Standard for collecting and reporting outcomes of IOL–based refractive surgery: update for enhanced monofocal, EDOF, and multifocal IOLs

For years, scientific journals have worked extensively to standardize the procedures for analyzing and reporting refractive results in refractive surgery.1–4 Currently, researchers have basic online resources outlining guidelines and examples for reporting results in laser or phakic refractive lens surgery and astigmatism analysis.5,6 In 2017, standards were extended to include outcomes of refractive intraocular lens (IOL) surgery and updated for IOL power calculation and toric IOLs.7–11 Some of these standards can also serve as a basis for reporting results of monofocal with enhanced intermediate vision (enhanced monofocal), extended depth-of-focus (EDOF), and multifocal IOLs. Given the high number of scientific publications involving these IOLs, the current document intends to offer an extension and update of the 2017 report to include distances other than far distance for efficacy, secondary endpoints, and patient-reported outcomes.7,8 Researchers should continue to review and comply with the statements included in the original standards of reporting refractive outcomes and the previous standards in the particular case of toric IOLs, which have not been repeated in this document.8,11 Furthermore, it is important to note that the guidelines for refractive surgery could potentially be used in subjects undergoing cataract or refractive lens exchange with normal or near-normal preoperative corrected distance visual acuity (CDVA).5,12,13 MANIFEST REFRACTION The procedure to obtain subjective manifest refraction (MR) can affect efficacy, predictability, and other visual performance measurements. Although it has been suggested to adjust MR to 6 m because of the clinical advantages at intermediate vision, especially in monofocal IOLs, MR is defined as the best spectacle correction at infinity; therefore, clinicians should adjust MR to infinity and not to 6 m (eg, adding −0.25 diopter [D] to the obtained refraction with a 4 m chart distance).14,15 Considering that MR is the refraction at infinity, clinicians should correct the vergence induced by the presentation distance of any visual performance test (eg, adding +0.25 D to the MR with a 4 m chart distance for obtaining CDVA). The dioptric power of the lens for correcting the vergence induced by the presentation distance is computed as 1/d, with d being the presentation distance in meters (eg, adding ½ = +0.50 D to the MR with a contrast sensitivity [CS] test presented at 2 m). For testing distance-corrected intermediate visual acuity (DCIVA) and distance-corrected near visual acuity (DCNVA), clinicians should remember to set the MR at infinity.12,13 EFFICACY In contrast to laser refractive surgery or refractive surgery with phakic IOLs, preoperative CDVA is not helpful for assessing efficacy outcomes after cataract surgery.7,8 Therefore, the efficacy index should only be used in refractive lens exchange.3 Two bar charts were proposed, one including the cumulative percentage at different levels of postoperative uncorrected distance visual acuity (UDVA) and CDVA and the other with the percentage for the difference between postoperative UDVA and CDVA in terms of the number of lines (Figure 1, A and B).7,8 The latter one has the advantage of normalizing the efficacy to the best measured VA of the sample, making this directly comparable between studies.7,8Figure 1.: Standard plots for primary outcomes in studies with presbyopia-correcting IOLs. Cumulative percentage of eyes achieving monocular CDVA and UDVA at (A) far, (C) intermediate, and (E) near distances. B: Difference on lines in far distance between postoperative UDVA and CDVA. D: Spherical equivalent prediction error distribution and (F) postoperative refractive cylinder distribution.The distances at which visual acuities were measured and the luminance levels should be clearly stated and defined. The main differences between presbyopia-correcting IOLs are not at far distance but are at intermediate and near distances; therefore, efficacy plots should be replicated at intermediate and near distances including the presentation distance and using the labels “eyes” or “subjects” when results are obtained from monocular or binocular vision, respectively (Figure 1, C and E). For studies comparing multiple IOLs, stacked bar charts could replace the previous ones for reducing the number of figures and enable comparison.16,17 Researchers should avoid using nonstandardized charts or scales (Jaeger) at intermediate and near distances, and instead, Early Treatment Diabetic Retinopathy Study (ETDRS) and logMAR should be used.18 If another test is used, a previous validation with the ETDRS should be conducted. The DCNVA and DCIVA should be used for the plots instead of CNVA and CIVA because the latter represent a displacement of the far distance focus with the addition of positive lenses, and it is expected, approximately, the same level of VA considering a possible small difference due to pupil size.12,13,19 Because the objective of presbyopia-correcting IOLs is to maximize spectacle independence, measuring the performance with a positive lens (addition) lacks clinical sense in this population. Although CNVA and CIVA might be considered in studies with enhanced monofocal and EDOF IOLs on which lower spectacle