- Email: [email protected]
Refractive outcome following radial keratotomy and combined radial and astigmatic keratotomy
Refractive outcome following radial keratotomy and combined radial and astigmatic keratotomy Gerald W Flanagan, OD, MPH, Perry S. Binder, MD ABSTRACT Purpose: To determine whether the visual and refractive outcomes of combined astigmatic and radial keratotomy (AK/RK) procedures was different from that following RK for the correction of naturally occurring compound myopic astigmatism and spherical myopia. Setting: Private professional practice, San Diego, California, USA. Methods: The computer database of all incisional procedures performed by one surgeon over 10 years was reviewed to compare the visual acuity outcome of AK/RK and RK procedures. Enhancement procedures were excluded. Only data from the last office visits were analyzed to establish the relationship between visual acuity and type of keratotomy procedure performed. A multiple regression model was constructed, which included covariates of age, postoperative keratometric cylinder, and postoperative refraction. Results: After controlling for covariates, the AK/RK population had significantly lower postoperative uncorrected visual acuity levels than the RK population (P < .03) after one operation (prior to enhancement surgery). Conclusions: Using the nomograms for myopia correction for unenhanced RK cases, combined AK and RK procedures appeared to reduce the expected visual results. Surgeons may consider modifying surgical nomograms to account for the expected spherical undercorrection that can occur when myopia and astigmatism are corrected simultaneously. J Cataract Refract Surg 1997; 23: 1057-1063
W
hen a refractive surgery candidate has more than 1.0 diopter (D) of astigmatism, a concurrent astigmatic procedure is considered to optimize the visual outcome.! Although incidence data are lacking, astigmatic keratotomy (AK) may be indicated in 30%
From the Ophthalmology Research Laboratory of the National Vision Research Institute and the Vision Surgery and Laser Center, San Diego, California, USA. Reprint requests to Perry S. Binder, MD, Suite 800, 8910 University Center Lane, San Diego, California 92122, USA.
or more of myopic eyes. 2 Despite numerous reports of the visual outcome after radial keratotomy (RK), little has been published comparing the refractive outcome of combined AKiRK with the RK outcome.! The ARC-T Study Group 3 recently studied the predictability of one nomogram for AK and found the predictive variables to be similar to those established for RK. Schneider and coauthors 4 report that 93% of eyes that had AK using transverse incisions achieved 20/40 or better postoperative uncorrected visual acuity (UCVA). Preoperative refractive status and patient selection criteria were not clearly described. In a se-
J CATARACT REFRACT SURG-VOL 23. SEPTEMBER 1997
1057
REFRACTIVE OUTCOME FOLLOWING RK
lected sample of 60 eyes, Thornton and Sanders 5 found that astigmatism was reduced to 0.4 D from a preoperative mean of 1.5 Dj however, the standard astigmatism deviation increased 50% (before = 0.43, after = 0.61). Visual acuity results were not reported. Using vector analysis to evaluate the efficacy of AK, Agapitos and coauthors6 conclude that the accuracy of astigmatic procedures was highly variable. Recent reports of simultaneous myopic and astigmatic correction with excimer laser photoablation (photo astigmatic refractive keratectomy [PARK]) indicate some degree of effectiveness in reducing astigmatism when using visual acuity as the outcome variable. In a study by Kim and coauthors/ fewer PARK patients had a UCVA of 20/25 or better than patients who had had PRK for myopia only. The efficacy of combined AKiRK using visual acuity as the main outcome measure has not been established. In many refractive surgery outcome studies, visual acuity is divided into "20/40 or better" and "worse than 20140." Boutque and coauthorsB suggest 20/20 acuity is a more sensitive criterion because it affects patients' lens wearing habits and determines satisfaction with surgery. Refraction was a less sensitive endpoint and failed to indicate differences between groups. Authorities agree that visual outcomes after keratotomy procedures are variable and complications, al-
though uncommon, are possible. 9,lo Most studies demonstrate that AK procedures reduce astigmatism as measured by keratometry or topography, but visual acuity data are sparse. In fact, the American Academy of Ophthalmology l6 did not comment on astigmatic procedures in their Radial Keratotomy Ophthalmic Procedures Assessment because of poor standardization and the inadequacy of evaluative literature. II In this study, we attempted to determine whether AK procedures incorporating transverse or arc-shaped incisions combined with RK produce similar visual acuity outcomes as RK procedures alone for the correction of naturally occurring myopia.
