Professor Robert Cundick
Textbooks: Hunink, H. and Glasziou, P. (2009) Decision Making in Health and Medicine: Integrating Evidence and Values (7th ed.). Cambridge England: Cambridge University Press
Berner, E.S. (Ed.) (2007). Clinical Decision Support Systems: Theory and Practice (2nd ed.). New York, NY: Springer
Osteroff, J.A., Teich, J.M., Levick, D., Saldana, L., Velasco, F.T., Sittig, D.F., Rogers, K.M. and Jenders, R.A. (2012) Improving Outcomes with Clinical Decision Support: An Implementers Guide (2nd ed.). Chicago, IL: HIMSS

Clinical Decision Support (CDS) was my favorite class despite requiring the most work of any course in the medical informatics degree program. Dr. Robb Cundick trained with Dr. Homer Warner in the Department of Medical Biophysics and Computing (now the Department of Medical Informatics) at the University of Utah Medical Center LDS Hospital in the 1970s when that department invented some of the very first electronic medical record and clinical decision
support systems in the United States. His passion for the field was palpable. CDS is the core value driver for most of the benefits expected from health information technology, since without it an EMR is essentially just an electronic filing cabinet. Presenting best practices recommendations and key supporting medical data to the clinician at the point of care is the probably most important single intervention for achieving the improvements in health care
quality and cost-effectiveness that we all strive for.
support systems in the United States. His passion for the field was palpable. CDS is the core value driver for most of the benefits expected from health information technology, since without it an EMR is essentially just an electronic filing cabinet. Presenting best practices recommendations and key supporting medical data to the clinician at the point of care is the probably most important single intervention for achieving the improvements in health care
quality and cost-effectiveness that we all strive for.
If I succeed in my goal of transitioning to a career in medical informatics, there will be innumerable opportunities to apply the lessons from
this course. We learned about CDS intervention types including order sets and smart documentation forms, critiques and warnings, selected relevant data summaries, evidence based guidelines, predictive and retrospective analytics, filtered reference information and knowledge resources, expert workup and management advisors, and result or event driven alerts. We studied the process of reasoning under conditions of uncertainty, Bayes theorem, and the use of decision trees to incorporate patient values into medical decision-making and optimize management strategies. We looked at approaches to knowledge representation, the use of machine reasoning to model human clinical decision-making, data warehousing, and data mining. We learned the CDS Five Rights: the right information delivered to the right person in the right intervention format through the right channel at the right point in workflow.
this course. We learned about CDS intervention types including order sets and smart documentation forms, critiques and warnings, selected relevant data summaries, evidence based guidelines, predictive and retrospective analytics, filtered reference information and knowledge resources, expert workup and management advisors, and result or event driven alerts. We studied the process of reasoning under conditions of uncertainty, Bayes theorem, and the use of decision trees to incorporate patient values into medical decision-making and optimize management strategies. We looked at approaches to knowledge representation, the use of machine reasoning to model human clinical decision-making, data warehousing, and data mining. We learned the CDS Five Rights: the right information delivered to the right person in the right intervention format through the right channel at the right point in workflow.
Importantly, the course also addressed the principles of developing and implementing CDS systems in the real world. There was emphasis on
involving a broad spectrum of stakeholders all throughout a careful process of project design, planning, and implementation and on avoiding potential pitfalls of CDS such as poor usability, alert fatigue, adverse effects on workflow, and inadequate change management. We also learned the importance of objectively evaluating the functioning and effectiveness of every CDS intervention, using that data along with user feedback to deploy an improved system, and then repeating that process in an iterative cycle of improvements. Finally, we addressed the legal, social, and ethical implications of CDS systems.
involving a broad spectrum of stakeholders all throughout a careful process of project design, planning, and implementation and on avoiding potential pitfalls of CDS such as poor usability, alert fatigue, adverse effects on workflow, and inadequate change management. We also learned the importance of objectively evaluating the functioning and effectiveness of every CDS intervention, using that data along with user feedback to deploy an improved system, and then repeating that process in an iterative cycle of improvements. Finally, we addressed the legal, social, and ethical implications of CDS systems.
As a practicing physician, I remain concerned and disappointed that virtually no CDS is incorporated into our inpatient or outpatient EMR’s and that the one system which is available, an adverse drug interaction checker, functions so poorly as to be virtually useless, producing endless alerts that are irrelevant or fail to take into account the patient's full clinical situation. On the other hand, I routinely use CDS systems available outside the EMR as third-party web applications in my practice. I can open the NCCN guidelines, UpToDate Medicine, or PubMed from one-click shortcuts on my desktop, and I refer to one or more of these resources every day, although it would be better if at least some of the information was available in the EMR in a “push” rather than a “pull” format. One reason we haven't achieved this is the sheer difficulty of the knowledge engineering in trying to model complex clinician reasoning.
I have attached a file combining four individual assignments: choosing and briefly evaluating three guideline sets and three performance measures, giving a brief critique of a paper describing a CDS System, using a decision tree to determine best therapy for a patient with an infected ankle fracture and doing a senstivity analysis on that tree, and using probability revision methods to give clinical perpective on testing for Lyme disease. The final team project challenged us to use everything we had learned to design and describe in detail how we would build and implement a CDS intervention. My teammates Erin Laney and Peter Gordon and I proposed an electronic population screening system capable of using family medical history data to identify women at increased risk of carrying harmful mutations in the BRCA1 and BRCA2 breast cancer susceptibility genes and therefore potentially at increased risk of developing breast or ovarian cancer. The system was rapid, simple, and automated enough to deploy in a primary care physician’s office or in a mammography center and was capable of appropriately prompting providers to initiate the referral of high risk patients to appropriate specialists in genetic counseling and DNA testing (see attached link).