Since the database contains many ``free text'' itemIDs, there is no unique method for representing certain parameters. For example, CPR is mentioned in 9 different itemIDs. Each itemID can contain a variety of values such as ``yes'', ``no'', ``done'', ``performed'', etc. In addition multiple different capitalization and spelling errors are found. This makes it extremely difficult to obtain accurate information on whether or not CPR was performed for a particular patient.
As and when certain data is extracted from these ``free text'' fields, it can be moved into more meaningful fields which will permit simpler data extraction. For example, the CPR itemIDs mentioned above could be translated into a binary field which simply states whether or not CPR was performed.
Examples of multiple mappings, plus code to merge them, can be found
in appendix 6.2.