Visual Acuity Outcomes in a Randomized Trial of Wavefront Metric Optimized Refractions in Adults with Down Syndrome

Abstract

Significance.

This study reports visual acuity outcomes from a clinical trial investigating an objective refraction strategy that may provide a useful tool for practitioners needing additional strategies to identify refractive corrections for adults with intellectual disability.

Purpose.

Determining refractions for individuals with Down syndrome is challenging due to the presence of elevated refractive error, optical aberrations, and cognitive impairment. This randomized clinical trial evaluated performance of spectacle corrections determined using clinical techniques and objective refractions derived from wavefront aberration measures.

Methods.

Thirty adults with Down syndrome had a clinical refraction determined by a single expert examiner using pre- and post-dilation techniques appropriate for this population. Objective refractions were determined from dilated wavefront aberration measures that were processed post-visit to identify refractions that optimized each of two image quality metrics: Pupil Fraction Tessellated and Visual Strehl Ratio in the spatial domain. The three refractions were dispensed in random order and worn for 2 months each. The primary outcome measure, binocular visual acuity, was obtained by a masked examiner administering a distance logMAR acuity test. To compare treatment types, mean acuity was compared using a 2-sided Type 3 F-test of the treatment effect in a linear mixed-effect regression model, where the final model included fixed-effects for treatment, period (1, 2, or 3), and first order carryover effects.

Results.

The two-month estimated least square means in binocular visual acuity (logMAR) was 0.34 (95% CI: 0.25, 0.39) for clinical refractions, 0.31 (0.25, 0.36) for Pupil Fraction Tesselated refractions, and 0.33 (0.27, 0.38) for Visual Strehl Ratio refractions. No statistically significant treatment effect was observed (F=1.10, P=.34).

Conclusions.

Objective refractions derived from dilated wavefront aberration measures resulted in acuity similar to expert clinician derived refractions, suggesting the objective method may be a suitable alternative for patients with Down syndrome.

Individuals with Down syndrome often have reduced visual acuity, even when corrected with spectacles.^1,2 Previous work suggests that standard spectacle prescribing techniques may not be optimal for this population and that optical deficits (including elevated levels of higher order aberration³) may play a role in reduced acuity.^4,5 Identifying the optimum spectacle correction for patients with Down syndrome can be challenging due to high refractive errors,⁶ large amounts of astigmatism,⁷ elevated higher order aberrations,³ and intellectual disability limiting participation in the subjective refraction process.

While elevated higher order aberrations cannot be corrected with sphere and cylinder spectacle lenses, the interactions between residual lower order aberrations and higher order aberrations play a role in the resultant image quality.^8–10 For this reason, objective refraction techniques that consider the combined impact of residual lower and higher order aberrations on image quality may better serve this population.⁴ One such technique is metric-optimized refraction.¹¹ This technique mathematically applies a large number of sphero-cylindrical refractions to uncorrected wavefront error measurements of the eye to calculate the resultant image quality with each refraction.^{11, 12} Changes in two such metrics, Visual Strehl Ratio and Pupil Fraction Tesselated,¹³ have been shown to correlate with change in visual acuity,^{14, 15} and thus prescribing refractions that optimize Visual Strehl Ratio or Pupil Fraction Tesselated may result in better visual outcomes for patients with Down syndrome than current clinical methods.

This trial quantified visual acuity outcomes in adults with Down syndrome with spectacles dispensed for each of three refraction methods: standard clinical refraction, Visual Strehl Ratio optimized refraction, and Pupil Fraction Tesselated optimized refraction. Specifically, this study accomplished the following objectives:

1. Measure visual acuity outcomes of metric optimized spectacle prescriptions versus clinically derived prescriptions at an initial dispense visit;

2. Measure visual acuity outcomes of metric optimized spectacle prescriptions versus clinically derived prescriptions after two months of spectacle wear;

3. Measure spectacle compliance (average hours of wear) of metric optimized spectacle prescriptions versus clinically derived prescriptions over two months.

