When demographics are involved in the process of patient matching, multiple attributes are compared on records from two sources. First name, last name, date of birth, gender, and address are typically always present on records, but all of these except date of birth are dynamic and/or hard to create standards for their capture. Other elements that are also changeable over time like email address and cell phone number are sometimes, but not always present. Most systems prefer not to use social security numbers since the identifier is of such value for identity theft. Because not everything is always present or entered in the same way, in order to get high match rates, algorithms that provide different weights to different attributes or that allow for “fuzzy” matches to correct for transposition or misspelling errors are usually used. Algorithms usually produce a “score” that represents the likelihood that the records are on the same person. When the score exceeds a threshold it can be used to probabilistically determine whether or not to treat two records as matching, that is, “this is probably the same person”. In deterministic matches, identifiers that are shared between systems are used to make the match. In the presence of such an identifier, probabilistic algorithms in use no longer apply as the identifier establishes the match reliably on its own, that is “this is the same person”.