Data Flow and Data Structure Overview

This is a brief overview of the data model and data flow related ot the LDR. You should also read the Data Structure Flexibility and Data Extraction

• The data originates in one or more proprietary LIS data structures.

• A process exists by which the data is transformed into a common and simpler data structure at a central location. This is the LDR data structure.

• The data can be queried and reported directly from this location or it can be transformed into another data structure the developer finds easier to use for reporting.

It should be noted that the original intent of the original LDR data structure was to be able to report directly from the LDR structure without the need for another structure. This can still be done; however, the way the data is stored in this data structure has been expanded in order to allow for the need to flexibly store the ever expanding list of required data elements.

Overview of LDR Data Model Manipulation for Flexibility in Adding Data Elements

This is not a venue to debate data modeling but the current methodology used by the LDR is a compromise between a horizontal and a vertical data model for flexibility reasons, rather than to have it purely one way or the other. At first glance the data model of the LDR is pretty simplistic. However, what you will notice there is a fixed number of columns beyond basic request and result information. A lot of individual data elements are collected at registration are not stored as explicit columns in the data model. A "wide" table with each individual data element as a column is a simple way to store and look at data. However, in that model if you need to collect new data elements you must update the data model with new columns and changes processes that transfer the data. The current LDR model handles these new data elements in a flexible way by adding a new Panel for the request which has rows in LabResults, each containing one of the data elements. For example, for viral load the registration screens ask for whether the patient is Pregnant and if they are BreastFeeding. Rather than add new columns to the data model; additional rows were added to the LabResult table with each row having a column that identifies what data is stored in that row.

The vertical structure of the data makes the data flow flexible without constant data structure changes. It is expected that for some reporting needs the data will be extracted and transferred to another database or table that is more optimized for specific reporting needs. There are also database Views which transform the data for easy retrieval for specific needs.