Evidence for Superior Efficacy and Safety of LASIK over Photorefractive Keratectomy for Correction of Myopia

Transcription

1 Evidence for Superior Efficacy and Safety of LASIK over Photorefractive Keratectomy for Correction of Myopia Alex J. Shortt, MSc, MRCOphth, Catey Bunce, DSc, Bruce D. S. Allan, MD, FRCOphth Purpose: To examine possible differences in efficacy and safety between LASIK and photorefractive keratectomy (PRK) for correction of myopia. Design: Meta-analysis/systematic review. Participants: Patient data from previously reported prospective randomized controlled trials (PRCTs) and a systematic review of prospective case series in the Food and Drug Administration (FDA) clinical trials database. Methods: A comprehensive literature search was performed using the Cochrane Collaboration methodology to identify PRCTs comparing LASIK and PRK for correction of myopia. A meta-analysis was performed on the results of PRCTs. In parallel, a systematic review of prospective data from FDA case series of LASIK and PRK for correction of myopia was undertaken. Main Outcome Measures: Key efficacy outcomes (uncorrected visual acuity [UCVA] 20/20, 0.50 diopters [D] of the target mean refractive spherical equivalent) and safety outcomes (loss of 2 lines of best spectacle-corrected visual acuity [BSCVA], final BSCVA 20/40, and final BSCVA 20/25 where preoperative BSCVA was 20/20). Results: Seven PRCTs were identified comparing PRK (683 eyes) and LASIK (403 eyes) for correction of myopia. More LASIK patients achieved UCVA 20/20 at 6 months (odds ratio, random effects model [95% confidence interval], 1.72 [ ]; P 0.009) and 12 months (1.78 [ ], P 0.01). Loss of 2 lines of BSCVA at 6 months was less frequent with LASIK (2.69 [ ], P 0.05). Data from 14 LASIK (7810 eyes) and 10 PRK (4414 eyes) FDA laser approval case series showed that more LASIK patients achieved UCVA of 20/20 or better at 12 months (1.15 [ ], P 0.01), significantly more LASIK patients were within 0.50 D of target refraction at 6 months (1.38 [ ], P ) and 12 months (1.21 [ ], P ) after treatment, and loss of 2 lines of BSCVA at 6 months was less frequent with LASIK (2.91 [ ], P ). Conclusions: LASIK appears to have efficacy and safety superior to those of PRK. However, the data examined are from studies conducted 5 years ago. It is therefore unclear how our findings relate to present-day methods and outcomes. Further trials comparing contemporary equipment and techniques are needed to reevaluate the relative merits of these procedures. Ophthalmology 2006;113: by the American Academy of Ophthalmology. The 2 most commonly used surgical techniques to correct myopia are photorefractive keratectomy (PRK) and LASIK. Both use the ophthalmic excimer laser to remove corneal Originally received: December 16, Accepted: August 9, Manuscript no From Moorfields Eye Hospital, London, United Kingdom. Presented at: American Society of Cataract and Refractive Surgery Annual Symposium, March 2006, San Francisco, California. The facilities used to conduct this study were provided by Moorfields Eye Hospital NHS Trust, London, United Kingdom. Two of the authors (AJS, BDSA) derive a private practice income from refractive surgery but have no other financial interest in this topic or the article. Dr Bunce has no financial interest in the subject matter. Correspondence and reprint requests to Dr Alex J. Shortt, Moorfields Eye Hospital, 162 City Road, London, EC1V 2PD, United Kingdom. a.shortt@ucl.ac.uk. tissue and reshape the cornea, reducing its refractive power. The number of PRK and LASIK procedures being performed has increased rapidly over the last 20 years. These interventions are performed on healthy eyes, and the vast majority of patients are under 60 years of age. Continued analysis of the relative safety and efficacy of these procedures is highly relevant to informed consent and evidencebased clinical practice in refractive surgery. Photorefractive keratectomy 1 3 gained Food and Drug Administration (FDA) approval before LASIK 4,5 and was initially more widely performed, but LASIK uptake grew rapidly in the late 1990s and LASIK quickly became the dominant modality in laser refractive correction for myopia This shift was based principally on the relatively rapid and painless visual rehabilitation delivered by LASIK. 13 More recently, flap complications and a higher risk for iatrogenic keratectasia in association 2006 by the American Academy of Ophthalmology ISSN /06/$ see front matter Published by Elsevier Inc. doi: /j.ophtha

2 Ophthalmology Volume 113, Number 11, November 2006 with LASIK have driven an increased popularity for surface treatments including PRK. The equation presented to many myopic patients before informed consent for laser refractive surgery balances the speed of recovery after LASIK with the elimination of flap complications in PRK. Assumptions of parity in the final visual outcome are based largely on influential prospective randomized controlled trials (PRCTs) conducted in the late 1990s These studies, taken individually, may lack statistical power to expose clinically significant differences in safety and efficacy that may exist. Corneal haze and refractive regression are more prominent after PRK than after LASIK, and the extent to which differential wound-healing patterns may influence final visual outcomes has not been examined in a systematic review. We performed a meta-analysis of existing PRCTs of PRK versus LASIK for the treatment of myopia with the