How do clinician and parent-reported data differ? An analysis of similarity and difference in the datasets from a cross-syndrome genetics cohort study (GenROC)

Background

Parent/patient-reported (PRD) datasets provide ready access to phenotypic data for monogenic neurodevelopmental disorders, yet their concordance with clinical data is unclear.

Methods

In the GenROC study, 547 children (mean age 7.6 years, balanced sex ratio) had parallel parent-reported web questionnaires and clinician-reported (CRD) Human Phenotype Ontology proformas. We compared the two sources per participant by system, gene and gene group and overall for quantity, detail and similarity.

Results

547 probands were analysed ranging in age from infancy to 16 years (mean 7.6) with similar gender distribution. PRD provided more terms for dental, gastroenterology, immunology and respiratory systems and for vision (p<0.001 for all) and to a lesser degree for cardiac (p=0.0012). CRD provides more detail than PRD for most gene subgroups, combined systems and for neurology (p<0.001). Similarity scores were low overall per participant (mean 0.38 for combined). Similarity scores were highest for cardiac (mean 0.74) and lowest for Ear/Nose/Throat(ENT) (mean 0.34). There was minimal difference in similarity scores across gene groups or between the top 10 genes—scaffold adaptor gene groups had the highest (mean 0.43) as did STXBP1 (mean 0.5) and CACNA1A (0.49). CRD is more similar to published syndrome phenotypes for syndromic genes.

Conclusions

Parents reported more common childhood phenotypes, such as asthma and dental issues, while clinicians provided clinical phenotype descriptors, such as brain morphology and seizure semiology. It is important to understand the differences when designing studies and using datasets to appreciate their strengths and limitations.