The detailed study methods and baseline characteristics of the study participants have previously been published.¹⁶ The results for the objectives listed above, as well as some additional study outcomes, are reported here.

METHODS

This study is a six-armed, double masked, randomized clinical trial with a triple cross-over design to compare outcomes with three spectacle prescription methods. Participants were randomized to wear each of the three corrections for a period of two months each. The study was approved by the University of Houston Committee for the Protection of Human Subjects and conducted at the University of Houston, College of Optometry. Parental/guardian permission was obtained for all participants, followed by participant assent. This study is registered on ClinicalTrials.gov (NCT03367793), in accordance with NIH policy. A data and safety monitoring board was not assigned to this trial, given the low risk of the intervention (spectacles), but a data and safety monitoring plan was submitted and participant safety monitored locally by the unmasked investigator. The information reported here conforms to the CONSORT guidelines.

Detailed descriptions of sample size calculation, participant recruitment, and eligibility screening have previously been published, along with baseline characteristics and study methods.¹⁶ In short, adults with a previous diagnosis of Down syndrome who were 18 years of age or older where eligible to participate, so long as they were developmentally able to sit for study measures and did not have ocular conditions (e.g. significant corneal scarring, cataracts, or nystagmus) that would result in difficulty obtaining optical imaging or limit best corrected acuity potential. A minimum level of refractive error was not a requirement for participation in the study due to the hypothesis that uncorrected higher order wavefront error is negatively impacting retinal image quality and thus the optimum sphero-cylindrical correction to balance the impact of the higher order aberrations may not follow expectations for the traditional correction of refractive error (e.g. an under- or over-correction of sphere and cylinder may provide the best image quality). In total, 30 adults with Down syndrome aged 18 – 52 years (mean = 29 ± 10 years) were randomized into one of six treatment arms (5 participants in each arm) to wear each of three spectacle prescriptions for 2 months each (Figure 1). Study recruitment began January 26, 2018, and the final primary outcome measure was collected on June 10, 2019. Participants included 21 habitual spectacle wearers and 9 who presented to the study unaided. Baseline binocular distance visual acuity with presenting correction (unaided if none) was collected at the initial study visit. Nine participants had strabismus, but no participants had monocular amblyopia (defined as 3 lines or greater difference in corrected visual acuity between the eyes). The 3-line difference definition of amblyopia was based on the common eligibility criteria for participation in clinical trials for the treatment of unilateral amblyopia conducted by the Pediatric Eye Disease Investigator Group.¹⁷

Figure 1.

CONSORT diagram depicting participation enrollment, randomization, and study completion. PFSt = pupil fraction tessellated; VSX = visual Strehl ratio in the spatial domain

Developmental ability was assessed with The Vineland Adaptive Behavior Scales Second Edition (NCS Pearson, Inc., Ontario, Canada). This assessment was selected rather than a cognitive assessment test given that it is appropriate for all ages, can be completed by the caregiver, and provides a holistic assessment of abilities in major categories of daily function. The survey was completed on paper by the parent/guardian and included categorical assessments of communication, daily living skills, socialization, and motor skills, which combined to provide an assessment of overall adaptive behavior represented by a standard score on a scale from 20 – 160. Standard scores falling 2 or more standard deviations below the mean (<70) were classified as Low adaptive functioning, scores falling between 1 and 2 standard deviations below the mean (70 – 85) were classified as Moderately Low adaptive functioning, and scores falling within 1 standard deviation of the mean (85 – 115) were classified as Adequate.

All participants underwent a comprehensive ocular examination, during which a single best clinical refraction was determined by a single, expert examiner with greater than 30 years of experience examining individuals with special needs. This examiner utilized common clinical techniques such as retinoscopy (pre and post cycloplegia), Grand Seiko autorefraction (Luneau Technology, Bensenville, IL), subjective refraction, and monocular estimate method (MEM) retinoscopy for assessment of accommodation. The determination of the final clinical prescription was left solely to the examiner’s discretion, as well as the decision of whether to prescribe a bifocal.