aim of determining whether any differences in safety or efficacy between the 2 techniques could be detected. In parallel, we examined archive data from prospective multicenter consecutive case series in the FDA trials database (http// These studies document safety and efficacy outcomes for LASIK and PRK using a variety of laser systems introduced since Data collection points and outcomes measures are relatively homogeneous, and large numbers of patients were enrolled. Both pooled data from all FDA trials of LASIK and PRK and results from FDA trials using the same laser system for both LASIK and PRK were analyzed. Materials and Methods Meta-analysis of Randomized Controlled Trials The Cochrane Central Register of Controlled Trials on the Cochrane Library, Medline, Embase, and LILACS (Latin American and Caribbean Literature on Health Sciences) were searched for studies on the correction of myopia by LASIK or PRK published up to and including September Two authors (AJS, BDSA) working independently assessed the titles and abstracts resulting from the searches. Full copies of all potentially or definitely relevant studies were obtained and assessed to determine whether they met standard quality criteria 36 for inclusion in the study. The reference lists of the included studies were searched for information about further trials. The Science Citation Index also was searched to find studies that have cited the identified trials. Included for further study were PRCTs of LASIK versus PRK for the correction of any degree of myopia and up to 3 diopters (D) of astigmatism conducted in adult patients 18 to 60 years old with no significant co-pathology, history of previous ocular surgery, or systemic disease associated with impaired or abnormal wound healing. Both authors (AJS, BDSA) independently assessed study quality according to the methods set out in the Cochrane Handbook for Systematic Reviews of Interventions. 36 Areas of methodological quality examined were allocation concealment, method of allocation to treatment, masking of outcome assessment, and completeness of follow-up. Based on these assessments and the Cochrane Collaboration guidelines, 36 the 2 authors independently devised a list of trials to be included in the meta-analysis. These were compared and found to be identical. In several instances, we contacted trial authors for clarification of issues regarding randomization and for additional data not contained in the original trial report. The original electronic searches identified 949 reports. Seven publications from 6 randomized controlled trials met the inclusion criteria. 24,27,29,37 40 Forseto et al 38 and Hjortdal et al 40 kindly provided additional previously unpublished data. One additional trial 41 was excluded because contact with the corresponding author revealed that assignment of eyes to treatment groups was performed on an ad hoc basis. Key safety and efficacy outcome measures were selected to facilitate data extraction, and proportions of patients achieving or affected by each criterion were compared. Efficacy measures examined were uncorrected visual acuity (UCVA) 20/20 and a manifest refractive spherical equivalent (SE) within 0.50 D of the target. Safety measures examined were loss of 2 lines of logarithm of the minimum angle of resolution best spectaclecorrected visual acuity (BSCVA), final BSCVA of 20/25 in patients with preoperative BSCVA 20/20, and final BSCVA of 20/40. Efficacy measures were examined at 6 and 12 months postoperatively. Data for safety measures were insufficient to allow separate analysis at the 6- and 12-month time points. Accordingly, safety data reported after a minimum of 6 months (range, 6 12) were pooled for comparison. Data were extracted independently by the 2 authors (AJS, BDSA) using a standardized record form and entered into a meta-analysis software package, RevMan 4.2 (Information Management System Group, Cochrane Collaboration, Oxford, United Kingdom), for analysis. Results were compared, and any discrepancies between the reviewers results again were resolved by discussion. Subanalysis for Low Myopia Some of the trials identified included treatments 6.00 D. To examine whether differences in key efficacy outcomes (UCVA 20/ 20; manifest refractive SE within 0.50 D of target) were attributable simply to poor results in higher level treatments, we performed a subanalysis of results from trials of LASIK versus PRK for low to moderate myopia ( 6.00 D sphere, 2.50 D astigmatism). Review of Food and Drug Administration Data A similar analysis, also using RevMan 4.2, was conducted using data extracted from prospective case series of myopic PRK and LASIK in patients with up to 3Dofastigmatism in the FDA database ( Initially, data from FDA approval case series of LASIK and PRK treatments performed using the same laser model were compared. Data were available for 5 laser types. Next, pooled data from all FDA case series in which LASIK treatment was performed (14 studies) were compared with pooled data from all FDA case series in which PRK was performed (10 studies). Safety and efficacy outcome measures examined were identical to those considered in the meta-analysis of randomized controlled trials detailed above; but where 12- month outcome data were not available for efficacy criteria, 9-month data were substituted. Statistical Analysis In the meta-analysis of randomized controlled trials, data for all outcomes were summarized using odds ratios (ORs). Heterogeneity between studies was assessed by review of trial reports, and statistical heterogeneity was tested for with a chi-square test. The I 2 statistic was calculated to assess inconsistency between studies. Summary ORs were calculated for each outcome measure using a random-effects model. 42 In the review of FDA data, a random 1898

