Keratoconus is a progressive corneal disease in which thinning of the corneal stroma causes the cornea to develop a cone-shaped ectasia, which leads to progressive myopia, irregular astigmatism, and often loss of visual acuity.1,2 Therefore, correctly measuring refractive errors is of paramount importance to assess visual performance, disease progression, and the prescription of visual aids in keratoconus.
Several methods to measure refractive errors are currently widely used. A manifest refraction is still considered the gold standard to prescribe spectacles.3,4 This subjective technique was first described by F.C. Donders in 1864 and consists of asking which trial lenses provide the best acuity on the vision chart.5 A properly performed manifest refraction neutralizes accommodation and takes binocular balance into account.6
With retinoscopy, autorefraction, or wavefront aberrometry, it is possible to objectively assess the refractive error of the eye. Autorefraction uses several different techniques to assess refractive error. It was first introduced in 1970 and has a high repeatability and low deviation for normally curved corneas.4 It is commonly used as a starting point for a manifest refraction. However, autorefractor measurements seem to be less accurate in keratoconic eyes with irregular corneas than in normal eyes.3 The repeatability of a manifest refraction in this patient group is also lower than in normal eyes.7
Most modern autorefractors are based on the Scheiner double pinhole method or automated retinoscopy.3 Wavefront aberrometry is the newest form of autorefraction. This method is often used for keratoconus research because of additional capacities such as tomography measurements and higher order aberration (HOA) measurements,8,9 which are more common in keratoconic eyes.10 Several devices that use different techniques are on the market.11 In addition, it is possible to calculate corneal aberrations using Scheimpflug tomography.12
The conical shape and its subsequent increased amount of irregular astigmatism in keratoconus potentially offsets the outcomes of automated measurements.13 The aim of this study was to compare the visual performance with autorefraction and manifest refraction in patients with keratoconus and to investigate which factors influence the difference between these two refraction methods.
Patients and Methods
Study Group and Protocol
This cross-sectional single-center observational study included patients with keratoconus who visited the outpatient clinic at the University Medical Center Utrecht, the Netherlands from September 2015 through February 2016. The study was approved by the University Medical Center Utrecht Ethics Review Board and informed consent was obtained from all patients prior to their participation. All procedures complied with the tenets of the Declaration of Helsinki and local laws regarding research on human subjects.
Inclusion criteria were an established diagnosis of keratoconus and a clear cornea. The diagnosis of keratoconus was based on tomography (Pentacam HR type 70900; Oculus Optikgeräte, Wetzlar, Germany). Keratoconus staging was defined by the Amsler–Krumeich classification and was based on the mean keratometry value of the Pentacam only; pachymetry and scarring were not taken into account.14 Patients who underwent a corneal cross-linking treatment within the past 6 months were excluded to avoid bias caused by any postoperative corneal haze. All patients were asked to avoid contact lens wear at least 2 weeks prior to the measurements. Participants were excluded if the autorefraction could not be performed after three attempts.
Measurements and Data Conversion
All monocular manifest refractions were performed by the same optometrist (NS) in the same dim room (30 lux) using an automatic phoropter (CV3000; Topcon Corporation, Tokyo, Japan) and a visual acuity screen (CC100P; Topcon Corporation), based on the refraction technique described by Donders.5 The autorefractions were performed with a keratorefractometer (KR8800; Topcon Corporation). The device provided one mean refraction value based on three measurements. The order in which patients were measured with autorefraction and manifest refraction was randomized: 43 eyes were measured with the autorefraction first and 47 eyes with manifest refraction first. All eyes were measured consecutively with both measurements on the same day.
Scheimpflug tomography was performed with the Pentacam and used to determine cone location, keratometry readings (Kmean and Kmax), and HOAs. The anterior corneal aberrations were calculated according to preset values for refractive index of the corneal tissue (n = 1.337) and entrance pupil size (6 mm). The results were transformed into absolute values for analysis. The cone location was obtained by using the Pentacam's distance coordinates (mm) from the corneal apex to the steepest corneal location (Kmax) and the distance from the corneal apex to the pupil center.
