- Email: [email protected]
Effect of instrument rotation on handheld keratometry
Effect of instrument rotation on handheld keratometry Andrew K.C. Lam, PhD, FAAO, Rufina Chan, MOptom, FAAO, Roger Chiu, HCert(Optom) Purpose: To study the effect of instrument rotation on handheld automated keratometry. Setting: Department of Optometry and Radiography, The Hong Kong Polytechnic University, Hong Kong SAR, China. Methods: In 33 recruited subjects, corneal curvature in 1 eye (randomly selected) was measured using a Medmont topographic keratometer and a Nidek handheld keratometer. In handheld keratometry, measurements were obtained holding the instrument in the upright position and rotating it anticlockwise 5 degrees, 10 degrees, and 15 degrees and then clockwise 5 degrees, 10 degrees, and 15 degrees. The sequence of measurements was randomly assigned. The steepest and flattest corneal curvatures as well as the axis along the flattest meridian were compared in different conditions. The corneal power was converted to vector representation (M, J0, J45) for a second comparison. Results: There was a significant difference in the steepest and flattest curvatures in different conditions (repeated-measures analysis of variation, P!.05). However, the mean difference between handheld and topographic keratometry was approximately 0.25 diopter (D) for the steepest curvature and 0.05 D for the flattest curvature regardless of the direction and amount of rotation. The intraclass correlation coefficients were nearly 1, which indicated good clinical reliability. The mean difference in axis determination followed the direction and amount of rotation. With vector representation, the difference between the devices increased with the amount of rotation, especially J0 and J45. Conclusions: Handheld keratometry was in agreement with topographic keratometry. However, practitioners should adjust the axis manually according to the direction and amount of rotation. The difference between handheld and topographic keratometry increased with the rotational effect, which was seen with vector representation. Practitioners are advised to use the handheld keratometer in the upright position. J Cataract Refract Surg 2004; 30:2590–2594 ª 2004 ASCRS and ESCRS
C
orneal refractive power can be measured by conventional keratometry and sophisticated topographic keratometry. Measurement of the corneal refractive power is essential in cataract extraction with implantation of an intraocular lens. Errors in measur-
Accepted for publication April 6, 2004. Reprint requests to Dr. Andrew Lam, Department of Optometry and Radiography, The Hong Kong Polytechnic University, Hong Kong SAR, China. 2004 ASCRS and ESCRS Published by Elsevier Inc.
ing the true corneal power can result in significant postoperative refractive errors.1–3 Handheld keratometry has the advantage of portability. Studies of handheld keratometers suggest that to measure the axis accurately, the keratometer should be in the upright position (6 o’clock to 12 o’clock).4,5 The manual keratometer is reported to provide better repeatability than the handheld keratometer in cataract extraction surgery.6 There have been few reports of the effect of rotation, clockwise or anticlockwise, on the handheld keratometer. A study by Lam4 was based on 0886-3350/04/$–see front matter doi:10.1016/j.jcrs.2004.04.069
TILTING EFFECT ON HANDHELD KERATOMETRY
an Alcon keratometer model. Investigation of the Nidek handheld keratometer confirms its clinical reliability,7 but the effect of rotation has not been studied. We looked at the effect of clockwise and anticlockwise rotation on the Nidek handheld keratometer, which is used in the Optometry Clinic of The Hong Kong Polytechnic University.
Subjects and Methods Thirty-three young, normal subjects were recruited. They had no ocular pathology. Written consent was obtained before keratometry measurements were performed. The study was approved by the departmental ethics committee and followed the tenets of the Declaration of Helsinki. One eye of each subject was randomly selected for the following corneal measurements in random order: The corneal curvature was measured with a Medmont E300 topographic keratometer. The keratometer took repeated captures and rated the accuracy of focus and centration. In each subject, 2 readings were taken by the same practitioner; only measurements with a rating higher than 90 were used.8 The simulated K-reading was used for analysis, and the result was treated as standard. A second practitioner performed the measurement with the Nidek KM-500 handheld keratometer. Subjects were required to sit in an examination chair with the head in the upright posture. The measurement with handheld keratometer was carried out with various random rotations. These involved no rotation of the keratometer (ie, upright position); anticlockwise rotation of 5 degrees, 10 degrees, and 15 degrees; and clockwise rotation of 5 degrees, 10 degrees, and 15 degrees. All rotations referred to the practitioner’s view. A third practitioner wrote down the results from the handheld keratometer so the second practitioner did not see the results shown on the screen of the device. The third practitioner also watched for head tilt and reminded subjects to keep their head straight. Three consecutive readings were taken under each rotational setting, and the mean of each setting was used for analysis. All the practitioners were masked to the keratometric results, and measurements in each subject were taken within 10 minutes.