3 Shortt et al Evidence for Superior Efficacy and Safety of LASIK over PRK effect model OR was calculated for efficacy and safety criteria in studies using the same laser model for PRK and LASIK. The Mantel Haenszel OR was calculated when synthesizing pooled data from all case studies. Considerable heterogeneity was detected between the FDA studies. This heterogeneity could be accounted for largely by differences between studies in the completeness of PRK or LASIK nomogram development. Study methodologies were otherwise consistent, and we proceeded to combine data to collate all available evidence. It must be noted, however, that where significant heterogeneity is present in the FDA data analyses, care should be taken in interpreting results. Results Uncorrected Visual Acuity of >20/20 Results of Meta-analysis of Randomized Controlled Trials. Seven publications from 6 randomized controlled trials that met the inclusion criteria compared PRK (683 eyes) and LASIK (403 eyes) for correction of myopia. 24,27,29,37 40 The first efficacy outcome examined was the proportion of patients achieving UCVA 20/20 at 6 months and at 12 months after treatment (Fig 1a). At the 6-month time point, data were available for a total of 691 eyes with myopia of 0 to 15 D, in 5 trials 24,29,37 40 of PRK (n 460) or LASIK (n 231). Examination of the forest plot showed that in one trial (Hjortdal et al 40 ) the proportion of participants having an UCVA of 20/20 or better at 6 months after treatment was higher in the PRK group than in the LASIK group. In the other 4 studies, a higher proportion of patients in the LASIK group achieved this outcome. Analysis of these data revealed that the LASIK group was statistically more likely to achieve UCVA of 20/20 or better at 6 months after treatment (OR, 1.72; 95% confidence interval [CI], ; P 0.009) [Fig 1a(i)]. At the 12-month time point, data were available for a total of 593 eyes randomized in 5 trials 27,29,37,38,40 to PRK (n 396) or LASIK (n 197). Again, examination of the forest plot demonstrated that, with the exception of Hjortdal et al, 40 a higher proportion of patients in the LASIK group achieved this outcome at 12 months. The odds of achieving UCVA of 20/20 or better at 12 months after treatment were significantly greater in the LASIK group (OR, 1.78; 95% CI, ; P 0.01) [Fig 1a(ii)]. A sensitivity analysis was performed to examine the effect of excluding Hjortdal et al, but this did not alter the above results. Results of Comparison of Food and Drug Administration Case Series Data from Studies Using the Same Laser. When the same model of laser was used to perform a series of LASIK procedures and a series of PRK procedures, the forest plot at the 6-month time point demonstrates a large amount of heterogeneity between outcomes. Statistical analysis of the data did not demonstrate any difference between groups (OR, 1.01; 95% CI, ; P 0.98) [Fig 1b(i)]. At the 12-month time point, data were available for 3 laser models. The outcomes again demonstrate marked heterogeneity, with results for 2 of the 3 laser models favoring LASIK and 1 favoring PRK [Fig 1b(ii)]. Results of Comparison of Pooled Data from All Food and Drug Administration Case Series. Analysis of pooled data from 14 LASIK (7810 eyes) and 10 PRK (4414 eyes) FDA approval case series showed that at 6 months after treatment no difference in the proportion of patients achieving this outcome was detectable between groups (OR, 0.98; 95% CI, ; P 0.65) [Fig 1c(i)]. In contrast to this, at the 12-month time point more LASIK patients achieved UCVA of 20/20 or better (OR, 1.15; 95% CI, ; P 0.01) [Fig 1c(ii)]. Postoperative Refraction within 0.50 Diopters of Target Refraction Results of Meta-analysis of Randomized Controlled Trials. The proportion of participants within 0.50 D of target refraction at 6 months and 12 months after treatment is shown in Figure 2a. At the 6-month time point, data were available for a total of 245 eyes with myopia of 0 to 15 D randomized in 4 trials 24,29,38 40 to PRK (n 125) or LASIK (n 120). There was no statistically significant difference between groups (OR, 0.83; 95% CI, ) [Fig 2a(i)]. At the 12-month time point, data were available for a total of 593 eyes randomized in 5 trials 27,29,37,38,40 to PRK (n 396) or LASIK (n 197). Examination of the forest plot revealed that, with the exception of Hjortdal et al, 40 a higher proportion of patients in the LASIK group achieved this outcome at 12 months. Statistical analysis of these data revealed that the difference between groups in favor of LASIK was not statistically significant (OR, 1.38; 95% CI, ; P 0.12) [Fig 2a(ii)]. A sensitivity analysis was performed to examine the effect of excluding Hjortdal et al, but this did not alter the above results. Results of Comparison of Food and Drug Administration Case Series Data from Studies Using the Same Laser. When the same model of laser was used in both PRK and LASIK case series, the forest plot at the 6-month time point demonstrates that for 4 of 5 models of laser, a greater proportion of patients in the LASIK group achieved a posttreatment refraction within 0.50 D of target refraction [Fig 2b(i)]. Although the likelihood