Patient's permanent identification and PMSI chaining system. Complementary or competing methods?

In order to "chain" all the information concerning their patients' files, hospitals have set up their own identification system. This system is not easy to manage and can encounter several difficulties. When entering the hospital, each patient is given an identification number resulting mainly from his ID card, his Health Care card data and other data... The clerk who registers a patient seeks for possible prior hospital stays of this patient: should the patient be unknown, a new record of identification will be created; but should this patient be already registered, the clerk will link the patient's new hospital stay to the already existing file and update the permanent patient's identification data if necessary. In spite of all the care taken to carry out this process, some errors may occur: creation of "doubles", which means creating a new patient's file for an already existing patient, or conversely wrong chaining for an already existing patient. In order to track down these kinds of errors hospitals have set up a quality system. Moreover, since January 2001, in the frame of the Prospective Payment System (in French PMSI), a hospital stay chaining system based on a calculated key (H coding) has been designed. If the use of this key is meant to provide statistics concerning patients' cares and not patients' stays it may also become part of the quality policy of an establishment concerning patients' permanent identification data. Considering the 61 486 hospital stays at the C.H.U of Nancy for the first six months in 2001, it has been possible to compare the results in term of a patient population calculated from the permanent identification system of the hospital with the results obtained by the PMSI chaining system. No important differences have been detected and the results are very close. Besides, in order to track down possible doubles in the C.H.U patient identification system, chaining differences have been analysed. For the 22 detected cases, no identification management anomalies can be found; the doubles are the result of the calculating system of the chaining key used by the PMSI. A single interesting case can be mentioned: it concerns an anonymous registration stay; despite a previous stay in the hospital, the patient was given another identification number to ensure her complete confidentiality.