To assess the refractive error of the eye, an objective refraction using an automated device (autorefractor) is performed to approximate the subjective refraction. Many types of autorefractors have been described since the first automated optometer was introduced in 1970, and objective refraction is currently a routine examination. Several studies1–3 have evaluated the accuracy of autorefractors in healthy eyes. However, it is not known whether autorefractors have the same accuracy and reliability after corneal refractive surgery. Photorefractive keratectomy (PRK)4,5 and laser in situ keratomileusis (LASIK)6 are now the most common refractive surgery techniques. It is therefore important to evaluate how the properties of the eye are altered by these procedures and how changes affect ophthalmic measurements such as automated refraction. Recently, Øyo-Szerenyi and coauthors7 demonstrated that the objective refraction differs more from the subjective refraction after PRK for myopia than before.
In addition to the ablation of corneal tissue as in PRK, LASIK involves a lamellar cut and thus profoundly alters the topographical and biophysical properties of the cornea. To our knowledge, no study has compared the objective and subjective refractions before and after LASIK. Since LASIK is becoming a popular alternative to PRK, we conducted a study to evaluate objective refraction before and after LASIK to correct myopia, hyperopia, and astigmatism.
Patients and Methods
In this prospective study, 125 eyes of 75 patients with myopia (32 women and 43 men with a mean age of 34.8 years ± 8.1 [SD] [range 19 to 55 years]) and 34 eyes of 19 patients with hyperopia (9 women and 10 men with a mean age of 44.3 ± 13.6 years [range 22 to 64 years]) were included. In all eyes, the cylindrical refractive error was less than the spherical error. Inclusion criteria were a best spectacle-corrected visual acuity (BSCVA) of 20/40 or better, a stable refraction for at least 6 months, 18 years of age or older at the time of surgery, and normal intraocular pressure (10 to 20 mm Hg). Exclusion criteria were pregnancy or a systemic disease that could interfere with wound healing or the synthesis of connective tissue or previous ocular surgery, disease, or injury. Hard contact lenses were discontinued 4 weeks before the preoperative examination and soft contact lenses, 2 weeks before. A general medical certificate attested to the patient's good health. All patients provided informed consent to the operation and inclusion in the study after they were given detailed information.
All surgery was performed by 1 surgeon (M.E.Z.) who attempted to correct the entire refractive error in each eye. The surgical procedure has been fully described.8,9 The Automated Corneal Shaper (Chiron Vision) was used to perform the lamellar cut. To correct hyperopia, the Chiron Keracor 117 excimer laser was used to ablate the cornea by scanning it with a 2 mm spot using the PlanoScan algorithm, software version 2.422. The repetition rate was 25 Hz and the energy density 120 mJ/cm2.8 If astigmatism was treated simultaneously, it was converted into a positive cylinder and the flat meridian was steepened by the ablation. To correct myopia, 63 eyes (50.4%) were treated with the Chiron Keracor 117 laser. If astigmatism was treated at the same time, it was converted into a negative cylinder and the cornea was flattened in the steep meridian. The remaining 62 myopic eyes (49.6%) were treated with the VISX Twenty-Twenty R excimer laser, which has an expanding iris diaphragm and ablates the cornea in a multizone procedure.9 To correct astigmatism (expressed as a negative cylinder), the cornea was ablated with an elliptical beam. The repetition rate was 10 Hz and the energy density 160 mJ/cm2. Both Chiron and VISX lasers emit ultraviolet light at a wavelength of 193 nm.
Assessment of Refraction
All refractions were assessed by clinicians who did not operate on the patients. The objective refraction was measured with the Nidek AR-K 900 autorefractor according to the manufacturer's guidelines. The device has an autofogging mechanism to relax accommodation. There are 2 separate infrared light sources within a certain distance of each other. The image of the light sources is projected through a small aperture onto the patient's retina, reflected onto a half mirror, and delivered to a photodetector in the optical receiving system of the autorefractor. If the eye is ametropic, the aperture image is not projected sharply on the retinaand the image on the photodetector is not in focus.The aperture circle diaphragm then varies its distance from the light sources until there is a sharp image on the retina and thus on the photodetector, and the distance of the diaphragm from the zero position is used to calculate the refractive error of the eye. Three measurements of each eye were taken and the values automatically averaged.
