Inter-Rater Reliability Between Critical Care Nurses Performing a Pediatric Modification to the Glasgow Coma Scale.

Objectives: Estimate the inter-rater reliability of critical care nurses performing a pediatric modification of the Glasgow Coma Scale in a contemporary PICU.

Design: Prospective observation study.

Setting: Large academic PICU.

Patients/subjects: All 274 nurses with permanent assignments in the PICU were eligible to participate. A subset of 18 nurses were selected as study registered nurses. All PICU patients were eligible to participate.

Interventions: None.

Measurements And Main Results: PICU nurses were educated and demonstrated proficiency on a pediatric modification of the Glasgow Coma Scale we created to make it more applicable to a diverse PICU population that included patients who are sedated, mechanically ventilated, and/or have developmental disabilities. Each study registered nurse observed a sample of nurses perform the Glasgow Coma Scale, and they independently scored the Glasgow Coma Scale. Patients were categorized as having developmental disabilities if their preillness Pediatric Cerebral Performance Category score was greater than or equal to 3. Fleiss' Kappa (κ), intraclass correlation coefficient, and percent agreement assessed inter-rater reliability for each Glasgow Coma Scale component (eye, verbal, motor) and age-specific scale (≥ 2 and < 2-yr-old). The overall percent agreement between study registered nurses and nurses was 89% for the eye, 91% for the verbal, and 79% for the motor responses. Inter-rater reliability ranged from good (intraclass correlation coefficient = 0.75) to excellent (intraclass correlation coefficient = 0.96) for testable patients. Agreement on the motor response was significantly lower for children with developmental disabilities (< 2 yr: 59% vs 95%; p = 0.0012 and ≥ 2 yr: 55% vs 91%; p = 0.0012). Agreement was significantly worse for intermediate range Glasgow Coma Scale motor responses compared with responses at the extremes (e.g., motor responses 2, 3, 4 vs 1, 5, 6; p < 0.05).

Conclusions: A pediatric modification of the Glasgow Coma Scale performed by trained PICU nurses has excellent inter-rater reliability, although reliability was reduced in patients with developmental disabilities and for intermediate range Glasgow Coma Scale responses. Further research is needed to determine the effectiveness of this Glasgow Coma Scale modification to detect clinical deterioration.