Cognitive radio sensor networks (CRSNs) with their dynamic spectrum access capability appear to be a promising solution to address the increasing challenge of spectrum crowding faced by the traditional WSN. In this paper, through maximizing the utility index of the CRSN, a node density-adaptive spectrum access strategy for sensor nodes is proposed that takes account of the node density in a certain event-driven region. For this purpose, considering the burst real-time data traffic, we analyze the energy efficiency (EE) and the packet failure rate (PFR) combining network disconnected rate (NDR) and packet loss rate (PLR) during the channel switching interval (CSI) for both underlay and interweave spectrum access schemes. Numerical results confirm the validity of our theoretical analyses and indicate that the adaptive node density threshold (ANDT) exists for underlay and interweave spectrum access scheme switching.