Table 1. The mean 6 SD of corneral curvatures along the steepest and flattest meridians and the axis along the flattest meridian. The vector representation included M, J0, and J45. Measurement
Medmont
Steepest (D)
44.36 6 1.66
Flattest (D)
42.75 6 1.44
Axis (degree)
176.2 6 9.9
M (D)
43.55 6 1.49
J0 (D)
ÿ0.76 6 0.44
J45 (D)
AC Z anticlockwise rotation at 5 , 10 , 15 ; C Z clockwise rotation at 5 , 10 , 15 ; M, J0, J45 Z vector representation
compared the steepest curvature, flattest curvature, and the axis along the flattest meridian. This required converting from vector representation back to 2 principal meridians and the axis. For the axis analysis, some adjustments were required because the axis could not be averaged directly. For example, an axis of 175 in 1 subject with an axis of 5 in another subject could result in a mean of 90. This would mistakenly lead to against-the-rule astigmatism. Therefore, for an axis within the first quadrant (1 to 89), 180 was added to transfer it to the third quadrant (181 to 269). In the example above, the result would be the mean of 175 and 185, ie, axis 180. The second approach was direct comparison of M, J0, and J45. The results were tested for normality (KolmogorovSmirnov test, PO.05) using parametric tests in the analysis. A repeated-measures analysis of variance (ANOVA) was used to compare results between the topographic keratometer and the handheld keratometer under different rotational settings. The intraclass correlation coefficient (ICC), a reliability coefficient calculated from variance estimates obtained with an ANOVA,10 was also calculated for each parameter by comparing the handheld keratometric results with the Medmont results. The mean difference between the topographic keratometer and handheld keratometer and the 95% limits of agreement (LoA) (1.96 multiplied by the standard deviation of the difference) were determined as suggested by Bland and Altman.11
Analysis After the 2 Medmont keratometric measurements or the 3 handheld keratometric readings were averaged, the corneal power was considered as a spherocylindrical lens using orthogonal power vector methods developed by Thibos and coauthors9: a spherical lens of power M [the mean spherical equivalent Z sphere C (cylinder/2)]; Jackson cross cylinder at axis 0 with power J0 [ÿ(cylinder/2) cos (2 ! axis)]; Jackson cross cylinder at axis 45 with power J45 [ÿ(cylinder/2) sin (2 ! axis)]. Two approaches were used to compare different settings. The first approach
0.08 6 0.26
Results The mean age of the 15 men and 18 women was 21.2 years 6 0.99 (SD). The mean spherical equivalent was ÿ3.41 6 2.49 diopters (D). Table 1 shows the mean results of each rotational setting and the vector representation. All the parameters demonstrated significant differences (repeated-measures ANOVA, P!.05) between different settings (Table 2). The ICC results were close to 1 except in axis and J45. The
J CATARACT REFRACT SURG—VOL 30, DECEMBER 2004
2591
TILTING EFFECT ON HANDHELD KERATOMETRY
Table 1
(cont.) Nidek
(No Rotation)
(AC5)
(AC10)
(AC15)
(C5)
(C10)
(C15)
44.09 6 1.67
44.08 6 1.68
44.11 6 1.68
44.11 6 1.70
44.11 6 1.68
44.12 6 1.69
44.10 6 1.71
42.71 6 1.39
42.71 6 1.40
42.69 6 1.39
42.71 6 1.38
42.71 6 1.40
42.71 6 1.38
42.69 6 1.38
177.6 6 12.3
171.3 6 12.1
166.1 6 11.6
160.8 6 14.6
181.2 6 14.4
186.2 6 15.2
191.5 6 13.7
43.40 6 1.47
43.40 6 1.48
43.40 6 1.48
43.41 6 1.48
43.41 6 1.48
43.41 6 1.48
43.40 6 1.48
ÿ0.66 6 0.44
ÿ0.64 6 0.45
ÿ0.63 6 0.43
ÿ0.56 6 0.41
ÿ0.66 6 0.44
ÿ0.64 6 0.43
ÿ0.58 6 0.40
0.007 6 0.21
0.15 6 0.19
0.27 6 0.21
0.37 6 0.28
ÿ0.11 6 0.24
ÿ0.22 6 0.27
ÿ0.31 6 0.34
agreement with topographic results was similar for the steepest and flattest curvatures at different rotations (Table 3); the axis showed greater deviation with increased amounts of anticlockwise and clockwise rotation. The agreement for M was similar among different conditions, but the differences increased with rotation in J0 and especially J45.
have poor agreement between the 2 instruments for the steepest curvature but not for the flattest curvature. This could be the disadvantage of considering only the steepest and flattest curvatures as corneal characteristics and disregarding the influence of axis. Although both Table 2.
Differences in the corneal parameters.