of achieving this outcome was greater in the LASIK group (OR, 1.25; 95% CI, ) [Fig 1b(i)], this finding was not statistically significant (P 0.06). At the 12-month time point, data were available for 3 models of laser. The outcomes for 2 of the 3 laser models favored LASIK. The results for the third model favored neither treatment [Fig 2b(ii)]. Results of Comparison of Pooled Data from Food and Drug Administration Case Series. Analysis of pooled data from 14 LASIK (7810 eyes) and 10 PRK (4414 eyes) FDA approval case series showed that at 6 months after treatment the proportion of patients within 0.50 D of target refraction was significantly greater in the LASIK group (P ) [Fig 2c(i)]. Again, at the 12-month time point more LASIK patients achieved this outcome (P ) [Fig 2c(ii)]. Outcomes in Low Myopia Subgroup (Myopia < 6.0 Diopters) The PRCT data permitted us to examine whether differences in visual (UCVA 20/20) and refractive ( 0.50 of target refraction) outcomes were present in the subgroup of patients with myopia 6.0 D and astigmatism 2.50 D. Proportion of Low Myopia Subgroup with Uncorrected Visual Acuity > 20/20 at 6 Months and 12 Months. At the 6-month time point, data were available for a total of 461 eyes with myopia of 0 to 6 D randomized in 2 trials 37,38 to PRK (n 344) or LASIK (n 117). Analysis of these data revealed that the LASIK group was statistically more likely to achieve UCVA of 20/20 or better at 6 months after treatment (OR, 1.79; 95% CI, ; P 0.03) [Fig 3a(i)]. At the 12-month time point, data were available for a total of 488 eyes randomized in 3 trials 27,37,38 to PRK (n 346) or LASIK (n 142). Statistical analysis of these data revealed that the LASIK group was statistically more likely to achieve UCVA of 20/20 or better at 12 months after treatment (OR, 1.95; 95% CI, ; P 0.008) [Fig 3a(ii)]. Proportion of Low Myopia Subgroup within 0.50 Diopters of Target Refraction at 6 Months and 12 Months. At the 6-month time point, data for this outcome were available for 1899

4 Ophthalmology Volume 113, Number 11, November 2006 Figure 1. Photorefractive keratectomy (PRK) versus LASIK for correction of myopia. Numbers of patients that achieve posttreatment uncorrected visual acuity (UCVA) of 20/20. In the forest plots, the odds ratios (OR; squares) and confidence intervals (CIs; horizontal lines) are graphically represented for each individual study. The size of the square represents the weight given to that study in the combined analysis of trials and is based upon the number of participants. The corresponding numerical data are displayed on the righthand side. The black diamond towards the bottom of each plot represents the OR (center of the diamond) and CIs (horizontal width of diamond) when data from all studies are pooled for analysis. The data on the bottom left of each plot relate to tests for statistical heterogeneity and the statistical significance of the overall effect estimate. A significant amount of heterogeneity is considered to be present when P 0.10, and under these circumstances, the estimate of the overall effect must be interpreted with a degree of caution. The test for overall effect assesses the statistical significance of the pooled data. D diopters; FDA Food and Drug Administration. 1900

5 Shortt et al Evidence for Superior Efficacy and Safety of LASIK over PRK Figure 2. Photorefractive keratectomy (PRK) versus LASIK for correction of myopia. Numbers of patients within 0.50 diopters of target refraction. For details on forest plot interpretation, see Figure 1 caption. FDA Food and Drug Administration. only one trial, 38 which did not demonstrate a difference in this outcome for the low myopia subgroup [Fig 3b(i)]. At the 12-month time point, data were available for a total of 488 eyes randomized in 3 trials 27,37,38 to PRK (n 346) or LASIK (n 142). Statistical analysis of these data revealed that the difference between groups in favor of LASIK was not statistically significant (OR, 1.55; 95% CI, ; P 0.06) [Fig 3b(ii)]. 1901

6 Ophthalmology Volume 113, Number 11, November 2006 Figure 3. Photorefractive keratectomy (PRK) versus LASIK for correction of myopia. Outcomes in low myopia subgroup (myopia 6.0 diopters [D]). a, Proportion of the low myopia subgroup with uncorrected visual acuity 20/20 at 6 months and 12 months. b, Proportion of the low myopia subgroup within 0.50 D of target refraction at 6 months and 12 months. For details on forest plot interpretation, see Figure 1 caption. Loss of >2 Lines of Best Spectacle-Corrected Visual Acuity Results of Meta-analysis of Randomized Controlled Trials. Five randomized controlled trials reported data for the proportion of patients losing 2 lines at 6 months or more after PRK or LASIK. 24,27,29,37,39,40 Examination of the forest plot demonstrates that in el Danasoury et al 27 no patient lost 2 lines of BSCVA (Fig 4a). Three of the 4 remaining studies showed that a greater proportion of participants in the PRK group lost 2 lines of BSCVA at 6 months or more after treatment. Analysis of these data revealed that significantly more patients in the PRK group lost 2 lines of BSCVA (OR, 2.69; 95% CI, ; P 0.05) (Fig 4a). Results of Comparison of Food and Drug Administration Case Series Data from Studies Using the Same Laser. When the same model of laser was used in both PRK and LASIK case series, the forest plot for this outcome demonstrates that, for 3 of 5 models of laser, a greater proportion of patients in the PRK group lost

