Clinical Optics 101

Cataract outcomes: Getting the best results, now and in the near future

Despite improvements in IOL calculation formulas, five variables continue to be responsible for most prediction errors.

Patient expectations to be less dependent or free of spectacles have become increasingly important with premium IOLs. In this article, I will explain common methods of assessing patient outcomes, show the progress we have made in the last decade, how to improve your patient outcomes, and show what the near future holds for our next big improvement.

The difference in the predicted and the actual spherical equivalent refraction is called prediction error and is what we use to measure our outcome. Most comparisons of formulas and procedures (femtosecond vs. manual) adjust the outcomes so that the mean error (sum of all the hyperopic and myopic errors) is zero. This is done by adjusting the lens constant (optimized lens constant). One can then look at three measures: standard deviation (SD), mean absolute error (MAE) and number of cases within ±0.5 D as a measure of the variability of quality of the outcome.

When the distribution of prediction error is Gaussian (normal or bell-shaped curve), which is almost always true when there are more than 100 cases, the SD also determines the other two (MAE and percentage within a certain limit). For example, if the standard deviation is 0.5 D, then 68% of the cases are within 0.5 D and the MAE is 80% of the SD, or 0.4 D. If we calculate one, the other two are predetermined. Figure 1 shows the performance of U.S. surgeons as of 2018. Less than 1% (“A” surgeon) have a SD of 0.31 D, a MAE of 0.24 D and 90% of their cases within 0.5 D. About 25% (“B” surgeon) have 80% of their cases within 0.5 D, and the average (“C” surgeon) has about 75% of the cases within 0.5 D. Figure 2 shows the results of a recent study we published in Ophthalmology with Ron Melles and Jerry Chang for 141 surgeons and 13,301 cases. The distribution is perfectly Gaussian and typical of our results today, an overall average of 77% within 0.5 D.

Figure 1. Percentage of surgeons within X% of ±0.5 D.

Source: Jack T. Holladay, MD, MSEE, FACS

Figure 2. Normal bell-shaped distribution of 13,301 cases showing typical results of approximately 77% of cases within 0.5 D in 2018.

In a study published in Journal of Cataract and Refractive Surgery by Sverker Norrby in 2008, the SD was 0.7 D with only 52% of eyes within 0.5 D. Dr. Norrby attributed most of the prediction error to five variables: predicted effective lens position (ELP) was 35%, postoperative refraction 27%, axial length 17%, keratometry 10% and pupil size 8% (Figure 3). Almost all of the 23% improvement in the past 10 years is due to optical axial length using OCT, which is far more accurate than ultrasound.

Every surgeon can improve their current outcomes by implementing the binocular and monocular screening of measurements shown in Figure 4. The technicians should be advised that if any of the three binocular screening conditions are exceeded, then the measurements must be repeated, and if either of the two monocular conditions are exceeded, then topography and/or ultrasound should be performed.

Contribution to PE
Figure 3. Percent contribution, standard deviation (D) and mean absolute error (D) of the top five variables for prediction error.
Data screening
Figure 4. Using binocular and monocular data screening can significantly improve outcomes. For monocular conditions, ultrasound should be performed if the signal-to-noise ratio is less than 2.0 and topography if the standard deviation for the keratometry value is greater than 0.2 D (0.03 mm or 30 µm).

Our IOL calculation formulas have improved over the past 10 years but are still limited by the tolerances on the five variables above. Older formulas such as the Holladay 1, Hoffer Q, SRK/T and Haigis were determined using ultrasound. Optical biometry overestimates the axial length in eyes over 25 mm, and a long eye adjustment such as the Wang-Koch linear or Holladay nonlinear adjustment should be used for these formulas to avoid hyperopic surprises. Intraoperative aberrometry (ORA, Alcon, and Holos, Clarity) has improved our results but varies significantly from one surgeon to another. The ELP must still be predicted with these devices, and so the ELP and postoperative refraction remain limiting factors.

A significant advance will occur with postoperative adjustment of the IOL spherical equivalent power, toricity and even spherical aberration in the near future. RxSight (formerly Calhoun Vision) received approval for its light adjustable IOL in November 2017. It is working on post-approval studies at this time and should be delivering the Light Delivery Device later this year. Two to three weeks after cataract surgery, the surgeon enters the desired refractive change into the device, and when the surgeon and the patient feel the vision is good, the vision is permanently locked in. With 91.8% of eyes achieving a result within 0.5 D of target manifest spherical equivalent refraction, we can all be “A” surgeons.

There is also another technology with the femtosecond laser that allows postoperative adjustment of acrylic IOLs. Perfect Lens (CEO, Steven Smathers) and Clerio Vision (CEO, Mikael Totterman) have been working on this technology for several years and will start clinical testing shortly. This technology means that the millions of patients who already have acrylic IOLs implanted can be adjusted.

The 92% of cases within 0.5 D of our target seems to be the magic number that we can achieve and is probably limited by our refraction. It is what we achieve with laser refractive surgery, and hopefully our cataract surgery will be the same in the not too distant future.

Disclosure: Holladay reports he is a consultant to Abbott Medical Optics, AcuFocus, Alcon Laboratories, ArcScan, RxSight, Carl Zeiss, Elenza, M&S Technologies, Oculus and Visiometrics.