Visual acuity was measured in Snellen decimals and converted for statistical analysis to logMAR units.
A refraction consists of the spherical, cylindrical, and axis components. These are interdependent factors. Spherical equivalent is a commonly used option to compare refractions; this method unfortunately does not consider the axis of the refraction and is therefore less suitable for research purposes. Another method is the Fourier analysis (ie, vector analysis), as reported by Thibos et al.15 and Raasch et al.16,17 The principle of this method is that each spherocylindrical power is a unique point that consists of three coordinates, making a compound coordinate for each refraction, as schematically shown in Figure A (available in the online version of this article).15,18 Calculating the distance between the corresponding points of the automated and manifest refraction results in the dioptric difference between the automated and manifest refraction, which is schematically shown in Figure B (available in the online version of this article). This method has been proven to be accurate while including all components of the refraction in one compound score.15
Schematic representation of a Cartesian notation for a refractive error using the Fourier analysis. The x- and y-axis are the cylinder direction and half of the magnitude. The z-axis is the spherical equivalent.
Schematic representation of the calculated dioptric difference using the Fourier analysis. The distance between the vector coordinates of the refractive error outcome is calculated to dioptric value, which represents the difference between the two measurements.
All data were tested for normality. Differences in visual acuity were analyzed using a paired sample t test. The relationship between dioptric difference, Kmax, and cone location was tested using univariable linear regression with generalized estimated equation correction, correcting for two eyes of the same patient. A multivariable regression analysis with generalized estimated equation correction was employed to determine which factors had a significant influence on dioptric difference. The factors that were included were gender, age, cone location, Kmax, and root mean square HOAs cornea, horizontal coma, vertical coma, horizontal trefoil, vertical trefoil, and spherical aberration. Data were collected in Microsoft Excel 2010 software (Microsoft Corporation, Redmond, WA) and analyzed using SPSS software (version 21.0; IBM Corporation, Armonk, NY).
Study Group Characteristics
In this study, 92 eyes from 62 patients were included. After exclusion of 2 eyes because of missing data, the data of 90 eyes were analyzed. The mean age of the study population was 27 ± 8.83 years (range: 14 to 51 years) and consisted of 41 males (66%) and 21 females (34%). Of the 90 eyes in the study group, 65 (72%) had stage I keratoconus, 19 (21%) had stage II keratoconus, 4 (4%) had stage III keratoconus, and 2 (2%) had stage IV keratoconus without scarring (staging classified according to the mean keratometry value of the Pentacam). There were 41 right and 49 left eyes analyzed. Sixty-six eyes (73%) had received a corneal cross-linking treatment more than 6 months before participating in this study.
Visual Performance of Automated and Manifest Refraction
Corrected distance visual acuity (CDVA) was achieved with the manifest refraction compared to autorefraction (0.06 vs 0.28 logMAR; 20/23 vs 20/38 Snellen; P < .0001). The CDVA improved with both automated and manifest refraction compared to the uncorrected distance visual acuity, but the increase in CDVA with manifest refraction was superior to the autorefraction. Details are shown in Table 1.
Study Group Characteristics
Multivariable Factor Analysis
The mean dioptric difference between the automated and manifest refraction was 4.83 diopters (P < .0001). Gender, age, cone location, Kmax, and root mean square HOAs cornea, horizontal coma, vertical coma, horizontal trefoil, vertical trefoil, and spherical aberration were factors that were included in the multivariable analysis. Kmax was found to be the only variable associated with dioptric difference (P < .0001); if Kmax was steeper, the dioptric difference was also steeper, meaning that the discrepancy between the two refraction methods increases with keratoconus severity. No independent relationships between dioptric difference and HOAs could be identified (Table 2).