The subjective refraction was determined by assessing BSCVA, using standard Snellen visual acuity charts and a standard trial frame in which loose lenses could be inserted so the lens with the highest refraction was next to the eye. The distance between the lens next to the cornea and the corneal vertex was 12.0 mm. The procedures were performed separately in each eye; all refractive measurements were done without cycloplegia. The refraction was noted as sphere (diopter [D]), negative cylinder (D) × axis (degree). The spherical equivalent was not used in this analysis.
The difference between the objective and subjective refractions (objective refraction [D] − subjective refraction [D] = difference [D]) and the mean difference (mean ± standard deviation [SD]) were calculated. As an indicator of the correlation between the objective and subjective refractions, the correlation coefficient (r2) was computed. These values were used to examine the accuracy of the objective refraction compared with the subjective refraction. To address the reliability of the objective refraction, the objective values were grouped within ±0.50, ±1.00, and ±2.00 D of the subjective refractions. The paired Student t test was used for within-group comparisons and the unpaired Student t test for between-group comparisons.
The magnitude of the cylinder (in diopters) was analyzed in the same manner as the sphere. For the axis, the axis difference (objective cylinder axis [degrees] − subjective cylinder axis [degrees] = axis difference [degrees]) was measured, the mean axis difference (mean ± SD) calculated, and the number of eyes in which the objective axis was within ±10 degrees of the subjective axis determined. The axis difference had to be less than 90 degrees. For example, the difference between an axis of 0 degree and another of 90 degrees was 90 degrees; the difference between an axis of 1 degree and another of 180 degrees was 1 degree. Only eyes with cylinder values of at least 0.25 D in the objective and subjective refractions were included in the axis analysis.
To determine whether the preoperative (baseline) refractive error influenced the preoperative and postoperative mean differences or the change between them, the myopic and hyperopic eyes were divided into 3 and 2 subgroups, respectively, based on the baseline ametropia. The refractive differences were plotted against BSCVA to assess whether lower BSCVA values were correlated with higher refractive differences.
Baseline Refractive Errors
In the myopic group, the mean subjective sphere was −6.43 ± 3.43 D (range −1 to −16 D) before LASIK and −0.20 ± 0.86 D (range +2 to −3 D) after LASIK, and the mean subjective cylinder was −0.88 ± 0.93 D (range 0 to −4 D) and −0.30 ± 0.73 D (range 0 to −5.5 D), respectively. In the hyperopic group, the mean subjective sphere was +4.49 ± 1.62 D (range +1.5 to +7.5 D) before and −0.47 ± 0.97 D (range −1.5 to +3.75 D) after, and the mean subjective cylinder was −0.83 ± 0.75 D (range 0 to −3 D) and −0.15 ± 0.34 D (range 0 to −1.75 D), respectively.
Differences Between Objective and Subjective Spherical Refractions
Figure 1, left, shows the mean objective refraction versus the mean subjective refraction in the myopic group before LASIK. There was a strong correlation between the measurements (Table 1). The objective refraction was less closely correlated with the subjective refraction 6 months after LASIK (Figure 1, right;Table 1). Many eyes had no subjective refractive error but had objective refractive errors. The difference between the objective and subjective refractions was not statistically significant before or after LASIK. Although the objective refraction was more reliable before LASIK (96.0% of the objective values within ±1.00 D of the subjective values), there was a reasonable reliability (88.8% within ±1.00 D) after LASIK.
Before LASIK, the high myopia subgroup had the greatest mean difference between the objective and subjective refractions, and it was significantly greater (P < .01) than in the low myopia subgroup (Table 2). This trend was not seen 6 months after LASIK. In contrast, the low myopia eyes had the highest mean difference. Thus, there was no postoperative tendency toward higher mean differences with higher baseline myopia.