Patients and Methods This study evaluated a refractive keratotomy database that includes demographic, procedural, and follow-up data on cases performed between 1983 and 1994. Procedures were performed in a hospital ambulatory care center (0.5%) or a private professional office (99.5%). All follow-up visits were at the private office. Patients were self-referred, professionally referred, or private patients of the surgeon (P.S.B.). Table 1 describes the population. Patients were eligible for the study if they were between 21 and 60 years of age and had a preoperative refraction between -1.00 and -6.50 D and if the
Table 1. Demographic and surgical statistics for keratotomy population, 1983 to 1994. Range Variable Age (years)
Number
Missing
Mean:<:: SO
749
12
37.4 :<:: 9.4
Median 37
Minimum 20
Maximum 76
Optical zone diameter (mm)
746
15
Incision number
745
16
5.86:<:: 2.63
4
Follow-up (months)
666
95
14.37 :<:: 21.5
5
0
Incision depth (mm)
706
55
0.56:<:: 0.06
0.55
0.46
0.74
Corneal curvature (D)
646
115
43.72:<:: 1.91
43.81
39.37
47.87
Baseline refraction (spherical equivalent [0])
533
115
-3.82:<:: 2.31
-3.5
-18.87
8.75
Postoperative refraction (spherical equivalent [0])
609
152
-1.02:<:: 1.59
-0.75
-14
2.67
Visual acuity
609
152
0.55:<:: 0.33 (20/36)
0.66 (20/30)
1058
3.61
J CATARACT REFRACT SURG-VOL 23, SEPTEMBER 1997
3.25
3
8 16
0.01
118
1.33
REFRACTIVE OUTCOME FOLLOWING RK
optical clear zone of the keratotomy procedure was between 3.0 and 5.0 mm. All patients had a preoperative best corrected Snellen acuity of at least 20/20 with no ocular pathology. Patients were excluded if the refractive procedure used fewer than four or more than eight radial incisions or if only astigmatic incisions were used. Only primary cases were used in the analysis; reoperations or enhancements to improve refractive results were excluded. The response to surgery by an individual eye was assumed to be independent of the other eye's response. Therefore, second eyes were included if they met the eligibility criteria. Technicians and assistants collected the data, which were then compiled by the surgeon in a spreadsheet software program (Microsoft Excel, version 5.0). The database included primary cases as well as secondary and tertiary cases (enhancements or reoperations). Data used in this analysis were recorded at the last follow-up visit of all patients. Refractions were recorded as the spherical equivalent of the noncycloplegic subjective refraction. Corneal curvature was measured with a standard Bausch & Lomb keratometer. Visual acuity was recorded using the decimal form of the Snellen fraction as obtained in a condensed 10 foot examination lane. In the surgical procedures, three anatomically similar front- and back-cutting diamond knives (Mastel, Inc.) were used to create centripetal incisions. Blade extension for centripetal incisions was set at 100% of the thinnest of four paracentral ultrasonic readings (Corneas can II, Storz Instruments) obtained at the edge of the optical clear zone. Blade extension was confirmed using a coin gauge observed at maximum magnification with monocular viewing under a Weck operating microscope equipped with a patient fixation light, the micrometer handpiece, andlor the Mastel microscope. A back-cutting, oblique diamond knife (K.O.I.) was used to create centrifugal incisions with the blade extension set at 115 to 120% of the thinnest paracentral reading. Centripetal incisions were created using a 45 degree front-cutting blade (Mastel Diamond, Inc.) with a blade extension set at 100% of the thinnest paracentral corneal pachymetry. Transverse or arc-shaped incisions were made between radial incisions in the corneal periphery at an optic zone between 6.0 and 8.0 mm in diameter to
reduce the astigmatic error; most were made at 7.0 mm. Incision depth for astigmatic procedures was set at 90% of the ultrasonic pachymetry reading at the optical zone diameter used for the astigmatism surgery. Jump radial or jump T-incisions were not performed. Surgery was performed by one surgeon (P.S.B). Patients received a topical antibiotic/steroid combination drop for 1 week after surgery. Occlusion pressure patching was discontinued in 1988. Once topical diclofenac was available, it was used every 4 to 6 hours after surgery for 1 day only. Biomicroscopy, refraction, and keratotomy data were collected at the 24 hour visit. Follow-up visits were at 1 week, 2 months, and 1 year. Reoperations to improve the refractive outcome were performed as early as 3 to 4 weeks and as long as 3 months after the initial operation. Visual acuity in this population was tested with the more common projection acuity system that uses the older-style Snellen lines (20/20, 20125, 20/30, etc.). Visual acuity was recorded as the decimal fraction (e.g., 20/20 = 1.0; 20/40 = 0.5) in condensed 10 foot lanes. Although the Snellen chart includes 20/15 and 20/10 lines, examiners failed to elicit these responses. The best achievable acuity as recorded in this database was therefore 20/20. The technician and surgeon used a standardized method of recording visual acuity. No partial credit was given when less than a full line was read by the patient. If half or more letters of a line were read properly (e.g., three of six on the 20/20 line or three of five on the 20/30 line), the visual acuity was recorded as the whole line. Ifless than half of a line was read properly, the next larger line was recorded (e.g., two of five on the 20/30 line recorded as 20/40). Although this approach involves some subjectivity, the surgeon believes the method was applied consistently to all patients. A similar method was used in the PERK studies. 12 The database recorded only keratometry readings and the preoperative and postoperative refractive manifest spherical equivalent refractions. Refractive astigmatism was performed on each operated eye but was not analyzed in the database because the quantitative amount of refractive astigmatism will vary with the amount of the spherical error; large myopic spherical errors will have more manifest atigmatism than small myopic errors with the same amount of keratometry-measured astigmatism. Reporting the results of manifest astigma-
J CATARACT REFRACT SURG-VOL 23, SEPTEMBER 1997
1059
REFRACTIVE OUTCOME FOLLOWING RK
tism is therefore considered inaccurate, whereas keratometry-measured astigmatism is the same regardless of the spherical error. 13 The null hypothesis of no difference in visual acuity outcome between the RK and RKIAK populations was tested using multiple linear regression. Confounding was minimized by applying the eligibility criteria and including postoperative corneal cylinder, patient age, and postoperative spherical equivalent refraction to the model. Age is known to affect refractive outcome in RK (approximately 0.75 greater effect per decade)14 and is also expected to reduce best corrected acuity because of cataracts. 15 ,16 Snellen decimal acuity, as the reciprocal transform of minimum angle of resolution, is the dependent variable in a multiple regression model used for hypothesis testing. Data analysis was performed using Systat 5.0 and BMDP New System 1.1 on a personal computer.
Results
1.00 to 2.00 2.12 to 3.00 3.12 to 4.00 4.12 to 5.00 5.12 to 6.50
There were no between-group differences in age, optical clear zone diameter, incision number, follow-up time, preoperative or postoperative spherical equivalent refraction (Table 2). Eligible cases were grouped by the preoperative spherical equivalent refraction to determine the resultant acuity for procedures with and without simultaneous astigmatic incisions. Figure 1 shows the median acuity in the two incision groups categorized by preop-
Preoperative equiv. sph.