Factors Associated With the Discrepancy Between Autorefraction and Manifest Refraction in Keratoconus Using Multivariable Regression Analysis
The correlation between Kmax and Kmean and the dioptric difference is shown in Figure 1. There was a strong and significant correlation between Kmax and dioptric difference (R2 = 0.682, P ≤ .0001), and a lower but significant correlation between Kmean and dioptric difference (R2 = 0.523, P ≤ .0001), meaning that the difference between both refraction methods increased when the cornea steepened and the keratoconus stage was worse. No clinically relevant or significant association between cone location and dioptric difference was identified (R2 = .036, P = .074).
Correlation between mean and maximum keratometry (Kmax and Kmean), and dioptric difference (dD) (P ≤ .0001).
This study shows that superior spectacle CDVA is achieved using manifest refraction compared to autorefraction in patients with keratoconus. Autorefraction and manifest refraction lead to significantly different results and this difference increases as Kmax increases.
A strength of this study is the standardization of the procedure, with one experienced optometrist (NS) performing all measurements in the same examination room with the same lighting conditions and equipment. We randomized the order in which patients were measured with autorefraction and manifest refraction to make sure that the visual acuity outcomes were not biased by fatigue or a learning effect. Another strength of this study is the application of a Fourier analysis for determining the dioptric difference between the automated and manifest refraction. This Fourier analysis is superior to spherical equivalent for research purposes because it takes into account the axis and the dependency of spherical and cylindrical measurements. In this study, patients were divided into keratoconus stages according to the Amsler–Krumeich classification and showed a large variation in dioptric difference between manifest and autorefraction, even in the low keratoconus group. Only 10% fell within a 1.00 diopter difference between manifest and autorefraction.
Some considerations should be mentioned. Accommodation could have biased the results, especially because the population in this study is relatively young and, in an attempt to correct for the visual distortion caused by the ectasia, accommodation may fluctuate more in keratoconus.16 A subpopulation in this study had received a corneal cross-linking treatment before participating in this study, which could have caused a subclinical residual haze that might have affected the measurements. However, all participants were checked for haze and a period of at least 6 months was maintained between the corneal cross-linking treatment and participation in this study; persistence of haze after 6 months is uncommon.19
There are several possible explanations for the discrepancy between the measurements with autorefraction and manifest refraction. The corneal ectasia in keratoconus reduces the image quality and causes a blur at the retinal plane. It is possible that these aberrations interfere with the measurement because the autorefractor relies on image quality for its measurements.3 Furthermore, the autorefractor takes the mean corneal strength after following a sine function to measure the meridional strength of the cornea.3 Extreme measurements can alter this mean meridional strength.
Wavefront aberrometry has been investigated as a promising option to provide a reliable autorefraction in keratoconic eyes, but a study that compared it to manifest refraction reported a lower visual acuity achieved with the aberrometer-derived refraction.13 Correction of HOAs in keratoconic eyes has also been proposed, but resulted in widely varying visual performance.20–23 It has been questioned whether correction for HOAs is helpful because the neural system seems to adapt to these HOAs.24
Another theory that can influence the visual acuity is a shift in the line of sight that patients with keratoconus could develop.16 A deviation in the retinal focal point emerges in reference to the corneal apex to the area where the best visual acuity is achieved. Milháltz et al.25 found evidence for a shift in the line of sight in keratoconic eyes using wavefront aberrometry. Moreover, vertical coma was significantly associated with this displacement.
Because manifest refraction is a subjective measurement, the patient will potentially identify the resultant effect of this displacement and vertical coma through a preference for an empirically achieved amount of regular astigmatic correction.
The repeatability of manifest refractions in patients with keratoconus is considered weak according to a study in 1998,7 but further investigations should be performed considering the better visual acuity found in our study compared to autorefraction.
Superior visual performance is achieved with manifest refraction compared to autorefraction in patients with keratoconus. Furthermore, the difference between the two refraction methods increases as the cornea steepens. According to this study, autorefraction is unreliable in patients with keratoconus and should be used with caution.
