To analyze the preoperative predictive value of retinometer visual acuity (VA) in eyes following Descemet membrane endothelial keratoplasty (DMEK). Patients undergoing DMEK between August 2011 and July 2020 were included.… Click to show full abstract
To analyze the preoperative predictive value of retinometer visual acuity (VA) in eyes following Descemet membrane endothelial keratoplasty (DMEK). Patients undergoing DMEK between August 2011 and July 2020 were included. Preoperative interference visual acuity was assessed using Heine Lambda 100 Retinometer. Depending on the presence or absence of concomitant ocular disease, the Retinometer was evaluated for its ability to preoperatively predict best-corrected visual acuity (BCVA) six months after surgery using correlation, simple and multiple linear regression, contingency analyses, and receiver operating characteristic (ROC) analysis. Preoperative corneal backscatter was correlated with Retinometer prediction accuracy. A total of 198 eyes were included in the analysis. There was a significant correlation between Retinometer VA and postoperative BCVA (r = 0.647, P < 0.001). Regardless of the presence or absence of concomitant ocular disease and the surgery procedure (DMEK & triple DMEK), Retinometer VA was the most significant predictor of postoperative BCVA (P < 0.001). ROC analysis revealed reliable diagnostic performance of the Retinometer (AUC = 0.829, P < 0.001). A Retinometer VA ≥ 0.5 accurately predicted a postoperative BCVA ≥ 0.5 in 91% of cases. No association was found between corneal backscatter and prediction accuracy (P = 0.566). Retinometer VA can be used for preoperative prediction of postoperative BCVA in DMEK and triple DMEK patients, independent of increased backscatter values and the presence or absence of concomitant ocular disease. By using this simple but effective tool, indication for DMEK can be facilitated and postoperative outcomes can be realistically predicted preoperatively.
               
Click one of the above tabs to view related content.