September 2019


EyeWorld Journal Club
Review of “Using the first eye prediction error in cataract surgery to refine the refractive outcome of the second eye”

by Tedi Begaj, MD, and Catherine Marando, MD

Alice Lorch, MD
Program director
Massachusetts Eye and Ear Harvard Medical School

Massachusetts Eye and Ear residents, December 2018
Source: Massachusetts Eye and Ear


In modern day cataract surgery, patients expect quick recovery and ideal refractive outcomes. The accuracy of intraocular lens (IOL) calculations has improved due to rapidly emerging technological advances in axial length measurement and keratometry. Advancements in IOL calculation formulas have increased success in reaching the desired target refraction.1 Furthermore, new and emerging IOL technologies, such as presbyopic and/or astigmatic-correcting lenses, are available to help accomplish complex refractive goals with decreasing spectacle dependence.2 However, despite the improvements seen over the last two decades, 10–20% of patients still exhibit greater or equal to 0.5 D of postoperative refractive error as compared to their predicted refractive outcome.3 The difference between the predicted postoperative refraction (PPOR) and the achieved postoperative refraction is known as prediction error (PE).
In the event that a cataract surgeon has not achieved target refraction in the first eye, how should they modify their IOL power selection when it comes time to operate on the second eye? Studies have shown that when extraneous sources of error have been excluded and both eyes have symmetric optical properties (e.g., corneal power, axial length), refining the second eye from the postoperative outcome of the first eye can decrease PE.4,5
While no standard recommendation for second eye IOL correction exists in the U.S., the National Institute for Health and Care Excellence (NICE) in the U.K. currently recommends that a 50% correction factor be used for second eye calculations when the first eye has resulted in a refractive surprise.6 At Massachusetts Eye and Ear (MEE), there is significant variability among the clinical faculty in refining IOL calculations for the second eye. However, delayed sequential bilateral cataract surgery (DSBCS) is the preferred approach by all faculty members in order to achieve the best binocular refractive outcome.
In the current study, Turnbull and Barrett provide both insight and guidance for second eye refinement. They systematically assess the utility of derived mathematical adjustment factors to the second eye IOL power based on results from the first eye cataract surgery.


Turnbull and Barrett retrospectively analyzed two large datasets of patients who underwent DSBCS, one from Australia and one from the U.K. The Australian dataset included 139 patients operated on by a single surgeon from October 2015 to March 2018. AcrySof SN60WF or SN6ATX lenses (Alcon) were used. Biometry was captured with LENSTAR LS 900 (Haag-Streit), with refraction by a trained optometrist at 4 weeks postop. Patients were excluded if there was a history of refractive surgery, additional combined surgeries (e.g., glaucoma or retinal surgery), or intra/postoperative complications. The U.K. dataset included 605 patients who had undergone DSBCS between May 2012 and May 2017 at Southampton Eye Unit, Southampton, U.K. They utilized Tecnis ZCB00 lenses (Johnson & Johnson Vision) and an IOLMaster 500 (Carl Zeiss Meditec) for biometry. Inclusion and exclusion criteria of patients were similar; however, a single surgeon performed all cataract surgeries in the Australian cohort, while a multitude of surgeons, including trainees, performed surgeries in the U.K. cohort.
Two methods were utilized for second eye refinement: (1) formula-specific and (2) patient-specific. With the formula-specific method, the authors performed a correlation analysis of the PE between the first and second eye for each individual formula (Barrett Universal II [BUII], Hoffer Q, Holladay, and SRK/T) and generated regression coefficients. They then retrospectively applied the formula-specific regression coefficients based on the PE of the first eye in order to adjust the PPOR of the second eye. Finally, they recalculated the percent of second eyes that were within 0.5 D, 0.75 D, and 1 D of the PPOR.
With the patient-specific method, the authors retrospectively calculated ideal IOL constants (IOLc) for BUII and SRK/T in both the first and second eye of each patient. They then derived a universal adjustment coefficient that could be applied to the first eye IOLc to arrive at an idealized second eye IOLc. Next, they recalculated the percent of second eyes that were within 0.5 D, 0.75 D, and 1 D of the PPOR after the adjustment.
In short, the formula-specific method generated a unique correction factor based on the IOL formula employed by utilizing the PE of the first eye for the second eye IOL calculation, whereas the patient-specific method generated a correction factor to optimize the IOL constant for the second eye based on the optimized IOLc of the first eye.
Finally, Turnbull and Barrett tested both validity and accuracy by applying the Australian coefficients from each method to the U.K. cohort and also separately calculating U.K. coefficients and comparing these to the Australian coefficients.


The study showed that application of the formula-specific correction to the Australian cohort significantly improved the number of second eyes that were within 0.5 D of the target refraction for all IOL formulas. Application of the Australian based formula-specific corrections to the U.K. cohort also resulted in statistically significant improved refractive outcomes for all formulas except the Hoffer Q (while not statistically significant, the trend showed improvement as well). Subsequently, U.K. formula-specific coefficients were calculated and compared to the Australian based cohort with similar coefficients for the BUII but slightly different coefficients for the remaining formulas.
Results for the patient-specific method showed that both cohorts had significant improvement in the number of second eyes within 0.5 D of PPOR. Overall, both the formula-specific and patient-specific methods improved refinement of the second eye, and there was no significant difference between the two methods. Of note, the formula-specific method appears to be more user friendly.


