ORIGINAL ARTICLE Year : 2000  Volume : 48  Issue : 4  Page : 27983 Comparison of optical and ultrasound pachometry S Korah, R Thomas, J Muliyil Schell Eye Hospital, Christian Medical College, Arni Road, Vellore632 001, India Correspondence Address:
Materials and Methods 1. Comparison of Optical versus Ultrasound Pachometry. At the initial examination, all study eyes were refracted to determine the best corrected visual acuity followed by a complete ocular examination. The ocular examination included slitlamp biomicroscopy, applanation tonometry, gonioscopy, indirect ophthalmoscopy and stereobiomicroscopic examination of the disc and nerve fibre layer using a + 60D or + 78D condensing lens. At the following visit, all normal subjects were randomized using the technique of permutated block randomization (computer generated blocks of four), to measurements in their right or left eye. The central corneal thickness was then measured with the HaagStreit slitlamp using the pachometer attachment no. 1 (Haag Streit AG, Koeniz, Switzerland). Five readings were taken in the described manner;[10] readjusting the scale to zero between readings. It has been suggested that the accuracy of optical pachometer readings using the HaagStreit attachment no. 1 can be increased by correcting for the corneal curvature.[10] Keratometric readings were obtained using the Bausch & Lomb keratometer (Rochester, NY, USA). These values were converted to the radius of corneal curvature and used to correct the optical pachometry measurements. The original measurements were compared to the corrected values. (In this article we have used the original term, pachometer, in favour of the later modification to pachymeter.[11]) The eye was anaesthetised with topical 4% lidocaine hydrochloride and the central corneal thickness was measured using the Tomey model AL 1000 pachometer (Tomey Corporation, Nagoya Aichi, Japan). Five consecutive readings were recorded to the nearest onethousandth of a millimeter and an average value obtained. In order to eliminate effects of diurnal variation on thickness, all measurements were taken between 2 PM and 5 PM. The statistical method used evaluates agreement using graphical techniques and simple calculations.[12] The mean difference (R1  R2) of the two values (R1, R2) taken with the two instruments (or repeated readings with the same instrument) was determined. The average of the two values (R1 + R2) / 2 was also determined. A scattergram was plotted with the mean difference on the "y" axis and the average values on the "x" axis. Agreement and type of error (systematic/random) was assessed; good agreement with a non systematic error would have readings that clustered randomly close to and on either side of the "zero" line as in [Figure:1]. A large deviation from the zero line would indicate a high disagreement in readings while clustering of points on one side of the zero line indicates the presence of a systematic error as in [Figure:2]. The standard error (SE) of the difference in means was determined and "mean difference ± 2 SE" calculated. If this range includes zero, we can conclude that systematic error is not involved in the two measurements. "Mean difference in values ± 1.96 x standard deviation (SD) of the difference" (simplified to "mean difference x 2 SD") provides the range of difference in readings that could occur 95% of the time. If this range (and resultant corrected IOP values[6]) are clinically acceptable, the variability is judged to be clinically not significant. 2. The interobserver and intraobserver variation in ultrasound and optical CCT measurement For logistical reasons, this was determined on another set of patients. Two different observers measured the CCT in 51 eyes of 26 patients using the HaagStreit attachment 1 in the manner described.[10] One observer then remeasured the CCT in all the 51 eyes in a random order. The observer was masked to the previously obtained recordings. The same method was used to study the inter and intraobserver variation for ultrasound pachometry. CCT was obtained in 34 eyes of 17 patients using the Tomey model AL 1000 ultrasound pachometer. The probe was centered over the undialated pupil. The statistical method used was the same as that described above. Results Part 1 a. Comparison of optical and ultrasound pachometers The difference (R1  R2) plotted against the mean (R1 + R2 /2) is shown in [Figure:1]. The points are randomly scattered on either side of the line representing zero. The mean difference between CCT using the two instruments was 0.001mm, with a standard deviation of 0.031mm. The standard error of the difference was 0.00439mm. The calculated range of mean difference ± 2SE is 0.00778 to +0.00878. This includes zero. The mean difference ± 2SD was0.062 to + 0.063. If used to "correct" IOP, this would represent a range of 4.4 mmHg to +4.5 mmHg. b. Optical pachometry values after correcting for radius of corneal curvature In this study the mean CCT was 0.538 ± 0.031mm. The value after correcting for the radius of curvature was practically the same (0.538 ± 0.034mm). The difference was not significant. Part 2 Interand intraobserver variability of the HaagStreit optical pachometer a. Interobserver variability of the HaagStreit pachometer. The mean interobserver difference was 0.019mm; standard deviation was 0.049mm. The standard error of the difference was 0.0069mm. The scatter plot depicting the difference between the two observers (R1 R2) against the average (R1 + R2 /2) is shown in [Figure:2]. There is clustering of points below the zero line indicating a systematic, nonrandom error in the interobserver variation of the optical pachometer. The calculated range of mean difference ± 2 SE is +0.0052mm to +0.0328mm; it does not include zero. The range, "mean difference ± 2 SD' is 0.079mm to +0.117mm representing a range of error in the "corrected" IOP of 5.6 mmHg to + 8.5 