Case Study

Better Medicine by Default

Problem

Computerized systems can be designed to aid physicians in making the best decisions and recommendations for their patients. Inherent in any system is the existence of defaults – preselected options adopted by a computer program or other mechanism that will be chosen when the user specifies no alternative. It is important to understand how opt-in and opt-out defaults in electronic order templates affect physicians’ treatment decisions.

Research

Participating medical residents were randomly assigned one of two different default displays for order sets (a group of related orders which a physician can place with a few keystrokes or mouse clicks). One default display had all potential orders visible and unselected, such that participants needed to “opt in” to make an order. The second default display had all of the same potential orders visible, but options were preselected such that participants needed to “opt out” to remove order.

Results

Opt-out defaults resulted in ordering more unnecessary items while opt-ins defaults led to a greater level of order omission (not ordering something that the patient needed). Errors were made with both default systems, revealing that the default display does indeed affect physicians’ decisions. However, the severity of these errors was low. Physicians were able to override incorrect defaults (and succeeded in making the right decisions) when the risk of error would lead to extreme consequences. However, in many low-risk cases, action was not taken on incorrect defaults. These results highlight the importance of designing systems with smart defaults, particularly concerning low-impact items where defaults are less likely to be overridden.

Why it matters

Substantial sums of federal funding are being allocated to the ‘meaningful use’ of electronic health. Defaults used in electronic order sets influence medical treatment decisions when the consequences to a patient’s health are low (but may still lead to unnecessary costs and treatments). This pattern suggests that physicians cognitively override incorrect default choices, but only to a point. Thus, defaults for low-impact items on electronic templates warrant careful attention because physicians are unlikely to override them.

More Information

The complete study can be accessed at the following link.