Evaluating Online Patient Information and AI-Generated Advice for Testicular Microlithiasis: Readability and Information Quality

BackgroundTesticular microlithiasis (TM) is commonly found incidentally on scrotal ultrasound, often resulting in considerable anxiety due to conflicting malignancy associations.Patients increasingly use the internet and large language models (LLMs) for information on health.Health Education England advises patient materials be written at a reading age of 11.We assessed the readability and quality of patient-facing resources and AI-generated adviceregardingTM. MethodsGoogle (Google LLC, Mountain View, CA, USA), Bing (Microsoft Corporation, Redmond, WA, USA), and Yahoo (Yahoo Inc., New York, NY, USA) were searched for "Testicular Microlithiasis," with the first two pages of results screened.After deduplication and exclusion of professional-facing sources, 13 English-language, patient-and public-facing websites were included.One response from each of GPT-5.2Instant (OpenAI, San Francisco, CA, USA), Gemini 3 Flash (Google LLC, Mountain View, CA, USA), and Claude Haiku 4.5 (Anthropic, San Francisco, CA, USA) was generated using a standardized patient-information prompt.Readability was assessed using the Flesch Reading Ease Score (FRES), Flesch-Kincaid Grade Level (FKGL), and Gunning Fog Index (GFI), while information quality was assessed using DISCERN.Exploratory comparisons between website and AI-generated content and between Google search subgroups used exact permutation tests based on the Mann-Whitney U statistic, with Holm adjustment across four outcomes separately for each comparison. ResultsThe 13 websites had poor readability (mean FRES 19.0; FKGL 15.3; GFI 17.8).AI responses had a higher mean FRES (41.2) and lower mean FKGL (11.1) and GFI (12.8), but none of the 16 resources met the adopted readability benchmarks.Mean summed DISCERN was 52.1.No website-AI comparison was statistically significant after Holm adjustment (all adjusted p≥0.157); no Google-subgroup comparison was significant either (all adjusted p=1.000).Two commercial resources contained potentially misleading symptom or dietary advice.None of the three AI responses provided verifiable source citations. ConclusionsWithin this evaluated sample, patient-facing TM resources exceeded both stringent and standard readability benchmarks and included occasional misleading commercial advice.Descriptively, the three AI responses had lower mean textual complexity and aligned with European Society of Urogenital Radiology (ESUR) management guidance, but exploratory statistical comparisons with websites were inconclusive, and all AI outputs lacked verifiable citations.Given the small sample, these findings should not be generalized to all online or AI resources; however, urologists should consider directing patients to clinician-reviewed, plain-English materials at diagnosis.