At the comprehensive examination, wavefront aberration measures were obtained with the COAS-HD wavefront aberrometer (Johnson & Johnson Vision, Santa Ana, CA) 30 minutes post-instillation of Tropicamide 1% and Phenylephrine 2.5%. A total of 3 – 5 good quality measures per eye were re-sized to the participant’s habitual pupil diameter (determined from infrared photorefraction measures in dim illumination), and averaged. A custom software program applied refractions to the average wavefront measure using a search range of 20,000 or more sphero-cylindrical combinations from at least ±3.00 DS in 0.25 DS steps surrounding the participant’s habitual sphere correction, and at least 0.00 to −4.00 DC in −0.25 DC steps (greater in cases of high habitual cylinder) for the entire range of cylindrical axes in 1-degree steps. For each of these applied refractions, the software calculated the residual wavefront error of the eye plus correction, as well as the resultant values of two separate metrics, Visual Strehl Ratio and Pupil Fraction Tesselated. Refractions were sorted by each respective metric and the refraction providing the best value for each metric were identified as the metric-optimized refractions. If the clinical examiner prescribed a bifocal power for a given participant, the same added bifocal power was prescribed for the metric-optimized refractions.

All three prescriptions were produced using a common spectacle frame selected by the study participant, yielding three duplicate frames each holding a different study treatment. Randomization order for the treatments was assigned prior to the second study visit, as previously described in the baseline and methods paper,¹⁶ and the un-masked examiner marked each frame with the treatment order. All three treatments were evaluated at the second study visit by a masked examiner to obtain initial visual acuity with each treatment (method described below). As described previously,¹⁷ prescriptions that resulted in greater than 7 letters poorer visual acuity than the presenting acuity were not dispensed. Of the 90 treatments produced for the study, only 1 clinically determined refraction was not dispensed due to reducing binocular distance visual acuity by 8 letters as compared to presenting acuity. All other spectacles were dispensed for 2 months of wear each, with participants returning at monthly intervals for a masked, clinical exam to assess adverse events, spectacle compliance, visual acuity, participant perceptions, and additional clinical measures described in the baseline publication.¹⁶ No participants discontinued the study, or prematurely stopped any of the treatments dispensed. The specific safety criteria, protocol for dispense, and interim stopping criteria have previously been published.¹⁶

The primary outcome of the study was binocular distance visual acuity obtained 2 months post dispense of each treatment. Binocular acuity was chosen over monocular acuity as the primary outcome given that people typically function in their daily environment with both eyes open and the goal of this study was to improve vision for daily function. Acuity was measured with a computerized logMAR style chart with 5 letters per line. Participants identified letters (either verbally or by pointing to a matching card) beginning at 0.8 logMAR and continued with decreasing logMAR lines in 0.1 logMAR steps until 5 total mistakes were made. Five total mistakes was selected as the standardized endpoint due to its successful and repeatable use in past studies,^{14, 16, 18, 19} as well as to avoid creating excessive frustration for the participants if they were required to read letters until missing an entire row for multiple acuity tests in a single study visit. The letter-by-letter scoring method was used with each letter equal to 0.02 logMAR. Twenty-nine participants were tested with charts composed of a 10-letter set of the British standard 1968 recommended letters and one participant was tested with a chart composed of a 4-letter set (H, O, T, V) due to difficulty reliably responding to the 10-letter set. The construction of the two chart types was identical with 5 letters per line and the same testing protocol was applied. Monocular distance visual acuity was obtained prior to binocular acuity with the order of eye tested randomized. Acuity measures were performed by a total of three individuals at each visit: a masked examiner who stood next to the participant and solicited responses, a masked examiner who stood by the monitor to point at letters and key in responses, and an unmasked examiner who wrote down responses to serve as a back-up to electronic data entry. The visual acuity protocol was consistently applied for all participants and study visits. Two different optometrists were trained to serve in the role of soliciting responses from the participant to allow greater flexibility in scheduling.