Measurement
Discussion The current study investigated the effect of rotation on handheld keratometry. Both the Medmont topographer keratometer8 and the Nidek handheld keratometer4,5,12 have good repeatability and reliability, but the performance of the Nidek handheld keratometer during rotation had not been investigated. Handheld keratometers have the flexibility to perform the measurement in different positions. Although it is recommended that they be held in the upright position,4 handheld keratometers have a close focusing distance that may require a small amount of rotation to prevent blocking by the subject’s nose. We found significant differences between the steepest and flattest curvature measurements with the Medmont keratometer and the handheld keratometer under different rotations. However, the ICC values were very close to 1. This indicated good validity for clinical measurements.10 The agreement analysis found a mean difference of 0.25 D between the measurements by the 2 devices along the steepest curvature regardless of the direction and amount of rotation (Table 3). The difference along the flattest curvature was even less. Although the mean difference was small, the 95% LoA varied by 60.42 D (1.96 ! 0.216 D) on the steepest curvature and 60.32 D (1.96 ! 0.163 D) on the flattest curvature. It is surprising to 2592
Repeated-Measures ANOVA
ICC
Steepest (D)
F Z 23.74, P!.01
0.996
Flattest (D)
F Z 2.19, P!.05
0.997
Axis (degree)
F Z 46.56, P!.01
0.573
M (D)
F Z 24.30, P!.01
0.998
J0 (D)
F Z 15.11, P!.01
0.958
J45 (D)
F Z 44.19, P!.01
0.351
ANOVA Z analysis of variance; ICC Z intraclass correlation coefficient
Table 3.
The mean difference (95% LoA) for the steepest and flattest meridians, axis along the flattest meridian, and vector representation.
Measurement
Medmont Vs Nidek (No Rotation)
Steepest (D)
0.273 6 0.177 (ÿ0.074 to 0.620)
Flattest (D)
0.040 6 0.132 (ÿ0.219 to 0.299)
Axis (degree)
ÿ1.4 6 11.3 (ÿ23.5 to 20.7)
M (D)
0.156 6 0.103 (ÿ0.046 to 0.358)
J0 (D)
ÿ0.104 6 0.151 (ÿ0.400 to 0.192)
J45 (D)
0.075 6 0.252 (ÿ0.419 to 0.569)
AC Z anticlockwise rotation at 5 , 10 , and 15 ; C Z clockwise rotation at 5 , 10 , 15 ; LoA Z limits of agreement; M, J0, J45 Z vector representation
J CATARACT REFRACT SURG—VOL 30, DECEMBER 2004
TILTING EFFECT ON HANDHELD KERATOMETRY
instruments report the steepest and flattest curvatures, the curvatures could be at different orientations. This makes direct comparison difficult. The power vector method developed by Thibos and coauthors9 is a better approach7 and will be discussed. It is interesting that there was a decrease of about 5 degrees in the axis measurement when the handheld keratometer was rotated 5 degrees anticlockwise and an increase of about 5 degrees when it was rotated 5 degrees clockwise (Tables 1 and 3). The mean differences matched the direction and amount of rotation (Table 3). It seems that the handheld keratometer is incapable of compensating for the effect of rotation. Practitioners should manually correct the measured axis for the angle of rotation. The 95% LoA varied up to 622 degrees (1.96 ! 11.3 degrees), even with the keratometer in the upright position. The deviation reached 627 degrees (1.96 ! 14.0 degrees) with 15 degrees of clockwise rotation. This finding could be the limitation of analyzing the results by the steepest and flattest curvatures and the axis along the flattest curvature. Although consideration of steepest and flattest curvatures as well as axis is simple for clinicians, analysis of the axis required some adjustment, as described. A better approach to analyzing the corneal curvature is to convert the principal meridians into rectangular Fourier form.9 This accounts for the effect of different axes but might be difficult for clinicians to understand. In contrast to a greater difference along the
Table 3
steepest meridian (Table 3), the analysis using vector representation demonstrated a similar order of agreement, that is 0.15 D, with poor agreement on J0 and J45 when rotation increased. Both the mean difference and standard deviation of the difference in J0 and J45 increased with the amount of rotation. This could not be demonstrated with the steepest and flattest meridians. We further analyzed the subjects’ corneal astigmatism using the Medmont results as a reference. The mean corneal astigmatism was 1.61 6 0.88 D (range 0.20 to 3.93 D). Our results included subjects with spherical corneas and those with moderate corneal astigmatism. Practitioners are advised to use the handheld keratometer in the upright position to obtain accurate results. One reason to hold the keratometer with some rotation is to eliminate blocking by the subject’s nose. Another way to avoid this is to ask the subject to look temporally. However, the narrower vertical palpebral aperture at the temporal side may cover part of the corneal surface and obstruct the measurement. Handheld devices with a long focusing distance, eg, an autorefractor, may not have this instrument rotation problem. One limitation of the study is that the subjects were asked to maintain their head in an upright posture without the use of a chin rest. Since the chin rest and forehead rest may obstruct handheld keratometry at different rotational settings, the head position of our subjects was closely monitored by the third practitioner. This practitioner kept reminding the subjects not to tilt
(cont.)