7 Shortt et al Evidence for Superior Efficacy and Safety of LASIK over PRK Figure 4. Photorefractive keratectomy (PRK) versus LASIK for correction of myopia. Loss of 2 lines of best spectacle-corrected visual acuity. For details on forest plot interpretation, see Figure 1 caption. FDA Food and Drug Administration. lines of BSCVA, and in one study, more LASIK patients lost 2 lines of BSCVA. In one study, there was no difference between groups (Fig 4b). Significant heterogeneity between study results was detected I 2 80% (Fig 4b). Results of Comparison of Pooled Data from Food and Drug Administration Case Series. Analysis of pooled data from 14 LASIK (7554 eyes) and 10 PRK (4412 eyes) FDA approval case series showed that significantly more patients in the PRK group lost lines of BSCVA (OR, 2.91; 95% CI, ; P ) (Fig 4c). Final Best Spectacle-Corrected Visual Acuity Less than 20/40 Results of Meta-analysis of Randomized Controlled Trials. Four studies reported data for this outcome. 24,27,29,39,40 The forest plot for this outcome demonstrates that in 2 of these studies (El- Maghraby et al 29 and Summit ) participants in the PRK group were more likely to have a final BSCVA of 20/40 or worse at 6 months after treatment. In the other 2 studies, no participants from either group had a final BSCVA of 20/40 or worse at 6 months after treatment (Fig 5a). Although the likelihood of this adverse outcome was greater in the PRK group (OR, 2.92; 95% CI, ), this finding was not statistically significant (P 0.36). Results of Comparison of Food and Drug Administration Case Series Data from Studies Using the Same Laser. Examination of the forest plot for the same laser model data demonstrates that, for 3 of 5 laser models, a greater proportion of patients in the PRK group had a BSCVA 20/40, and in one study, a greater proportion of LASIK patients had this outcome (Fig 5b). Analysis of these data revealed that, though the likelihood of this adverse outcome was greater in the PRK groups, this finding was not statistically significant (P 0.15; OR, 2.13; 95% CI, ) (Fig 5b). 1903

8 Ophthalmology Volume 113, Number 11, November 2006 Figure 5. Photorefractive keratectomy (PRK) versus LASIK for correction of myopia. Final best spectacle-corrected visual acuity 20/40. For details on forest plot interpretation, see Figure 1 caption. FDA Food and Drug Administration. Results of Comparison of Pooled Data from Food and Drug Administration Case Series. Analysis of pooled data from 14 LASIK (7810 eyes) and 10 PRK (4414 eyes) FDA approval case series did not demonstrate a difference between treatments (Fig 5c). Final Best Spectacle-Corrected Visual Acuity Less than 20/25 When Preoperative Best Spectacle- Corrected Visual Acuity Was >20/20 Results of Meta-analysis of Randomized Controlled Trials. Four studies reported data for this outcome. 25,27 29,40 The proportion of participants with BSCVA of 20/20 or better preoperatively who had BSCVA of 20/25 at 6 months after treatment was marginally higher for the LASIK group in one study. In the remaining 3 studies, there was no difference between groups (Fig 6a). Analysis of these data showed no difference between groups (OR, 0.93; 95% CI, ) (Fig 6a). Results of Comparison of Food and Drug Administration Case Series Data from Studies Using the Same Laser. The forest plot for this outcome demonstrates a large amount of heterogeneity between laser models (Fig 6b). For 4 of 5 laser models, a greater proportion of patients in the PRK group had final BSCVA 20/25 when preoperative BSCVA was 20/20, whereas this was the case for LASIK with only one laser model. Analysis of these data revealed that, though the likelihood of this adverse outcome was greater in the PRK groups, this finding was not statistically significant (P 0.38; OR, 2.24; 95% CI, ) (Fig 6b). Results of Comparison of Pooled Data from Food and Drug Administration Case Series. Analysis of pooled data from 14 LASIK (7612 eyes) and 10 PRK (4299 eyes) FDA approval case 1904

9 Shortt et al Evidence for Superior Efficacy and Safety of LASIK over PRK Figure 6. Photorefractive keratectomy (PRK) versus LASIK for correction of myopia. Final best spectacle-corrected visual acuity (BSCVA) 20/25 when preoperative BSCVA 20/20. For details on forest plot interpretation, see Figure 1 caption. FDA Food and Drug Administration. series showed that significantly more patients in the PRK group lost BSVCA according to this outcome (P ) (Fig 6c). Discussion This study provides evidence that, with available evidence from early randomized controlled trials, the efficacy and safety of LASIK appears to be superior to PRK. Results of the meta-analysis of PRCTs are supported by similar findings from an analysis of large numbers of patients enrolled in FDA laser approval trials (prospective case series). A problem with systematic reviews of a technique in evolution, as opposed to a drug treatment, is that changes in technology and technique may influence results significantly. The data included in our analyses are from studies conducted 5 years ago, with many of the trials conducted in the late 1990s. There is evidence that technological advances have improved the safety and accuracy of these techniques. We recently summarized evidence for improved safety in LASIK after 2000, 35 and developments, including the use of adjunctive intraoperative mitomycin C 43 and smoother ablation profiles, may have improved outcomes in PRK and other surface laser treatments. It is therefore unclear how our findings relate to the outcomes associated with current technology and techniques. It cannot be assumed, however, that presentday LASIK and PRK (or other surface treatments) have comparable safety and efficacy. Further trials employing contemporary equipment and techniques are needed to reevaluate the relative merits of these procedures. 1905

