Patient classification systems: a new approach to computing reliability.

Patient classification systems are an integral part of managing nursing care that requires accuracy and consistency. Assessing interrater reliability is essential in monitoring and evaluating patient classification systems. Unfortunately, little attention has been paid to statistically assessing reliability. The most commonly used method, percentage of agreement, is limited and tends to be inflated due to chance variation. The ANOVA method, although more complex computationally, offers more accuracy and flexibility.