One way to think of the EMPI is that it is not a "number" or unique identifier. Rather it is a function that matches a candidate identify of a person standing in front of the counter or a candidate identity of a result coming across an interface with a known identity in the universe of persons in the institution's database.
It does that by matching as much information about the person as needs be. The incoming result should have several pieces of identifying information attached. The person in front of the desk must be asked several pieces of information also. These pieces of information are fed to an identifying system which compares the exact information AND IT'S COMMON MISTAKES AND MIS-TYPINGS with all persons in the database after including ALL THE COMMON MISTAKES AND MIS-TYPINGS of the entries in the database. A list of probable matches in the database is generated including the statistical probabilities that they match the candidate. A human operator or an algorithm picks the best match if it is over a certain threshold. Over and over again articles and books about data modeling and database administration admonish against using a primary key that contains information about the row in question. This applies just as well to unique identifiers in an EMPI system. FUPI could be a clever and useful tool in some circumscribed environment. IMHO, using it in a large enterprise would be fraught with pitfalls logistically and for privacy reasons. Best of luck in the implementation. Bruce PS - I loved the piece at http://www.omp.de.vu/mismanagement.html. I am already thinking of how to write up my experience! ----- Original Message ----- From: <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Wednesday, May 15, 2002 10:48 PM Subject: Re: customizable vs. fixed schema, was Re: FUPI :-), was Re: new subject > Thank you Dr. David Forslund: > > Would you please be so kind as to indicate the places I can find all these > articles > as I will need to print them out, read them (or re-read them as I think I > have seen some of them) and present these options and instructions to the > Stamford IPA IT Users group that is now being finalized and we need to begin > implementation > quite soon. Hopefully all these ideas and recommendations made over the past > few years will provide us with the best solution. > If you have the URLs for those articles it would be greatly appreciated. > > Thanks again. > > Sincerely, > Alan Falkoff, M.D., D.A.B.F.P. > [EMAIL PROTECTED] > High Ridge Family Practice, LLC > 992 High Ridge Road > Stamford, CT 06905 > (203) 322-7070 > FAX (203) 322-2389 > Assistant Chief of Department of Family Practice The Stamford Hospital > Chairman of the Stamford IPA Medical Information Technology Search Committee > President, CEO of Medical Information Technology Consultants, LLC > MIT Consultants, LLC > MIT [EMAIL PROTECTED] > >
