Abstract

The predictability of a theoretical, computer-assisted IOL calculation method and of the Sanders-Retzlaff-Kraff (SRK) method was evaluated from pre- and postoperative biometry in 110 cataractous patients subjected to a routine IOL calculation. With the theoretical method and the preoperative data the refraction was on the average 0.72 D (SD 0.78) more myopic than expected, an error which could be corrected for by (1) substituting the assumed postoperative chamber depth by the actual mean value and (2) adding 0.16 mm to the preoperative axial length. With the postoperative data the error was 0.05 D (SD 0.78) (p greater than 0.05). By the SRK method the refraction was 0.15 D (SD 0.71) and 0.37 D (SD 0.69) more hyperopic than expected with the pre- and the postoperative data respectively. These offset errors could be corrected for in either case by adjusting the assumed A constant in the SRK formula. With appropriate empirical corrections for offset errors the theoretical and the SRK method were similar in accuracy, that is, about 82% of the predictions within +/- 1 D, 99% within +/- 2 D, and 100% within +/- 3 D error. The variation in postoperative refraction after computerised IOL calculation was about one-third of the variation previously seen after implantation of standard power IOL.