mmHg. b. Intraobserver variability of the HaagStreit optical pachometer. The mean intraobserver difference was 0.003mm; standard deviation was 0.017mm and the standard error of the difference was 0.0024 mm. The scatter plot depicting the difference between the two readings (R1 R2) against the average (R1 + R2 /2) is shown in [Figure:3]. There is no clustering of points on one side of the zero line indicating a nonsystematic, random error in the intraobserver variation of the optical pachometer. The calculated range of "mean difference ± 2SE" is 0.0018mm to +0.078 mm; it includes zero. The range, "mean difference ± 2 SD" is 0.031 mm to +0.037 mm. This would represent a range of error in the "corrected" IOP of 2.2 mmHg to +2.6 mmHg. Part 3 Inter and intraobserver variability of the Tomey AL 1000 Ultrasound Pachometer. a. Interobserver variability of the ultrasound pachometer. The mean interobserver difference was 0.001mm; standard deviation was 0.009mm and the standard error of the difference was 0.0015mm. The calculated range of "mean difference ± 2SE" is 0.002 to +0.004, which includes zero. The range, "mean difference ± 2SD "is 0.017mm to 0.019mm. This represents a range of error in the "corrected" IOP of 1.2 mmHg to +1.4 mmHg. b. Intraobserver variability of the ultrasound pachometer. The mean intraobserver difference was 0.002mm; standard deviation was 0.011mm and the standard error of the difference was 0.0019mm. The calculated range of "mean difference ± 2 SE" is 0.0038 to +0.0058 mm; it includes zero. The range, "mean difference ± 2 SD' is 0.02mm to 0.024 mm. This represents a range of error in the "corrected" IOP of 1.4 mmHg to +1.7 mmHg. Discussion Our results suggest a fairly good agreement between the optical and ultrasound pachometry. However, two standard deviations of the difference in CCT, translates to IOP difference of 4.4 mmHg to +4.5 mmHg, this is not clinically insignificant. Depending on the clinical situation, this range may or may not be acceptable. Further, as the error was random, it probably cannot be improved upon. In order to determine which of the two methods would be the more reproducible (reliable) for clinical use, we studied the intra and inter observer variability for both the instruments. In this study, both the instruments showed an equally low intraobserver variability [mean difference optical: 0.003mm(SD 0.017) vs ultrasound: 0.002mm (SD 0.011)]. This is not surprising since the same observer would be expected to use the same technique and measuring conditions for every patient. In measurement of the interobserver variability that of ultrasound pachometry was low (mean difference 0.001 ± 0.009mm). It was unacceptably high for the optical method (0.019 ± 0.049mm). There may be several explanations, and include: (a) There may have been a difference in the thickness of the optical section used by the two observers. The optics of the Pachometer 1 instrument are such that the readings of corneal thickness are most accurate when the optical section used for the measurement is infinitely thin.[13] It is best to use the thinnest, brightest image for clear visualization and measurement. Any variation in this thickness could lead to error. (b) Inclusion of the tear film in the measuring process is another potential source of error. To avoid this, the tear film was lightly stained using a fluorescein strip before taking the measurements. (c) Use of the "overlap"method by one observer and the "touch" method used by the second observer is yet another possibility; but both observers in this study used the touch method. Other studies too have found that the ultrasound pachometer seems to fare better than the optical pachometer; the optical pachometer has almost two to three times as much intersession variability as compared to the ultrasound method.[1416] In order to improve the accuracy of the HaagStreit attachment 1, it has been suggested that a correction for the radius of curvature of the cornea (HaagStreit Berne conversion scale supplied with the instrument) should be used. As can be seen from our results as well as the literature,[10] this is probably unnecessary for most clinical purposes but could be obligatory if the corneal thickness lies outside the limits of normal (0.48 mm  0.56 mm). Clinically, the addition of the correction factor may become important when "ocular hypertensives" with borderline thick corneas are encountered. In these cases, addition of the correction factor may clarify the situation. These considerations become irrelevant when ultrasound pachometry is performed. Ultrasound pachometry is not without disadvantages. Reproducible alignment at the same corneal location is essential for comparision of values. The major disadvantage of the ultrasound pachometer is that reproducible alignment is more difficult to achieve than for optical measurements.[17] Optical pachometry measures central corneal thickness by centering the slit beam over the non dilated pupil; mild eccentric measurements have minimal effect.[18] The Mishima Hedby modification of the optical pachymeter allows even more reliable centration but is not available for clinical use.[19] However, for purposes of "correction" of IOP, ultrasound pachometer measurements are obtained from 23 mm of the central cornea. The variation of corneal thickness in the central 23mm has been found to be minimal[20] and any effect on the correction factor should be insignificant. In conclusion, determination of CCT to correct applanation IOP is more reliable with the ultrasound pachometer than the HaagStreit attachment 1. Where more than one observer is involved in obtaining these measurements, the ultrasound should be the instrument of choice. However, where only one observer is expected to obtain readings, the optical method may be a reasonable substitute, provided the potential difference from ultrasound correction is acceptable. References