10 Ophthalmology Volume 113, Number 11, November 2006 A major difficulty in combining the results of PRCTs was the diversity of outcome measures and follow-up intervals reported. Currently, there is no generally accepted method for reporting visual and refractive results of trials involving refractive procedures. Waring 47,48 has suggested a standardized format. Based on this and the safety and efficacy measures used in FDA clinical trials of refractive surgery, we defined a framework of key outcome measures at fixed time points for use in this study. This framework enabled us to extract and compare data from relevant randomized controlled trials in a systematic manner. Contact with the authors of several trials resulted in our obtaining additional previously unpublished data. Despite this, none of the trials reported a complete data set for all of the defined outcomes. Hence, not all trials could be included in each of the outcome analyses. Accepting these limitations, we were able to combine the data from these different trials and perform statistical analysis on the results. The follow-up duration reported in the PRCTs was limited. Only 4 of the 7 studies reported 12-month follow-up data. It is possible that this duration of follow-up is inadequate to determine the final visual outcomes for either or both of these procedures. There is evidence that improvements in vision and the resolution of corneal haze may continue beyond the first 12 months after treatment. 49 Recovery of corneal innervation and restoration of a normal tear film and ocular surface also may take longer than 12 months. 50,51 To facilitate comparison between future trials in refractive surgery, a standardized framework of outcome measures and follow-up intervals should be evolved further to create an evidence base that is more relevant to contemporary clinical practice. This should include later time points and a questionnaire instrument designed to measure visionrelated quality of life One of the PRCTs included in the meta analysis (Wang et al 37 ) was found to be at risk of bias. After randomization in this study, 49% of the participants assigned to the LASIK group instead insisted on having PRK because they could not afford to pay for LASIK. Otherwise, the methodological quality of this PRCT is satisfactory. This study is the largest PRCT reported to date and, accordingly, carries a significant weight in our analyses. Its findings were similar to those of the other studies included. A sensitivity analysis was performed to determine the effect of excluding this trial from the meta-analysis. Exclusion of this RCT resulted in the likelihood of achieving the relevant outcome being similar or unchanged, but in some analyses, the result was no longer statistically significant. Considered in isolation, existing PRCTs do not reveal clear differences in the final visual outcomes for PRK and LASIK treatment in similar myopic patients. For this reason, it would be inappropriate to criticize physicians for having recommended one procedure over the other in the past. However, our meta-analysis of these trials detected differences and suggests that patients are almost twice as likely to achieve a final UCVA 20/20 after LASIK [Fig 1a(i)]. Food and Drug Administration data also show that final UCVA after LASIK appears to be superior to that after PRK (Fig 1b, c). There are 2 possible explanations for these unexpected findings: LASIK may have produced a more accurate refractive outcome, or BSCVA may have been relatively reduced in PRK. It is difficult to comment on relative accuracy for PRK and LASIK in the trials analyzed, because many of the observed differences apparent in trial data (Fig 2) may be attributable to disparities in nomogram development rather than genuine intrinsic differences in accuracy. Incomplete nomogram development for either the PRK or the LASIK arm of any given trial is a plausible explanation for the heterogeneity in both accuracy and consequent UCVA findings evident in both PRCT (Figs 1a, 2a) and FDA data comparing trials of PRK and LASIK using similar lasers (Figs 1b, 2b). Analysis of early postoperative ( 1 month) results in PRCTs (data not shown) revealed wide variations between trials in the initial average manifest refractive SE (LASIK, 1.19 to 0.90 D; PRK, 0.66 to 0.6 D). This heterogeneity cannot be accounted for by differences in the target refraction: in all but one study, the target refraction was emmetropia. All but one of the PRCTs were conducted in the late 1990s. At that time, nomogram development for PRK was generally at a more advanced stage than that for LASIK. The most recent trial by Hjortdal et al also may have been affected by incomplete nomogram development. The authors conducted a high-quality trial, but examination of their postoperative mean refractive SE revealed that their LASIK nomogram resulted in a mean 1.19 D of undercorrection, as opposed to their PRK nomogram, which resulted in 0.66 D of undercorrection. 