Different approaches in the hybridization of constraint programming and local search techniques have been recently proposed in the literature. In this paper we investigate two of them, namely the employment of local search to improve a solution found by constraint programming and the exploitation of a constraint model to perform the exploration of the local neighborhood. We apply the two approaches to a real-world personnel rostering problem arising at the department of neurology of the Udine University hospital and we report on computational studies on both real-world and randomly generated structured instances. The results highlight the benefits of the hybridization approach w.r.t. their component algorithms.