A secondary outcome of the study was spectacle wear time quantified in average hours per day for the 2-month dispense period for each treatment. Wear time was monitored objectively with the Smart Button data logger (ACR Systems Inc., Surrey, BC Canada) temperature sensor mounted in a silicone sleeve to the temple of the spectacles. Temperature versus time plots obtained from the data logger were scored by two investigators to calculate the average daily wear time.²⁰

Participant perceptions of each spectacle treatment also served as a secondary outcome. Perceptions were obtained with a survey read to study participants by a masked examiner at the outcome visit for each treatment, occurring 2 months after dispensing. Surveys included three questions rated on a five-item scale to determine participant perceptions about distance and near vision, and whether the participant ‘liked’ wearing the spectacles. Responses were obtained using a pictorial rating scale showing two levels of frowning faces, a neutral face, and two levels of smiling faces. A fourth question, ‘Do you see better with these glasses than without glasses?’ was answered ‘yes’ or ‘no’ after showing the participant their vision both with and without glasses and asking which was better.

Data Analysis

Initial distance visual acuity at treatment dispense and two-month adapted distance visual acuity outcomes were analyzed using a mixed-effects linear modeling approach to compare differences in visual acuity among the three spectacle prescription types. For this analysis, sequence and period were fixed effects and participant was a random effect to account for within-subject and between-subject variability, as well as to evaluate period and period-by-treatment effects. An overall F test, which accounts for the co-variance structure of the variance-covariance matrix, was used in SAS (PROC GLIMMIX; SAS Institute, Cary, NC) to test the overall effect of treatment. Tukey post hoc analysis was used to further elucidate differences between experimental prescriptions.

The same statistical approach as the visual acuity analysis was used to analyze spectacle compliance. Patient perceptions on the survey were analyzed as a binary, categorical response to have sufficient power to detect differences with a sample size of 30 and the 3×3 crossover design. To create a binary response, scores from 4 to 5 were assigned satisfactory and all others (1–3) scored not satisfactory. Linear correlation was used to determine whether participant developmental ability (Vineland adaptive behavior standard score) was predictive of two-month treated binocular distance visual acuity outcomes, controlling for treatment type.

While the primary outcome was binocular visual acuity, analysis was also performed for monocular visual acuity. Monocular visual acuity was not a planned secondary outcome, but given that some study participants had alternating strabismus, it is feasible that the participant switched fixation during binocular visual acuity testing which could have impacted the overall findings. Thus, analysis was performed for monocular visual acuity using a similar approach as described above, but with visual acuity measures from both right and left eyes included while controlling for inter-eye correlations.

RESULTS

Baseline binocular distance visual acuity with presenting correction for the cohort averaged 0.39 ± 0.17 logMAR (range = 0.12 to 0.80). The standard scores for the Vineland Adaptive Behavior test ranged from 22 to 99 with 18 participants classified as having low adaptive functioning, 11 as having moderately low adaptive functioning, and 1 as having adequate functioning.

Initial Visual Acuity with Treatment at Dispense

Initial, non-adapted, distance visual acuity was obtained with each of the three treatments in the randomization order at the second study visit prior to dispensing the first randomized treatment for extended wear. Based on the linear mixed-effect model, the initial visual acuity (logMAR) was 0.35 (95%: 0.28, 0.41) for Pupil Fraction Tesselated refractions, 0.33 (0.26, 0.39) for Visual Strehl Ratio refractions, and 0.34 (95% CI: 0.28, 0.41) for clinical refraction. There was no statistically significant effect of treatment at the initial dispense (F=0.93, P=.410).

Study Visit Completion

For each treatment dispensed, participants were scheduled 28 ± 7 days after dispensing for the interim follow-up and 56 ± 14 days after dispensing for the primary outcome visit. All study participants completed both visits for each treatment dispensed. Of the 89 outcome visits, only 1 visit was completed out of window (86 days post initial dispense) due to the glasses breaking during the first week of the treatment period. For this participant, it took 3 weeks to obtain and dispense the replacement glasses, after which the treatment period was started over.

