AB1097 Development of Statistical Analysis and Computer Tablet Based Clinical Score Input System on the Electronic Medical Record for Rheumatoid Arthritis

Background In clinical research, statistical analysis of clinical data is a costly process. Its cost is composed of data input, data store and statistical analysis. As usually clinical score is recorded on paper, manual data input is needed. For the protection of personal information, anonymization of data and isolation of database is needed. For statistical analysis, statistical software and statistical operation are needed. Objectives In order to solve these problems, we developed a computer tablet based clinical data input and statistical analysis system which can work on the electronic medical record system. Methods Our system consists of three modules. First module is a score input system, which can create and custom the clinical input screen for each clinical score (such as HAQ, DASH, and DAS28) and perform the calculation such as total score. Second module is a data store system, which can store the anonymized clinical data and connect to electronic medical record system (development of connecting module for each electronic medical manufacturer is needed). Third module is an analysis system, which can perform various kind of statistical procedure and record the statistical methods. For statistical analysis, we selected R, which is open source software for statistical computing. R can perform many kinds of statistics and describe statistical procedure. With combination of these modules, we can visualize the statistical results and graphics immediately after the data input. In order to verify the utility of this system, HAQ score input using tablet computer system was performed in 26 RA patients and 23 healthy controls. We recorded the score and time of entering data (input time), and examined the correlation between the score, time used for data input, input experience and age. These statistics were performed immediately after the data input. Results Input time was 170±83 seconds in RA patients and 88±38 seconds in healthy individuals. Age and input time has good correlation in healthy controls (r=0.63, p=0.05), but poor correlation in RA patients (r=0.11, p=0.57). HAQ and input time has good correlation in RA patients (r=0.50, p=0.01). Input time is shortened with an increase of the number of input experience. Real-time display of statistical results and graph drawing, and the connection to the electronic medical record was possible. Conclusions The cost of this system is low, because this system was all created by open source software and work on a