Before LASIK, there was no statistically significant difference between the 2 groups treated with different lasers. The mean preoperative difference was −0.08 ± 0.45 D (range –1.25 to +1.00 D) in the 62 eyes treated with the VISX laser and –0.04 ± 0.40 D (range –1.25 to +0.75 D) in the 63 eyes treated with the Chiron laser. Six months after LASIK, the mean difference was +0.30 ± 0.66 D (range –1.25 to +2.50 D) in the VISX group and –0.16 ± 0.80 D (range –3.75 to +1.00 D) in the Chiron group (P < .001). The change in the mean difference before and after LASIK was greater in the VISX group (+0.38 D; P < .001) than in the Chiron group (−0.12 D).
In the hyperopic group, the objective refraction was less accurate and reliable than in the myopic group (Figure 2, left;Table 1) before LASIK. However, the difference between the refractions was not statistically significant. Six months after LASIK, there was a significant difference between the refractions (P < .000 001) and the correlation decreased considerably (from r2 = 0.71 before to r2 = 0.19 after). The objective refraction was also less reliable: 73.5% were within ±1.00 D of the subjective values before LASIK but only 38.2% after. The 2 subgroups based on baseline hyperopia are shown in Table 3. Before LASIK, the mean difference between the refractions was smaller in the high hyperopia subgroup and was statistically significant in the low hyperopia subgroup (P = .011). After LASIK, the difference was statistical significant in both subgroups. The change in the mean difference from before to after was statistically significant in the high hyperopia subgroup and considerably greater (+2.46 D; P < .001) than in the low hyperopia subgroup (+0.58 D; P = .052).
There was no statistical correlation between BSCVA and the refraction difference before and after LASIK.
Differences Between Objective and Subjective Cylindrical Refractions
In the myopic group, the magnitude of the cylinder in the objective refraction was significantly different from that in the subjective refraction before and after LASIK (Table 4). However, the correlation was stronger before than after (Table 2, Figure 3), and more objective measurements were within ±1.00 D of the subjective values before than after (96% and 87.2%, respectively). Many eyes with zero subjective refractions had different values in the objective refraction after LASIK.
In the hyperopic group, the objective and subjective refractions were significantly different before and after LASIK. The correlation between them was greater, and more objective values were within ±1.00 D of the subjective refraction before than after (Table 2, Figure 4). Many eyes with zero subjective cylinder had objective values ranging from 0 to −2.5 D after LASIK, and there was no correlation between the magnitude of the cylinder in the objective and subjective refractions.
The highest mean axis difference was in the hyperopic group after LASIK (+6.0 ± 45.5 degrees; range −86 to +82 degrees) (Table 4) and the fewest objective axes were within ±10 degrees of the subjective measurements (16.7%). In the myopic group, accuracy of the axis measurement in the objective refraction was comparable before and after LASIK.
Corneal refractive procedures are widely applied to correct ametropia. Since these methods alter the optical, biohistochemical,10 and biophysical11 properties of the cornea, they may affect the reliability and accuracy of autorefraction and autokeratometry measurements.
Our preoperative findings agree with those in other studies of the precision and reliability of the Nidek autorefractor.2,3 Minor differences are present, however. Kinge and coauthors3 found that the Nidek AR-1000 autorefractor, which is comparable to the device used in our study, yields a more myopic value than the subjective refraction. Preoperatively, this was true in our myopic group but not in the hyperopic group. This is remarkable since we performed all measurements without cycloplegia. Thus, patients were able to accommodate during examinations, and we would have expected the hyperopia to produce more negative refractions by autorefraction because of accommodation, an effect termed instrument myopia. Since our preoperative results are comparable to those in earlier studies, the greater differences between objective and subjective refractions postoperatively must be caused by the refractive procedure.
Refractive and keratometric readings vary more in eyes treated with PRK for myopia than in untreated control eyes.7 To quantitatively examine whether this is also the case after LASIK, we conducted this prospective study and also included hyperopic eyes treated with LASIK. The results show that objective refractions obtained with the Nidek AR-K 900 autorefractor differed significantly more from the subjective refractions after LASIK than before. The type of ametropia appeared to influence the objective refraction after LASIK, because the objective refraction was less reliable after LAISK for hyperopia. In the hyperopic group, the postoperative mean difference was highest in eyes with high baseline ametropia. A deeper ablation appeared to be associated with a loss of accuracy and reliability in the objective refraction in this group, but this was not observed in the myopic group. Also, the type of excimer laser appeared to influence the autorefraction results. While the accuracy of autorefraction was comparable preoperatively, it was better in the Chiron-treated eyes after LASIK. A separate study must be performed to confirm whether this is a typical effect of the lasers.
