Study objective and hypotheses
In this study we aim to demonstrate that the refractive predictability of SMILE is not inferior to the established procedure (LASIK) to treat myopia. Refractive predictability is one of several standard refractive outcomes, defined as the proportion (number) of eyes achieving a postoperative spherical equivalent (SE) within ±0.50 diopter (D) of the intended target. Our primary hypothesis (stated as null) in this non-inferiority trial would be that SMILE differs from LASIK in adults (>21 years old) with myopia (> −3.00 D) at a tertiary eye center in terms of refractive predictability at 3 months postoperatively. Our secondary hypothesis (stated as null) in this non-inferiority trial would be that SMILE differs from LASIK in adults (>21 years old) with myopia (> −3.00 D) at a tertiary eye center in terms of other refractive outcomes (efficacy, safety, higher-order aberrations) at 3 months postoperatively.
This trial is a single tertiary center, parallel group, single-masked, paired-eye design, non-inferiority, randomized controlled trial. We will use a paired-eye study design, with subjects randomly assigned to undergo one procedure in each eye (SMILE in one eye, LASIK in the other eye). All procedures will be performed in the SNEC Refractive Center by the same fully qualified refractive surgeons, who are co-investigators in this study. Each surgeon has performed more than 30 similar surgeries and 20 cases of SMILE to ensure that each surgeon is adept at performing this procedure. After randomization and random allocation to treatment group, each subject will undergo either SMILE or LASIK in one eye, followed by LASIK or SMILE in the fellow eye on the same day (either the left or right eye will be randomized to decide which eye is operated on first). We have obtained ethics approval from our Institutional Review Board (CIRB Ref No: 2011/109/A) and this trial has been registered (Clinical Trials Registration NCT01216475).
Participants and recruitment
All participants with bilateral myopia will be recruited at the Singapore National Eye Center (SNEC) with the inclusion and exclusion criteria detailed in Table 1. All subjects will be recruited and provide written informed consent that explains the details of the trial, interventions, and study protocol in accordance with the principles of the Declaration of Helsinki.
Each SMILE procedure will be performed using an established, described technique . After application of topical anesthesia, standard sterile draping, and insertion of the speculum, the patient’s eye will be centered and docked with the curved interface cone before application of suction fixation. The laser will then be activated for photo-dissection in the following sequence: first the posterior surface of the refractive lenticule (spiral in), then the lenticule border is created. The anterior surface of the refractive lenticule (spiral out) is then formed which extended beyond the posterior lenticule diameter by 0.5 mm to form the anterior flap and is followed by a rim cut. We will use the following FS laser parameters: 120 μm flap thickness, 7.5 mm flap diameter, 6.5 mm optical zone of lenticule, 145 nj of power with side cut angles at 90°. A superior hinge, 50° in cordal length, will be made in all cases. The spot distance and tracking spacing are 3/3 μm for the lenticule, 2.5/2.5 μm for the lenticule side cut, 3/3 μm for the flap, and 2/2 μm for the flap side cut. After the suction is released, a Siebel spatula (Rhein Medical, Heidelberg, Germany) is inserted under the flap near the hinge before the flap is separated and reflected. The edge of the refractive lenticule is separated from the stromal bed with a sinsky hook and the posterior border of the lenticule gently separated with the Siebel spatula. The lenticule is then grasped with non-toothed serrated forceps through the small incision.
Each LASIK procedure will be performed using a standard, established technique. Under topical anesthesia and standard draping, a lid speculum is used to retract the eyelids, and polyvinyl acetate surgical spears (Ivalon, New London, CT) to dry the conjunctival fornices. A superiorly hinged 120/140 μm thick flap will be created using the Visumax femtosecond laser (Carl Zeiss). Excimer laser ablation is then performed using Wavelight Allegretto WAVE Eye-Q 400 Hz excimer laser (Wavelight GmbH, Alcon, Fort Worth, TX, USA). After ablation, the flap will be carefully repositioned, and postoperative medications are commenced.
We plan to use standard primary and secondary outcomes measures at 3 months postoperatively, which are reported in any assessment refractive surgical technique and standard outcomes in refractive studies. Measurements and outcomes are based on visual acuity (VA) and refraction that are performed by trained refractive optometrists and are repeatedly tested to ensure accuracy and reproducibility. Our primary outcome measure is refractive predictability, which is defined as the proportion number of eyes achieving a postoperative spherical equivalent (SE) within ±0.50 D of the intended target.
