Simplifying Orthopedic Patient Education: A ChatGPT-Based Readability Analysis
TL;DR
ChatGPT revisions significantly improved readability scores, suggesting its potential as a tool to enhance patient comprehension and readability improvements were observed across all pathologies.
Abstract
Health literacy plays a crucial role in patient decision-making. National guidelines from the NIH and AMA recommend that patient education materials (PEMs) be written at or below a 6th-grade reading level. Four common foot and ankle conditions were studied: bunions, diabetic foot ulcers, broken ankles, and arthritis of the big toe. Each condition was searched on Google, and the first ten non-sponsored, non-academic websites were selected. Readability was assessed using Readable.com for grade level, Flesch-Kincaid Grade Level, and Flesch Reading Ease, both before and after prompting ChatGPT to “translate the following to a 6th-grade reading level.” Paired t-tests and chi-square tests were used for statistical analysis in Stata v17. Before ChatGPT revision, 55% of websites received an “A” grade. The mean Flesch-Kincaid Grade Level was 7.69 (95% CI: 7.12–8.25), and the mean Flesch Reading Ease was 62.01 (95% CI: 58.65–65.37). After ChatGPT revision, the proportion of “A” grades rose to 85% (p=0.008). The average Flesch-Kincaid level dropped to 5.75 (95% CI: 5.18–6.31, p<0.001), and the Flesch Reading Ease score improved to 74.93 (p<0.001). Significant readability improvements were observed across all pathologies. Online PEMs related to foot and ankle conditions often exceed recommended readability standards. ChatGPT revisions significantly improved readability scores, suggesting its potential as a tool to enhance patient comprehension.