Identifying developmental stuttering and associated comorbidities in electronic health records and creating a phenome risk classifier.

The study used hospital records, not school or private speech pathology files, so many people who stutter may be missing. The control group may include hidden stuttering cases, because speech problems can go unnoticed or unrecorded during medical visits. Enriched diagnosis groups are not confirmed clinical relationships. They can arise from broad billing codes, coding habits, or shared medical care. The study cannot show that stuttering causes any listed condition, or that any condition causes stuttering. A very conservative correction for many comparisons may have missed real associations, including anxiety and ADHD. The classifier estimate cannot be converted into exact sensitivity, specificity, or negative predictive value because true absence of stuttering in controls is uncertain. Race recording was done by a third party and was not uniform, so it may have introduced some bias.

The authors report no declarations of interest. The paper lists NIH and non-U.S. government research support.

Do not read the enriched diagnosis groups as proof that stuttering causes sleep, hearing, allergy, infection, neurological, or developmental conditions. The study only compared records from a hospital database, and the authors state that causality cannot be assumed.