Medmont Vs Nidek (AC5)
Medmont Vs Nidek (AC10)
Medmont Vs Nidek (AC15)
Medmont Vs Nidek (C5)
Medmont Vs Nidek (C10)
Medmont Vs Nidek (C15)
0.277 6 0.164 (ÿ0.044 to 0.598)
0.251 6 0.151 (ÿ0.045 to 0.547)
0.250 6 0.207 (ÿ0.157 to 0.656)
0.249 6 0.190 (ÿ0.123 to 0.621)
0.239 6 0.216 (ÿ0.184 to 0.662)
0.259 6 0.186 (ÿ0.105 to 0.624)
0.042 6 0.137 (ÿ0.227 to 0.311)
0.061 6 0.141 (ÿ0.215 to 0.337)
0.042 6 0.150 (ÿ0.252 to 0.336)
0.031 6 0.113 (ÿ0.190 to 0.252)
0.045 6 0.146 (ÿ0.241 to 0.331)
0.056 6 0.163 (ÿ0.263 to 0.375)
4.9 6 12.1 (ÿ18.8 to 28.6)
10.1 6 11.1 (ÿ35.4 to 55.6)
15.4 6 13.3 (ÿ10.7 to 41.5)
ÿ5.7 6 13.6 (ÿ32.4 to 21.0)
ÿ10.0 6 13.6 (ÿ36.7 to 16.7)
ÿ15.3 6 14.0 (ÿ42.7 to 12.1)
0.159 6 0.108 (ÿ0.053 to 0.371)
0.156 6 0.095 (ÿ0.030 to 0.342)
0.146 6 0.117 (ÿ0.083 to 0.375)
0.140 6 0.106 (ÿ0.068 to 0.348)
0.142 6 0.128 (ÿ0.109 to 0.393)
0.157 6 0.125 (ÿ0.088 to 0.402)
ÿ0.119 6 0.116 (ÿ0.346 to 0.108)
ÿ0.133 6 0.113 (ÿ0.354 to 0.088)
ÿ0.200 6 0.140 (ÿ0.474 to 0.074)
ÿ0.105 6 0.152 (ÿ0.403 to 0.193)
ÿ0.126 6 0.173 (ÿ0.465 to 0.213)
ÿ0.178 6 0.182 (ÿ0.179 to 0.179)
ÿ0.066 6 0.256 (ÿ0.568 to 0.436)
ÿ0.193 6 0.246 (ÿ0.767 to 0.289)
ÿ0.285 6 0.293 (ÿ0.859 to 0.289)
0.208 6 0.235 (ÿ0.253 to 0.669)
0.306 6 0.282 (ÿ0.247 to 0.859)
0.392 6 0.351 (ÿ0.296 to 1.080)
J CATARACT REFRACT SURG—VOL 30, DECEMBER 2004
2593
TILTING EFFECT ON HANDHELD KERATOMETRY
their head, to keep it straight against the chair. Previous studies show that the eyes are capable of compensating for a small amount of head tilt through incycloduction or excycloduction.13,14 We completed the measurements with handheld keratometers within 5 minutes in each subject. Therefore, errors from subjects’ head tilting should be very small. To conclude, we found that rotating the handheld keratometer affected the corneal measurements. The difference may not be great when one looks at the steepest and flattest curvatures and the axis along the flattest meridian. However, use of vector representation may show an increase in deviation by the amount of rotation. Practitioners are reminded to use the handheld keratometer in the upright position.
References 1. Seitz B, Langenbucher A. Intraocular lens power calculation in eyes after corneal refractive surgery. J Refract Surg 2000; 16:349–361 2. Feiz V, Mannis MJ, Garcia-Ferrer F, et al. Intraocular lens power calculation after laser in situ keratomileusis for myopia and hyperopia; a standardized approach. Cornea 2001; 20:792–797 3. Odenthal MTP, Eggink CA, Melles G, et al. Clinical and theoretical results of intraocular lens power calculation for cataract surgery after photorefractive keratectomy for myopia. Arch Ophthalmol 2002; 120:431– 438 4. Lam AKC. A hand-held keratometer. Ophthalmic Physiol Opt 1995; 15:227–230 5. Noonan CP, Rao GP, Kaye SB, et al. Validation of a handheld automated keratometer in adults. J Cataract Refract Surg 1998; 24:411–414
2594
6. Leyland M, Benjamin L. Clinical assessment of a handheld automated keratometer in cataract surgery. Eye 1997; 11:854–857 7. Lam AKC, Chan R, Chiu R. Effect of posture and artificial tears on corneal power measurements by handheld automated keratometer. J Cataract Refract Surg 2004; 30:645–652 8. Cho P, Lam AKC, Mountford J, Ng L. The performance of four different corneal topographers on normal human corneas and its impact on orthokeratology lens fitting. Optom Vis Sci 2002; 79:175–183; errata, 462 9. 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 10. Portney LG, Watkins MP. Foundations of Clinical Research. Applications to Practice. Norwalk, CT, Appleton & Lange, 1993; 505–528 11. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307–310 12. Edwards MH, Cho P. A new, hand-held keratometer: comparison of the Nidek KM-500 auto keratometer with the B&L keratometer and the Topcon RK-3000A keratometer. J Br Contact Lens Assoc 1996; 19:45–48 13. Linwong M, Herman SJ. Cycloduction of the eyes with head tilt. Arch Ophthalmol 1971; 85:570–573 14. Lam AKC, Chung E, Kho J, Wong S. Digital measurement of torsional eye movement due to postural change and its effect on reading performance. Curr Eye Res 2000; 21:763–766 From the Department of Optometry and Radiography, The Hong Kong Polytechnic University, Hong Kong SAR, China. None of the authors has a proprietary or financial interest in any product mentioned.