Figure 1. (Flanagan) Median postoperative UCVA in AKIRK and RK eyes related to preoperative spherical equivalent myopia (1983- 1994) (0 = AK/RK; • = RK).
erative refraction. In only one preoperative group (-4.12 to - 5.00 D) was UCVA equivalent in RK and AKlRK cases. Uncorrected acuity after AKIRK failed to reach the same level as after RK in all other categories.
Table 2. Mean demographic and surgical statistics for AK/RK and RK groups.
Age (years)
RK Group
AKIRK Group (n = 120)
95%CI
(n = 356)
36.21
95%CI
36.12
34 .85 , 37.38
Optical clear zone (mm)
3.54
3.44,3.55
3.48
3.42,3.54
Number of incisions
5.57
5.21,5.92
5.97
5.76,6.18
Follow-up (months)
11 .72
8.39, 15.05
13.9
35.32, 37.10
11.45, 16.36
Preoperative corneal cylinder (0)
1.61
1.46, 1.75
0.75
0.69,0.81
Postoperative corneal cylinder (0)
1.05
0.88, 1.22
0.78
0.69,0.86
Preoperative refraction (spherical equivalent [OJ)
-3.71
-3.99, -3.42
-3.76
-3.91, -3.61
Postoperative refraction (spherical equivalent [OJ)
-0.77
-0.98, -0.55
-0.77
-0.90, -0.64
Visual acuity (mean Snellen decimal)
0.55 (20/36)
0 .50, 0.60 (20/40, 20/33)
0.63 (20/32)
0.60,0.67 (20/33, 20/30)
Visual acuity (median Snellen decimal)
1060
0.50 (20/40)
J CATARACT REFRACT SURG-VOL 23, SEPTEMBER 1997
0.66 (20/30)
REFRACTIVE OUTCOME FOLLOWING RK
c:::
o
:e &. e D.
20140 or worse
20/25 -20/30
20/20 or better
Visual acuity Figure 2. (Flanagan) Postoperative UCVA in RK and AK/RK eyes (1983--1994) (B = RK; D = AK/RK).
24 23 22 21 20 19 18 17 16 15 14 13 12 11 10
Simultaneous AK was performed in 31.0% of cases. Enhancements (reoperations) accounted for 14.6% of procedures; 2.6% were tertiary procedures to improve the initial enhancement surgery. No more than three surgeries were performed on a single eye. However, no enhancment data were included in this analysis. Figure 2 shows the percentages of AKiRK and RK eyes that achieved a UCVA of 20120 or better, 20/25 to 20/30, and 20/40 or worse postoperatively. Only 7% of the AK eyes achieved 20/20 or better UCVA after one procedure (prior to enhancements), whereas 23.2% of the RK eyes reached that level (P < .03). The characteristics of the population distribution were ascertained to determine the best method for testing the null hypothesis of no difference in acuity between the two groups. The Snellen decimal visual acuity in the keratotomy population is shown in Figure 3. The histogram appears skewed to the right because patient responses better than 20/20 Snellen decimal fraction (= 1.0) were not elicited. A truer Gaussian distribution would have been followed had smaller Snellen letters been presented to the patients. In the multiple linear regression model, dummy variables created for type of procedure were 0 for presence of astigmatic incisions with radial incisions and 1 when only radial incisions were used. The coefficient estimate was 0.07. The P-value for the coefficient was 0_03. The model explained 32% of the
-
Figure 3. (Flanagan) Postoperative UCVA (Snellen decimal form) in all keratotomy eyes (1983-1994). Midpoints of the Snellen decimal range are indicated by hash marks on the x-axis (width of each bar 0.05).
987-
~I
1-'" 0.0250
0.1750
0.3250
0.4750
0.6250
0.7750
0_9250
Snellen decimal groups, UCVA, postoperative
J CATARACT REFRACT SURG-VOL 23, SEPTEMBER 1997
1061
REFRACTIVE OUTCOME FOLLOWING RK
variability in visual acuity. Therefore, after adjusting for the covariates of age, postoperative spherical equivalent refraction, and postoperative corneal cylinder, the alternative hypothesis of different acuity levels in the keratotomy groups was accepted.
Discussion The analysis indicates that UCVA outcomes in the two groups differed most in the 20/20 or better category. After adjusting for age, postoperative spherical equivalent refraction, and postoperative corneal cylinder in a multiple regression model, the alternative hypothesis of different acuity levels in the two keratotomy groups was accepted at P < .03. This agrees with the conclusions of Kim and coauthors l6 in their review of myopic excimer laser patients receiving simultaneous astigmatic corrections. They found that when astigmatic procedures were used, fewer patients achieved a UCVA of 20/25 or better than patients who had PRK for myopia only. In this cohort, 67% of AKiRK eyes achieved 20/40 or better UCVA. Schneider and coauthors 4 report that 93% of eyes with preoperative astigmatism between 0.50 and 6.00 D treated with transverse incisions achieved a UCVA of 20/40 or better and 43% achieved 20/20 or better. Many surgeons consider a 0.50 D astigmatic error too insignificant to warrant correction, which may explain why such a large number of patients in their study achieved 20/20 or better. In our study, 207 of 284 (72.9%) RK procedures and 65 of 97 (67.0%) AKiRK procedures were corrected to the 20/40 level prior to any enhancement procedure. Schneider and coauthors 4 included enhancement procedures in 11 % of their cases. Our cohort included primary cases; subsequently, 14.6% of all eyes (RK and AKlRK) had enhancement procedures. The postoperative astigmatism results indicate a reduction in mean corneal cylinder (keratometry measured) in AK eyes from 1.60 to 1.06 D. The amount of reduction is less than others report using refractive astigmatism as a measure. In their sample of 60 eyes, Thornton and Sanders 5 report that a preoperative mean of 1.5 D of refractive astigmatism was reduced to 0.4 D. These reports use refractive astigmatism, which is known to differ from the keratometric cylinder measured for our cohort and used as a surrogate for it. 1062
