Purchase this article with an account.
M. Busuioc, R. T. Smith, R. Allikmets, N. Lee, A. Laine, J. Shi; Controlled Field Entries versus Free Text in a Clinical/ Research Emr. Invest. Ophthalmol. Vis. Sci. 2010;51(13):1804.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To present a database for integration of clinical, imaging and genetic data, that can allow a faster approach to data entry by clinical personnel.
The database development challenge was to offer controlled information entry for easy retrieval and analysis but also the full free text fields that would allow busy clinicians to have their comments and allow us to still be able to extract useful information without limiting the options for some entries.The system intended for a Stargardt disease research project asked for integration of clinical information, images and genetic data. The database consists of one main table with demographic information and multiple children tables with clinical visits, imaging and genetic information. We opted initially for a controlled entry model for easy analysis, but this proved to be a never ending task of expanding the available options (e.g.: more alternative diagnoses). From previous published studies we established that free entries for additional diagnoses are usually prone to error and even if we try disambiguation using a large vocabulary support, the rate of error was still high. Hence we decided a modified approach, with a pop-up screen, where whenever necessary a new concept (diagnostic) was introduced in our restricted list with assistance from a UMLS (Unified Medical Language System) derived list of retinal diseases. If the disease was in realm of acceptable entries a study coordinator could update the list on the spot, otherwise a physician would have to do it through a waiting list for approval, adding a new concept to the local vocabulary. In the updated list we keep the userid of the upgrader , the CUI ( unique concept id ), SUI (unique string id ), and LUI (unique linguistic identifier ). If later somebody shall try to introduce a synonym, a warning is popping up suggesting the former term if appropriate. This approach allowed us to assign worker responsibilities in the process of updating our related vocabulary and keep consistency of our data.
Our approach, using limited expandable vocabulary for specific fields proved to be a reasonable approach.
In an ideal system, the study coordinator or the physician can enter scientifically rigorous information directly into the software application without the limitation of controlled entries. Ours can deal just with limited variability, in a couple of fields, but the added flexibility was a welcome improvement.
This PDF is available to Subscribers Only