Corneal power determination after previous corneal refractive surgery for intraocular lens calculation

To assess the accuracy of different corneal power determination methods in patients who had undergone myopic laser in situ keratomileusis (LASIK), photorefractive keratectomy (PRK), and radial keratotomy (RK).

The results for 208 eyes of 116 patients who had had corneal refractive surgery (LASIK, PRK, RK) for myopia were analyzed retrospectively. Keratometry measurements, i.e., autokeratometry readings (K(meas)), simulated keratotopography readings (Sim-K), and topographically measured average central corneal power in a 3-mm zone (ACP) were compared with calculated refraction-derived keratometric value. Correction factors based on the difference between measured and calculated keratometric powers were rated.

Direct power measurements with standard keratometers and planokeratotopography systems overestimate corneal power after myopic PRK and LASIK. The average K(meas) and K(topo) were significantly greater than the average calculated refraction-derived keratometric values. Corneal power overestimation correlated significantly with the spherical equivalent change after refractive surgery (p < 0.001). After RK, there is no significant correlation of the difference between all measured K values and refraction-derived power with the spherical equivalent change. In these cases, the Sim-K value seems the most accurate among measured keratometric powers. The precision of measurement significantly depends on the parameters of the autokeratometer (i.e., measurement place, number of measurement points, keratometric index of refraction).

To avoid underestimation of intraocular lens power after cataract surgery in the eyes that had previously undergone myopic corneal refractive surgery, the measured corneal power must be corrected. Although correction factors may be calculated for cases after PRK and LASIK, there are no universal and absolutely reliable methods to determine corneal power in these cases. More than one accessible method should be used, and the lowest, most reliable data should be used.