J CATARACT REFRACT SURG—VOL 30, DECEMBER 2004
C
orneal refractive power can be measured by conventional keratometry and sophisticated topographic keratometry. Measurement of the corneal refractive power is essential in cataract extraction with implantation of an intraocular lens. Errors in measur-
Accepted for publication April 6, 2004. Reprint requests to Dr. Andrew Lam, Department of Optometry and Radiography, The Hong Kong Polytechnic University, Hong Kong SAR, China. 2004 ASCRS and ESCRS Published by Elsevier Inc.
ing the true corneal power can result in significant postoperative refractive errors.1–3 Handheld keratometry has the advantage of portability. Studies of handheld keratometers suggest that to measure the axis accurately, the keratometer should be in the upright position (6 o’clock to 12 o’clock).4,5 The manual keratometer is reported to provide better repeatability than the handheld keratometer in cataract extraction surgery.6 There have been few reports of the effect of rotation, clockwise or anticlockwise, on the handheld keratometer. A study by Lam4 was based on 0886-3350/04/$–see front matter doi:10.1016/j.jcrs.2004.04.069
TILTING EFFECT ON HANDHELD KERATOMETRY
an Alcon keratometer model. Investigation of the Nidek handheld keratometer confirms its clinical reliability,7 but the effect of rotation has not been studied. We looked at the effect of clockwise and anticlockwise rotation on the Nidek handheld keratometer, which is used in the Optometry Clinic of The Hong Kong Polytechnic University.
Subjects and Methods Thirty-three young, normal subjects were recruited. They had no ocular pathology. Written consent was obtained before keratometry measurements were performed. The study was approved by the departmental ethics committee and followed the tenets of the Declaration of Helsinki. One eye of each subject was randomly selected for the following corneal measurements in random order: The corneal curvature was measured with a Medmont E300 topographic keratometer. The keratometer took repeated captures and rated the accuracy of focus and centration. In each subject, 2 readings were taken by the same practitioner; only measurements with a rating higher than 90 were used.8 The simulated K-reading was used for analysis, and the result was treated as standard. A second practitioner performed the measurement with the Nidek KM-500 handheld keratometer. Subjects were required to sit in an examination chair with the head in the upright posture. The measurement with handheld keratometer was carried out with various random rotations. These involved no rotation of the keratometer (ie, upright position); anticlockwise rotation of 5 degrees, 10 degrees, and 15 degrees; and clockwise rotation of 5 degrees, 10 degrees, and 15 degrees. All rotations referred to the practitioner’s view. A third practitioner wrote down the results from the handheld keratometer so the second practitioner did not see the results shown on the screen of the device. The third practitioner also watched for head tilt and reminded subjects to keep their head straight. Three consecutive readings were taken under each rotational setting, and the mean of each setting was used for analysis. All the practitioners were masked to the keratometric results, and measurements in each subject were taken within 10 minutes.