- Rabinowitz Y. Keratoconus. Surv Ophthalmol. 1998;42:297–319. doi:10.1016/S0039-6257(97)00119-7 [CrossRef]
- Romero-Jiménez M, Santodomingo-Rubido J, Wolffsohn JS. Keratoconus: a review. Contact Lens Anterior Eye. 2010;33:157–166. doi:10.1016/j.clae.2010.04.006 [CrossRef]
- Elliott DB. What is the appropriate gold standard test for refractive error?Ophthalmic Physiol Opt. 2017;37:115–117. doi:10.1111/opo.12360 [CrossRef]
- Bullimore MA, Fusaro RE, Adams CW. The repeatability of automated and clinician refraction. Optom Vis Sci. 1998;75:617–622. doi:10.1097/00006324-199808000-00028 [CrossRef]
- Donders FC. On the Anomalies of Accommodation and Refraction of the Eye. , trans. LondonSW: The New Sydenham Society; 1864.
- Momeni-Moghaddam H, Goss DA. Comparison of four different binocular balancing techniques. Clin Exp Optom. 2014;97:422–425. doi:10.1111/cxo.12198 [CrossRef]
- Davis LJ, Schechtman KB, Begley CG, Shin JA, Zadnik KThe CLEK Study Group. Repeatability of refraction and corrected visual acuity in keratoconus. Optom Vis Sci. 1998;75:887–896. doi:10.1097/00006324-199812000-00011 [CrossRef]
- Radhakrishnan H, Jinabhai A, O'Donnell C. Dynamics of ocular aberrations in keratoconus. Clin Exp Optom. 2010;93:164–174. doi:10.1111/j.1444-0938.2010.00471.x [CrossRef]
- Bruce AS, Catania LJ. Clinical applications of wavefront refraction. Optom Vis Sci. 2014;91:1278–1286. doi:10.1097/OPX.0000000000000377 [CrossRef]
- Colak HN, Kantarci FA, Yildirim A, et al. Comparison of corneal topographic measurements and high order aberrations in keratoconus and normal eyes. Contact Lens Anterior Eye. 2016;39:380–384. doi:10.1016/j.clae.2016.06.005 [CrossRef]
- Rozema JJ, Van Dyck DEM, Tassignon MJ. Clinical comparison of 6 aberrometers: Part 1. Technical specifications. J Cataract Refract Surg. 2005;31:1114–1127. doi:10.1016/j.jcrs.2004.11.051 [CrossRef]
- Wisse RPL, Gadiot S, Soeters N, Godefrooij DA, Imhof SM, van der Lelij A. Higher-order aberrations one year after corneal crosslinking for keratoconus and their independent effect on visual acuity. J Cataract Refract Surg. 2016;42:1046–1052. doi:10.1016/j.jcrs.2016.04.021 [CrossRef]
- Jinabhai A, O'Donnell C, Radhakrishnan H. A comparison between subjective refraction and aberrometry-derived refraction in keratoconus patients and control subjects. Curr Eye Res. 2010;35:703–714. doi:10.3109/02713681003797921 [CrossRef]
- Krumeich JH, Daniel J. Live epikeratophakia and deep lamellar keratoplasty for I–III stage-specific surgical treatment of keratoconus [article in German]. Klin Monbl Augenheilkd. 1997;211:94–100. doi:10.1055/s-2008-1035103 [CrossRef]
- Thibos LN, Wheeler W, Horner D. Power vectors: an application of Fourier analysis to the description and statistical analysis of refractive error. Optom Vis Sci. 1997;74:367–375. doi:10.1097/00006324-199706000-00019 [CrossRef]
- Raasch TW, Schechtman KB, Davis LJ, Zadnik K. Repeatability of subjective refraction in myopic and keratoconic subjects: results of vector analysis. Ophthalmic Physiol Opt. 2001;21:376–383. doi:10.1046/j.1475-1313.2001.00596.x [CrossRef]
- Raasch TW. Clinical refraction in three-dimensional dioptric space revisited. Optom Vis Sci. 1997;74:376–380. doi:10.1097/00006324-199706000-00020 [CrossRef]
- Keating MP. Geometric, Physical, and Visual Optics. Philadelphia: Elsevier Health Sciences; 1988.
