Validity of claims-based algorithms to identify neurodevelopmental disorders in children

Area of Science:

• Pediatric Health
• Medical Informatics
• Neurodevelopmental Disorders

Background:

• Accurate identification of neurodevelopmental disorders (NDDs) is crucial for timely intervention and research.
• Healthcare claims data offer a large-scale resource but require validation for specific conditions.

Purpose of the Study:

• To validate the accuracy of healthcare claim-based algorithms for identifying various NDDs in children.
• To assess the positive predictive values (PPVs) of these algorithms against medical record reviews.

Main Methods:

• Children aged ≤14 years with NDDs were identified using claims data from two Boston hospitals (2010-2014).
• Fifty cases per NDD were randomly sampled for independent medical record review by two physicians.
• Positive predictive values (PPVs) and 95% confidence intervals (CIs) were calculated.

Main Results:

• High PPVs were observed for autism spectrum disorder (94%), ADHD (88%), learning disability (98%), speech/language disorder (98%), intellectual disability (82%), and behavioral disorder (92%).
• Developmental coordination disorder (DCD) algorithms showed a low PPV (38%), with many false positives due to varying criteria and co-occurring conditions.

Conclusions:

• Claims-based algorithms are generally reliable for studying common NDDs in children.
• Algorithm refinement, particularly for DCD, is necessary to improve classification accuracy and clinical utility.