Combining ratings from multiple physician reviewers helped to overcome the uncertainty associated with adverse event...

Area of Science:

• Medical safety and quality improvement
• Health services research
• Patient outcome analysis

Background:

• Adverse events (AEs) represent significant poor patient outcomes stemming from medical care.
• Accurate detection of AEs is crucial for improving patient safety and healthcare quality.
• Existing AE detection methods may suffer from misclassification, leading to an overestimation of risks.

Purpose of the Study:

• To quantify the misclassification rate of current AE detection methods.
• To evaluate the impact of combining physician AE ratings on classification accuracy.
• To assess factors influencing the correct classification of potential AEs.

Main Methods:

• Latent class analysis was employed to estimate AE prevalence and reviewer accuracy.
• Simulations involving 10,000 cases rated by five independent reviewers were conducted.
• The study analyzed the influence of AE prevalence, reviewer accuracy, and agreement levels on correct classification probability.

Main Results:

• Individual reviewer sensitivity and specificity for AE classification were 0.86 and 0.94, respectively.
• With a 3% AE prevalence, the positive predictive value increased from 31% (1/3 reviewers) to 51% (2/3 reviewers).
• Positive predictive values for AE occurrence were positively correlated with AE prevalence, reviewer accuracy, and the number of reviewers.

Conclusions:

• Current AE detection methodologies tend to overestimate the actual risk of adverse events.
• Enhancing diagnostic certainty for AEs can be achieved by increasing the number of independent physician reviews.
• This highlights the importance of consensus-based review in accurate AE identification.