Quality Management in Health Care thanks this issue's Guest Editor, Dr Farrokh Alemi of the Center for Health Administration and Policy of the College of Health and Human Services, George Mason University, for organizing this collection of articles dealing with probabilistic risk assessment of the occurrence of rare adverse events. Dr Alemi and the colleagues whom he has enlisted are providing readers with valuable insights into a useful but sometimes overlooked methodology that can be notably effective in the presence of small databases and samples.
In the introductory article, Dr Alemi points out that, while it is imperative to seek out the patterns and causes of adverse events in the course of patient care, such sentinel events are relatively rare, creating daunting technical problems in collecting enough data to justify drawing conclusions about risk. He proposes probabilistic risk assessment as a practical research strategy in the presence of very small samples. In the second article, Alemi presents an example of the use of probabilistic risk assessment, taking into account the variety of sources of hazards.
Focusing on medication errors, Hovor and Walsh ask how changes in the process of care could be detected with the use of control charts based on data reflecting the omission of medication.
They conclude that "time-to-event" can be used to measure the frequency of medication omission errors, despite the relative rarity of such an event. Kitsantas, Kang, and Yang point to the value of Bayesian network formulation as a tool for encoding and demonstrating probabilistic relationships among variables. This approach is extremely valuable in the presence of small samples consisting of sentinel events. Using a sample consisting of 9 incident reports (falls), in an assisted living setting, Song and Chila demonstrate the use of probabilistic risk analysis in identifying the risk factors associated with falls in this patient population.
The statistical approach known as failure mode effects analysis (FMEA) is being applied to an increasing extent in analyzing and avoiding sources of risk to patients. Day, Dalto, Fox, Allen, and Ilstrup report on their use of FMEA in detecting factors in a hospital's trauma patient registration process that may have enhanced risk to the patients. Their sample consisted of near-miss events over a 5-year period involving trauma patient registration and the effectiveness of the associated electronic data system.
Hovor and O'Donnell, the authors of another study on medication errors reported in this issue, present an overview of the use of the Bayesian Causal Network Model in identifying the systemic and behavioral factors that may increase the risk of medication errors.
Jean Gayton Carroll, PhD Editor