Digit preference and biased conclusions in cardiac arrest studies

Area of Science:

• Emergency Medicine
• Medical Informatics
• Biostatistics

Background:

• Cardiac arrest (CA) prognosis is time-dependent, with delayed cardiopulmonary resuscitation increasing mortality.
• Reported times in CA management studies may be subject to

Purpose of the Study:

• To investigate the presence of digit preference bias in times of day related to cardiac arrest management.
• To assess if reported times in a national registry exhibit non-random patterns.

Main Methods:

• Analysis of 47,211 times-of-day from 6,131 cardiac arrests in the French National Electronic Registry.
• Comparison of observed frequencies of times ending in 0 or 5 with expected random distribution.
• Statistical evaluation of overrepresentation of times as multiples of 15.

Main Results:

• Significant overrepresentation of times ending in 0 (e.g., :00) and 30 minutes.
• Times reported as multiples of 15 minutes (e.g., :00, :15, :30, :45) were significantly overrepresented.
• Observed frequencies deviated significantly from expected random distribution (p < 0.0001).

Conclusions:

• Digit preference phenomenon demonstrably influences prospectively collected times in cardiac arrest management.
• Caution is advised when interpreting studies relying on non-automatic time recordings without bias assessment.
• This bias can affect the accuracy of time-sensitive data in emergency medicine research.