Table 1. The mean 6 SD of corneral curvatures along the steepest and flattest meridians and the axis along the flattest meridian. The vector representation included M, J0, and J45. Measurement
Medmont
Steepest (D)
44.36 6 1.66
Flattest (D)
42.75 6 1.44
Axis (degree)
176.2 6 9.9
M (D)
43.55 6 1.49
J0 (D)
ÿ0.76 6 0.44
J45 (D)
AC Z anticlockwise rotation at 5 , 10 , 15 ; C Z clockwise rotation at 5 , 10 , 15 ; M, J0, J45 Z vector representation
compared the steepest curvature, flattest curvature, and the axis along the flattest meridian. This required converting from vector representation back to 2 principal meridians and the axis. For the axis analysis, some adjustments were required because the axis could not be averaged directly. For example, an axis of 175 in 1 subject with an axis of 5 in another subject could result in a mean of 90. This would mistakenly lead to against-the-rule astigmatism. Therefore, for an axis within the first quadrant (1 to 89), 180 was added to transfer it to the third quadrant (181 to 269). In the example above, the result would be the mean of 175 and 185, ie, axis 180. The second approach was direct comparison of M, J0, and J45. The results were tested for normality (KolmogorovSmirnov test, PO.05) using parametric tests in the analysis. A repeated-measures analysis of variance (ANOVA) was used to compare results between the topographic keratometer and the handheld keratometer under different rotational settings. The intraclass correlation coefficient (ICC), a reliability coefficient calculated from variance estimates obtained with an ANOVA,10 was also calculated for each parameter by comparing the handheld keratometric results with the Medmont results. The mean difference between the topographic keratometer and handheld keratometer and the 95% limits of agreement (LoA) (1.96 multiplied by the standard deviation of the difference) were determined as suggested by Bland and Altman.11
Analysis After the 2 Medmont keratometric measurements or the 3 handheld keratometric readings were averaged, the corneal power was considered as a spherocylindrical lens using orthogonal power vector methods developed by Thibos and coauthors9: a spherical lens of power M [the mean spherical equivalent Z sphere C (cylinder/2)]; Jackson cross cylinder at axis 0 with power J0 [ÿ(cylinder/2) cos (2 ! axis)]; Jackson cross cylinder at axis 45 with power J45 [ÿ(cylinder/2) sin (2 ! axis)]. Two approaches were used to compare different settings. The first approach
0.08 6 0.26
Results The mean age of the 15 men and 18 women was 21.2 years 6 0.99 (SD). The mean spherical equivalent was ÿ3.41 6 2.49 diopters (D). Table 1 shows the mean results of each rotational setting and the vector representation. All the parameters demonstrated significant differences (repeated-measures ANOVA, P!.05) between different settings (Table 2). The ICC results were close to 1 except in axis and J45. The
J CATARACT REFRACT SURG—VOL 30, DECEMBER 2004
2591
TILTING EFFECT ON HANDHELD KERATOMETRY
Table 1
(cont.) Nidek
(No Rotation)
(AC5)
(AC10)
(AC15)
(C5)
(C10)
(C15)
44.09 6 1.67
44.08 6 1.68
44.11 6 1.68
44.11 6 1.70
44.11 6 1.68
44.12 6 1.69
44.10 6 1.71
42.71 6 1.39
42.71 6 1.40
42.69 6 1.39
42.71 6 1.38
42.71 6 1.40
42.71 6 1.38
42.69 6 1.38
177.6 6 12.3
171.3 6 12.1
166.1 6 11.6
160.8 6 14.6
181.2 6 14.4
186.2 6 15.2
191.5 6 13.7
43.40 6 1.47
43.40 6 1.48
43.40 6 1.48
43.41 6 1.48
43.41 6 1.48
43.41 6 1.48
43.40 6 1.48
ÿ0.66 6 0.44
ÿ0.64 6 0.45
ÿ0.63 6 0.43
ÿ0.56 6 0.41
ÿ0.66 6 0.44
ÿ0.64 6 0.43
ÿ0.58 6 0.40
0.007 6 0.21
0.15 6 0.19
0.27 6 0.21
0.37 6 0.28
ÿ0.11 6 0.24
ÿ0.22 6 0.27
ÿ0.31 6 0.34
agreement with topographic results was similar for the steepest and flattest curvatures at different rotations (Table 3); the axis showed greater deviation with increased amounts of anticlockwise and clockwise rotation. The agreement for M was similar among different conditions, but the differences increased with rotation in J0 and especially J45.
have poor agreement between the 2 instruments for the steepest curvature but not for the flattest curvature. This could be the disadvantage of considering only the steepest and flattest curvatures as corneal characteristics and disregarding the influence of axis. Although both Table 2.
Differences in the corneal parameters.
Measurement
Discussion The current study investigated the effect of rotation on handheld keratometry. Both the Medmont topographer keratometer8 and the Nidek handheld keratometer4,5,12 have good repeatability and reliability, but the performance of the Nidek handheld keratometer during rotation had not been investigated. Handheld keratometers have the flexibility to perform the measurement in different positions. Although it is recommended that they be held in the upright position,4 handheld keratometers have a close focusing distance that may require a small amount of rotation to prevent blocking by the subject’s nose. We found significant differences between the steepest and flattest curvature measurements with the Medmont keratometer and the handheld keratometer under different rotations. However, the ICC values were very close to 1. This indicated good validity for clinical measurements.10 The agreement analysis found a mean difference of 0.25 D between the measurements by the 2 devices along the steepest curvature regardless of the direction and amount of rotation (Table 3). The difference along the flattest curvature was even less. Although the mean difference was small, the 95% LoA varied by 60.42 D (1.96 ! 0.216 D) on the steepest curvature and 60.32 D (1.96 ! 0.163 D) on the flattest curvature. It is surprising to 2592
Repeated-Measures ANOVA
ICC
Steepest (D)
F Z 23.74, P!.01
0.996
Flattest (D)
F Z 2.19, P!.05
0.997
Axis (degree)
F Z 46.56, P!.01
0.573
M (D)
F Z 24.30, P!.01
0.998
J0 (D)
F Z 15.11, P!.01
0.958
J45 (D)
F Z 44.19, P!.01
0.351
ANOVA Z analysis of variance; ICC Z intraclass correlation coefficient
Table 3.
