RetroGuide is a name of a research project in medical informatics focusing on using workflow technology in healthcare to analyze retrospective data about patient care.
The project was designed during 2004-2007. The methodology can be adapted to use data from any large healthcare institution. During the initial development phase the project was tested on data from Intermountain Health Care in Utah, USA.
The methodology requires two inputs from the user
- a specification of a cohort of patients to be analyzed
- a model of events (=scenario) which is to be tested against retrospective data
RetroGuide's output is then a set of hierarchically linked reports. Some of them are generated by default, some of them can be optional. RetroGuide enables export of results or original data elements into other programs via XML files and CSV files.
The most used tool to analyze retrospective terminology-coded retrospective data are SQL queries on a population datase. RetroGuide's approach is different from a query language since the tested scenario is executed on every patient in the analyzed population. It enables procedural features as opposed to declarative nature of SQL.
RetroGuide can be used in three main domains
RetroGuide has been used on several clinical problems
- investigating blood pressure control in diabetics patients
- investigating pregnancy rate in female patients after treatment of Hodgkin lymphoma
- investigating HEDIS quality improvement measures from NCQA
- osteoporosis measure (OMW)
- cholesterol control in cardiovascular patients (CMC)
- investigating alternative glucose protocol logic for blood glucose control in ICU patients
- course of care for diabetes
- course of care for AMI
- course of care for chronic kidney disease
- detecting adverse drug events (respiratory failure after use of narcotics, Naloxone)
RetroGuide graphical approach to model queries has been formally evaluated in a study involving 18 human subjects with limited database expertise . The study compared RetroGuide technology with SQL
. Each subject had to solve 14 analytical tasks using both compared technologies. The qualitative comparison of average test scores showed that the study subjects achieved significantly higher scores using the RG technology.
Each subject also filled a follow-up questionnaire which compared both technologies qualitatively. The results of this qualitative study showed that 94% of subjects preferred RG to SQL because RetroGuide was easier to learn, it better supported temporal tasks, and it seemed to be a more logical modeling paradigm. The second part of the follow-up qualitative questionnaire also asked RG-specific questions based on validated constructs from the Unified Theory of Acceptance and Use of Technology. The results of this second part suggested that a fully developed, RetroGuide-like technology would be well accepted by users.
- Huser V, Rocha RA, James BC, "Use of Workflow Technology Tools to Analyze Medical Data," pp. 455-460, 19th IEEE Symposium on Computer-Based Medical Systems (CBMS'06), 2006. article (doi)
- Huser V, "Running Decision Support Logic Retrospectively to Determine Guideline Adherence: a Case Study With Diabetes," Spring AMIA2007 symposium poster conference
- Huser V, Rocha, RA, "Analyzing medical data from multi-hospital healthcare information system using graphical flowchart models," BMIC Symposium, Orlando, 2007 (accepted).
- Huser V, Rocha RA, Huser M, "Conducting Time Series Analyses on Large Data Sets: a Case Study With Lymphoma," Medinfo 2007, Brisbane, 2007.
- Huser V, Rocha RA, "Retrospective Analysis of the Electronic Health Record of Patients Enrolled in a Computerized Glucose Management Protocol," CBMS 2007 article (doi)
- Huser V, Rocha RA, "Graphical Modeling of HEDIS Quality Measures and Prototyping of Related Decision Support Rules to Accelerate Improvement", AMIA Fall Symposium, 2007, Chicago, USA poster