Residents' numeric inputting error in computerized physician order entry prescription.

Area of Science:

• Human-Computer Interaction (HCI)
• Health Informatics
• Patient Safety

Background:

• Computerized physician order entry (CPOE) systems with clinical decision support (CDS) reduce some prescription errors.
• However, prescription errors persist, with limited research on human-computer interaction factors like numeric input methods.
• Understanding these factors is crucial for enhancing CPOE usability and safety.

Purpose of the Study:

• To investigate the impact of numeric inputting methods and urgency levels on prescription errors.
• To categorize the types of numeric inputting errors occurring in CPOE systems.
• To identify specific digits or input methods associated with higher error rates.

Main Methods:

• Thirty residents performed prescribing tasks manipulating numeric input method (keyboard row vs. keypad) and urgency level (urgent vs. non-urgent).
• Prescribing behavior and errors were measured under controlled, sober conditions.
• Statistical analysis, including generalized linear models, was used to assess the effects of the manipulated factors.

Main Results:

• Urgent situations significantly increased errors, particularly when using the keyboard's numeric row.
• The effect of input method diminished when controlling for sober performance, highlighting urgency as a key factor.
• Omission and substitution were common error types, but transposition and intrusion errors were notably higher than previously reported. Digits 3, 8, and 9 showed higher error rates.

Conclusions:

• Urgency is a more significant driver of CPOE numeric typing errors than typing proficiency or habits.
• Employing a numeric keypad is advisable in urgent scenarios to minimize errors.
• Future CPOE design improvements could involve adjusting key sensitivity for less frequent digits and decimals, and leveraging error spatial incidence data for better error detection.