INFLUENCE OF AN OPTICAL BIOMETER REPEATABILITY IN THE PREDICTION ERROR ACHIEVED WITH A THICK LENS FORMULA

Influence of an optical biometer repeatability in the prediction error achieved with a thick lens formula

Qvision, Ophthalmology Department, Vithas Virgen del Mar Hospital, Almería, Spain

A literature search was conducted to identify repeatability studies of the IOL Master 700 (Carl Zeiss AG) optical biometer reporting within-subject standard deviation (Sw). The mean corneal keratometry (Km), anterior chamber depth (ACD), axial length (AXL), lens thickness (LT), central corneal thickness (CCT) and white-to-white (WTW) of the population was estimated from the results reported by each study sample. Two set of random measurements normally distributed with a standard deviation of Sw were generated for all the variables and these were introduced in a thick lens formula to evaluate the prediction error (PE) attributed to measurement bias.

14 studies accomplished the inclusion criteria. Estimated population repeatability coefficients (R=2.77xSw) derived from the 14 studies were 0.25 D for Km and 0.04, 0.05, 0.07, 0.007, and 0.29 mm for ACD, AXL, LT, CCT and WTW, respectively. The R for the PE combining all parameters was 0.29 D [95%CI: 0.24 – 0.34] and the R for the PE attributed to each isolated parameter considered by the formula were 0.24 D [95%CI: 0.21-0.28], 0.03 [95%CI: 0.02-0.04] D and 0.13 [95%CI: 0.07-0.19] D for Km, ACD and AXL.     

The biometer resulted in a very good repeatability with up to 95% of eyes achieving a PE attributed to device repeatability below 0.29 D. For a thick lens formula, which predicted the ELP based on ACD and AXL, the Km first and followed by AXL, explained the major part of the PE. The influence of ACD repeatability on PE was negligible. This work demonstrates that even though with a perfect formula, >95% of eyes could result in a PE in +/- 0.50 D but never in +/- 0.25 D due to biometer repeatability.