Databases Reference
In-Depth Information
FIGURE B.32
Data grouping in a
data warehouse.
FIGURE B.33
Data model for patient.
to placing data physically in a database, another important database design practice is that of insuring
that any data elements that are operational only (such as telephone number) are removed from the
data warehouse data model. And summary data is normally removed as well. (Only on rare occasions
is it permissible or advisable to store summary data in a data warehouse.)
The data model for the patient in the healthcare data warehouse is almost always divided into
three distinctive parts.
Figure B.33
shows this division.
Figure B.33
shows that the healthcare data model for the patient is divided into three parts: one
part for patient identification, one part descriptive, and one part for body parts and functions. This
division makes sense for a variety of reasons. From a functional standpoint there is little or no rela-
tionship between a patient's heart rate and the address of the patient. From a database design stand-
point, to normalize the data, this division is natural in any case.
So the data model is created with normalization and with a distinct separation of data types.
Despite the separation of data, in each grouping of data is always found patient ID. It is necessary to
have patient ID as a data element to bring together the many different types of data that are likely to
exist for a patient. The data that populates the data model comes from many sources, many of which
are generated by the recording of an episode of care.
The data begins as a recording of the activities found in an episode of care. The data is trans-
formed into a form and structure suitable for an individual data warehouse, the kind of data ware-
house that a hospital or a clinic might have for its own use.
Figure B.34
shows the architecture that is
created.
Once the individual or local data warehouse is created, the data flows into the collective common
data warehouse. At this point the data is stripped of its identity. The data either becomes anonymous
or the data becomes encrypted. The data in the collective common data warehouse is organized by
subject area. Then data is taken from the subject areas to form special subject areas. These special
subject areas can encompass any type of information useful to research for healthcare.
Note that data can go directly from an episode of care to the collective common data warehouse.
In the eventuality that there is no individual provider data warehouse, the data can flow from the epi-
sode of care directly into the collective common data warehouse.