The mean difference (95% LoA) for the steepest and flattest meridians, axis along the flattest meridian, and vector representation.
Measurement
Medmont Vs Nidek (No Rotation)
Steepest (D)
0.273 6 0.177 (ÿ0.074 to 0.620)
Flattest (D)
0.040 6 0.132 (ÿ0.219 to 0.299)
Axis (degree)
ÿ1.4 6 11.3 (ÿ23.5 to 20.7)
M (D)
0.156 6 0.103 (ÿ0.046 to 0.358)
J0 (D)
ÿ0.104 6 0.151 (ÿ0.400 to 0.192)
J45 (D)
0.075 6 0.252 (ÿ0.419 to 0.569)
AC Z anticlockwise rotation at 5 , 10 , and 15 ; C Z clockwise rotation at 5 , 10 , 15 ; LoA Z limits of agreement; M, J0, J45 Z vector representation
J CATARACT REFRACT SURG—VOL 30, DECEMBER 2004
TILTING EFFECT ON HANDHELD KERATOMETRY
instruments report the steepest and flattest curvatures, the curvatures could be at different orientations. This makes direct comparison difficult. The power vector method developed by Thibos and coauthors9 is a better approach7 and will be discussed. It is interesting that there was a decrease of about 5 degrees in the axis measurement when the handheld keratometer was rotated 5 degrees anticlockwise and an increase of about 5 degrees when it was rotated 5 degrees clockwise (Tables 1 and 3). The mean differences matched the direction and amount of rotation (Table 3). It seems that the handheld keratometer is incapable of compensating for the effect of rotation. Practitioners should manually correct the measured axis for the angle of rotation. The 95% LoA varied up to 622 degrees (1.96 ! 11.3 degrees), even with the keratometer in the upright position. The deviation reached 627 degrees (1.96 ! 14.0 degrees) with 15 degrees of clockwise rotation. This finding could be the limitation of analyzing the results by the steepest and flattest curvatures and the axis along the flattest curvature. Although consideration of steepest and flattest curvatures as well as axis is simple for clinicians, analysis of the axis required some adjustment, as described. A better approach to analyzing the corneal curvature is to convert the principal meridians into rectangular Fourier form.9 This accounts for the effect of different axes but might be difficult for clinicians to understand. In contrast to a greater difference along the
Table 3
steepest meridian (Table 3), the analysis using vector representation demonstrated a similar order of agreement, that is 0.15 D, with poor agreement on J0 and J45 when rotation increased. Both the mean difference and standard deviation of the difference in J0 and J45 increased with the amount of rotation. This could not be demonstrated with the steepest and flattest meridians. We further analyzed the subjects’ corneal astigmatism using the Medmont results as a reference. The mean corneal astigmatism was 1.61 6 0.88 D (range 0.20 to 3.93 D). Our results included subjects with spherical corneas and those with moderate corneal astigmatism. Practitioners are advised to use the handheld keratometer in the upright position to obtain accurate results. One reason to hold the keratometer with some rotation is to eliminate blocking by the subject’s nose. Another way to avoid this is to ask the subject to look temporally. However, the narrower vertical palpebral aperture at the temporal side may cover part of the corneal surface and obstruct the measurement. Handheld devices with a long focusing distance, eg, an autorefractor, may not have this instrument rotation problem. One limitation of the study is that the subjects were asked to maintain their head in an upright posture without the use of a chin rest. Since the chin rest and forehead rest may obstruct handheld keratometry at different rotational settings, the head position of our subjects was closely monitored by the third practitioner. This practitioner kept reminding the subjects not to tilt
(cont.)
