October 2019


YES Connect
Crunching numbers for IOL optimization

by Liz Hillman EyeWorld Senior Staff Writer

You’ve performed cataract surgery, implanted your selected IOL, and conducted postop visits. While it might sound like your job is done at this point, if you want to improve your outcomes further, now’s the time to start crunching numbers.
EyeWorld reached out to experts H. John Shammas, MD, and Jack T. Holladay, MD, to get their thoughts about systems for selecting IOLs and managing calculations, following up on refractive outcomes, and tips for conducting clinical research with IOL data.

At what point postoperatively do you recommend collecting variables for outcome analysis? What would you say are the most relevant variables to collect from each patient? 
Dr. Shammas: After surgery, we refract the operated eye on day 1 and day 7. These early refractions will ensure that no errors in the IOL calculations have been made, [which] might require immediate intervention. However, the best time to check the final outcome is 4 to 6 weeks after the surgery; this is when we prescribe glasses. If the surgeon or his/her staff want to conduct an outcome analysis, they would record the IOL power and the final refraction. 

Dr. Holladay: Six months after surgery is ideal for stability in the spherical equivalent refraction as well as the residual astigmatism. Although 3 months is often used, it is too soon for changes in the capsular bag to be complete and axial changes in the IOL are still occurring. FDA clinical trials have visits at 1 day, 1 month, 3 months, 6 months, and 1 year. The 6-month data is usually the visit where refraction is used for analysis. There is a long-term drift in against-the-rule astigmatism of about 0.25 D per decade that is the same for both cataract surgery and controls found by Ken Hayashi, MD, in his article following patients for 20 years.1

How do you collect your postop outcomes data for later optimization analysis? What systems do you use?
Dr. Shammas: You do not need any system. You have to record the power of the IOL used and the spherical equivalent of the final refraction. 

Dr. Holladay: The Holladay IOL Consultant software does this automatically when you enter the surgical and postoperative refraction. The optimization excludes outliers that are more than two standard deviations from the mean because they are usually measurement errors and cases with less than 20/50 best-corrected vision because the refraction is suspect. The software does not actually use the personalized lens constant (PLC) until it reaches statistical significance, which is usually from 30 to 50 cases. It looks at the data through a “rearview window,” looking at the data that is most recent first and going back historically until statistical significance is reached. This keeps the PLC “fresh” in case the surgeon makes changes in his surgical technique that may unknowingly affect the PLC.
What is your process of optimization of surgical outcomes derived from your collection of data? What is the process of how you go from a collection of postop outcomes data to optimization of your future IOL calculations?
Dr. Shammas: There are different ways to do that. Some biometers will allow you to enter the power of the IOL used and the spherical equivalent of the final refraction and it will continuously optimize the constants being used. 
An easy way to do this optimization is to collect at least 50 cases (surgeons can start with 20 cases and update later). You then compare the final refraction to the one predicted by the formula. You subtract the predicted refraction from the obtained refraction and average all these differences. If the average difference is a negative value, that means that the surgeon is using a higher power IOL than what is required and he/she should decrease the A constant by the (average difference x 1.35). If the average difference is a positive value, then the A constant has to be increased by (the average difference x 1.35).
Dr. Holladay: The optimization (or personalization) is performed for each surgeon, each IOL, and each procedure for a specific IOL. For example, you may have surgeon1, IOL1, and manual and femto as procedures. The personalized lens constants for manual and femto are usually different. Another common procedure is post-refractive. This keeps these cases separate for analysis.

Do you have any tips to improve the accuracy and reliability of data collection for outcome analysis and clinical research? 
Dr. Shammas: Data collection for outcome purposes cannot be indiscriminate. In other words, you cannot include all surgeries. My personal opinion is to include the cases that achieve a stable postop refraction with a corrected visual acuity of 20/40 or better. Do not include eyes with co-morbidities such as macular degeneration, advanced glaucoma, or corneal scarring. Diabetics present a challenge since the refraction fluctuates if the diabetes is out of control. Eliminate any cases with advanced diabetic retinopathy. 

Dr. Holladay: Simply entering the surgical and postoperative data is all you need, but you must have a system setup to do this. As we move forward, the Holladay IOL Consultant Software will obtain this information directly from the electronic medical record (EMR). The software already imports directly from Optical Biometers (IOLMaster, LenStar, …) avoiding data entry errors, but integrating with the scores of EMRs is a monumental task.

How do you incorporate clinical research into your practice without compromising clinical flow?
Dr. Shammas: I personally find clinical research to be a challenging part of my professional life. As a busy ophthalmologist, I rarely have time to do any research during clinic hours. I have always taken a half day per week to teach or to work on my research projects. Delegating data collection to medical assistants is not a good idea. They do not understand the importance of data accuracy, for example, separating monofocal from multifocal IOLs or transcribing the astigmatism in negative or positive values. A lot of the data analysis is done by myself at home during the evening hours and the weekends. In the past few years, this process has been facilitated by having our EMR data in the cloud, which I can access from my home computer. 

Dr. Holladay: It takes time and personnel to enter the data for now, and that costs money. However, patients want to be spectacle-free and determining you PLC improves outcomes. This is not the only factor; you must have monocular and binocular screening in place to identify patients preoperatively who are at the highest risk for having a refractive surprise. Monocular screening requires a standard deviation of the keratometry less than 0.20 D (0.030 mm or 30 µm) and a signal-to-noise ratio of the axial length greater than 2.0. Binocular screening requires the axial length difference to be less than 0.3 mm between eyes and the keratometry and recommended IOL power for a specific target be less than 1.0 D difference between eyes.

About the doctors

Jack T. Holladay, MD,
Clinical professor of ophthalmology
Baylor College of Medicine

H. John Shammas, MD
Clinical professor of ophthalmology
Keck School of Medicine,
University of Southern California
Los Angeles


1. Hayashi K, et al. Changes in corneal astigmatism during 20 years after cataract surgery. J Cataract Refract Surg. 2017;43:615–21.

Relevant financial interests

Holladay: Holladay Consulting Inc
Shammas: Shammas post-LASIK formulas are licensed to optical biometers and ultrasound units

Contact information

Holladay: holladay@docholladay.com
Shammas: HShammas@aol.com

Crunching numbers for IOL optimization Crunching numbers for IOL optimization
Ophthalmology News - EyeWorld Magazine
283 110
283 110
True, 10