Validating a Global Measure of Severity in Children with Chronic Conditions

Abstract

Background: Approximately 20% of children live with a chronic physical condition, such as asthma, epilepsy, or diabetes. These conditions place considerable burden on children, their families, clinicians, and the health system. However, these burdens are reduced when conditions are effectively managed, typically accomplished by appropriately monitoring the severity and progression of the condition. Several condition-specific scales exist for measuring severity in children but are limited in their clinical utility for general practitioners or pediatricians who care for children with different conditions. Objectives: This study aimed to validate the Global Assessment of Severity of Illness (GASI)—a single-item scale that can be used to measure severity in children with different chronic physical conditions. Study objectives were to examine the construct validity, test-retest reliability, responsiveness, and sensitivity/specificity of the GASI. Methods: Clinicians assessed severity of asthma, food allergy, epilepsy, diabetes, and juvenile arthritis in 56 children using the GASI, Duke Severity of Illness Scale (DUSOI; the external clinical anchor), and a general visual analogue scale (VAS). Parents reported on child health-related quality of life using the KIDSCREEN-27. Kendall’s Tau-c and area under the receiver operating characteristic curve (AUC) determined the strength of association between measures. Fisher’s Exact test indicated whether the GASI could discriminate between children with and without multimorbidity. McNemar’s test, the Kappa coefficient, and weighted Kappa assessed stability in GASI ratings over time. The standardized response mean and Guyatt’s responsiveness index examined internal and external responsiveness, respectively. AUC determined sensitivity/specificity. Clinician characteristics, as potential confounders, were investigated within AUC regression models. Results: The GASI demonstrated strong correlations with the DUSOI composite score (τ-c= 0.57-0.63; AUC= 0.83-0.96) and VAS (τ-c= 0.78) and weak correlation with health-related quality of life (τ-c < 0.1). Lack of discrimination between children with and without multimorbidity was indicated by Fisher’s Exact test (p-value > 0.05). Moderate to substantial test-retest reliability was supported by McNemar’s test (p-value > 0.05), Kappa (κ= 0.79; CI= 0.51-1.00), and weighted Kappa (κ-w= 0.57; CI= 0.36-0.78). The GASI was largely responsive (Cohen’s d= 0.84; CI= 0.68-1.11) and the magnitude of sensitivity/specificity was low to moderate (AUC= 0.62-0.81). Construct validity was excellent regardless of whether regression modeling accounted for type of diagnosis, clinician, or child age. Conclusion: Initial evidence supports validity of the GASI to make meaningful comparisons of severity between different chronic conditions in children. Future research using larger samples should aim to replicate these findings and test inter-rater reliability between different health professionals. Such work is needed to fill knowledge gaps in comparative pediatric research and, potentially, simplify clinical practice.