Refractive astigmatism values are more valuable in evaluating efficacy of astigmatic incisions because refractive astigmatism dictates whether transverse incisions are to be used. However, refractive astigmatism does not always reflect the same amount of keratometric astigmatism, which depends on the spherical refractive error. 13 In a separate database of lamellar refractive procedures, we found the ratio of keratometric astigmatism to refractive astigmatism to range between 1.2 and 1.5 (unpublished observations, September 1996). Assuming these ratios are true for the RK and AKiRK cases we report, our results would agree with previous literature reports. In many refractive surgery outcome studies, visual acuity is divided into "20/40 or better" and "worse than 20/40." In our analysis, the 20/40 cutoff does not adequately assess a between-group difference. Bourque and coauthors 8 suggest that 20/20 acuity is a more sensitive criterion because it drives patients' lens-wearing habits and determines satisfaction with surgery. In our population, the 20/20 criterion was more sensitive in demonstrating differences between populations than the customary 20/40 cutoff. Refraction data as a surrogate endpoint failed to show a distinction between the groups. Our study, therefore, provides additional evidence illustrating that visual acuity may be a more important determinant in assessing refractive surgery outcomes. Why did the AKiRK eyes have poorer UCVA than the RK eyes? The major reason was cylindrical undercorrection. Although we found no significant difference in the mean postoperative spherical equivalent refractive errors, most of the enhancement procedures were performed to correct residual myopia. This phenomenon has occurred when astigmatism and myopia are simultaneously corrected with excimer laser photoablation (PARK).? It is not clear whether the RK procedure reduces the effect of the AK procedure or the reverse. We did record the "coupling ratio" in this study, and we plan to analyze the ratios to determine whether one procedure has a negative impact on the other. We agree with Agapitos and coauthors,6 who used vector analysis and found that the accuracy of astigmatic procedures was highly variable. If so, visual acuity levels would be expected to worsen in RK procedures in which astigmatic incisions were judged to be needed.
J CATARACT REFRACT SURG-VOL 23. SEPTEMBER 1997
REFRACTIVE OUTCOME FOLLOWING RK
Certain limitations of this study should be identified. A potential confounding variable that can diminish the strength of the relationship between UCVA and type of keratotomy procedure is postoperative refractive error for patients 40 years and older. Should one type of keratotomy produce more eyes with postoperative refractions greater than + 1.00 D, this group would have poorer overall acuity because of the presbyope's inability to accommodate for vision improvement. To determine whether there was a difference between the number of presbyopic patients with refractions greater than or equal to + 1.00 0 in the two surgery groups for patients 40 years and over, another contingency table analysis was performed. No difference in frequency of postoperative hyperopic refraction was detected between the two keratotomy groups. The two-by-two frequency analysis chi square has a P-value of 0.40. The data did not allow an estimation of the effect on refractive astigmatism. Many investigations studying the effects of astigmatic procedures accept postoperative spherical equivalent refraction and ignore refractive astigmatism. We find this acceptable since our goal was not to determine whether there was a difference in best corrected vision but only whether a postoperative acuity difference existed that was not explained by a difference in preoperative spherical equivalent refraction. The mean postoperative spherical refraction was equal in both groups. A more meaningful comparison might be between best corrected preoperative visual acuity and best corrected postoperative acuity data in the AKIRK and RK groups. This would call for spherocylindrical refraction data. This analysis would reveal whether AKIRK patients lost best corrected vision when compared to RK patients. Many refractive surgery studies use loss and gain of best corrected acuity to assess safety and effectiveness of refractive surgery. The data in our study were not suited to this type of review.
References 1. Binder PS, Waring GO III. Keratotomy for astigmatism. In: Waring GO III, ed, Refractive Keratotomy. St Louis, MO, CV Mosby, 1992; 1085-1198
2. Lindstrom R. The surgical correction of astigmatism: a clinician's perspective. Refract Corneal Surg 1990; 6:441454 3. Price FW, Grene RB, Marks RB, et al. Astigmatism Reduction Clinical Trial: a multicenter prospective evaluation of the predictability of arcuate keratotomy; evaluation of nomogram predictability. Arch Ophthalmol, 1995; 113:277-282; correction 577 4. Schneider 0, Draghic T, Murthy R. Combined myopia and astigmatism surgery. Review of 350 cases. J Cataract Refract Surg 1992; 18:370-374 5. Thornton SP, Sanders DR. Graded nonintersecting transverse incisions for correction of idiopathic astigmatism. J Cataract Refract Surg 1987; 13:27-31 6. Agapitos pJ, Lindstrom RL, Williams PA, et al. Analysis of astigmatic keratotomy. J Cataract Refract Surg 1989; 15:13-18 7. Kim YJ, Sohn J, Tchah H, Lee CO. Photoastigmatic refractive keratectomy in 168 eyes: six-month results. J Cataract Refract Surg 1994; 20:387-391 8. Bourque LB, Lynn MJ, Waring GO III, Cartwright C. Spectacle and contact lens-wearing six years after radial keratotomy in the Prospective Evaluation of Radial Keratotomy Study. Ophthalmology 1994; 101 :421431 9. Binder PS. Optical problems following refractive surgery. Ophthalmology 1986; 93:739-745 10. Rashid ER, Waring GO III. Complications of radial and transverse keratotomy. Surv Ophthalmol 1989; 34:73106 11. American Academy of Ophthalmology. Ophthalmic procedures assessment: radial keratotomy for myopia. Ophthalmology 1993; 100:1103-1115 12. Santos, YR, Waring GO III, Lynn MJ, et al. Relationship between refractive error and visual acuity in the Prospective Evaluation of Radial Keratotomy (PERK) study. Arch Ophthalmol 1987; 105:86-92 13. Sampson WG. Applied optical principles: keratometry. Ophthalmology 1979; 86:347-351 14. Lynn MJ, Waring GO III, Sperduto RD, PERK Study Group. Factors affecting outcome and predictability of radial keratotomy in the PERK study. Arch Ophthalmol 1987; 105:42-51 15. Whitaker, 0, Elliott DB, MacVeigh D. Variations in hyperacuity performance with age. Ophthalmic Physiol Opt 1992; 12:29-32 16. Kim JH, Hahn Tw, Lee yc, Sah WI. Clinical experience of two-step photo refractive keratectomy in 19 eyes with high myopia. J Refract Corneal Surg 1993; 9(suppl):S44-S47
J CATARACT REFRACT SURG-VOL 23, SEPTEMBER 1997
1063
W
hen a refractive surgery candidate has more than 1.0 diopter (D) of astigmatism, a concurrent astigmatic procedure is considered to optimize the visual outcome.! Although incidence data are lacking, astigmatic keratotomy (AK) may be indicated in 30%
From the Ophthalmology Research Laboratory of the National Vision Research Institute and the Vision Surgery and Laser Center, San Diego, California, USA. Reprint requests to Perry S. Binder, MD, Suite 800, 8910 University Center Lane, San Diego, California 92122, USA.