- Dhawan S, Rao K, Natrajan S. Complications of corneal collagen cross-linking. J Ophthalmol. 2011;2011:869015. doi:10.1155/2011/869015 [CrossRef]
- Katsoulos C, Karageorgiadis L, Vasileiou N, Mousafeiropoulos T, Asimellis G. Customized hydrogel contact lenses for keratoconus incorporating correction for vertical coma aberration. Ophthalmic Physiol Opt. 2009;29:321–329. doi:10.1111/j.1475-1313.2009.00645.x [CrossRef]
- Okamoto C, Okamoto F, Samejima T, Miyata K, Oshika T. Higher-order wavefront aberration and letter-contrast sensitivity in keratoconus. Eye (Lond). 2008;22:1488–1492. doi:10.1038/sj.eye.6702902 [CrossRef]
- Sabesan R, Yoon G. Visual performance after correcting higher order aberrations in keratoconic eyes. J Vis. 2009;9:6.1–10. doi:10.1167/9.5.6 [CrossRef]
- Applegate RA, Marsack JD, Ramos R, Sarver EJ. Interaction between aberrations to improve or reduce visual performance. J Cataract Refract Surg. 2003;29:1487–1495. doi:10.1016/S0886-3350(03)00334-1 [CrossRef]
- Artal P, Chen L, Fernández EJ, Singer B, Manzanera S, Williams DR. Neural compensation for the eye's optical aberrations. J Vis. 2004;4:281–287. doi:10.1167/4.4.4 [CrossRef]
- Miháltz K, Kránitz K, Kovács I, Takács Á, Németh J, Nagy ZZ. Shifting of the line of sight in keratoconus measured by a Hartmann-Shack sensor. Ophthalmology. 2010;117:41–48. doi:10.1016/j.ophtha.2009.06.039 [CrossRef]
Study Group Characteristics
|Type of Measurement||Value (Mean ± SD)|
|UDVA (logMAR) (Snellen)||0.65 ± 0.49 (20/90)|
|CDVA autorefraction (logMAR) (Snellen)||0.28 ± 0.29 (20/38)|
|CDVA manifest refraction (logMAR) (Snellen)||0.06 ± 0.18 (20/23)|
|Kmean (D)||46.70 ± 3.84|
|Kmax (D)||54.10 ± 6.00|
|Cone location (mm)||1.73 ± 1.02|
|Horizontal coma (Z31)||−0.02 ± 0.95|
|Vertical coma (Z3−1)||−1.98 ± 1.22|
|Horizontal trefoil (Z33)||−0.004 ± 0.30|
|Vertical trefoil (Z3−3)||0.12 ± 0.30|
|Spherical aberration (Z40)||−0.28 ± 0.62|
|SE autorefraction (D)||−5.71 ± 3.42|
| Spherical autorefraction (D)||−3.90 ± 0.50|
| Cylindrical autorefraction (D)||−3.62 ± 2.31|
|SE manifest refraction (D)||−1.73 ± 3.85|
| Spherical manifest refraction (D)||0.23 ± 3.43|
| Cylinder manifest refraction (D)||−3.90 ± 2.16|
Factors Associated With the Discrepancy Between Autorefraction and Manifest Refraction in Keratoconus Using Multivariable Regression Analysis
|RMS HOAs cornea||−1.822||0.611||.43|
|Horizontal coma (Z31)||0.692||0.530||.47|
|Vertical coma (Z3−1)||2.148||0.941||.33|
|Horizontal trefoil (Z33)||−1.494||0.403||.53|
|Vertical trefoil (Z3−3)||−0.365||0.019||.89|
|Spherical aberration (Z40)||1.625||1.267||.26|