Evaluation of 9 IOL power calculation formulas using a heteroscedastic statistical method and a new method of IOL
Daniel Romero-Valero1, Alicia Cárceles1, Jorge L Alió2
1Department of Opthalmology, General University Hospital of Elche, Elche, Spain.
European Journal of Ophthalmology
|February 7, 2024
Summary
The Kane formula showed the lowest prediction error standard deviation, outperforming older formulas. New generation formulas, including Kane and EVO 2.0, demonstrated comparable accuracy for intraocular lens (IOL) power calculations.
Area of Science:
- Ophthalmology
- Biomedical Engineering
- Statistical Analysis
Background:
- Accurate intraocular lens (IOL) power calculation is crucial for achieving desired refractive outcomes after cataract surgery.
- Various IOL power calculation formulas exist, each with unique algorithms and constants.
- Optimizing IOL constants and employing robust statistical methods are essential for evaluating formula accuracy.
Purpose of the Study:
- To compare the prediction accuracy of nine different IOL power calculation formulas.
- To utilize a heteroscedastic statistical analysis for evaluating prediction errors.
- To implement a novel method for optimizing IOL constants prior to analysis.
Main Methods:
- Retrospective analysis of 278 eyes from 278 patients undergoing cataract surgery.
- Preoperative biometry performed using the LenStar LS900 device.
- Calculation of predicted refractive error using nine formulas: Barrett Universal II, EVO-2.0, Hill RBF-3.0, Hill-RBF 2.0, Kane, PEARL-DGS, SRK-T, Hoffer-Q, and Holladay-1.
- Optimization of IOL constants and application of heteroscedastic statistical methods to compare standard deviations of prediction errors (PE).
Main Results:
- The Kane formula exhibited the lowest standard deviation (SD) of prediction error (0.4214D).
- Kane and EVO 2.0 formulas showed significantly lower SD of PE compared to SRK-T, Holladay 1, and Hoffer-Q.
- PEARL formula's SD was significantly lower than SRK-T and Hoffer-Q.
- No significant difference in SD of PE was observed among newer generation formulas (Hill RBF 3.0, Hill-RBF 2.0, Kane, EVO 2.0, Barrett Universal II, PEARL).
Conclusions:
- The Kane formula demonstrated the highest accuracy, followed closely by EVO 2.0 and PEARL-DGS.
- No statistically significant differences were found between the prediction errors of the analyzed new-generation formulas.
- Further research is recommended to assess the accuracy of these formulas in eyes with unusual biometric characteristics (extreme eyes).
More Related Videos
Related Concept Videos
01:24One-Compartment Open Model: Wagner-Nelson and Loo Riegelman Method for ka Estimation
507
This lesson introduces two critical methods in pharmacokinetics, the Wagner-Nelson and Loo-Riegelman methods, used for estimating the absorption rate constant (ka) for drugs administered via non-intravenous routes. The Wagner-Nelson method relates ka to the plasma concentration derived from the slope of a semilog percent unabsorbed time plot. However, it is limited to drugs with one-compartment kinetics and can be impacted by factors like gastrointestinal motility or enzymatic degradation.
On...
On...
507
01:14Standard Deviation of Calculated Results
6.2K
Standard deviation measures the spread of data around the mean value. Many large data sets follow a Gaussian distribution, also known as a normal distribution. This distribution is bell-shaped curved, with the most frequently observed value (mean or central value) in the middle. The farther away from the central value, the greater the deviation from the central value, and the lower the frequency.
A broad Gaussian distribution curve has a wider standard deviation, representing a data set with...
A broad Gaussian distribution curve has a wider standard deviation, representing a data set with...
6.2K