or more of myopic eyes. 2 Despite numerous reports of the visual outcome after radial keratotomy (RK), little has been published comparing the refractive outcome of combined AKiRK with the RK outcome.! The ARC-T Study Group 3 recently studied the predictability of one nomogram for AK and found the predictive variables to be similar to those established for RK. Schneider and coauthors 4 report that 93% of eyes that had AK using transverse incisions achieved 20/40 or better postoperative uncorrected visual acuity (UCVA). Preoperative refractive status and patient selection criteria were not clearly described. In a se-
J CATARACT REFRACT SURG-VOL 23. SEPTEMBER 1997
1057
REFRACTIVE OUTCOME FOLLOWING RK
lected sample of 60 eyes, Thornton and Sanders 5 found that astigmatism was reduced to 0.4 D from a preoperative mean of 1.5 Dj however, the standard astigmatism deviation increased 50% (before = 0.43, after = 0.61). Visual acuity results were not reported. Using vector analysis to evaluate the efficacy of AK, Agapitos and coauthors6 conclude that the accuracy of astigmatic procedures was highly variable. Recent reports of simultaneous myopic and astigmatic correction with excimer laser photoablation (photo astigmatic refractive keratectomy [PARK]) indicate some degree of effectiveness in reducing astigmatism when using visual acuity as the outcome variable. In a study by Kim and coauthors/ fewer PARK patients had a UCVA of 20/25 or better than patients who had had PRK for myopia only. The efficacy of combined AKiRK using visual acuity as the main outcome measure has not been established. In many refractive surgery outcome studies, visual acuity is divided into "20/40 or better" and "worse than 20140." Boutque and coauthorsB suggest 20/20 acuity is a more sensitive criterion because it affects patients' lens wearing habits and determines satisfaction with surgery. Refraction was a less sensitive endpoint and failed to indicate differences between groups. Authorities agree that visual outcomes after keratotomy procedures are variable and complications, al-
though uncommon, are possible. 9,lo Most studies demonstrate that AK procedures reduce astigmatism as measured by keratometry or topography, but visual acuity data are sparse. In fact, the American Academy of Ophthalmology l6 did not comment on astigmatic procedures in their Radial Keratotomy Ophthalmic Procedures Assessment because of poor standardization and the inadequacy of evaluative literature. II In this study, we attempted to determine whether AK procedures incorporating transverse or arc-shaped incisions combined with RK produce similar visual acuity outcomes as RK procedures alone for the correction of naturally occurring myopia.
Patients and Methods This study evaluated a refractive keratotomy database that includes demographic, procedural, and follow-up data on cases performed between 1983 and 1994. Procedures were performed in a hospital ambulatory care center (0.5%) or a private professional office (99.5%). All follow-up visits were at the private office. Patients were self-referred, professionally referred, or private patients of the surgeon (P.S.B.). Table 1 describes the population. Patients were eligible for the study if they were between 21 and 60 years of age and had a preoperative refraction between -1.00 and -6.50 D and if the
Table 1. Demographic and surgical statistics for keratotomy population, 1983 to 1994. Range Variable Age (years)
Number
Missing
Mean:<:: SO
749
12
37.4 :<:: 9.4
Median 37
Minimum 20
Maximum 76
Optical zone diameter (mm)
746
15
Incision number
745
16
5.86:<:: 2.63
4
Follow-up (months)
666
95
14.37 :<:: 21.5
5
0
Incision depth (mm)
706
55
0.56:<:: 0.06
0.55
0.46
0.74
Corneal curvature (D)
646
115
43.72:<:: 1.91
43.81
39.37
47.87
Baseline refraction (spherical equivalent [0])
533
115
-3.82:<:: 2.31
-3.5
-18.87
8.75
Postoperative refraction (spherical equivalent [0])
609
152
-1.02:<:: 1.59
-0.75
-14
2.67
Visual acuity
609
152
0.55:<:: 0.33 (20/36)
0.66 (20/30)
1058
3.61
J CATARACT REFRACT SURG-VOL 23, SEPTEMBER 1997
3.25
3
8 16
0.01
118
1.33
REFRACTIVE OUTCOME FOLLOWING RK
optical clear zone of the keratotomy procedure was between 3.0 and 5.0 mm. All patients had a preoperative best corrected Snellen acuity of at least 20/20 with no ocular pathology. Patients were excluded if the refractive procedure used fewer than four or more than eight radial incisions or if only astigmatic incisions were used. Only primary cases were used in the analysis; reoperations or enhancements to improve refractive results were excluded. The response to surgery by an individual eye was assumed to be independent of the other eye's response. Therefore, second eyes were included if they met the eligibility criteria. Technicians and assistants collected the data, which were then compiled by the surgeon in a spreadsheet software program (Microsoft Excel, version 5.0). The database included primary cases as well as secondary and tertiary cases (enhancements or reoperations). Data used in this analysis were recorded at the last follow-up visit of all patients. Refractions were recorded as the spherical equivalent of the noncycloplegic subjective refraction. Corneal curvature was measured with a standard Bausch & Lomb keratometer. Visual acuity was recorded using the decimal form of the Snellen fraction as obtained in a condensed 10 foot examination lane. In the surgical procedures, three anatomically similar