Medmont Vs Nidek (AC5)
Medmont Vs Nidek (AC10)
Medmont Vs Nidek (AC15)
Medmont Vs Nidek (C5)
Medmont Vs Nidek (C10)
Medmont Vs Nidek (C15)
0.277 6 0.164 (ÿ0.044 to 0.598)
0.251 6 0.151 (ÿ0.045 to 0.547)
0.250 6 0.207 (ÿ0.157 to 0.656)
0.249 6 0.190 (ÿ0.123 to 0.621)
0.239 6 0.216 (ÿ0.184 to 0.662)
0.259 6 0.186 (ÿ0.105 to 0.624)
0.042 6 0.137 (ÿ0.227 to 0.311)
0.061 6 0.141 (ÿ0.215 to 0.337)
0.042 6 0.150 (ÿ0.252 to 0.336)
0.031 6 0.113 (ÿ0.190 to 0.252)
0.045 6 0.146 (ÿ0.241 to 0.331)
0.056 6 0.163 (ÿ0.263 to 0.375)
4.9 6 12.1 (ÿ18.8 to 28.6)
10.1 6 11.1 (ÿ35.4 to 55.6)
15.4 6 13.3 (ÿ10.7 to 41.5)
ÿ5.7 6 13.6 (ÿ32.4 to 21.0)
ÿ10.0 6 13.6 (ÿ36.7 to 16.7)
ÿ15.3 6 14.0 (ÿ42.7 to 12.1)
0.159 6 0.108 (ÿ0.053 to 0.371)
0.156 6 0.095 (ÿ0.030 to 0.342)
0.146 6 0.117 (ÿ0.083 to 0.375)
0.140 6 0.106 (ÿ0.068 to 0.348)
0.142 6 0.128 (ÿ0.109 to 0.393)
0.157 6 0.125 (ÿ0.088 to 0.402)
ÿ0.119 6 0.116 (ÿ0.346 to 0.108)
ÿ0.133 6 0.113 (ÿ0.354 to 0.088)
ÿ0.200 6 0.140 (ÿ0.474 to 0.074)
ÿ0.105 6 0.152 (ÿ0.403 to 0.193)
ÿ0.126 6 0.173 (ÿ0.465 to 0.213)
ÿ0.178 6 0.182 (ÿ0.179 to 0.179)
ÿ0.066 6 0.256 (ÿ0.568 to 0.436)
ÿ0.193 6 0.246 (ÿ0.767 to 0.289)
ÿ0.285 6 0.293 (ÿ0.859 to 0.289)
0.208 6 0.235 (ÿ0.253 to 0.669)
0.306 6 0.282 (ÿ0.247 to 0.859)
0.392 6 0.351 (ÿ0.296 to 1.080)
J CATARACT REFRACT SURG—VOL 30, DECEMBER 2004
2593
TILTING EFFECT ON HANDHELD KERATOMETRY
their head, to keep it straight against the chair. Previous studies show that the eyes are capable of compensating for a small amount of head tilt through incycloduction or excycloduction.13,14 We completed the measurements with handheld keratometers within 5 minutes in each subject. Therefore, errors from subjects’ head tilting should be very small. To conclude, we found that rotating the handheld keratometer affected the corneal measurements. The difference may not be great when one looks at the steepest and flattest curvatures and the axis along the flattest meridian. However, use of vector representation may show an increase in deviation by the amount of rotation. Practitioners are reminded to use the handheld keratometer in the upright position.
References 1. Seitz B, Langenbucher A. Intraocular lens power calculation in eyes after corneal refractive surgery. J Refract Surg 2000; 16:349–361 2. Feiz V, Mannis MJ, Garcia-Ferrer F, et al. Intraocular lens power calculation after laser in situ keratomileusis for myopia and hyperopia; a standardized approach. Cornea 2001; 20:792–797 3. Odenthal MTP, Eggink CA, Melles G, et al. Clinical and theoretical results of intraocular lens power calculation for cataract surgery after photorefractive keratectomy for myopia. Arch Ophthalmol 2002; 120:431– 438 4. Lam AKC. A hand-held keratometer. Ophthalmic Physiol Opt 1995; 15:227–230 5. Noonan CP, Rao GP, Kaye SB, et al. Validation of a handheld automated keratometer in adults. J Cataract Refract Surg 1998; 24:411–414
2594
6. Leyland M, Benjamin L. Clinical assessment of a handheld automated keratometer in cataract surgery. Eye 1997; 11:854–857 7. Lam AKC, Chan R, Chiu R. Effect of posture and artificial tears on corneal power measurements by handheld automated keratometer. J Cataract Refract Surg 2004; 30:645–652 8. Cho P, Lam AKC, Mountford J, Ng L. The performance of four different corneal topographers on normal human corneas and its impact on orthokeratology lens fitting. Optom Vis Sci 2002; 79:175–183; errata, 462 9. 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 10. Portney LG, Watkins MP. Foundations of Clinical Research. Applications to Practice. Norwalk, CT, Appleton & Lange, 1993; 505–528 11. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307–310 12. Edwards MH, Cho P. A new, hand-held keratometer: comparison of the Nidek KM-500 auto keratometer with the B&L keratometer and the Topcon RK-3000A keratometer. J Br Contact Lens Assoc 1996; 19:45–48 13. Linwong M, Herman SJ. Cycloduction of the eyes with head tilt. Arch Ophthalmol 1971; 85:570–573 14. Lam AKC, Chung E, Kho J, Wong S. Digital measurement of torsional eye movement due to postural change and its effect on reading performance. Curr Eye Res 2000; 21:763–766 From the Department of Optometry and Radiography, The Hong Kong Polytechnic University, Hong Kong SAR, China. None of the authors has a proprietary or financial interest in any product mentioned.
J CATARACT REFRACT SURG—VOL 30, DECEMBER 2004