Abstract:
Background: Humans often prefer certain terminal digits (e.g., even numbers, zeros, or fives), while automated systems (machines) theoretically do not. This study compared the terminal-digit distribution of human-recorded versus machine-recorded neonatal heart rate and oxygen saturation and examined terminal digit preferences in human- recorded respiratory rate.
Method: Data consisting of approximately 200 observations per vital sign per neonate from six infants was collected, including hourly human-recorded and continuous machine-captured values. This data was evaluated to assess if recorded digits matched expected theoretical distributions (50% even; 20% ending in 0 or 5). Fisher’s exact tests compared human vs. machine prevalences of even digits, and confidence intervals
identified deviations in respiratory rate terminal-digit frequencies.
Results:
Heart rate:
1199 human observations (52.5% even vs. 47.5% odd)
1002 machine observations (47.2% even vs. 52.8% odd)
Absolute difference of 5.3% between human vs machine recording for both odd and even numbers, p = 0.0131 (Fisher’s exact test).
O2 saturation:
1103 human observations (52.8% even vs. 47.2% odd)
1002 machine recordings (50.8% even vs. 49.2% odd)
Absolute difference of 2% between human vs machine recording for both odd and even numbers, p = 0.3911 (Fisher’s exact test).
Achieved respiratory rate:
Only human recordings used. 1154 observations with 47.6% (95% CI:44.7–50.5%) of values ending in five or zero. This strongly exceeds the theoretical zero/five count of 20% (95% CI:17.7–22.3%).
Conclusion: Human recordings exhibit clear digit bias toward even numbers in heart rate and numbers ending in zeros/fives in respiratory rate, while machines show a bias towards odd numbers in heart rate recordings. Such biases may reflect cognitive rounding or ventilator settings influencing the data being recorded, rather than true physiological values. These findings suggest a need for larger studies and potential revisions to neonatal monitoring protocols to ensure accuracy in respiratory-rate documentation and support.