front- and back-cutting diamond knives (Mastel, Inc.) were used to create centripetal incisions. Blade extension for centripetal incisions was set at 100% of the thinnest of four paracentral ultrasonic readings (Corneas can II, Storz Instruments) obtained at the edge of the optical clear zone. Blade extension was confirmed using a coin gauge observed at maximum magnification with monocular viewing under a Weck operating microscope equipped with a patient fixation light, the micrometer handpiece, andlor the Mastel microscope. A back-cutting, oblique diamond knife (K.O.I.) was used to create centrifugal incisions with the blade extension set at 115 to 120% of the thinnest paracentral reading. Centripetal incisions were created using a 45 degree front-cutting blade (Mastel Diamond, Inc.) with a blade extension set at 100% of the thinnest paracentral corneal pachymetry. Transverse or arc-shaped incisions were made between radial incisions in the corneal periphery at an optic zone between 6.0 and 8.0 mm in diameter to
reduce the astigmatic error; most were made at 7.0 mm. Incision depth for astigmatic procedures was set at 90% of the ultrasonic pachymetry reading at the optical zone diameter used for the astigmatism surgery. Jump radial or jump T-incisions were not performed. Surgery was performed by one surgeon (P.S.B). Patients received a topical antibiotic/steroid combination drop for 1 week after surgery. Occlusion pressure patching was discontinued in 1988. Once topical diclofenac was available, it was used every 4 to 6 hours after surgery for 1 day only. Biomicroscopy, refraction, and keratotomy data were collected at the 24 hour visit. Follow-up visits were at 1 week, 2 months, and 1 year. Reoperations to improve the refractive outcome were performed as early as 3 to 4 weeks and as long as 3 months after the initial operation. Visual acuity in this population was tested with the more common projection acuity system that uses the older-style Snellen lines (20/20, 20125, 20/30, etc.). Visual acuity was recorded as the decimal fraction (e.g., 20/20 = 1.0; 20/40 = 0.5) in condensed 10 foot lanes. Although the Snellen chart includes 20/15 and 20/10 lines, examiners failed to elicit these responses. The best achievable acuity as recorded in this database was therefore 20/20. The technician and surgeon used a standardized method of recording visual acuity. No partial credit was given when less than a full line was read by the patient. If half or more letters of a line were read properly (e.g., three of six on the 20/20 line or three of five on the 20/30 line), the visual acuity was recorded as the whole line. Ifless than half of a line was read properly, the next larger line was recorded (e.g., two of five on the 20/30 line recorded as 20/40). Although this approach involves some subjectivity, the surgeon believes the method was applied consistently to all patients. A similar method was used in the PERK studies. 12 The database recorded only keratometry readings and the preoperative and postoperative refractive manifest spherical equivalent refractions. Refractive astigmatism was performed on each operated eye but was not analyzed in the database because the quantitative amount of refractive astigmatism will vary with the amount of the spherical error; large myopic spherical errors will have more manifest atigmatism than small myopic errors with the same amount of keratometry-measured astigmatism. Reporting the results of manifest astigma-
J CATARACT REFRACT SURG-VOL 23, SEPTEMBER 1997
1059
REFRACTIVE OUTCOME FOLLOWING RK
tism is therefore considered inaccurate, whereas keratometry-measured astigmatism is the same regardless of the spherical error. 13 The null hypothesis of no difference in visual acuity outcome between the RK and RKIAK populations was tested using multiple linear regression. Confounding was minimized by applying the eligibility criteria and including postoperative corneal cylinder, patient age, and postoperative spherical equivalent refraction to the model. Age is known to affect refractive outcome in RK (approximately 0.75 greater effect per decade)14 and is also expected to reduce best corrected acuity because of cataracts. 15 ,16 Snellen decimal acuity, as the reciprocal transform of minimum angle of resolution, is the dependent variable in a multiple regression model used for hypothesis testing. Data analysis was performed using Systat 5.0 and BMDP New System 1.1 on a personal computer.
Results
1.00 to 2.00 2.12 to 3.00 3.12 to 4.00 4.12 to 5.00 5.12 to 6.50
There were no between-group differences in age, optical clear zone diameter, incision number, follow-up time, preoperative or postoperative spherical equivalent refraction (Table 2). Eligible cases were grouped by the preoperative spherical equivalent refraction to determine the resultant acuity for procedures with and without simultaneous astigmatic incisions. Figure 1 shows the median acuity in the two incision groups categorized by preop-
Preoperative equiv. sph.
Figure 1. (Flanagan) Median postoperative UCVA in AKIRK and RK eyes related to preoperative spherical equivalent myopia (1983- 1994) (0 = AK/RK; • = RK).
erative refraction. In only one preoperative group (-4.12 to - 5.00 D) was UCVA equivalent in RK and AKlRK cases. Uncorrected acuity after AKIRK failed to reach the same level as after RK in all other categories.
Table 2. Mean demographic and surgical statistics for AK/RK and RK groups.