Postoperatively, there were many eyes with zero subjective refraction and various objective refractions. In other words, the autorefractor detected deviations from plano refraction that could not be corrected with spherical or cylindrical lenses in the subjective refraction. Either the autorefractor gave a false result or the examiner was biased. To minimize the latter, patients were seen by clinicians other than the surgeon. However, the examiners were informed about the patients' medical history and bias cannot be completely excluded.
There are several possible explanations for the loss of accuracy and reliability in autorefraction and the clustering of plano manifest refractions with various values in autorefraction that were seen in this study. First, the anterior curvature of the cornea is markedly altered by the ablation. In normal eyes, central corneal power averages 43.75 D.12 In the eyes in this study, the central corneal power was changed by up to 16 D (the most myopic eye). Not only the central but also the paracentral part of the cornea is affected. Transition zones, creating a smooth change from the fully corrected optical zone in the center to the untreated peripheral cornea, are shaped. For example, if a 5.0 mm optical zone is created with the PlanoScan algorithm, the entire ablated area will measure 8.6 mm in diameter. The important factor is where the infrared rays of the autorefractor pass the cornea. Russel and coauthors13 note that the pupil size can influence the precision of the objective refraction in an ablated eye. When the infrared rays pass through the transition or unablated zone, the measurement will be incorrect. Because higher corrections require smaller optical zones, we analyzed the effect of baseline refractive error on the objective refraction. Only in the hyperopic group was there a trend toward less accurate autorefraction in eyes with high ametropia preoperatively. In the myopic group, the contrary was found, but standard deviation was highest in the eyes with high myopia originally, indicating less reliable measurements.
Second, corneas are usually prolate surfaces (i.e., they flatten toward the periphery).14 After LASIK for myopia, the cornea becomes an oblate surface (steeper in the periphery than in the center), whereas after LASIK for hyperopia, the prolate shape increases. Thus, the anterior shape is markedly altered by LASIK. It is not clear whether an automated device like the Nidek AR-K 900 autorefractor can compensate for significant deviations from the normal anterior shape of the cornea.
Third, the lamellar cut and the laser ablation may create an irregular corneal surface. This can also affect the accuracy of the objective refraction. Furthermore, the cornea undergoes healing changes after keratorefractive procedures. During wound healing, the refractive index of the cornea might be altered.15,16 Although we did not find clinically relevant corneal haze after LASIK,8,9 we cannot exclude subtle healing changes that might interfere with accurate autorefraction. An irregular corneal surface or intensive stromal opacities17 may create a situation in which a refractive error is measured in autorefraction but no subjective correction is accepted. In this case, the BSCVA should decrease because of the irregular features of the cornea. Indeed, a reduction in BSCVA was found in some eyes postoperatively.8,9 However, we did not find any correlation between postoperative BSCVA and the difference between objective and manifest refractions. Also, we did not detect excessive wound healing or postoperative corneal haze more intensive than grade 1, nor did we find clinically relevant optical zone decentrations or keratoconus in postoperative corneal topographies.8,9 Further investigation is needed to explain the decreased accuracy of autorefraction after refractive procedures.
The findings of this study have practical implications. First, there is the question of postoperative care after refractive surgery. Although most clinical studies use subjective refraction, others may prefer objective refraction.18 Based on our findings, we wonder whether autorefraction results are comparable with subjectively measured outcomes. We therefore recommend using subjective refraction exclusively in clinical trials, and clinicians should be aware that autorefraction can be misleading after LASIK. Although the objective refraction can be considered a reasonable method to approximate subjective refraction after LASIK for myopia (88.8% of objective values within ±1.00 D of the subjective values), it was not reliable after LASIK for hyperopia (38.2% of objective values within ±1.00 D of the subjective values). Being unaware of this fact can lead to uncertainty among patients and physicians and even to unnecessary retreatments.