Primary Outcome Measure Adapted Distance Visual Acuity

Change from habitual binocular distance visual acuity with each treatment type is shown for each participant in Figure 2. Acuity gains with at least one of the treatments prescribed were achieved in 87% (n=26) of the participants. Half of participants (n=15) achieved an acuity gain of at least 1 line or more with one or more of the treatments.

Figure 2.

Individual participant change in visual acuity from habitual presenting acuity with each of the three treatment types dispensed in the clinical trial. Change in acuity is calculated based on binocular distance acuity after two months of treatment wear. One of 30 clinical treatments was not dispensed due to failure of safety criteria.

The estimated least square means in binocular visual acuity and monocular visual acuity (logMAR) following adapted wear time based on the linear mixed-effect model are shown in Table 1. A descriptive summary of visual acuity reported by randomization group for each treatment is also shown in Appendix Table A1. There was no significant treatment effect observed (F=1.10, P-value=.341) for binocular acuity. There was also no significant linear correlation between the Vineland developmental ability standard score and binocular acuity (F = 0.11, P-value = .736). For monocular acuity, there was no significant effect of eye (F=0.04, I-value=0.85), but there was a significant effect of treatment (F=5.00, P-value=.01) with Visual Strehl Ratio refractions resulting in acuity with a 0.04 logMAR improvement over the other two treatments (Pupil Fraction Tesselated versus Visual Strehl Ratio t=2.73, p-value=0.03; Visual Strehl Ratio versus Clinical t=−2.74, P-value=.02).

Table 1.

Visual acuity outcomes following 2 months adapted wear time for each treatment type.

Viewing Condition	Treatment	Estimate (logMAR)	95% Confidence Interval
Binocular Acuity	PFSt	0.31	0.26, 0.36
	VSX	0.33	0.27, 0.38
	Clinical*	0.34	0.28, 0.39
Monocular Acuity	PFSt	0.41	0.35, 0.46
	VSX	0.37	0.31, 0.42
	Clinical*	0.41	0.35, 0.47

PFSt = pupil fraction tessellated; VSX = visual Strehl ratio in the spatial domain

^*One of 30 clinical treatments was not dispensed due to failure of safety criteria

Secondary Outcome Measure Spectacle Wear Time

Spectacle wear time was variable, but all participants wore each treatment for some portion of the day. Average wear time ranged from approximately 2 hours per day to as much as 18 hours per day across participants and treatment types. A descriptive summary of spectacle wear time reported by randomization group for each treatment is shown in Appendix Table A2. There was no statistically significant effect of treatment on average wear time (hours per day) (F=1.09, P=0.35). Based on the linear mixed-effect model, the estimated wear-time in hours per day was 11.0 (95%: 9.3, 12.7) for Pupil Fraction Tesselated, 10.9 (9.2, 12.6) for Visual Strehl Ratio, and 11.2 (95% CI: 9.5, 12.9) for clinical refractions (Table 2).

Table 2.

Objectively measured spectacle wear time in hours per day for each treatment type.

Treatment	Estimate (hours per day)	95% Confidence Interval
PFSt (n=30)	11.0	9.3, 12.7
VSX (n=30)	10.9	9.2, 12.6
Clinical (n=29)*	11.2	9.5, 12.9

PFSt = pupil fraction tessellated; VSX = visual Strehl ratio in the spatial domain

^*One of 30 clinical treatments was not dispensed due to failure of safety criteria

To explore whether spectacle wear time related to the magnitude of a participant’s refractive error, average wear time was plotted by refractive error measures obtained from non-dilated, distance Grand Seiko autorefraction (Figure 3). Due to the large magnitudes of astigmatism present in this group, comparisons were plotted for both spherical equivalent (sphere power + ½ cylinder power) (Figure 3A) and cylinder power alone (Figure 3B). As would be predicted, participants with the largest magnitudes of spherical equivalent refractive error (either plus power or minus power) tended to wear the spectacle treatments for the most hours per day, with increasing variability in spectacle wear across participants with lesser amounts of spherical equivalent refractive error. This same trend was not observed with respect to cylinder magnitude.

