Development of a Clinical Decision Support Tool for Facilitating Naturalistic Decision Making and Improving Blood Culture Utilization

Sepsis is a life-threatening bloodstream infection that affects 8 percent of infants in the pediatric intensive care unit (PICU) and is responsible for 25 percent of PICU mortality. Blood cultures are the only way to diagnose sepsis; however, overuse of blood cultures may result in unnecessary laboratory tests, unnecessary antibiotic use, prolonged hospitalization, and increased healthcare costs. Existing clinical decision support (CDS) tools for optimizing blood culture utilization rely on analytic approaches to determine the pretest probability of bloodstream infections. Thus far these tools have not included pattern match (i.e., the prior experience of the clinician) nor have these tools incorporated the evaluation of potential actions to be considered for a given situation. These types of strategies to reach an informed decision are termed naturalist decision making. Not addressing the naturalistic and intuitive nature of certain clinical decision making results in low adoption of CDS tools by providers when these additional cognitive steps are necessary.

This study developed an electronic health record (EHR)-embedded CDS tool that draws upon the strengths of analytical and naturalistic decision making. The team applied a sociotechnical systems approach and a user-centered design method to guide development of the CDS tool. The tool was implemented in the PICU at the Johns Hopkins Hospital (JHH), and an interrupted time series study design was used to evaluate its impact on blood culture utilization and patient outcomes.

The specific aims of the research were as follows:

• Examine individual and team cognitive work associated with obtaining a blood culture. 
• Develop an EHR-embedded CDS tool to reduce unnecessary blood culture utilization. 
• Implement the CDS tool in the PICU at JHH and assess its use and impacts on blood culture utilization and patient outcomes. 

The research team conducted interviews with clinicians in the PICU at JHH to examine the decision-making and cognitive work associated with obtaining a blood culture. In addition, the team assessed the existing EHR system to identify challenges to collecting key information required for blood culture decision making and sepsis diagnosis. Based on findings from the interviews and review of the EHR, the team then collaborated with technology experts within Johns Hopkins and engaged clinicians in the PICU at JHH (e.g., attending physicians, fellow physicians, resident physicians, nurses) to iteratively develop the electronic CDS tool. The team pilot-tested the CDS tool in JHH’s PICU and has been continuously collecting data on patient outcomes and blood culture use. The results of the research show that the use of a clinical decision tool that relies on analytical and naturalistic decision making reduced blood culture utilization in a PICU and may be generalizable to a broader set of healthcare issues.

The research team has demonstrated the feasibility of including naturalistic processes within the CDS tool that was developed to access the need for a blood culture. The implication for CDS design includes the need to undertake the cognitive work associated with obtaining a blood culture that includes: (1) knowing patient history, (2) recognizing patient with change of status, (3) ruling out causes of change of status, and (4) making and validating decisions on blood culture ordering. The process and architecture used in the development of this CDS tool can be applied to other conditions in the PICU, such as venous thromboembolism, acute respiratory distress syndrome, traumatic brain injury, ventilator management, and central line-associated blood stream infection.