P1.041: Variability and Educational Gaps in Public Information Resources on Myelomeningocele: An Analysis of Web-Based Resources and AI-Powered Large Language Model Responses

Background: Myelomeningocele (MMC) is a congenital diagnosis that requires multidisciplinary collaboration for determination of best treatment pathway, with prenatal and postnatal repair representing standard of care options. Approximately 20 centers in the United States offer fetal surgery for MMC, limiting access to care for these patients. We sought to investigate the availability and content of public resources about MMC.

Methods: Internet searches using Google were performed using 3 primary terms (myelomeningocele, spina bifida, myeloschisis) and 6 modifying terms to search common concerns and questions. A series of questions were similarly used to elicit responses from three AI-Powered Large Language Models (LLM), ChatGPT, Copilot and Gemini. A coding scheme was developed to evaluate the content of websites which contained 49 items across 4 categories: basic information, potential complications, treatment, and support system information. Two coders (NS and JL) independently rated the resources after achieving interrater reliability. Descriptive statistics were performed.

Results: 345 discrete websites were identified, of which 82 met inclusion criteria. Websites most commonly discussed the types of neural tube defects (n=67, 82.7%), description of MMC (n=69, 85.2%), and risk of hydrocephalus (n=65, 80.2%). Possibility of death (n=12, 14.7%), need for intubation (n=5, 6.2%), possibility of termination of pregnancy (n=7, 8.6%), and housing support options for families (n=8, 9.9%) were least commonly included. Search resources most commonly discussed option for postnatal surgery (n=67, 82.7%) compared to fetal/prenatal surgery (n=55, 67.9%). Of the 55 sites that discussed fetal/prenatal surgery, only 56.4% discussed specific evidence or risks pertaining to the procedure. On average, internet sources discussed 21 of the elements from our coding scheme (range 1, 37). AI generated responses were overall more comprehensive, with 100% (n=9) of AI generated responses discussing MMC diagnosis, risk factors, fetal surgery, and postnatal surgery. (Table 1) All nine AI generated responses discussed specific evidence or risks pertaining to fetal surgery for MMC. While all AI generated information informed parents that damage in MMC is ongoing throughout pregnancy, the majority of websites did not (72.8%).

Conclusion: Publicly available information about MMC is highly variable, with a noticeable gap in critical information regarding fetal versus postnatal treatment options particularly among website-based sources. AI-generated responses were more comprehensive in amalgamating information when targeted with specific questions about the diagnosis. There is a need for more standardized, detailed, and accessible public educational resources about MMC so families can inform themselves prior to making treatment decisions.