Intraocular lens power calculations after refractive surgery: Consensus-K technique

PurposeTo describe a new strategy for intraocular lens (IOL) power calculations after laser in situ keratomileusis (LASIK), determine the accuracy of the method, and compare the results with those of previously described techniques.SettingEmory Eye Center, Atlanta, Georgia, USA.MethodsThis retrospective comparative series compared eyes having cataract extraction after laser in situ keratomileusis (LASIK) and eyes having cataract extraction without previous surgery (control group) from January 1997 to December 2005. In the LASIK group, 2 strategies were used to determine the appropriate corneal curvature (K) value for IOL calculation: (1) the nonconsensus method and (2) the consensus-K technique. Postoperative outcomes were compared and included refraction, deviation from target refraction, and deviation from the ideal K or ideal IOL value for several techniques.ResultsThere were 43 LASIK eyes (14 nonconsensus and 29 consensus K) and 50 control eyes. The mean absolute deviation from target refraction in the nonconsensus group (1.47 diopters [D]) was significantly higher than that in the consensus-K group (0.52 D) (P = .02) or control group (0.44 D) (P = .01); the mean was not statistically different between the consensus-K group and the control group (P = .5). Compared with values with the consensus-K technique, the absolute deviation from back-calculated K values was significantly higher (P<.05) with all other K-generating methods tested except the classic refractive history method (0.56 D versus 0.65 D) (P = .4). When compared with the IOL value generated using the consensus-K technique, the absolute deviation from back-calculated IOL values was significantly higher for all other methods (P<.05).ConclusionsThe consensus-K technique generated refractive outcomes similar to those in the control group and was better than with all other K- or IOL-generating techniques except the classic refractive history method. The consensus method showed less variability and higher predictability than all other methods tested.