Today, our patients expect excellent refractive outcomes after cataract surgery. The work by Turnbull and Barrett helps cataract surgeons get closer to meeting their goals by refining second eye outcomes based on first eye results. Their study provides a methodological way for second eye refinement that is applicable to two heterogeneous populations.
There are several novel aspects to be highlighted. First, this study generates formula-specific adjustment coefficients, whereas previous studies have only offered general adjustment factors. Second, the primary cohort data is validated in a different patient population that was operated on by a heterogeneous group of surgeons, including trainees. Third, the authors provide a criteria dependent guideline that is simple to follow and can be implemented by any surgeon who aims to adjust their IOL choice for cataract surgery in the second eye.
The methods are quite robust given the size of the cohorts and number of formulas evaluated. However, if the validation was performed in another single-surgeon cohort, the improvement in the second eye PE may have been even larger. Furthermore, subgroup analysis to separate more symmetric eyes from less symmetric eyes may corroborate the specific recommendations from the authors regarding the axial length and cornea power differences.
The authors obviously have done an enormous amount of data processing and analysis for this study. However, for the general audience, who may be less familiar with this type of analysis as compared to the experienced authors, it would be helpful to provide a hypothetical example, describing step by step how the adjustment coefficient was generated. For example, was the surgeon-intended refractive outcome (rather than the formula-predicted refractive outcome) utilized in any of the calculations? Such examples would also help the audience to understand how the second-eye refinement, which may have been utilized in the U.K. series, potentially affects the mean absolute error (MAE).
The BUII formula outperforms the other three formulas (in both Australian and U.K. cohorts), even without any correction factor. The third generation vergence formulas such as Hoffer Q, Holladay, and SRK/T, are still widely used by cataract surgeons around the world. However, the result of this study is consistent with the consensus that the latest formulas (the Ladas Super Formula, the Barrett Universal II, and the Hill-RBF) outperform the third and fourth generation formulas regardless of axial length.7,8 Furthermore, in the BUII formula-specific correction, the number of patients within 0.5 D of target increased by 2.16% and 1.99% in the Australian and U.K. cohorts, respectively. Thus, there is opportunity for further improvement even in the formulas with the best outcomes. We wonder if the 90% mark may be surpassed in the short future since the artificial intelligence model (Hill-RBF9) continually improves its prediction algorithms as the database expands over time.
The work herein also indirectly supports waiting minimally 4 weeks between successive surgeries. It is unclear why the 4-week timeframe was chosen, but we postulate that the authors considered the passage of such time as necessary before stable refraction. At MEE, some surgeons find that even by the first postoperative week, a fairly stable refractive outcome can be achieved in a selected group of patients who have undergone uncomplicated cataract surgeries.10 Patients expect faster recovery, an accurate refractive outcome and minimal disturbance in daily activity. Delaying the second eye surgery by more than 4 weeks may meet strong resistance as well as affect activities of daily living. Future research is needed to determine the optimal postoperative wait time for stable refraction.

Bottom line

Based on this study and specific recommendations from the authors, cataract surgeons can correct the second eye PPOR by applying 30% of the first eye PE to the second IOL calculation when employing the BUII formula and by 50% of the PE when using Hoffer Q, Holladay I, and SRK/T formulas. This recommendation is useful in the following circumstances:
1. First eye PE >0.5 D
2. Interocular symmetry: <0.7 mm difference in axial length and <0.9 D difference in mean corneal power
3. Biometry errors and IOL malposition are excluded
Though this work included all eyes, irrespective of interocular symmetry, we agree with the authors’ warning to consider repeating biometry in cases of asymmetry; caution should be used if applying a second eye correction factor under such circumstances.
The clinicians at MEE would like to thank authors Turnbull and Barrett, as well as the EyeWorld Journal Club, for sparking lively and thoughtful discussion on how to think about second eye cataract surgery outcomes and provide the best possible care for our patients.

Contact information

Lorch: Alice_Lorch@MEEI.HARVARD.EDU

One question about separate vs. same-day, sequential bilateral surgery is the effect on refractive outcomes when the first eye result is known. I’ve asked the MEE residents to review this study that is published in the September JCRS.

—David F. Chang, MD, EyeWorld Journal
Club Editor


1. Narváez J, et al. Accuracy of intraocular lens power prediction using the Hoffer Q, Holladay 1, Holladay 2, and SRK/T formulas. J Cataract Refract Surg. 2006;32:2050–3.
2. Greenstein S, Pineda R. The quest for spectacle independence: A comparison of multifocal intraocular lens implants and pseudophakic monovision for patients with presbyopia. Semin Ophthalmol. 2017;32:111–115.
3. Melles RB, et al. Accuracy of intraocular lens calculation formulas. Ophthalmology. 2018;125:169–178.
4. Covert DJ, et al. Intraocular lens power selection in the second eye of patients undergoing bilateral, sequential cataract extraction. Ophthalmology. 2010;117:49–54.
5. Aristodemou P, et al. First eye prediction error improves second eye refractive outcome: results in 2129 patients after bilateral sequential cataract surgery. Ophthalmology. 2011;118:1701–9.
6. National Institute for Health and Care Excellence. Cataracts in adults: management. NICE Guideline (NG77). October 2017.
7. Roberts TV, et al. Comparison of Hill-radial basis function, Barrett Universal and current third generation formulas for the calculation of intraocular lens power during cataract surgery. Clin Exp Ophthalmol. 2018;46:240–246.
8. Kent C. In Search of the Perfect IOL Formula. Review of Ophthalmology. January 2017.
9. ASCRS Announces Hill-RBF Calculator for Cataract Surgery IOL Power Calculations. Updated June 2, 2016. Accessed July 31, 2019.
10. McNamara P, et al. Refractive stability following uncomplicated cataract surgery. Clin Exp Optom. 2019;102:154–159.

Review of “Using the first eye prediction error in cataract surgery to refine the refractive outcome of the second eye” Review of “Using the first eye prediction error in cataract surgery to refine the refractive outcome of the
Ophthalmology News - EyeWorld Magazine
283 110
220 58
True, 9