independence is achieved, the ANSI Z80.35-2018 standard includes neither measurement as required.20 In studies with a large enough sample size that allows the subanalysis to be stratified by age, efficacy should be separately reported considering a cutoff age of ≥65 years.15 A possible bias between studies might be found owing to the test used or the background luminance of the testing chart.21,22 The methods section should include the test, the manufacturer, and the luminance of the chart for all the distances. If the authors do not have the luminance information because of the lack of a luminance meter, at least the ambient illumination should be measured with a lux meter, especially when printed charts are used. A background luminance for the chart of 85 cd/m2 should be selected in photopic vision and approximately 2.5 to 3.2 cd/m2 for mesopic vision.15 It is usual to find bias in the results because of the background luminance of the chart, as the recommended background luminance is 200 cd/m2 in standard clinical practice.23 SAFETY In terms of visual performance, the safety section should include the preoperative and postoperative mean CDVA and the percentage of eyes in which the postoperative CDVA was equal to or worse than the preoperative CDVA in a 0.2 logMAR difference with the possible causes.7,8,15 ISO-11979-7:2018 also considers including subject-by-subject analysis of reasons if the subject failed to achieve 0.3 logMAR for the CDVA.15 Adverse events (AEs) should be described in this section, and a list of the possible AEs can be found in the ISO-11979-7:2018 and in the American Academy of Ophthalmology Task Force consensus.15,24 In studies with a large enough sample size that allows the subanalysis to be stratified by age, rates of cumulative AEs and rates of persistent AEs should be reported, considering a cutoff age of ≥65 years.15 Eyes with AEs that affect the CDVA as previously described, 0.2 logMAR worse than the preoperative CDVA or failed to achieve 0.3 logMAR for the CDVA, should be considered as outliers and excluded from the efficacy analysis. These cases should be described separately through descriptive statistics when there are a considerable number of cases (eg, providing the median VA and interquartile range of cases with AE). PREDICTABILITY In addition to the predictability plot adjusting the postoperative spherical equivalent to the intended target refraction, the authors should provide in the methods section the formula selected for IOL calculation, the constant used by the formula, and the biometer used for obtaining the biometric eye data (Figure 1D).7,8 Further details on whether the constant from the chosen formula was optimized or not should be clearly defined. In the results section, the mean manifest refraction spherical equivalent, the mean refractive astigmatism, and the mean intended target refraction should be provided. Differences can be found between studies depending on the test distance for which refraction was obtained, but the authors should correct the MR at infinity regardless of the distance used for test refraction (eg, −0.25 D for a 4 m chart distance).15 PUPIL SIZE The pupil size should be measured preoperatively only when it has been included in the inclusion criteria for a pupil-dependent IOL.15 In other cases, postoperative pupil diameter should be measured preferably beyond the 2-month follow-up.15 Pupil diameter is usually measured under extreme light conditions and might not be at the illumination levels associated with the tests used for measuring visual performance.15 The photopic pupil diameter can be measured with a simple ruler during the testing procedure. More precise measurement of scotopic pupil size is possible with a digital infrared pupillometer.25 This procedure can correlate more with visual performance than photopic and mesopic pupil size measured with some devices.26 The authors should provide the illumination (lux) at which photopic and mesopic pupil diameters were measured. In studies with a large enough sample size that allows the subanalysis to be stratified by pupil size, a stratification of visual performance results into 3 groups can help the reader understand the pupil dependence of the IOL. Different cutoff points have been proposed by the ISO (small [≤2.5 mm], medium [>2.5 to <4.0 mm], and large [≥4.0 mm]) and the task force consensus (small [<3.0 mm], medium [≥3.0 to ≤4.0 mm], and large [>4.0 mm]) for pupil size stratification in defocus curves (DCs), although stratification could also be conducted in efficacy.15,24 Therefore, either of both that results in a better balanced-sized groups after stratification could be used. Furthermore, including correlations between pupil diameter and visual performance with best distance correction could help to the same purpose. CONTRAST SENSITIVITY CS with IOLs is generally reported in the current practice only for far distance in photopic and mesopic levels, with and without glare. The authors should be aware that differences between 2 IOLs in CS could completely change depending on the presentation distance, and we cannot say that CS of an IOL is superior than other IOLs if all defocus ranges are not measured. Therefore, if CS is measured at a single distance, the results should report the distance, as it is done for efficacy in terms of VA. Systems that use glare conditions should be previously validated in such a way that