Secondary outcome measures include: (1) Efficacy: defined as the proportion number of eyes achieving a unaided visual acuity (UAVA) of 20/20 or better postoperatively; (2) Safety: defined as the proportion number of eyes that lost or gained one or more lines of postoperative best-corrected visual acuity (BCVA) relative to the preoperative BCVA; (3) Higher-order aberrations (HOAs): measured using the Bausch and Lomb Technolas Zywave aberrometer with Zywave software version 4.45 (ZYOPTIX Diagnostic Workstation, Bausch & Lomb); (4) Contrast sensitivity: tested using the Vision Contrast Test System (VCTS) chart (VCTS 6500 contrast sensitivity chart) in six spatial frequencies.
As this is a paired design, non-inferiority trial with a binary outcome, we calculate the required sample size using the maximum likelihood method for large sample proposed by Nam (1997) . A review of current literature reveal that the reported refractive predictabilities in LASIK and SMILE range from 78.2% to 96.7% [1, 14, 15] and from 90.0% to 95.6%, [5–8], respectively. The results from our own audit department estimate our refractive predictability at 82% for LASIK (2011, unpublished). We therefore assumed the refractive predictabilities in LASIK and SMILE in this study are 82% and 92%, respectively. Thus, a sample of 67 subjects (134 eyes) will be sufficient to confirm non-inferiority with a power of ≥80% and at a 5% significance level using a 10% non-inferiority margin, which is the clinically significant difference from our preliminary data. To account for a lost to follow-up rate of 5%, 70 subjects will be recruited.
Randomization and blinding
The random allocation sequence will be generated by a computer with no blocks or restrictions, and implemented by concealing the number-coded surgery within sealed envelopes until just before the procedure. This randomization process will be performed by a research assistant masked to the study subjects and participants will be enrolled by co-investigator surgeons who will assign participants to their groups after opening the sealed envelope, that is each subject will receive a different procedure in each eye at random. In this single-masked trial, subjects and their caregivers will be masked to the assigned treatment in each eye. Both procedures will be performed within the SNEC Refractive Suite, using the femtosecond laser machine. Clinically, it is impossible to detect any difference between each procedure postoperatively to the untrained eye without the slit-lamp microscope. While the surgeons cannot be masked as they will be performing the intervention, the outcome assessors such as nurses, research assistants, and trained optometrists will also be masked to the assigned treatment to improve the objectivity of the research outcomes, as well as to minimize bias. In the event of adverse events (please see below), a code-breaking envelope for each subject will be available.
All patients will have data collection forms outlining each follow-up visit and data to be collected at each visit, which include visual acuity, refraction results, clinical examination findings, and the outcome measures as described above. All data will be securely stored in the SNEC Refractive Suite and then entered into a password-secure desktop computer locked within the Singapore Eye Research Institute, with data back-up into hard drives done daily. Only the named investigators will have access to the research data. All data access will be monitored and controlled by the PI. At the end of the study, the research data will be entered by the research assistant and stored for up to 5 years in compliance with any integrity issues that may arise from any subsequent publications. Following that time period the data will be kept under the control of the PI.
All subjects will be monitored during enrolment into the study for adverse events. All adverse events or serious adverse event (SAE) will be reported to both the centralized institution review board and institution heads (Singhealth) according to the guidelines (http://research.singhealth.com.sg).
Demographic and baseline information will be described, and eye-specific characteristics will be described for each treatment. To study the non-inferiority of SMILE to LASIK, a 90% confidence interval of the difference in predictability between the two treatments (LASIK minus SMILE) will be constructed by a method using score intervals with continuity correction (method #10 in Newcombe RG, 1998) . If the upper limit of the 90% confidence interval does not exceed the pre-defined non-inferiority margin of 5%, non-inferiority is confirmed. Similarly, for each of the two secondary outcomes, efficacy, and safety, a 90% confidence interval of the difference between the two treatments using the above-mentioned method will be constructed and then compared with a non-inferiority margin of 5%. Assuming the other secondary outcome, HOA, follows a normal distribution, a 90% confidence interval of the difference between the two treatments will be constructed through a paired t-test, and then compared with a non-inferiority margin of 10%. We would not be performing any interim analyses due to the short duration of follow-up for each outcome measure.