Figure 3.

Participant wear time for each treatment type by spherical equivalent (A) and cylindrical (B) refractive error. Refractive error powers for this analysis were obtained from distance autorefraction measures. PFSt = pupil fraction tessellated; VSX = visual Strehl ratio in the spatial domain.

Secondary Outcome Measure Subjective Quality

Overall, there were no statistical differences in satisfaction across treatments among items on the spectacle comparison survey administered at the two-month visit for each treatment type (Table 3). Satisfaction was high among the study cohort for all treatment types.

Table 3.

Number of participants responding with a satisfactory rating for items 1–3 and a ‘yes’ for item 4.

Survey Item	PFST N (% out of 30)	VSX N (% out of 30)	Clinical N (% out of 29)*	P-value
1. Do you like wearing this pair of glasses?	29 (96.7)	28 (93.3)	26 (89.7)	0.56
2. How well do you see with this pair of glasses when looking up close?	29 (96.7)	29 (96.7)	28 (96.6)	>0.99
3. How well do you see with this pair of glasses when looking far away?	29 (96.7)	29 (96.7)	28 (96.6)	>0.99
4. Do you see better with this pair of glasses than without glasses (yes/no)?	29 (96.7)	30 (100)	26 (89.7)	0.359

^*One of 30 clinical treatments was not dispensed due to failure of safety criteria.

Adverse Events

Adverse event data were systematically collected through phone calls 1 day and 1 week after dispense of each new treatment, and at the start of each study visit. Participants and their family member were asked to report changes in health or medication use, concerns about vision, or concerns about the glasses. No participants in the study reported serious adverse events, but there were 36 reports of non-serious adverse events. Fifteen of the non-serious adverse events were classified by the study team as ‘unrelated’ to study treatment and included complaints such as respiratory symptoms. Eleven non-serious events were judged ‘probably’ related to study treatment and 10 judged ‘possibly’ related. The most common non-serious adverse events (n = 14) were comments related to a preference for a prior correction (either study treatment or presenting) due to a difference in some aspect of vision with the newly dispensed treatment. This was reported for 12 different participants in the study in reference to at least one of their treatments (9 reports when switching to clinical, 2 when switching to Pupil Fraction Tesselated, and 3 when switching to Visual Strehl Ratio). One participant also reported eyestrain and headaches when wearing the Pupil Fraction Tesselated correction. None of these events was severe or persistent enough to necessitate discontinuation of treatment.

DISCUSSION

This study evaluated three prescription types (1 clinical and 2 objective) in adult patients with Down syndrome and found no significant difference in binocular distance visual acuity outcomes between treatment types at both initial dispense and after 2 months of treatment wear. While the clinically derived prescription was considered the standard of care method, it should be noted that the examiner performing the clinical refractions had significant expertise working with individuals with special needs and thus is unlikely representative of the average practitioner when examining this population. Thus, the equivalent performance of all three treatment types demonstrates that the objective method is a promising tool in determining refractions for this population and might benefit patients examined by practitioners with less experience. While this study provides some proof of concept for the use of metric-optimized refractions for patients with Down syndrome, additional work is needed before this treatment strategy could be readily adopted in clinical practice. Most notably, the current software that analyzes images can take multiple hours to search a refractive correction range large enough to identify the metric-optimized refraction, which would make it unlikely that a metric-optimized refraction could be offered the same day as an initial vision examination.

Binocular distance visual acuity was selected as the primary outcome measure given acuity is commonly used to quantify vision and that the binocular testing condition reflects the normal, two-eyed viewing environment. Monocular visual acuity measures did show a significant effect in favor of Visual Strehl Ratio optimized refractions, but the average acuity gain of two letters was not clinically significant. We suspect that these differences were observed due to the controlled monocular viewing environment versus binocular testing whereby a participant with alternating strabismus could switch fixation during testing.

