Does a standardised scoring system of clinical signs reduce variability between doctors' assessments of the potentially dehydrated child?

AIMS: Clinical assessment of dehydration in children is often inaccurate. We aimed to determine if a scoring system based on standardised clinical signs would reduce the variability between doctors' assessment of dehydration.
METHODS: A clinical scoring system was developed using seven physiological variables based on previously published research. Estimated percentage dehydration and severity scores were recorded for 100 children presenting to a Paediatric Emergency Department with symptoms of gastroenteritis and dehydration by three doctors of different seniority (resident medical officer, registrar and consultant). Agreement was measured using intra-class correlation coefficient (ICC) for percentage ratings and total clinical scores and kappa for individual characteristics.
RESULTS: Estimated percentage dehydration ranged from 0-9%, mean 2.96%, across the three groups. Total clinical scores from 0-10, mean 2.20. There was moderate agreement amongst clinicians for the percentage dehydration (ICC 0.40). The level of agreement on the clinical scoring system was identical (ICC 0.40). Consultants gave statistically lower scores than the other two groups (Consultant (Con) vs. Resident P = 0.001, Con vs. Registrar P = 0.013). There was a marked difference in agreement across characteristics comprising the scoring system, from kappa 0.02 for capillary refill time to 0.42 for neurological status.
CONCLUSION: The clinical scoring system used did not reduce the variability of assessment of dehydration compared to doctors' conventional methods. In order to reduce variability improving education may be more important than production of a scoring system as experience appears to be a key determinant in the assessment of a potentially dehydrated child.