Ontologies are currently more and more frequently used to represent knowledge in distributed heterogeneous environments. This approach supports knowledge sharing and knowledge reuse. In order to increase the effectiveness of such solutions, a method should be developed which would enable us to integrate ontologies coming from various sources. The article presents a concept for integration of knowledge, based on structural and lexical similarity measures, including the Similarity Flooding algorithm. The proposed concepts are demonstrated on the basis of a selected area of medical studies: the analysis of the incidence of hospital infections. Sample ontologies (exhibiting structural or lexical similarities) have been developed and for each case a suitable algorithm is proposed.