Spectacle wear also did not differ by treatment type; however, given that participants only had one prescription available to them at a time, the alternative to wearing a prescription was not wearing any prescription. Therefore, the data regarding spectacle compliance cannot be interpreted as a participant’s preference for one prescription over another, but instead demonstrates consistent treatment compliance throughout the duration of the study.

One would expect the magnitude of refractive error to correlate with uncorrected visual acuity,²¹ and thus spectacle wear could be expected to increase for patients whose refractive corrections are larger magnitude. We did observe a trend of greater spectacle wear with greater spherical equivalent refractive error, both myopic and hyperopic, but did not observe the same trend relative to the magnitude of astigmatism. It is challenging to analyze astigmatism in isolation; however, because it is possible that patients with higher amounts of astigmatism also had spherical equivalent refractive error closer to zero. It is also conceivable that not only the magnitude, but also the orientation of the astigmatism plays a role in spectacle compliance. An exploratory analysis to calculate average wear time by astigmatism category (with the rule, against the rule, and oblique) showed no difference in spectacle wear by astigmatism category; however, this analysis did not account for astigmatism magnitude.

Although high contrast acuity provides quantitative information about visual performance, it does not capture the qualitative aspects of the visual experience, such as perceptions of image blur, ghosting, or the level of contrast of an image. In other words, equivalent visual acuity does not necessarily mean equivalent visual performance. Our previous simulation studies conducted with aberration measures obtained from the eyes of individuals with Down syndrome demonstrate that these eyes have sizeable quantities of blur, ghosting, and contrast reduction in the predicted retinal image.²² When control observers read charts derived from these aberration measures and rated perceived image quality, our analysis demonstrated that blur and ghosting influence acuity more than contrast; however, contrast still plays an important role in perceived image quality.²² While we attempted to capture subjective feedback about each treatment type from the adults with Down syndrome in the present study, satisfaction was overall high with each treatment and did not yield any discriminative findings related to treatment type. The adverse event reporting did point to some initial participant preferences for one treatment over the other that occurred when switching to a new prescription; however, this experience is not uncommon when adapting to a new correction. We are limited in our ability to make conclusions about subjective preferences based on the information participants offered, or were able to communicate. While this limitation is true of any study seeking participant feedback, it may have been magnified in our study due to our participants’ intellectual disabilities. Side by side comparisons and ratings of each treatment were performed later in the study to inform clinical decision making about which prescription to dispense for long-term wear, and measures of contrast sensitivity were performed at the adapted outcome visits for each treatment. These data will be presented in future publications.

Most study participants benefitted from participation in this trial in that one or more of the treatments provided a gain in acuity over their presenting correction. That said, the final level of acuity achieved after two months wear was still reduced compared to age-matched, typically-sighted controls. The participants in this study all have elevated levels of higher order aberrations which likely contribute to the overall reduction in visual acuity with standard spectacle corrections.¹⁶ However, another possible explanation for the reduced visual acuity levels is long-standing amblyopia. Individuals with unilateral amblyopia (greater than three lines difference between eyes) were excluded from participation, but the presence of bilateral amblyopia cannot be ruled out for this cohort given that historical best acuity performance during childhood is unknown. The treatment for bilateral amblyopia is refractive correction (as was prescribed in this clinical trial), but it is well documented that success in treating amblyopia is greatest in childhood and diminishes in adulthood, demonstrated by the increasing odds ratios for 5 age bins spanning 0–3 years up to >20+ years published in a meta-analysis of 961 amblyopic patients from 23 previously published studies.²³ Previous studies demonstrate that elevated higher order aberrations and abnormal corneal topography findings are present at a young age in individuals with Down syndrome,^{3, 24} and thus applying wavefront optimized corrections to younger participants may help to determine whether reduced neural plasticity limited the ultimate visual performance of the participants in the present study.

