Accessibility, the ease of reaching a given destination from a given location, is a key indicator for identifying populations whose quality of life is constrained by limited access to essential goods and opportunities such as medical institutions. In Japanese geography, accessibility studies have rarely examined the relative effectiveness of analytical methods and tools, and the construction of data spanning multiple transport modes has remained difficult. This study employs the General Transit Feed Specification (GTFS), a publictransport data format that has become mainstream in recent years, to compute travel times that reflect actual departure and arrival times, including transfers, at a fine temporal resolution. Taking older adults' trips to medical institutions in the Yamagata urban area, a region with an exceptionally high car modal share and a proactive medical policy, as a case study, we compare three tools, ArcGIS Pro, OpenTripPlanner (OTP) and R5, in terms of travel-time accuracy and processing time. R5 proved both fast and generally accurate, whereas structural differences between the underlying routing algorithms (A-Star in OTP and RAPTOR in R5) produce systematic travel-time discrepancies that concentrate at specific origins and destinations. We organize these discrepancies into four structural factors and, to support quality control after tool selection, propose an open-source QGIS plugin that automatically flags origins and destinations requiring inspection.