Validity of the keratometric index: large population-based study.

Purpose: To determine the accuracy of the keratometric index of 1.3315 based on the Gullstrand model eye in predicting the power of the posterior cornea, Gullstrand's model was compared to a calculated keratometric index derived from actual measurements of the cornea.

Setting: Eye Institute, Tan Tock Seng Hospital, Singapore.

Methods: One eye of 2429 subjects with a mean spherical equivalent of -5.32 diopters (D) +/- 2.88 (SD) was measured with the Orbscan II (Bausch & Lomb). The following variables were analyzed: anterior radius of curvature (r(anterior)), posterior radius of curvature (r(posterior)), radius of keratometry (r(simK)), and central pachymetry.

Results: The r(anterior), r(posterior), and r(simK) were normally distributed, with a mean of 7.87 +/- 0.25 mm (95% confidence interval [CI], 7.38-8.36), 6.46 +/- 0.26 mm (95% CI, 5.95-6.97), and 7.71 +/- 0.27 mm (95% CI, 7.18-8.24), respectively. The mean ratio between the anterior corneal curvature and posterior corneal curvature was 1.22 +/- 0.03 (95% CI, 1.16-1.28). Based on the measurements of each eye, the mean calculated keratometric index, N(calc), was 1.3273 +/- 0.0013 (95% CI, 1.3248-1.3298). Using N(calc), the posterior corneal power was predicted to within +/-0.50 D of the actual posterior power in 98.3% of eyes. The mean absolute error between the actual and calculated posterior power was 0.157 +/- 0.123 D using N(calc) and 0.326 +/- 0.133 D using the Gullstrand model.

Conclusion: Modifying the keratometric index increased the accuracy of predicting the posterior corneal power.