J. Park1, N.R. Jikaria1, J.M. Tabbaa1, R.Y. Kim2, J.S. Oh1,3, M.M. Boltz1,3 1Penn State Health Milton S. Hershey Medical Center, Department Of Surgery, Hershey, PA, USA 2Penn State Health Milton S. Hershey Medical Center, Department Of Orthopaedics And Rehabilitation, Hershey, PA, USA 3Penn State Health Milton S. Hershey Medical Center, Division Of Trauma, Acute Care & Critical Care Surgery, Hershey, PA, USA
Introduction: Low health literacy is associated with poor surgical outcomes, but targeted interventions may help mitigate these challenges. The average American adult reads at the 6-8th grade reading level. Consequently, the American Medical Association and National Institutes of Health recommend that patient education material (PEM) should not exceed the readability of sixth grade. However, multiple studies suggest that publicly available PEMs remain at a more complex reading level than recommended. Artificial intelligence (AI) has emerged as a promising tool for improving patient education. AI may offer a novel solution to improving the readability of PEM in surgery.
Methods: PEM on acute cholecystitis, acute appendicitis, cholecystectomy, and appendectomy, published online by surgical and medical societies and academic institutions were identified, analyzed, and compared. Available PEMs were entered into an AI dialogue platform with the prompt “translate to fourth-grade reading level” to convert pre-existing PEM. The readability of each text was analyzed using validated systems, Flesch Kincaid Grade Level (FKGL) and Flesch Reading Ease Score (FRES), for each description before and after the conversion. Each conversion was reviewed for factual inaccuracies and retention of the appropriate level of detail for each subject.
Results: A total of 20 PEMs were gathered from professional medical and surgical societies and academic institutions in the U.S. The original descriptions of acute cholecystitis and acute appendicitis had median (range) FKGLs of 10.19 (8.42-11.1), and 10.35 (9.4-11.4), respectively. Following conversion using the AI platform, the FKGLs of acute cholecystitis, and acute appendicitis were 6.2 (4.6-7.2) and 5.1 (4.7-6.21), respectively. The FRESs followed a similar trend, improving from 55.0 (48.8-61.1) and 51.9 (48.6-58.4) to 69.3 (66.2-70.1) and 76.7 (74.7-77.5), respectively. Similarly, FKGLs of cholecystectomy and appendectomy improved from 11.02 (9.7-13.7) and 10.35 (9-10.8) to 5.5 (4.1-6.4) and 5.7 (5.4-6.2), respectively. The FRESs followed a similar trend, improving from 54.8 (50.8-63.6) and 53.4 (49.7-60.5) to 72.2 (58.7-82.2) and 75.1 (69.5-76.9), respectively.
Conclusion: Publicly available PEM often exceeds the recommended reading levels of 6-8th grade, which may contribute to barriers to receiving surgical care. AI offers a promising solution to improve the readability of PEM in surgery efficiently and effectively. Further research is needed to determine whether AI technology can be applied to other educational materials, such as surgical consents or post-hospitalization discharge instructions.