OBJECTIVES: Time-to-event analyses can be complex as they involve the use of multiple variables which include study start and end dates, index dates, event and censoring dates, and handling of missing data. Data cleaning and manipulation is often time-consuming and can lead to delayed insights and incomplete data interpretation. To address these limitations, we evaluated the use of a novel tool to allow a rapid data exploration of time-to-event patterns.

METHODS: The tool allows users to select any specific event interactively within the dataset, reindex patients based on the chosen event, and display other pre-specified events in relation to the new index. The results are visualised with the time of each event in the x axis and frequency displayed in a colour scale. Pre-specified events (outcomes and covariates, e.g Diagnosis, Chemotherapy, Surgery, Death, Progression, etc) were collected and classified along with patient IDs and dates from a Gastric Cancer Cohort in Medical Data Vision (MDV) data source from Japan. The tool was developed using Plotly package in a R Shiny application and will be made open source in GitHub.

RESULTS: In a practical application, with surgery as the indexing event, the tool demonstrated its efficiency by identifying commonly used drugs in neo-adjuvant or adjuvant settings and their frequencies in various care settings. The instant reindexing by the tool enabled meaningful time-to-event insights to be drawn this large dataset. The tool's applications can extend to exploratory analyses and treatment pattern studies, making it an asset for researchers while also significantly reducing the time and input required for generating time-to-event analysis insights.

CONCLUSIONS: This tool enhances the identification of time-to-event patterns in Real-World Data. Its dual role in data exploration and reporting accelerates insight generation.