Second, the planning of other surgical procedures is affected. Before cataract surgery, the radius of the cornea is measured (often with automated devices also used for the assessment of objective refraction). The precision of these measurements may be decreased, as is the objective refraction since the same principle applies. A previous study7 revealed that autokeratometric readings varied more from those obtained with subjective methods after PRK for myopia. Accordingly, autokeratometry should not be used to calculate intraocular lens (IOL) power after LASIK. Instead, corneal topography should be studied to identify the fully corrected optical zone on the cornea. The readings in this area can be inserted into the IOL power calculation if the cornea has a regular shape. Some topography devices automatically show the values for the 3 mm central zone of the cornea,14 which often represents the fully corrected optical zone. Another option is to perform autokeratometry before the refractive procedure and add the achieved subjective correction to the measurement.
In conclusion, this study demonstrates that the objective refraction can differ significantly from the subjective refraction after LASIK, especially after LASIK for hyperopia. The physician should therefore not rely on automated devices to assess the refraction or anterior curvature in a PRK- or LASIK-treated cornea. Further studies are necessary to determine the causes of the decreased accuracy and reliability of the objective refraction after refractive corneal surgery.
1. Wood ICJ. A review of autorefractors. Eye 1987; 1:529-535
2. Sunder Raj P, Villada JR, Lewis AE, et al. Comparative evaluation of the Allergan Humphrey 570 and Canon RK-1 autorefractors: I: objective autorefraction in normal subjects. Eye 1992; 6:284-286
3. Kinge B, Midelfart A, Jacobsen G. Clinical evaluation of the Allergan Humphrey 500 autorefractor and the Nidek AR-1000 autorefractor. Br J Ophthalmol 1996; 80:35-39
4. Trokel SL, Srinivasan R, Braren B. Excimer laser surgery of the cornea. Am J Ophthalmol 1983; 96:710-715
5. Munnerlyn CR, Koons SJ, Marshall J. Photorefractive keratectomy: a technique for laser refractive surgery. J Cataract Refract Surg 1988; 14:46-52
6. Pallikaris IG, Papatzanaki ME, Stathi EZ, et al. Laser in situ keratomileusis. Lasers Surg Med 1990; 10:463-468
7. Øyo-Szerenyi KD, Wienecke L, Businger U, Schipper I. Autorefraction/autokeratometry and subjective refraction in untreated and photorefractive keratectomy-treated eyes. Arch Ophthalmol 1997; 115:157-164
8. Salchow DJ, Zirm ME, Stieldorf C, Parisi A. Laser in situ keratomileusis for myopia and myopic astigmatism. J Cataract Refract Surg 1998; 24:175-182
9. Salchow DJ, Zirm ME, Stieldorf C, Parisi A. Laser-in-situ Keratomileusis (LASIK) zur Myopie- und Astigmatismuskorrektur. Ophthalmologe 1998; 95:142-147
10. Assil KK, Quantock AJ. Wound healing in response to keratorefractive surgery. Surv Ophthalmol 1993; 38:289-302
11. Peacock LW, Slade SG, Martiz J, et al. Ocular integrity after refractive procedures. Ophthalmology 1997; 104:1079-1083
12. Rabinowitz YS, Yang H, Brickman Y, et al. Videokeratography database of normal human corneas. Br J Ophthalmol 1996; 80:610-616
13. Russel GE, Berymanson JPG, Barbeito R, Cross WD. Differences between objective and subjective refractions after radial keratotomy. Refract Corneal Surg 1992; 8:290-295
14. Holladay JT. Corneal topography using the Holladay Diagnostic Summary. J Cataract Refract Surg 1997; 23:209-221
15. Rogers CM, Lawless MA, Cohen PR. Photorefractive keratectomy for myopia of more than −10 diopters. J Refract Corneal Surg 1994; 10(suppl):S171-S173
16. McDonnell PJ. Excimer laser corneal surgery: new strategies and old enemies. Invest Ophthalmol Vis Sci 1995; 36:4-8
17. Heitzmann J, Binder PS, Kassar BS, Nordan LT. The correction of high myopia using the excimer laser. Arch Ophthalmol 1993; 111:1627-1634
18. Maldonado Bas A, Onnis R. Excimer laser in situ keratomileusis for myopia. J Refract Surg 1995; 11(suppl.):S229-S233