Data integration processes that combine sensitive personal information from different institutions are usually subject to strict privacy regulations. Privacy-preserving record linkage (PPRL) methods can be used in such projects to conceal the identities. However, the quality of such linkages may be low as the parametrization mostly must be done blindly or based on estimations from previous supposedly similar linkage problems. Our framework, SecUREmatch, integrates a multi-layer clerical review system to adapt the matching algorithm to the actual linked data in order to achieve high linkage quality while following the privacy-by-design principle.