Evaluating AI chatbots in enuresis nocturna information: A comparative analysis of readability, reliability and...

Nocturnal enuresis, defined as involuntary urination during sleep in children aged five years and older, is a prevalent condition affecting millions of children worldwide with significant psychosocial implications. While artificial intelligence-powered chatbots have rapidly emerged as accessible health guidance tools, the clarity, consistency, and reliability of their information remain uncertain.

This study systematically compared the quality, readability, and clinical reliability of responses from three leading large language models (OpenAI GPT-4o, Google Gemini 2.5 Pro, and DeepSeek R1) to the 40 most frequently asked questions about childhood nocturnal enuresis. Designed as a cross-sectional observational study, questions identified through decade-long search engine data analysis were organized into four thematic categories and posed in Turkish to the chatbots. As the study exclusively analyzed publicly available AI outputs without involving human participants, institutional review board approval was not required. Responses were evaluated using the Flesch Reading Ease Score (FRES) and Flesch-Kincaid Grade Level (FKGL) for readability; Ensuring Quality Information for Patients (EQIP) and modified DISCERN (mDISCERN) tools for quality, applying a double-blind methodology. Group comparisons were conducted using ANOVA and post-hoc tests.

All chatbots generated texts in the "difficult to read" range (FRES: 33.6-40.9), requiring university-level comprehension (FKGL: 20.3-21.9), thereby substantially limiting accessibility for the target parent audience. DeepSeek demonstrated significantly superior performance on EQIP criteria (70.4±9.2), outperforming both Gemini (57.8±6.3) and GPT-4o (54.7±4.8) (P=0.003). However, mDISCERN scores remained low across all models (2.10-2.30, P=0.183).

Current AI chatbots offer only limited potential as reliable and accessible health information sources on nocturnal enuresis and are not yet adequate for clinical use. Future developments must prioritize plain language implementation, structured information delivery, and alignment with current pediatric urology guidelines to transform digital health tools into genuinely beneficial and clinically reliable resources for families.

I.