Editor's note: This article was updated June 11, 2018, to include Figures 3 and 4.

Patient expectations to be less dependent or free of spectacles have become increasingly important with premium IOLs. In this article, I will explain common methods of assessing patient outcomes, show the progress we have made in the last decade, how to improve your patient outcomes, and show what the near future holds for our next big improvement.

The difference in the predicted and the actual spherical equivalent refraction is called prediction error and is what we use to measure our outcome. Most comparisons of formulas and procedures (femtosecond vs. manual) adjust the outcomes so that the mean error (sum of all the hyperopic and myopic errors) is zero. This is done by adjusting the lens constant (optimized lens constant). One can then look at three measures: standard deviation (SD), mean absolute error (MAE) and number of cases within ±0.5 D as a measure of the variability of quality of the outcome.

When the distribution of prediction error is Gaussian (normal or bell-shaped curve), which is almost always true when there are more than 100 cases, the SD also determines the other two (MAE and percentage within a certain limit). For example, if the standard deviation is 0.5 D, then 68% of the cases are within 0.5 D and the MAE is 80% of the SD, or 0.4 D. If we calculate one, the other two are predetermined. Figure 1 shows the performance of U.S. surgeons as of 2018. Less than 1% (“A” surgeon) have a SD of 0.31 D, a MAE of 0.24 D and 90% of their cases within 0.5 D. About 25% (“B” surgeon) have 80% of their cases within 0.5 D, and the average (“C” surgeon) has about 75% of the cases within 0.5 D. Figure 2 shows the results of a recent study we published in Ophthalmology with Ron Melles and Jerry Chang for 141 surgeons and 13,301 cases. The distribution is perfectly Gaussian and typical of our results today, an overall average of 77% within 0.5 D.

Figure 1. Percentage of surgeons within X% of ±0.5 D.

Source: Jack T. Holladay, MD, MSEE, FACS

Figure 2. Normal bell-shaped distribution of 13,301 cases showing typical results of approximately 77% of cases within 0.5 D in 2018.
PAGE BREAK

In a study published in Journal of Cataract and Refractive Surgery by Sverker Norrby in 2008, the SD was 0.7 D with only 52% of eyes within 0.5 D. Dr. Norrby attributed most of the prediction error to five variables: predicted effective lens position (ELP) was 35%, postoperative refraction 27%, axial length 17%, keratometry 10% and pupil size 8% (Figure 3). Almost all of the 23% improvement in the past 10 years is due to optical axial length using OCT, which is far more accurate than ultrasound.

Every surgeon can improve their current outcomes by implementing the binocular and monocular screening of measurements shown in Figure 4. The technicians should be advised that if any of the three binocular screening conditions are exceeded, then the measurements must be repeated, and if either of the two monocular conditions are exceeded, then topography and/or ultrasound should be performed.

Contribution to PE
Figure 3. Percent contribution, standard deviation (D) and mean absolute error (D) of the top five variables for prediction error.
Data screening
Figure 4. Using binocular and monocular data screening can significantly improve outcomes. For monocular conditions, ultrasound should be performed if the signal-to-noise ratio is less than 2.0 and topography if the standard deviation for the keratometry value is greater than 0.2 D (0.03 mm or 30 µm).

Our IOL calculation formulas have improved over the past 10 years but are still limited by the tolerances on the five variables above. Older formulas such as the Holladay 1, Hoffer Q, SRK/T and Haigis were determined using ultrasound. Optical biometry overestimates the axial length in eyes over 25 mm, and a long eye adjustment such as the Wang-Koch linear or Holladay nonlinear adjustment should be used for these formulas to avoid hyperopic surprises. Intraoperative aberrometry (ORA, Alcon, and Holos, Clarity) has improved our results but varies significantly from one surgeon to another. The ELP must still be predicted with these devices, and so the ELP and postoperative refraction remain limiting factors.

A significant advance will occur with postoperative adjustment of the IOL spherical equivalent power, toricity and even spherical aberration in the near future. RxSight (formerly Calhoun Vision) received approval for its light adjustable IOL in November 2017. It is working on post-approval studies at this time and should be delivering the Light Delivery Device later this year. Two to three weeks after cataract surgery, the surgeon enters the desired refractive change into the device, and when the surgeon and the patient feel the vision is good, the vision is permanently locked in. With 91.8% of eyes achieving a result within 0.5 D of target manifest spherical equivalent refraction, we can all be “A” surgeons.

There is also another technology with the femtosecond laser that allows postoperative adjustment of acrylic IOLs. Perfect Lens (CEO, Steven Smathers) and Clerio Vision (CEO, Mikael Totterman) have been working on this technology for several years and will start clinical testing shortly. This technology means that the millions of patients who already have acrylic IOLs implanted can be adjusted.

The 92% of cases within 0.5 D of our target seems to be the magic number that we can achieve and is probably limited by our refraction. It is what we achieve with laser refractive surgery, and hopefully our cataract surgery will be the same in the not too distant future.

Disclosure: Holladay reports he is a consultant to Abbott Medical Optics, AcuFocus, Alcon Laboratories, ArcScan, RxSight, Carl Zeiss, Elenza, M&S Technologies, Oculus and Visiometrics.

Editor's note: This article was updated June 11, 2018, to include Figures 3 and 4.