Stakeholder perspectives and requirements for environmental-data-driven capacity forecasting in stroke care: a national survey of German stroke units
TL;DR
By integrating atmospheric factors with real-time clinical data to generate short-term demand forecasts, such systems could optimize resource allocation, enhance coordination among stroke units, and support timely, high-quality patient care.
Abstract
Stroke remains a major health burden in Europe, with rising absolute case numbers. Atmospheric factors such as air pollution and temperature extremes are increasingly recognized as stroke risk determinants. Many hospitals face growing staff shortages, highlighting the need for predictive management of stroke care resources. This study surveyed directors of German stroke units to explore current strategies, challenges, and the perceived usefulness of a digital, environmental data-driven forecasting tool for resource management. As part of the ALERT-ITS project, a cross-sectional, web-based survey was conducted among certified stroke units in Germany. The questionnaire was developed using the REDCap® software and pre-tested by senior neurologists. The survey was distributed with support from the German Stroke Society in June and July 2025. 58 valid responses were analyzed. Most respondents had over 15 years of clinical experience. High occupancy rates (>80%) were common, and nursing shortages were the leading cause of bed closures. Although 48.3% ( n = 28) of stroke units were part of a stroke network, structured coordination of resources in such networks was rarely established: only about one in five networked units (21.4%, n = 6) exchanged capacity data at least weekly, and nearly half (46.4%, n = 13) never did so. Communication within networks for acute stroke care primarily relied on telephone contact (89.3%, n = 25), while digital tools (39.3%, n = 11) or video consultations (35.7%, n = 10) were less common. Key requirements for an atmospheric factor–based prediction tool included short forecast horizons of 24–48 h, a reliability threshold (>80%), and detection of increases of three to four beds. Overall, opinion on the need for a digital forecasting system was evenly divided [44.8% ( n = 26) perceived a need, 44.8% ( n = 26) did not, and 10.3% ( n = 6) were undecided]. By integrating atmospheric factors with real-time clinical data to generate short-term demand forecasts, such systems could optimize resource allocation, enhance coordination among stroke units, and support timely, high-quality patient care. This even split underscores that clearly defined user requirements, rather than a broad consensus on demand, are the central contributions of this work.
