelial cancer are among the most common cancer. There is a lack of data on treatment patterns in patients with locally advanced or metastatic (La/m) urothelial cancer (UC) from real-world setting. Therapeutic strategies include different chemotherapy families or monoclonal antibodies. The objective of the study was to apply an IA-based algorithm to identify chemotherapy protocols and treatment patterns used in patients with La/mUC.METHODS:
Using the French National Healthcare database (SNDS), adult patients with La/mUC initiating a first line (1L) chemotherapy in 2019 were included. Chemotherapy protocols were identified using an adaptation of the Smith-Waterman alignment sequences algorithm, which is an IA-based algorithm that evaluates similarities between observed and theorical frequency chemotherapy administration accounting for delays and gaps. Protocols included GC (gemcitabine, cisplatin) as gold standard, MVAC (methotrexate, vinblastine, adriamycin, and cisplatin), and PCG (paclitaxel, cisplatin, gemcitabine). Similarity scores were calculated on a 28-day basis for each theoretical protocol and normalized between 0 and 1, to establish an empirical ranking for 1L and potential second line (2L) treatment.RESULTS:
Among the 4,700 patients with La/mUC initiating a 1L in 2019,38% were attributed GC, 38% MVAC, 19% PCG and 5% were not evaluable because of conflicting scores. Most of patients (75%) had their best elevated score >0.7 and their second best <0.5 suggesting no conflict in the final protocol ranking and acceptable predictability. For 20%, similarity scores of both, best and second best, were between 0.6 and 0.8 requiring expert opinion to establish final ranking. A small percentage (10%) of patient with 2L were identified.CONCLUSIONS:
The use of a well-known IA-based sequence algorithm adapted and developed from a claims database can provide insights on patient treatment course from real-world clinical settings and their associate outcomes.Code
MSR39
Topic
Methodological & Statistical Research, Real World Data & Information Systems, Study Approaches
Topic Subcategory
Artificial Intelligence, Machine Learning, Predictive Analytics, Health & Insurance Records Systems
Disease
SDC: Urinary/Kidney Disorders