A comparison of two methods to calculate axial length.

PURPOSE

To compare prediction accuracy with the axial length (AL) calculation method of the Lenstar biometer (traditional AL) and that of the ARGOS biometer (sum-of-segments AL).

SETTING

Private practice clinic.

DESIGN

Comparative case series.

MAIN OUTCOME MEASURE

Mean absolute error (MAE).

METHODS

Predictions were developed for nine formulas, grouping them into those derived with ultrasound (US) (SRK/T, Holladay 1 and 2, Hoffer Q, Haigis) and those derived with optical biometry (Barrett, OKULIX, Olsen from PhacoOptics, and Olsen from Lenstar). Formulas were ranked by MAE using sum-of-segments AL and traditional AL, in short eyes (traditional AL <22.0 mm), long eyes (traditional AL >26.0 mm), and all eyes.

RESULTS

The study comprised 1442 eyes (54 short eyes and 67 long eyes) of 1070 patients. The best-ranking formula for long eyes was Haigis using sum-of-segments AL. For short eyes and for all eyes, OKULIX using sum-of-segments AL was best. Using sum-of-segments AL instead of traditional AL, Holladay 2 improved the most; Olsen from PhacoOptics worsened the most.

CONCLUSIONS

Some biometers used traditional AL, and at least one used sum-of-segments AL. Formula accuracy varied depending on how various commercial biometers internally calculate AL. Using sum-of-segments AL instead of traditional AL improved predictions for formulas designed on US data (SRK/T, Holladay 1, Holladay 2, Hoffer Q, and Haigis), although it worsened the Barrett and Olsen formulas. OKULIX was generally improved with sum-of-segments AL. When ranking by MAE, OKULIX ranked first.