Background and Objective: Deployment of Body Area Networks (BAN) in hospitals can enable real time data collection and analysis for patient health. Such networks have a two-tier structure: the primary layer consists of nodes that are installed on hospital facilities, while the second tier contains sensors on patient body. The unique two-tier network structure poses challenges to secure and continuous information transmission between sensors and data servers, especially when patients are moving. We plan to design a suite of mechanisms to support power efficient and secure data collection and smooth hand-over of sensors when patients move freely in a hospital.Methods: We assume that both group and individual secrets can be provided to BAN sensors when a patient checks in to the hospital. We first design mechanisms that use hash chains and double exclusive-or operations to protect data confidentiality and authenticity from BAN sensors. When a patient moves, the top tier nodes installed on the hospital facilities can provide smooth hand-over of the BAN sensors through secret updates. Our mechanisms can support network dynamics and changes of sensors in BAN networks.Results: We present the proposed mechanisms in details. We analyze the power efficiency of the approaches. We compare the power consumption of the proposed approach to those of the three security levels of IEEE 802.15.6 standard. Using formal methods, we prove the safety of the mechanisms. We also study the robustness of the approaches against various attacks.Conclusions: In summary, we design a suite of mechanisms to support secure and continuous data collection from body sensors on patients in hospitals. The approaches are secure and efficient, which satisfy the requirements of future smart health applications.