[A general solution to the linkage of anonymous medical data]

The necessity for confidentiality entails that most medical data on files do not allow for a precise identification of patients. A statistical pattern has been defined to assess the average difference between the real (unknown) number of patients in the file and the approximate number derived by linkage from indirect available data (date of birth, sex,...). An example is here given, which allows for the epidemiological use of anonymous data concerning hospitalized patients at the Assistance Publique-Hôpitaux de Paris. This method is of general application for all files of anonymous data.