Age (years)
RK Group
AKIRK Group (n = 120)
95%CI
(n = 356)
36.21
95%CI
36.12
34 .85 , 37.38
Optical clear zone (mm)
3.54
3.44,3.55
3.48
3.42,3.54
Number of incisions
5.57
5.21,5.92
5.97
5.76,6.18
Follow-up (months)
11 .72
8.39, 15.05
13.9
35.32, 37.10
11.45, 16.36
Preoperative corneal cylinder (0)
1.61
1.46, 1.75
0.75
0.69,0.81
Postoperative corneal cylinder (0)
1.05
0.88, 1.22
0.78
0.69,0.86
Preoperative refraction (spherical equivalent [OJ)
-3.71
-3.99, -3.42
-3.76
-3.91, -3.61
Postoperative refraction (spherical equivalent [OJ)
-0.77
-0.98, -0.55
-0.77
-0.90, -0.64
Visual acuity (mean Snellen decimal)
0.55 (20/36)
0 .50, 0.60 (20/40, 20/33)
0.63 (20/32)
0.60,0.67 (20/33, 20/30)
Visual acuity (median Snellen decimal)
1060
0.50 (20/40)
J CATARACT REFRACT SURG-VOL 23, SEPTEMBER 1997
0.66 (20/30)
REFRACTIVE OUTCOME FOLLOWING RK
c:::
o
:e &. e D.
20140 or worse
20/25 -20/30
20/20 or better
Visual acuity Figure 2. (Flanagan) Postoperative UCVA in RK and AK/RK eyes (1983--1994) (B = RK; D = AK/RK).
24 23 22 21 20 19 18 17 16 15 14 13 12 11 10
Simultaneous AK was performed in 31.0% of cases. Enhancements (reoperations) accounted for 14.6% of procedures; 2.6% were tertiary procedures to improve the initial enhancement surgery. No more than three surgeries were performed on a single eye. However, no enhancment data were included in this analysis. Figure 2 shows the percentages of AKiRK and RK eyes that achieved a UCVA of 20120 or better, 20/25 to 20/30, and 20/40 or worse postoperatively. Only 7% of the AK eyes achieved 20/20 or better UCVA after one procedure (prior to enhancements), whereas 23.2% of the RK eyes reached that level (P < .03). The characteristics of the population distribution were ascertained to determine the best method for testing the null hypothesis of no difference in acuity between the two groups. The Snellen decimal visual acuity in the keratotomy population is shown in Figure 3. The histogram appears skewed to the right because patient responses better than 20/20 Snellen decimal fraction (= 1.0) were not elicited. A truer Gaussian distribution would have been followed had smaller Snellen letters been presented to the patients. In the multiple linear regression model, dummy variables created for type of procedure were 0 for presence of astigmatic incisions with radial incisions and 1 when only radial incisions were used. The coefficient estimate was 0.07. The P-value for the coefficient was 0_03. The model explained 32% of the
-
Figure 3. (Flanagan) Postoperative UCVA (Snellen decimal form) in all keratotomy eyes (1983-1994). Midpoints of the Snellen decimal range are indicated by hash marks on the x-axis (width of each bar 0.05).
987-
~I
1-'" 0.0250
0.1750
0.3250
0.4750
0.6250
0.7750
0_9250
Snellen decimal groups, UCVA, postoperative
J CATARACT REFRACT SURG-VOL 23, SEPTEMBER 1997
1061
REFRACTIVE OUTCOME FOLLOWING RK
variability in visual acuity. Therefore, after adjusting for the covariates of age, postoperative spherical equivalent refraction, and postoperative corneal cylinder, the alternative hypothesis of different acuity levels in the keratotomy groups was accepted.
Discussion The analysis indicates that UCVA outcomes in the two groups differed most in the 20/20 or better category. After adjusting for age, postoperative spherical equivalent refraction, and postoperative corneal cylinder in a multiple regression model, the alternative hypothesis of different acuity levels in the two keratotomy groups was accepted at P < .03. This agrees with the conclusions of Kim and coauthors l6 in their review of myopic excimer laser patients receiving simultaneous astigmatic corrections. They found that when astigmatic procedures were used, fewer patients achieved a UCVA of 20/25 or better than patients who had PRK for myopia only. In this cohort, 67% of AKiRK eyes achieved 20/40 or better UCVA. Schneider and coauthors 4 report that 93% of eyes with preoperative astigmatism between 0.50 and 6.00 D treated with transverse incisions achieved a UCVA of 20/40 or better and 43% achieved 20/20 or better. Many surgeons consider a 0.50 D astigmatic error too insignificant to warrant correction, which may explain why such a large number of patients in their study achieved 20/20 or better. In our study, 207 of 284 (72.9%) RK procedures and 65 of 97 (67.0%) AKiRK procedures were corrected to the 20/40 level prior to any enhancement procedure. Schneider and coauthors 4 included enhancement procedures in 11 % of their cases. Our cohort included primary cases; subsequently, 14.6% of all eyes (RK and AKlRK) had enhancement procedures. The postoperative astigmatism results indicate a reduction in mean corneal cylinder (keratometry measured) in AK eyes from 1.60 to 1.06 D. The amount of reduction is less than others report using refractive astigmatism as a measure. In their sample of 60 eyes, Thornton and Sanders 5 report that a preoperative mean of 1.5 D of refractive astigmatism was reduced to 0.4 D. These reports use refractive astigmatism, which is known to differ from the keratometric cylinder measured for our cohort and used as a surrogate for it. 1062
Refractive astigmatism values are more valuable in evaluating efficacy of astigmatic incisions because refractive astigmatism dictates whether transverse incisions are to be used. However, refractive astigmatism does not always reflect the same amount of keratometric astigmatism, which depends on the spherical refractive error. 13 In a separate database of lamellar refractive procedures, we found the ratio of keratometric astigmatism to refractive astigmatism to range between 1.2 and 1.5 (unpublished observations, September 1996). Assuming these ratios are true for the RK and AKiRK cases we report, our results would agree with previous literature reports. In many refractive surgery outcome studies, visual acuity is divided into "20/40 or better" and "worse than 20/40." In our analysis, the 20/40 cutoff does not adequately assess a between-group difference. Bourque and coauthors 8 suggest that 20/20 acuity is a more sensitive criterion because it drives patients' lens-wearing habits and determines satisfaction with surgery. In our population, the 20/20 criterion was more sensitive in demonstrating differences between populations than the customary 20/40 cutoff. Refraction data as a surrogate endpoint failed to show a distinction between the groups. Our study, therefore, provides additional evidence illustrating that visual acuity may be a more important determinant in assessing refractive surgery outcomes. Why did the AKiRK eyes have poorer UCVA than the RK eyes? The major reason was cylindrical undercorrection. Although we found no significant difference in the mean postoperative spherical equivalent refractive errors, most of the enhancement procedures were performed to correct residual myopia. This phenomenon has occurred when astigmatism and myopia are simultaneously corrected with excimer laser photoablation (PARK).? It is not clear whether the RK procedure reduces the effect of the AK procedure or the reverse. We did record the "coupling ratio" in this study, and we plan to analyze the ratios to determine whether one procedure has a negative impact on the other. We agree with Agapitos and coauthors,6 who used vector analysis and found that the accuracy of astigmatic procedures was highly variable. If so, visual acuity levels would be expected to worsen in RK procedures in which astigmatic incisions were judged to be needed.
