From the 2018 HVPAA National Conference
Matthew Johnson (Michigan Medicine), Katie Schwalm (Michigan Medicine), Robert Chang (Michigan Medicine), Christopher Petrilli (Michigan Medicine)
Background
Clinical decision-making is a nuanced process that relies on gathering relevant information to formulate the most appropriate decision for a patient. Current electronic health records (EHR) can access vast, potentially overwhelming volumes of data. However, the ability to identify high value care opportunities requires system level insight functionality.
Objectives
The use of industrial and systems engineering principles at Michigan Medicine has improved the ability to define discordant clinical decision making and potential opportunities to improve clinical decision support (CDS) to deliver higher value care. A standardized framework model was developed to identify opportunities for CDS that utilize fundamental engineering principles and adapt those to clinical situations. A review of this framework will use highlighted case studies to define key components of developing effective clinical decision support (Figure 1).
Methods
Systems engineering methodology works through defining and verifying a problem and effected population, scoping/resourcing the project, developing the current state, assessing gaps and appropriateness of care, developing a future statemodel of care, and implementing and sustaining interventions. This standard methodology, accompanied by a multidisciplinary team approach, allows a team to work together to achieve CDS mechanisms for both increased quality and productivity. The framework sets the foundation for incorporation of discrete event modeling and probabilistic modeling that can be translated to improved CDS and deliver higher value care.
Results
Case studies at Michigan Medicine have demonstrated that clinician anecdotes regarding care delivery limitations can be an excellent source for system level, CDS initiatives. The framework has contributed to translating these anecdotes to CDS interventions. Some examples have included reduced ED length of stays by 15-20% for low risk chest pain patients to providing automated notifications to oncologists on inpatient clinical status changes on their patients. This framework has contributed to CDS interventions to control laboratory utilization that have reduced downstream ordering of serotonin release assays by 75% and reduced CBC ordering by 25% in presence of clinical stability.
Conclusion
Making a clinical decision is individual to the provider, but by providing relevant clinical information we can facilitate interventions that deliver higher value. The CDS framework model has developed a foundation for evaluating these clinical opportunities to deliver more robust decision support while reducing utilization and cost.
Implications for the Patient
In various case studies, the Department of Internal Medicine at Michigan Medicine has demonstrated that you can provide high value results feedback, reduce laboratory utilization and provide targeted clinical communication on patients.