Saccade tasks: A non-invasive approach for predicting postoperative delirium in elderly arthroplasty patients.

Summary

In 316 elderly arthroplasty patients, preoperative saccadic eye-movement features predicted postoperative delirium with AUROC 0.81 (logistic) and 0.89 (MLP), outperforming MMSE/MoCA and serum NfL. Five saccadic parameters showed significant group differences.

Key Findings

• POD incidence was 8.2% (26/316) in elderly arthroplasty patients.
• Five saccadic parameters significantly differed between POD and non-POD groups.
• Saccade-based models outperformed MMSE/MoCA and serum NfL (AUROC 0.81 logistic; 0.89 MLP).

Clinical Implications

Saccadic testing could be integrated into preoperative assessments to identify high-risk patients for enhanced delirium prevention (e.g., multicomponent nonpharmacologic bundles), though external validation and workflow integration are needed.

Limitations

• Single surgical domain (arthroplasty) and likely single-center design; generalizability uncertain.
• Potential overfitting without external validation; effects of anesthesia types and perioperative variables need exploration.

Future Directions

External, multi-center validation; development of pragmatic screening cut-offs; integration with multimodal risk models and evaluation of impact on delirium prevention outcomes.