40 It is unlikely, therefore, that the potential greater accuracy in LASIK (Fig 2) is simply underpinned by a systematic bias relating to poor nomogram development for PRK. Interindividual differences in the wound-healing response to a given treatment are often cited as a limiting factor in the accuracy of laser refractive surgery. Because wound-healing responses are relatively upregulated in surface laser treatments, these differences are likely to be amplified. LASIK may be more accurate than PRK, but clinical trials conducted after documented nomogram optimization in both arms will be required before any definitive conclusions can be reached. In addition to producing more accurate results, our data demonstrated that LASIK appears to be safer than PRK. Meta-analysis of PRCTs showed that significantly more patients lost 2 lines of BSCVA after PRK (Fig 4a). Hersh et al 24 comment on superior safety in LASIK, but the results of individual trials failed to demonstrate statistically significant safety gains over PRK. The statistical significance of this result is sensitive to the exclusion of the trial by Wang et al, 37 but even after exclusion of this study, a greater likelihood of losing BSCVA in PRK is still evident. This finding is clearly supported by the same laser (Fig 4b) and pooled (Fig 4c) FDA data. The pooled FDA data are particularly relevant, given the large numbers of patients in both the PRK and the LASIK groups analyzed in this comparison, and the low frequency of poor visual outcomes. From this, it would appear that significant loss of vision ( 2 lines of BSCVA) is 2 to 4 times more common after PRK. Data for other safety outcome measures (Figs 5, 6) are less conclusive, but the likelihood of greater safety with LASIK is apparent. Pooled data from all FDA case series indicate 1906

11 Shortt et al Evidence for Superior Efficacy and Safety of LASIK over PRK that loss of vision to 20/25 is significantly more likely to occur after PRK. Caution is required in interpreting safety data reported as early as 6 months after treatment because, as argued above, continued visual improvement can be expected in both the PRK and the LASIK groups. From the data we present, it would appear that persisting visual disability in the first year after treatment is more likely after PRK. To examine whether the differences we observed were related to the degree of myopia treated, we performed an analysis of visual (UCVA 20/20) and refractive ( 0.50 D of target refraction) outcomes in the subgroup of patients with myopia of 6.0 D and astigmatism 2.5 D. The outcomes for this subanalysis did not differ from those for the analyses of all degrees of myopia. We have found that data from early trials comparing PRK and LASIK for correction of myopia suggest that LASIK is more accurate and safer than PRK. Changes in technology and techniques over time are likely to influence or alter this finding, but equivalence of safety and final visual outcomes cannot be assumed. Our study highlights the need for further trials comparing current-day LASIK and surface treatments to determine the relative merits of these procedures. References 1. Munnerlyn CR, Koons SJ, Marshall J. Photorefractive keratectomy: a technique for laser refractive surgery. J Cataract Refract Surg 1988;14: Goodman GL, Trokel SL, Stark WJ, et al. Corneal healing following laser refractive keratectomy. Arch Ophthalmol 1989;107: Epstein D, Fagerholm P, Hamberg-Nystrom H, Tengroth B. Twenty-four-month follow-up of excimer laser photorefractive keratectomy for myopia: refractive and visual acuity results. Ophthalmology 1994;101: Pallikaris IG, Papatzanaki ME, Stathi EZ, et al. Laser in situ keratomileusis. Lasers Surg Med 1990;10: Pallikaris IG, Papatzanaki ME, Siganos DS, Tsilimbaris MK. A corneal flap technique for laser in situ keratomileusis: human studies. Arch Ophthalmol 1991;109: Leaming DV. Practice styles and preferences of ASCRS members 1998 survey. J Cataract Refract Surg 1999;25: Leaming DV. Practice styles and preferences of ASCRS members 1999 survey. J Cataract Refract Surg 2000;26: Leaming DV. Practice styles and preferences of ASCRS members 2000 survey. J Cataract Refract Surg 2001;27: Leaming DV. Practice styles and preferences of ASCRS members 2001 survey. J Cataract Refract Surg 2002;28: Leaming DV. Practice styles and preferences of ASCRS members 2002 survey. J Cataract Refract Surg 2003;29: Leaming DV. Practice styles and preferences of ASCRS members 2003 survey. J Cataract Refract Surg 2004;30: Sandoval HP, de Castro LE, Vroman DT, Solomon KD. Refractive surgery survey J Cataract Refract Surg 2005; 31: Sugar A, Rapuano CJ, Culbertson WW, et al. Laser in situ keratomileusis for myopia and astigmatism: safety and efficacy. A report by the American Academy of Ophthalmology. Ophthalmology 2002;109: Gimbel HV, Penno EE, van Westenbrugge JA, et al. Incidence and management of intraoperative and early postoperative complications in 1000 consecutive laser in situ keratomileusis cases. Ophthalmology 1998;105: Stulting RD, Carr JD, Thompson KP, et al. Complications of laser in situ keratomileusis for the correction of myopia. Ophthalmology 1999;106: Lin RT, Maloney RK. Flap complications associated with lamellar refractive surgery. Am J Ophthalmol 1999;127: Tham VM, Maloney RK. Microkeratome complications of laser in situ keratomileusis. Ophthalmology 2000;107: Jacobs JM, Taravella MJ. Incidence of intraoperative flap complications in laser in situ keratomileusis. J Cataract Refract Surg 2002;28: Seiler T, Koufala K, Richter G. Iatrogenic keratectasia after laser in situ keratomileusis. J Refract Surg 1998;14: Geggel HS, Talley AR. Delayed