The SAGE guideline model

From Clinfowiki
Jump to: navigation, search

The SAGE Guideline Model. Insinuating best practices into clinical information systems.

Tu SW, Campbell JR, Glasgow J, Nyman MA, McClure R, McClay J, Parker C, Hrabak KM, Berg D, Weida T, Mansfield JG, Musen MA, Abarbanel RM. The SAGE Guideline Model: Achievements and Overview. J Am Med Inform Assoc. 2007 September-October;14(5):589-598.


Question

What attributes of an environment support implementation of effective decision support based on clinical guidelines, in commercial clinical information systems?


Design

Implementation project, a pilot with qualitative assessment.


Setting

Over four years, design of SAGE (Standards- Based Active Guideline Environment) was studied at several institutions including GE healthcare, the University of Nebraska Medical Center, Stanford University, and the Mayo Clinic. These institutions represented various disparate commercial CIS implementations.


Intervention

The SAGE project includes creation approach and infrastructure for implementation of decision support. The model was used to build decision support based on specific best practices guidelines for immunizations, and management of community acquired pneumonia, diabetes and hypertension in diabetes.


Main Outcome Measures

The outcome is a description of a guideline modeling system. The system was verified with the testing of decision support using simulations. Features were tested and refined through implementation of this decision support. .


Main Results

Information models are described. The Virtual Medical Record (VMR), a standardized and abridged set of patient data using standard terminologies and conventions provides for the unambiguous application of decision support. Specific clinical scenarios are encoded to provide workflow context and trigger events for decision support at the right moment, and of the right type. Medical concepts are pulled from the guidelines by clinicians and translated to standard terminology to accomplish this task Events in SAGE are driven by an action (for example, accessing the record), a situation (a condition not being met), or by time (a situation continuing to not be met after a period of time). When an event occurs, SAGE must look at patient data. SAGE guidelines use a normalized structure of patient data using Health Level 7 data types, to get around variations in CIS. SAGE represents clinical data with a virtual medical record- a simplified record that contains just data that is relevant to the decision support and uses HL 7 Reference Information Model standards. A recommendation set is comprised of the various recommendations of decision support that may be made in a given clinical setting. Recommendation types are classified as activity (such as asking for more information), or decision (providing direction in orders). SAGE allows for selection of order sets, selection of items within an order set, and annotation of orders based on clinical decision support directions. Evidence statements provide detailed information for a given situation such as drug interaction or contraindication information. SAGE must make such assessments to avoid contradicting such decision support from within the CIS. A Virtual Knowledge Base allows SAGE to efficiently access outside knowledge sources in a way analogous to the VMR.


Conclusions

The system endeavors to find commonalities between various guideline models for standard representation and to provide decision support that is effective by being relevant, timely and non-intrusive. SAGE accomplishes this through components in a decision support environment that 1) incorporates workflow awareness, 2) employs information and terminology standards, 3) incorporates simple flow of control standards, and 4) integrates with commercial clinical information systems. The authors assert that this system uniquely accomplishes all of these tasks


Commentary

Clinical guidelines extract best practices out of a complex mass of medical literature that varies in quality and relevance. To date, using clinical guidelines to improve care, by changing practice behaviors has been problematic. Presenting best practices to clinicians has rarely been effective in changing behaviors. Practice continues based on anecdote. Clinicians are skeptical that best practices are relevant to their particular situation. Information is not available when needed. New developments are slow to catch on. The amount of information presented to clinicians is impossible to remember. Clinical decision support in the EHR could make a difference by presenting up to date best practices at the right time and place. However, it is burdensome to keep decision support up to date. Clinicians hate intrusive irritating pop-ups. Clinicians learn to ignore them. The SAGE model explores ways to create consistency through standardization, relevance through detail, and efficiency by being timely. Efficient dissemination of new information can be made possible through interoperability with EHRs. Decision support associated with CPOE is inherently an interruption of flow. For physicians to welcome such interruptions, to see value in decision support, remains an elusive goal. If I can learn from the system -perhaps avoid pop-ups by anticipating and applying knowledge, or get some feedback on how I am doing, perhaps I could experience satisfaction from learning a system- a sense that I am becoming a better doctor. It would then become worthwhile. The system would be my teacher, rather than my critic. The EHR would be my ally, my partner in caring for patients. (Perhaps the computer game development industry could help.)

John Butler, MD