M.S. Candidate: Ramis Korkmaz
Program: Information Systems
Date: 03.09.2026 / 10:30
Place: A-212

Abstract: Incident management systems play an important role in documenting adverse events and supporting organizational learning in healthcare. However, while incident reporting and investigation quality have received considerable attention, the operational performance of the investigation process, particularly investigation duration, remains insufficiently examined in home healthcare. This study aims to define and analyze prolonged incident investigations using data-driven process performance metrics and to examine whether prolonged investigations vary over time and across incident and organizational characteristics. A retrospective, observational quantitative study was conducted using routinely collected incident management data from a home healthcare provider covering July 2023 to December 2025. Following data preparation and completeness criteria, 493 completed mandatory investigations were included in the analysis. Investigation duration was measured as the number of days between investigation start and completion, and an empirical 80th-percentile threshold was used to identify prolonged investigations. Mann–Whitney U testing, temporal trend analysis, and nested logistic regression were employed to examine differences and potential determinants of prolonged investigations. The 80th percentile corresponded to 14.6 days and was operationalized as a threshold of 15 days, identifying 99 prolonged investigations (20.1%). Prolonged investigations had a median duration of 23 days compared with 6 days for non-prolonged investigations. The prolonged proportion remained statistically stable over the study period. Organizational region contributed more to model fit than incident category, although the overall improvement was modest. Regional prolonged shares ranged from 9.5% to 24.7%, with county-level shares ranging from 0% to 58.3%. The findings indicate that prolonged investigations constitute a persistent operational pattern more closely associated with organizational variation than incident type. The study demonstrates that routine system timestamps can support data-driven monitoring of investigation performance in home healthcare without requiring additional data collection.