Initial assessment of a binocular handheld autorefractor in a pediatric population

1.Introduction

Measuring refractive error in children can be challenging due to several factors such as accommodation during the measurements or lack of cooperation1 . To deal with accommodation, cyclopegic drugs are commonly employed, which lengthens the overall vision exam and causes discomfort for the patients. The objective of this study is to evaluate the performance of measuring refractive error in a non-cyclopegic pediatric population with the QuickSee autorefractor (PlenOptika Inc. MA, USA).

2. Methods

QuickSee The QuickSee is a novel open-view, binocular handheld wavefront aberrometry-based autorefractor. It records 10 second videos while the patient holds and looks through the device2 . The system is able to process data of both eyes in real time to obtain a set of measurements from which the most suitable prescription is calculated using statistical methods. An important difference between the QuickSee and other refraction systems is the detection of changes in accommodation during the measurement in order to provide objective refraction results with minimal influence of accommodation. Specifically, if moderate accommodation is present, the software compensates for it, and if the accommodation is severe the software provides an error message suggesting the measurement be retaken. Experimental Protocol Participants were recruited from Hospital Fundación Jiménez Díaz (Madrid, España). The criteria for inclusion were (1) between 5 and 16 years old, (2) best-corrected visual acuity (VA) of -0.3 LogMAR units (6/12) or better in each eye and (3) a refractive error between -11 D and +10 D of Spherical Equivalent (SE) as determined by subjective refraction (SR). Study subjects first underwent an objective measurement with a desktop autorefractor (Topcon KR 8800) followed by subjective refinement with trial lenses and visual acuity measurement with the resulting prescription. After this initial step, a different optometrist performed three recordings of the patient using the QuickSee and measured the VA based on the objective refraction measurements. Data Analysis To evaluate the agreement between the QuickSee and SR we performed a Bland-Altman analysis of the power vectors M, J0, and J45 obtained with each method. Differences with SR were also evaluated by determining the number of eyes in which agreement between both refraction techniques was within 0.25 and 0.5 D thresholds (S, C, M, J0, and J45) and 5 and 15 degrees for the axis of astigmatism. Lastly, VA achieved with each refraction method was translated into LogMAR units and analyzed for further comparison. The analysis uses the measurement number that the optometrist used for measuring objective refraction-based VA, which was the measurement number showing the median result for SE. Patients without at least a valid measurement for each eye with the QuickSee were excluded from the data analysis. The remaining videos are intended to be used for further validation of the device (e.g. repeatability) and for refinement of the data-processing algorithms.

3. Results

Of the 27 study participants, 3 were excluded. Two of these patients were excluded because they were uncooperative during QuickSee refraction (the total study time for these patients was up to 3.5 hours and at this point they indicated that they were tired to continue with the study). The other was excluded because after 3 measurement attempts with the QuickSee a valid video of each eye was not able to be obtained due to an error message indicating accommodation. The average age of the 24 patients finally enrolled was 9.74 ± 2.34 years (ranging from 6 to 14 years) with a non-corrected SE ranging from -7.25 to 3.5 D. Average SE refraction for these patients was -0.33 ± 1.94 D and -0.40 ± 1.72 D for SR and QuickSee, respectively, and no statistically significant differences were found for SE between both refraction methods (p = 0.4). The Bland-Altman analyses showed a bias between SR and QuickSee of 0.07, 0.02, and 0.01 D (Figure 1) for M, J0, and J45, respectively, with 95 % limits of agreement of ± 0.98, ± 0.37, and ± 0.31 D. It is interesting to note that these results were 0.19, 0.02, and 0.02 D, and ± 1.09, ± 0.27, and ± 0.27 respectively, for the Topcon KR 8800 autorefractor used during the standard clinical refraction procedure. The QuickSee showed remarkable agreement with SR (Figure 2A) in steps of 0.5 D, obtaining accordance of 83.3% and 89.6 % for S and C, respectively. For the astigmatism axis, the QuickSee agreed with SR in 66.7 % and 96.7 % for ≤ 5 and ≤ 15 degrees, respectively. Lastly, the measurements given by the QuickSee reached 20/20 or better VA in 75.0 % of the eyes in comparison to 87.5 % obtained by SR. Furthermore, the QuickSee obtained a result equal to or better than SR in 81.3 % of the evaluated eyes. (Figure 2B).

4. Conclusions

The QuickSee obtained excellent preliminary results in terms of agreement with SR in a relatively challenging population (non-cyclopegic children with SE ranging from -7.25 to 3.5 D and average cylinder of -1.12 D). It is important to note, however, that this pilot study only provides moderate statistical power due to the small number of subjects tested so far. As a continuation of this research, the sample size is being increased to 100 children and will also include patients under the effects of cycloplegia, which will serve as an additional validation of this study. An interesting result of this analysis is that for SE refraction the QuickSee obtained slightly better numbers than the commercial Topcon KR 8800 autorefractor, which is surprising given that the KR 8800 measurement was used as starting point for SR. It is also interesting to note that the QuickSee detected significant accommodation in 8 videos from a total of 4 patients which is 6.0 % of all the videos acquired during this study with the QuickSee. This result together with the agreement with SR achieved in the patient population may indicate that the open view design of the device is effectively helping to control accommodation problem in most of the patients, however, further research is still required. Finally, this study has found an important increase in VA derived from QuickSee measurement which is consistent with the agreement found between QuickSee and SR. Specifically, the number of people able to achieve 20/20 VA increased from 33.3 % to 75.0 % using only the QuickSee measurement without SR.