The interrater reliability of a validated bronchiolitis severity assessment tool.

BACKGROUND

We previously constructed and tested a bronchiolitis severity assessment tool in 2 independent hospitals. The model uses age, work of breathing, dehydration and tachycardia to successfully predict disease severity.

OBJECTIVE

To prospectively measure the interrater reliability of a bronchiolitis severity assessment tool and of its component variables.

DESIGN

Prospective observational survey.

SETTING

A county teaching hospital emergency department serving a mixed urban and rural population with an emergency medicine residency program in 2-3-4 format.

SUBJECTS: Thirty-two physicians evaluated a convenience sample of children aged less than 18 months presenting to the emergency department with a clinical diagnosis of bronchiolitis during a single season.

METHODS

Two physicians independently examined each patient. Each physician completed a physical examination template that included the variables used in the severity assessment tool. Interrater agreement was measured for the variables work of breathing and dehydration and for the tool as a whole using a weighted kappa statistic.

RESULTS

One hundred and forty-six cases were enrolled. Twenty-five were dropped for incomplete data collection. The actual weighted agreement on overall classification was 92%; expected, 73%, kappa = 0.676; P < 0.0001. The actual weighted agreement for dehydration was at 95%; expected, 92%, kappa = 0.305; P = 0.0001. The agreement for work of breathing was 95%; expected, 86%; kappa = 0.611; P < 0.0001. The overall model showed better interrater reliability than its individual components.

CONCLUSIONS

Overall interrater reliability for this bronchiolitis severity assessment tool is substantial.