The Universal Patient Identifier: a discussion and proposal

The broad-based need for a standard, robust patient identifier is reviewed. Advantages and difficulties of the Social Security Number are considered. An alternative number, composed of birthdate, location, and check digits, is proposed. Logistic extensions of the geographically based schema are made for provider and facility identification. Several variations of representation are described in base 34, which make the composite number short, easily derived and decoded, and robust.

The accuracy of the National Death Index when personal identifiers other than Social Security number are used

This study analyzed the accuracy of the National Death Index when personal identifiers were used that included or excluded Social Security number. Computerized records of the Department of Veterans Affairs were used for comparison. Different combinations of identifiers other than Social Security number correctly identified from 83 to 92 percent of dead and 92 to 99 percent of living persons. These results should prove useful in ascertaining the mortality status of patient populations without information on Social Security numbers.

Improvement in transfusion safety using a new blood unit and patient identification system as part of safe transfusion practice

A new patient and blood unit identification system designed to confirm the identity of crossmatched blood products and that of the intended recipient was evaluated. Six hundred seventy-two red cell concentrates were transfused to 312 patients. Participating hospital personnel and patients were interviewed regarding the use and benefit of this unique system, which incorporates a "lock-box" approach to the identification process. The product and procedure were accepted unanimously and enthusiastically, and three potential mistransfusions were avoided by use of the system during the limited period of observation. This type of approach to the identification process affords greater security than conventional practices and minimally burdens staff.

Validating patient names in an integrated clinical information system

Methods for validating patient names during the upload of clinical records are described. Exact string matching, Soundex method and a pattern matching algorithm (LCS method) are described and compared to a manual analysis of 10000 patient name pairs. In addition, the types of spelling and typographical errors that occur in patient names in the pathology database at CPMC are described. The data analysis shows that the LCS method performs better than the other techniques when compared to manual analysis.