Algorithms to Define Abnormal Growth in Children: External Validation and Head-To-Head Comparison

Area of Science:

• Pediatric endocrinology
• Clinical decision support systems
• Growth monitoring

Background:

• Current growth monitoring algorithms for early disease detection offer contradictory definitions of abnormal growth.
• Existing algorithms have a low level of validation, necessitating external validation studies.

Purpose of the Study:

• To externally validate seven growth abnormality algorithms.
• To compare the performance of World Health Organization (WHO) versus national growth charts in algorithm application.
• To assess the impact of algorithm choice and growth chart type on early disease detection.

Main Methods:

• A case-referent approach was used, retrospectively applying seven algorithms to growth data.
• Data included children with Turner syndrome, GH deficiency, celiac disease (n=341) and apparently healthy children (n=3406).
• Sensitivity, specificity, and theoretical time to diagnosis reduction were calculated using WHO and national growth charts.

Main Results:

• The Grote clinical decision rule demonstrated higher sensitivity (4.6%–54%) than the Coventry consensus (0%–8.9%) among high-specificity algorithms.
• The Grote rule offered a better theoretical reduction in time to diagnosis (median 0.0–0.9 years vs 0 years).
• World Health Organization (WHO) growth charts yielded significantly higher sensitivity but lower specificity compared to national charts.

Conclusions:

• The Grote clinical decision rule showed the best performance for early detection of the studied diseases.
• However, its limited impact on reducing time to diagnosis highlights the need for superior algorithms.
• Further development of growth monitoring algorithms, utilizing appropriate growth charts, is recommended for improved clinical practice.