Background <p>Modifications and adaptations to evidence-based interventions are common, and of special relevance to complex interventions in healthcare. Although they play an important role in scale-up and sustainment, the potential exists for negatively affecting the core functions of an intervention. This study explores modifications to rapid response systems (RRSs), using the established Framework for Reporting Adaptations and Modifications – Expanded (FRAME). RRSs are patient safety interventions developed to identify and respond to hospital patients in clinical deterioration. Despite widespread use, little evidence-based guidance exists for necessary adaptations to local context. Applying adaptation frameworks is a novel perspective to improve RRS intervention design and implementation guidance. We aimed to explore which modifications and adaptations to RRSs that have taken place in Norwegian hospital units, how they occurred, and what the underlying reasons were by using FRAME.</p> Methods <p>Nine hospital units across six hospitals, which had initiated the implementation of RRSs 4 to 12 years previously, were included. Data was collected through focus group and individual interviews with clinicians and leaders. Analysis involved two steps: a conventional, inductive content analysis to identify and categorize modifications, followed by further characterization of these modifications through deductive analysis employing FRAME.</p> Results <p>Inductive analysis identified 5 categories and 24 subcategories of modifications to the RRS intervention. Application of FRAME revealed modifications to be mainly reactive and occurring in the maintenance/sustainment phase, decided at the unit level and with varying fidelity consistency. Both structured and informal processes were identified. The goals of modifications were improvement of feasibility, effectiveness and fit, and reasons were related to available resources, service structure, clinical judgment and patient factors. Minor adaptations to FRAME were necessary to fit the RRS intervention and the methods of data collection.</p> Conclusions <p>Studying real-life implementations of RRSs provides insight in modification processes, highlights which intervention elements are modified for better fit and feasibility, and which modifications are prone to fidelity inconsistency. Our findings underline the ubiquity of modifications to RRSs, and the need to systematically anticipate them throughout all implementation stages. Further exploration of RRS core functions and application of FRAME within collectively implemented patient safety interventions could advance the field.</p>