Estimation of Goldmann applanation tonometer intraocular pressure (IOP) from scleral Schiotz IOP values in eyes with type-1 keratoprostheses

Purpose To validate estimation of Goldmann applanation tonometer (GAT) intraocular pressure (IOP) from scleral Schiotz IOP measurements using a regression model in normal eyes and eyes with type-1 keratoprostheses. Methods In this prospective cross-sectional study, cohort-1 had 253 normal anterior segment eyes, and cohort-2 had 100 eyes with type-1 keratoprostheses. Scleral Schiotz IOP measurements were used (in a non-linear model) to predict GAT IOP values for these eyes. Accuracy of predicted GAT IOP values was assessed using actual GAT IOP values for normal eyes, while for type-1 keratoprosthetic eyes, finger tension (FT) IOP assessments by an experienced glaucoma specialist were used. Primary outcome was agreement between FT IOP (assessed by an experienced glaucoma specialist) and predicted GAT IOP-derived clusters. Results The actual values of GAT IOP measurements in normal eyes (n=253; mean age +/- SD, 51.35 +/- 15.56 years) ranged between 6 mm Hg and 62 mm Hg (mean=22 +/- 10.05 mm Hg). Estimated and actual GAT IOP values for normal eyes were very similar (mean difference=0.05 mm Hg with limits of agreement: -5.39 to 5.5 by Bland-Altman plot). Of the 100 eyes with type-1 keratoprostheses, 68 were classified as having digitally normal IOP, 28 as borderline and 4 as high. The agreement between classification by FT assessment and model-predicted GAT IOP values was substantial (Kappa=0.81, 95% CI 0.69 to 0.93). The accuracy of the model in assessing IOP was found to be 91% (95% CI 0.84 to 0.96). Conclusion Scleral Schiotz IOP values along with our predictive model can be an alternative objective method to FT IOP in assessing IOP in eyes with type-1 keratoprostheses.

Automated summary

The study aimed to validate the estimation of Goldmann applanation tonometer (GAT) intraocular pressure (IOP) from scleral Schiotz IOP measurements using a regression model in normal eyes and eyes with type-1 keratoprostheses, showing that the accuracy of the model in assessing IOP was found to be 91%.