Generalize the integration approaches. Consider two generic electronic health record (EHR) systems with medication orderables: Entity 1 and Entity 2. If both entities were mapped to controlled medical terminologies (CMT), ontological matching could be achieved through CMT (path A). If one of the two systems were to only contain mappings to a commercial vocabulary, then matching could still be accomplished through path B, assuming that the commercial vocabulary had been mapped to a CMT (path C), which would then allow vocabulary-based matching between Entity 1 and Entity 2. If CMTs were not available, but the systems had common metadata attributes (e.g. Attribute 1), which could be validated against an entity (Entity 3) in an external reference system (Reference System 1), then metadata-based matching is possible between Entity 1 and Entity 2 through path D. Matching could also be performed if the systems had other metadata attributes in common (e.g. Attribute 2), which matched with an attribute in another reference system (path E), as could direct matching between the two, although the latter approach would be less than ideal.