A strength of the present study is the 100% completion of study visits and participant retention which we attribute to strong study team staff support and an outstanding commitment to and interest in the scientific process on the part of the participant families. In addition, wavefront aberration measurements, corneal topography measurements and a rigorous visual acuity protocol were completed on all participants, despite a large range of developmental ability.¹⁶ Most participants tolerated treatments well and had excellent treatment compliance. While there may have been selection bias because participants volunteering for the study had previously had eye examinations and families were motivated to try spectacle corrections, there were some participants who presented to the study unaided and had past resistance to wearing spectacles. One of these participants became a regular, full-time spectacle wearer, suggesting to us that past poor compliance with spectacles does not necessarily indicate future failure to wear spectacles.

The results of this study are based upon a cohort of adults with Down syndrome with a relatively large range of developmental abilities. Some of the participants had limited speech with significant developmental delays, whereas others were relatively high functioning and pursuing post high school education or working full-time service industry jobs. Thus, the findings should be generalizable to the spectrum of intellectual disability observed with Down syndrome. Individuals with nystagmus were excluded from participation, despite nystagmus being a relatively common finding in individuals with Down syndrome.^{25, 26} This decision was made based upon the need to enroll participants who would be able to fixate accurately to obtain the wavefront aberration measures required to derive the objective refractions. Thus, the findings of this study cannot be extended to those individuals with nystagmus. The findings of this study may also be applicable for other populations that are challenging to refract, such as those with intellectual disability from other conditions, or those with elevated aberrations in the absence of intellectual disability.

One limitation of the present study is that the clinical examiner determined the clinical prescription while considering that the patient would be viewing with two eyes. The metric-optimized refractions, however, were determined for each eye individually without consideration for binocular viewing. Clinical practice involves adjusting the final prescription to account for differences in accommodation that may occur during monocular testing and thus balance the stimulus to accommodation under binocular viewing. In addition, the clinician had knowledge about the binocular status of the patients and knew that 9 of the patients had strabismus at both distance and near viewing.¹⁶ The metric-optimized refractions were determined from wavefront aberration measures obtained post-dilation, and thus monocular differences in accommodative tonus during wavefront measures should not have occurred, making the binocular balance procedure of less concern. That said, many of the study participants had hyperopic refractive error, a common refractive finding in Down syndrome,⁶ and the metric-optimized refractions would not have incorporated the clinical practice of binocularly ‘cutting-back’ a hyperopic correction to account for baseline accommodative tonus to improve outcomes.²⁷ Future studies that allow a clinical examiner to modify (or potentially subjectively refine) the metric-optimized refraction may provide the best visual outcomes.

A second limitation of the present study is that the clinical prescriptions were determined from measurements encompassing a range of pupil diameters (both with a natural and dilated pupil), whereas metric-optimized prescriptions were determined from a fixed pupil diameter (that was unique for each subject). The practice of utilizing both dilated and non-dilated findings in clinical refraction replicates the standard of care, whereas the metric-optimized refraction method requires the selection of a single pupil diameter to complete analysis, and thus the two practices were necessarily different with respect to considerations of pupil diameter. However, prior to conducting the clinical trial we performed analysis comparing metric-optimized refractions from wavefront aberrations re-sized to 4mm and 6mm pupil diameters from adults with Down syndrome and found that the dioptric difference between the resultant refractions was approximately 0.50D, on average, for the group.²⁸ These small differences in refraction based on differences in pupil diameter would not be expected to impact the findings of this study.

CONCLUSIONS

Overall, this study demonstrates that objective, wavefront-optimized refractions provide distance visual acuity equivalent to expert clinician refractions in adults with Down syndrome with both myopic and hyperopic refractive errors. With additional methodological development, metric-optimized refraction may be a beneficial tool in the future for practitioners who are not experienced in the examination of adults with Down syndrome, or who wish to have additional refractive data to determine their prescriptions.

Supplementary Material

Appendix tables A1-A2 final

Appendix Table A1. This table provides comparisons of primary outcome data by randomization group by showing the mean and standard deviation binocular distance visual acuity (logMAR) at two month outcome visit by randomization group and treatment type.

Appendix Table A2. This table provides comparisons of secondary outcome data by randomization group by showing the mean and standard deviation spectacle wear time (hours per day) averaged for 2 months of wear by randomization group and treatment type.