the glare source produces a decrease of 0.1 logCS at 6 cycles per degree in young healthy subjects.15,27 The spatial frequencies used should vary in photopic (3 cycles per degree [cpd], 6 cpd, 12 cpd, and 18 cpd) and mesopic (1.5 cpd, 3 cpd, 6 cpd, and 12 cpd) conditions. The procedure should be conducted by correcting the vergence induced by the test distance and preferably in monocular vision with MR for detecting the loss of CS that may be associated with the IOL. The procedure can be conducted in monocular (eyes) or binocular (subjects) vision with or without MR, and it should be described in the methods section using the corresponding labeling in the plots (Figure 2).28Figure 2.: Standard contrast sensitivity plots for studies with intraocular lenses. Error bars should be provided in the plot, describing in the legend if these correspond to SD, if the purpose is to represent the sample variation, or 95% CI if the purpose is to represent the confidence in mean population estimation. Shadowed areas represent the normal range.Although it was not specified by the international standard normative, CS should be separately reported for ages ≥65 years in studies with a large enough sample size to allow an age-stratified subanalysis.15 Results should be reported in log10 units (logCS), and if the authors include the normal ranges for comparison purposes, these should correspond to the testing conditions and for the approximate age of the patients because normal range can change with age. The authors should cite in the manuscript the source from which they obtained the normal ranges. We encourage researchers to develop future studies to provide reference CS ranges for 5 patient groups of interest: (1) unoperated phakic eyes for all ages, (2) unoperated phakic eyes for ages <65 years, (3) operated eyes with monofocal IOLs for all ages, (4) operated eyes with monofocal IOLs for ages <65 years, and (5) operated eyes with monofocal IOLs for the subgroup ≥65 years. For the normal ranges operated with monofocal IOLs, the IOL model for which the range was obtained should be provided. The aim of these normal ranges is first to evaluate the change of vision in comparison with preoperative conditions and second to understand the possible decrease due to the use of a presbyopia-correcting IOL instead of a monofocal IOL. In tests that measure the CS in discrete steps, if the patient does not see the first patch, the result provided for data aggregation should be the first patch level and not an empty value or a zero value.29 DEFOCUS CURVES DCs should be measured with MR and by correcting the vergence induced by the presentation distance of the test, as it was described above for the CDVA (Figure 3).30 Take note that performing the refraction at 6 m and measuring the DC at 6 m with the obtained refraction would produce a labeling construct error. The 0 D location in the plot would really correspond to −0.17 D (1/−6), and for this reason, it is recommended to take the refraction at a distance that allows the exact calculation at infinity adding lenses (ie, at 4 m adding −0.25 D). If a static chart is used for measuring the DC, the procedures followed for preventing memorization should be described, starting for the highest power negative lens, defocus lens randomization, or chart replacement after achieving a peak of best vision. The background luminance of the chart and environmental light, if this was different from the efficacy testing, should also be indicated.Figure 3.: Standard defocus curve plot for studies with presbyopia-correcting IOLs. AUCs correspond to the areas under the curve for values below 0.3 logMAR. The AUCs are computed for total range (T), far (F), intermediate (I), and near (N) distances. Error bars should be provided in the plot, describing in the legend if these correspond to SD, if the purpose is to represent the sample variation, or 95% CI if the purpose is to represent the confidence in mean population estimation.The defocus lens range and steps will be determined by the design of the IOL. In bifocal or trifocal multifocal IOLs, a range from +1.00 to −4.00 D is enough to cover the effective visual performance from infinity (0 D) to 25 cm (−4 D), although ISO recommends +2.00 to −4.00 D.15 We have to consider that above 0 D we are already at infinity, and if a good MR is conducted, extending the DC to values above +1.00 D would extend the testing time without including relevant information to clinical practice. In EDOF and enhanced monofocal IOLs, the positive range should be extended because the target can be modified in the nondominant eye to extend the binocular range of vision; therefore, the positive range will offer relevant information about the DC before selection of a myopic target. Task force consensus recommends a range from +1.50 to −2.50 D for EDOF IOLs.24 The areas under the curve (AUCs) below 0.3 logMAR can be easily calculated by means of the trapezoidal numerical integration method by normalizing the 0.5 D defocus steps to the unit.31 The AUCs can be stratified for total (T: only comparable between studies that share the same total range), far (F: +0.50 to −0.50 D), intermediate (I: −0.50 to −2.00 D), and near (N: −2.00 to −4.00 D) ranges (Figure 3).32 Although CS DCs have been proposed in recent years, this type of DC does not replace VA DCs and should be included only as a secondary endpoint in studies for