J CATARACT REFRACT SURG-VOL 23. SEPTEMBER 1997
REFRACTIVE OUTCOME FOLLOWING RK
Certain limitations of this study should be identified. A potential confounding variable that can diminish the strength of the relationship between UCVA and type of keratotomy procedure is postoperative refractive error for patients 40 years and older. Should one type of keratotomy produce more eyes with postoperative refractions greater than + 1.00 D, this group would have poorer overall acuity because of the presbyope's inability to accommodate for vision improvement. To determine whether there was a difference between the number of presbyopic patients with refractions greater than or equal to + 1.00 0 in the two surgery groups for patients 40 years and over, another contingency table analysis was performed. No difference in frequency of postoperative hyperopic refraction was detected between the two keratotomy groups. The two-by-two frequency analysis chi square has a P-value of 0.40. The data did not allow an estimation of the effect on refractive astigmatism. Many investigations studying the effects of astigmatic procedures accept postoperative spherical equivalent refraction and ignore refractive astigmatism. We find this acceptable since our goal was not to determine whether there was a difference in best corrected vision but only whether a postoperative acuity difference existed that was not explained by a difference in preoperative spherical equivalent refraction. The mean postoperative spherical refraction was equal in both groups. A more meaningful comparison might be between best corrected preoperative visual acuity and best corrected postoperative acuity data in the AKIRK and RK groups. This would call for spherocylindrical refraction data. This analysis would reveal whether AKIRK patients lost best corrected vision when compared to RK patients. Many refractive surgery studies use loss and gain of best corrected acuity to assess safety and effectiveness of refractive surgery. The data in our study were not suited to this type of review.
References 1. Binder PS, Waring GO III. Keratotomy for astigmatism. In: Waring GO III, ed, Refractive Keratotomy. St Louis, MO, CV Mosby, 1992; 1085-1198
2. Lindstrom R. The surgical correction of astigmatism: a clinician's perspective. Refract Corneal Surg 1990; 6:441454 3. Price FW, Grene RB, Marks RB, et al. Astigmatism Reduction Clinical Trial: a multicenter prospective evaluation of the predictability of arcuate keratotomy; evaluation of nomogram predictability. Arch Ophthalmol, 1995; 113:277-282; correction 577 4. Schneider 0, Draghic T, Murthy R. Combined myopia and astigmatism surgery. Review of 350 cases. J Cataract Refract Surg 1992; 18:370-374 5. Thornton SP, Sanders DR. Graded nonintersecting transverse incisions for correction of idiopathic astigmatism. J Cataract Refract Surg 1987; 13:27-31 6. Agapitos pJ, Lindstrom RL, Williams PA, et al. Analysis of astigmatic keratotomy. J Cataract Refract Surg 1989; 15:13-18 7. Kim YJ, Sohn J, Tchah H, Lee CO. Photoastigmatic refractive keratectomy in 168 eyes: six-month results. J Cataract Refract Surg 1994; 20:387-391 8. Bourque LB, Lynn MJ, Waring GO III, Cartwright C. Spectacle and contact lens-wearing six years after radial keratotomy in the Prospective Evaluation of Radial Keratotomy Study. Ophthalmology 1994; 101 :421431 9. Binder PS. Optical problems following refractive surgery. Ophthalmology 1986; 93:739-745 10. Rashid ER, Waring GO III. Complications of radial and transverse keratotomy. Surv Ophthalmol 1989; 34:73106 11. American Academy of Ophthalmology. Ophthalmic procedures assessment: radial keratotomy for myopia. Ophthalmology 1993; 100:1103-1115 12. Santos, YR, Waring GO III, Lynn MJ, et al. Relationship between refractive error and visual acuity in the Prospective Evaluation of Radial Keratotomy (PERK) study. Arch Ophthalmol 1987; 105:86-92 13. Sampson WG. Applied optical principles: keratometry. Ophthalmology 1979; 86:347-351 14. Lynn MJ, Waring GO III, Sperduto RD, PERK Study Group. Factors affecting outcome and predictability of radial keratotomy in the PERK study. Arch Ophthalmol 1987; 105:42-51 15. Whitaker, 0, Elliott DB, MacVeigh D. Variations in hyperacuity performance with age. Ophthalmic Physiol Opt 1992; 12:29-32 16. Kim JH, Hahn Tw, Lee yc, Sah WI. Clinical experience of two-step photo refractive keratectomy in 19 eyes with high myopia. J Refract Corneal Surg 1993; 9(suppl):S44-S47
J CATARACT REFRACT SURG-VOL 23, SEPTEMBER 1997
1063