Module Components

• Overview
• Assignments

Topics

• Decision Theories
• Clinical Decision Support Systems
• Clinical Decision Support Systems - Examples
• Clinical Decision Support Systems - Benefits and Challenges
• References

Nurs 460 - HEALTH INFORMATICS FOR REGISTERED NURSES

Module 6: Technology to Support Decision-Making

Clinical Decision Support Systems

Learning Activities

What is CDSS at http://www.healthit.gov/policy-researchers-implementers/clinical-decision-support-cds

Clinical Decision Support (2.43) at https://www.youtube.com/watch?v=fOQuPfiaq_8

Criteria for CDSS

A Clinical Decision Support System encompasses any method used by a decision maker that assists in synthesizing information, such that s/he can weigh trade-offs and make a health related decision. You must keep in mind that CDSSs support, but do not MAKE decisions. There are several criteria for a clinically useful CDSS:

• Knowledge based on best evidence
• Comprehensive
• Knowledge regularly updated
• Information drawn from current sources
• Rigorous performance validation
• System improves clinical practice
• Clinician is in control
• User-friendly

Application Areas for CDSS

A CDSS does not have to be a computer; however, it must provide for user inputs, make the decision maker an integral part of the system, define alternatives, provide a structured decision-making model, offer reproducible alternatives, be consistent, and balance variables, such as time and cost. Therefore, it usually is a computer. Overall, important decisions usually require huge quantities of information. Technologies like the electronic health record, Geographical Information Systems (GIS), public health databases, simulators, etc. make huge quantities of information available to use in making better decisions.

Types of Computerized CDSS

There are numerous types of CDSS, which fall into three general categories: active, semi active, and passive.

Active systems are simple closed loop systems that can be used to monitor internal parameters, such as venous backpressure in an IV system. These systems are active, because they are programmed to make a predetermined decision based on information. They require little input from the user and give little guidance beyond an annoying noise and/or predetermined change in status (e.g., stop flow).

Semi-active systems are the next level of CDSS. These systems are the most common. They monitor parameters, orders, and other inputs against a pre-defined metric (e.g., time, ST segment length, etc.) and notify the user of violations of the parameters. These systems include Reminders, Alarms, and Alerts.

• Reminders simply remind a clinician to perform an activity, such as renew and antibiotic order.
• Alarms tell clinician of a violation of parameters. For example, residents of an Alzheimer’s unit may wear wristbands that notify nurse if they enter the wrong room.
• Alerts come in two types, synchronous and asynchronous. Synchronous alerts immediately alert the user to violations of a certain preset parameter. For instance, while entering a warfarin Na+ order into the CPOE, an alert notifies the provider that the patient is also taking ibuprofen. The problem with synchronous alerts is that they invoke “noise” with repeated alerts about the same parameter, which can lead to alerts being ignored. Asynchronous alerts monitor parameters over time and alert clinicians to violations. For instance, the CDSS notifies a psychiatrist that serum Na+ and serum Li+ levels have been higher on each of the last three lab tests, indicating a possible renal problem. The system may suggest a decrease in dosage, change in prescription, or additional patient education.

These systems can also invoke noise. To overcome noise, the system should be based on excellent programming, including access for user inputs, and be set up properly by the users. Some programs give users the ability to select whether they want to be alerted about a certain parameter again. For example, an alert system is programmed to notify the user of BUN>20 mg/dl. Each day, it alerts a nurse that a person with chronic renal failure has an elevated BUN. A good system will offer a checkbox that says, “Do not alert me again about this parameter on this patient.”

Passive systems require a great deal of user input, but give a very informed decision in return. For instance, Consultative Systems take clinical information from the user and give a list of alternatives, whereas, Critical Systems take clinical information and the user’s planned interventions and provide feedback on the best alternative. They are passive, because they do nothing until asked by the user. Standard CDSS, Expert Systems (ES), & Artificial Intelligence (AI) or cognitive computing use combinations of heuristic models, databases, neural networks, geographic information systems (GIS), gaming simulations, and other technologies to assist decision makers in choosing alternatives.