detecting small changes in optical quality and approximately doubling the sample size of studies with VA DCs.31,33 PATIENT-REPORTED OUTCOMES Unfortunately, there is not a questionnaire that collects all the clinically relevant domains validated in patients implanted with presbyopia-correcting IOLs. For instance, some popular questionnaires, such as the Catquest-9SF and Visual Function Index, were developed for measuring benefits of cataract surgery and those may not be sensitive to detect differences between presbyopia-correcting IOLs as they were not developed including this population.32 The Near Activity Visual Questionnaire and Near Activity Visual Questionnaire–Presbyopia might be more sensitive to detect differences between IOLs, but they lack clinically relevant information about dysphotopsia and spectacle independence at several distances provided by questionnaires measuring other domains, such as the Quality of Vision or the Patient-Reported Spectacle Independence Questionnaire.35–37 This fact explains why researchers usually use their own nonvalidated questionnaires, or they extract single questions from validated questionnaires to collect data of clinical relevance in this population.38 The researcher should decide the validated questionnaire that better fits the domain that is intended to measure, but from a clinically relevant perspective and for enabling the between-study comparison, there are single questions for which reporting the proportion of subjects achieving a particular level in a 5-point Likert scale could be of great relevance for studies with short-term follow-ups (<6 months). Some examples are: During the past 7 days, how satisfied were you with your vision without spectacles at far distance (1.5 m or longer)? (Not at all Satisfied/Slightly Satisfied/Neutral/Satisfied/Very Satisfied) During the past 7 days, how satisfied were you with your vision without spectacles at intermediate distance (between 1.5 m and 45 cm)? (Not at all Satisfied/Slightly Satisfied/Neutral/Satisfied/Very Satisfied) During the past 7 days, how satisfied were you with your vision without spectacles at near distance (shorter than 45 cm)? (Not at all Satisfied/Slightly Satisfied/Neutral/Satisfied/Very Satisfied) During the past 7 days, how bothersome did you find when you experienced visual artifacts around the lights at night, such as streetlights, car front lights, or car brake lights? (Not at Bothersome/Slightly Bothersome/Moderately/Very Bothersome/Extremely Bothersome) If you were to be operated again, how likely is it you would decide to be implanted with the same IOL with which you had the operation? (Not at All Likely/Slightly Likely/Neutral/Likely/Very Likely) These questions are usually used in clinical studies to obtain clinically relevant information, although they are not from validated questionnaires. For measuring specific psychometric domains, a validated questionnaire should always be used, but for reporting the proportion of subjects achieving a particular outcome, these single questions could be of great value for the reader and for enabling the comparison of outcomes. In the medium and long-term (≥6 months), in addition to the previous questions, the PRSIQ should be considered to also evaluate spectacle independence at distances for which the IOL correction is intended. Despite the limitations already described for detecting differences between IOLs for presbyopia, the difficulties in daily tasks should also be included, and we recommend the Catquest‐9SF and Visual Function Index because those have been validated for multiple languages.28,39 In conclusion, in addition to the mean score of the validated questionnaires selected to measure the desired domain, we encourage the authors to report the percentage of subjects achieving satisfaction with vision without spectacles at 3 distances in the short-term, spectacle independence at 3 distances in the medium and long-term, the percentage of subjects bothered by any photic phenomena, and the percentage of those who would decide to be implanted with the same IOL. METHODS OF AGGREGATION Studies should be reported with the awareness that they may be included in systematic reviews and meta-analyses. Therefore, it is critical that both testing and reporting methods are standardized. Systematic reviews filter the studies included in a meta-analysis according to the uniform criteria of study-level and outcome-level data. For this reason, the methods section should clearly answer the study-level questions, and the results section should include the outcome-level data.28 Statistical indices of centrality and variation, mean, and SD should be provided including only the first implanted eye of each subject. If this eye is excluded as an outlier (ie, by an AE), it should be replaced by the second operated eye and detailed in the statistical analysis section.15,28 This is applicable for VA, CS, and DC. A common practice is to report binocular results with best distance correction instead of monocular results, especially in DCs and CS. The authors should between the performance of the IOL and the performance of the procedure. The performance of the IOL should be reported through monocular testing with best distance correction and including only eye in binocular results represent the performance of the procedure and should be reported without best distance correction and in with patient-reported