onset keratectasia following laser in situ keratomileusis. J Cataract Refract Surg 1999;25: Amoils SP, Deist MB, Gous P, Amoils PM. Iatrogenic keratectasia after laser in situ keratomileusis for less than 4.0 to 7.0 diopters of myopia. J Cataract Refract Surg 2000;26: Pallikaris IG, Kymionis GD, Astyrakakis NI. Corneal ectasia induced by laser in situ keratomileusis. J Cataract Refract Surg 2001;27: Randleman JB, Russell B, Ward MA, et al. Risk factors and prognosis for corneal ectasia after LASIK. Ophthalmology 2003;110: Hersh PS, Brint SF, Maloney RK, et al. Photorefractive keratectomy versus laser in situ keratomileusis for moderate to high myopia: a randomized prospective study. Ophthalmology 1998;105: , discussion Hersh PS, Scher KS, Irani R, Summit PRK-LASIK Study Group. Corneal topography of photorefractive keratectomy versus laser in situ keratomileusis. Ophthalmology 1998;105: Steinert RF, Hersh PS, Summit Technology PRK-LASIK Study Group. Spherical and aspherical photorefractive keratectomy and laser in-situ keratomileusis for moderate to high myopia: two prospective, randomized clinical trials. Trans Am Ophthalmol Soc 1998;96: el Danasoury MA, el Maghraby A, Klyce SD, Mehrez K. Comparison of photorefractive keratectomy with excimer laser in situ keratomileusis in correcting low myopia (from 2.00 to 5.50 diopters). A randomized study. Ophthalmology 1999;106: Hersh PS, Abbassi R, Summit PRK-LASIK Study Group. Surgically induced astigmatism after photorefractive keratectomy and laser in situ keratomileusis. J Cataract Refract Surg 1999;25: El-Maghraby A, Salah T, Waring GO III, et al. Randomized bilateral comparison of excimer laser in situ keratomileusis and photorefractive keratectomy for 2.50 to 8.00 diopters of myopia. Ophthalmology 1999;106: Seiler T, Wollensak J. Myopic photorefractive keratectomy with the excimer laser: one-year follow-up. Ophthalmology 1991;98:

12 Ophthalmology Volume 113, Number 11, November Krueger RR, Talamo JH, McDonald MB, et al. Clinical analysis of excimer laser photorefractive keratectomy using a multiple zone technique for severe myopia. Am J Ophthalmol 1995;119: Alio JL, Artola A, Claramonte PJ, et al. Complications of photorefractive keratectomy for myopia: two year follow-up of 3000 cases. J Cataract Refract Surg 1998;24: Kuo IC, Lee SM, Hwang DG. Late-onset corneal haze and myopic regression after photorefractive keratectomy (PRK). Cornea 2004;23: Sun L, Liu G, Ren Y, et al. Efficacy and safety of LASIK in 10,052 eyes of 5081 myopic Chinese patients. J Refract Surg 2005;21(suppl):S Watson SL, Bunce C, Allan BD. Improved safety in contemporary LASIK. Ophthalmology 2005;112: Higgins JPT, Green S, eds. Cochrane handbook for systematic reviews of interventions [updated May 2005]. Sections 5 7. Available at: hbook.htm. Accessed May 31, Wang Z, Chen J, Yang B. Comparison of laser in situ keratomileusis and photorefractive keratectomy to correct myopia from to diopters. J Refract Surg 1997;13: Forseto AS, Nosé RA, Nosé W. PRK versus LASIK for correction of low and moderate myopia [in Portuguese]. Arq Bras Oftalmol 2000;63: Hersh PS, Steinert RF, Brint SF, Summit PRK-LASIK Study Group. Photorefractive keratectomy versus laser in situ keratomileusis: comparison of optical side effects. Ophthalmology 2000;107: Hjortdal JO, Moller-Pedersen T, Ivarsen A, Ehlers N. Corneal power, thickness, and stiffness: results of a prospective randomized controlled trial of PRK and LASIK for myopia. J Cataract Refract Surg 2005;31: Lee JB, Kim JS, Choe C, et al. Comparison of two procedures: photorefractive keratectomy versus laser in situ keratomileusis for low to moderate myopia. Jpn J Ophthalmol 2001;45: DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7: Gambato C, Ghirlando A, Moretto E, et al. Mitomycin C modulation of corneal wound healing after photorefractive keratectomy in highly myopic eyes. Ophthalmology 2005; 112: O Brart DP, Corbett MC, Lohmann CP, et al. The effects of ablation diameter on the outcome of excimer laser photorefractive keratectomy: a prospective, randomized, double-blind study. Arch Ophthalmol 1995;113: Corbett MC, Verma S, O Brart DP, et al. Effect of ablation profile on wound healing and visual performance 1 year after excimer laser photorefractive keratectomy. Br J Ophthalmol 1996;80: O Brart DP, Corbett MC, Verma S, et al. Effects of ablation diameter, depth, and edge contour on the outcome of photorefractive keratectomy. J Refract Surg 1996;12: Waring GO III. Conventional standards for reporting results of refractive surgery. Refract Corneal Surg 1989;5: Waring GO III. Standard graphs for reporting refractive surgery. J Refract Surg 2000;16: Rajan MS, Jaycock P, O Brart D, et al. A long-term study of photorefractive keratectomy: 12-year follow-up. Ophthalmology 2004;111: Murphy PJ, Corbett MC, O Brart DP, et al. Loss and recovery of corneal sensitivity following photorefractive keratectomy for myopia. J Refract Surg 1999;15: Calvillo MP, McLaren JW, Hodge DO, Bourne WM. Corneal reinnervation after LASIK: prospective 3-year longitudinal study. Invest Ophthalmol Vis Sci 2004;45: Massof RW, Rubin GS. Visual function assessment questionnaires. Surv Ophthalmol 2001;45: Nichols JJ, Mitchell GL, Saracino M, Zadnik K. Reliability and validity of refractive error-specific quality-of-life instruments. Arch Ophthalmol 2003;121: Garamendi E, Pesudovs K, Elliott DB. Changes in quality of life after laser in situ keratomileusis for myopia. J Cataract Refract Surg 2005;31: