Event Interval Probability Explained with Healthcare Case Study

SUMMARY & CONCLUSIONSThe case study involves 25 deaths and near-death episodes of babies in a neonatal unit. A nurse in the unit has been convicted of murder and attempted murder; therefore, the dataset is an extreme example of recognizing causal system unreliability and describing the actual reliability of that system without any knowledge of what is inside the causal system. The presence of an unknown serial killer poses a challenge to traditional reliability approaches; however, the new data analysis method, Event Interval Probability (EIP), does not rely on knowledge of what is inside the system producing events.When attempted murder is combined with murders, the reliability of the neonatal unit drops from normal and acceptable to dangerously unreliable on the very first day of the 383-day-long dataset, with a 0.00683 false positive. The false positive is the probability of being wrong if we declare the neonatal unit unreliable. Without probabilities, the nurse was allowed to stay in the neonatal care unit for an additional year, which resulted in 23 more deaths and near-death events. The case study reveals that the system-generated events exhibit irrational time intervals, suggesting that the causes and effects behind these events are also irrational, perhaps a characteristic one could expect from a serial killer.Reliability engineers understand that the reliability of complex, steady-state, repairable systems, where repairs restore the system to a condition as good as new, can be expressed as e-λt, with λ representing the failure rate and t representing time. Importantly, understanding the causal basis of λ is unnecessary. This concept is used in EIP as the null hypothesis that λ remains constant, with its value based on a comparative population average, a goal, historical performance, or an acceptable standard. The concept of a null hypothesis is borrowed from statistics; however, it is either rejected or not based on probability values (p-values) calculated using probability theory, which assesses the likelihood that the null hypothesis could have produced the observed event intervals.EIP often detects a system change from reliable to unreliable with just a single data point; therefore, analysis must be performed at every event to ensure unreliability is identified and addressed immediately, preventing future events. The study also identifies a zero-point chaos theory attractor within the dataset (inside the neonatal unit), where the interval between events approaches zero. This neonatal unit dataset is contrasted with two social system precursor datasets, both involving individuals within the causal systems charged with murder.