In complex domains with features of collaboration and integration, changes taking place in one process may have positive or negative impacts on other collaborative partners and processes. Therefore, supporting changes propagation as well as analysis the impact of changes are desirable functionalities of Business Process Management Systems (BPMSs) and have been investigated in many researches. This paper presents a method to analyze the impacts of process changes at run-time in order to help process practitioners deciding to adopt or not a change.We propose an approach based on Process Dependency Graph (PDG) to represent and monitor the dependencies among running process instances managed by a BPMS. When achange happens, we analyze the PDG graph to deduce the affected process elements and then assess the impact of change by using quantitative metrics extended from the workflow Quality of Service (QoS). Our approach thus provides a generic framework that can be adapted to a specific process domain and a BPMS. The work presented here was validated on some processes in Health-care and Software Development domains with the use of jBPM as aBPMS and Neo4j as a graph database to store and traverse the PDG.