Purpose
To assess the safety of VR 3D headset (virtual reality 3-dimensional binocular-stereoscopic near-eye display) use in young children. Product safety warnings that accompany VR headsets ban their use in children under age 13 years.
Design
Prospective, interventional, before-and-after study.
Methods
Recordings were obtained in 50 children (29 boys) aged 4-10 years (mean 7.2 ± 1.8 years). Minimum binocular corrected distance visual acuity (CDVA) was 20/50 (logarithm of the minimum angle of resolution [logMAR] 0.4) and stereoacuity 800 seconds of an arc or better. A Sony PlayStation VR headset was worn for 2 sequential play sessions (of 30 minutes each) of a first-person 3D flying game (Eagle Flight) requiring head movement to control flight direction (pitch, yaw, and roll axes). Baseline testing preceded VR exposure, and each VR session was followed by post-VR testing of binocular CDVA, refractive error, binocular eye alignment (strabismus), stereoacuity, and postural stability (imbalance). Visually induced motion sickness was probed using the Simulator Sickness Questionnaire modified for pediatric use (Peds SSQ). Visual-vestibulo-ocular reflex (V-VOR) adaptation was also tested pre- vs post-trial in 5 of the children. Safety was gauged as a decline or change from baseline in any visuomotor measure.
Results
Forty-six of 50 children (94%) completed both VR play sessions with no significant change from baseline in measures of binocular CDVA ( P = .89), refractive error ( P = .36), binocular eye alignment ( P = .90), or stereoacuity ( P = .45). Postural stability degraded an average 9% from baseline after 60 minutes of VR exposure ( P = .06). Peds SSQ scores increased a mean 4.7%—comparing pretrial to post-trial—for each of 4 symptom categories: eye discomfort ( P = .02), head/neck discomfort ( P = .03), fatigue ( P = .03), and motion sickness ( P = .01). None of the children who finished both trial sessions (94%) asked to end the play, and the majority were disappointed when play was halted. V-VOR gain remained unaltered in the 5 children tested. Three children (6% of participants) discontinued the trial during the first 10 minutes of the first session of VR play, 2 girls (aged 5 and 6 years) and 1 boy (aged 7 years). The girls reported discomfort consistent with mild motion sickness; the boy said he was bored and the headset was uncomfortable. No child manifested aftereffects (“flashbacks”) in the days following the VR exposure.
Conclusion
Young children tolerate fully immersive 3D virtual reality game play without noteworthy effects on visuomotor functions. VR play did not induce significant post-VR postural instability or maladaption of the vestibulo-ocular reflex. The prevalence of discomfort and aftereffects may be less than that reported for adults.
Introduction
Virtual reality (VR) is computer-generated simulation of a visual environment. Adults who engage in VR game play or VR training in vehicle simulators may experience eyestrain and/or visually induced motion sickness. The effects tend to be more pronounced when using an immersive VR 3D headset (a device that displays 3-dimensional stereoscopic scenes and precludes all visual orientation to the real world). Use of VR 3D technology for children’s entertainment has elicited concern that exposure may damage the immature visual system. Product safety warnings that accompany VR headsets ban their use in children under age 13 years. Quantitative studies as of this writing have not addressed VR headset effects specifically in a sizeable pediatric population. The purpose of this study was to assess the visuomotor effects of VR headset use in young children.
Subjects and Methods
Subjects and Inclusion/Exclusion Criteria
Recordings were obtained in 50 children (29 boys, 58%) aged 3-10 years (mean 7.2 ± 1.8 years). Inclusion criteria were as follows: logarithm of the minimum angle of resolution (logMAR) binocular corrected distance visual acuity (CDVA) 0.4 (Snellen 20/50) or better; stereo acuity of 800 seconds of an arc or better; and binocular eye alignment graded as orthotropic or no greater than 4 PD heterotropic. Nine of the children (18%) wore prescription glasses; the spherical equivalent refractive error (SEQRE) of these children was a mean +2.7 ± 1.6 diopter (D). Exclusion criteria were as follows: major neurodevelopmental disability; vestibulocochlear (cranial nerve 8) dysfunction; nystagmus; heterotropia (strabismus) greater than 4 PD, and/or stereoacuity less than 800 seconds of an arc. Familiarity with video game play was not a requirement, though all of the children reported some experience with 2D video game play. None of the children had experience using an immersive VR headset. Children were recruited with the aid of Washington University’s Volunteer For Health Office. All testing was conducted in the Visual Diagnostics Laboratory at St Louis Children’s Hospital. The study was approved by the Institutional Review Board of Washington University in St Louis and informed consent from parents for minor participants was obtained prior to study enrollment.
Study Design and Methods
A prospective, interventional single-subject design study was conducted in which each child acted as his or her own control. The intervention was exposure to immersive 3D VR. Baseline testing was conducted as follows: (1) binocular corrected distance visual acuity (CDVA) using an electronic early treatment diabetic retinopathy study protocol ; (2) refraction of each eye (dry) using a Plusoptix S16 (Plusoptix Inc, Atlanta, Georgia, USA) and Topcon KR-800 Auto Kerato-Refractometer (Topcon Medical Systems Inc, Oakland, New Jersey, USA); (3) stereoacuity measured by Titmus Stereo Fly or Randot (Stereo Optical Inc, Chicago, Illinois, USA); and (4) eye alignment assessed by cover-uncover and alternate-cover testing as the child viewed an accommodative target at 6 m. Binocular alignment was also measured using the interpupillary distance (strabismus) metric of the Plusoptix device.
Postural stability was quantified using Sway Balance software (Sway Medical LLC, Aledo, Texas, USA) on a triaxial accelerometer mobile device (iPad, Apple Inc, Cupertino, California, USA). The child held the iPad to its chest while performing a modified Romberg balance test. With eyes open, the child stood in 5 postures for 10 seconds each: (1) double-leg stance (both feet together), (2) tandem (heel-to-toe) left foot forward, (3) tandem right foot forward, (4) single-leg stance left foot, and (5) single-leg stance right foot. Thoracic sway scores for each stance were calculated by the device on a 0-100 scale, with 0 = most extreme sway and 100 = no sway (perfect balance).
Discomfort was assessed using a Likert scale Simulated Sickness Questionnaire (SSQ). The query language of the adult SSQ was simplified for children and modified Wong-Baker faces added to create a Peds SSQ ( Appendix 1 ). The Peds SSQ, like the adult SSQ, contains queries that probe 4 symptom categories: (1) eye strain (queries 1-4); (2) head and neck discomfort (queries 5 and 6); (3) sleepiness (solipsism) or fatigue (queries 7 and 8); and dizziness or nausea (visually induced motion sickness, queries 9-13). The Peds SSQ was administered to the child in the presence of the parent and addressed the experience of the child with or without the aid of the parent as proxy-reporter.
Measurement of the visual-vestibulo-ocular reflex (V-VOR) 23 was performed in 5 children (chosen randomly) using SMI 3D VOG corneal-reflection infrared eye-head tracking goggles and software (SensoMotoric Instruments, Teltow, Germany). The child viewed a stationary accommodative target (0.5°) binocularly located at 1.5 m straight ahead under normal room illumination. The child was instructed to rotate the head horizontally in sinusoidal, metronomic fashion (0.3 Hz) over an excursion of approximately 40° while maintaining continuous fixation on the target. Twenty cycles of rotation were recorded at a sampling rate of 60 Hz and minimum angle of resolution of 0.25°. Eye position was converted to eye velocity and linear head acceleration to head velocity. The gain (performance) of the V-VOR was calculated as peak eye velocity/head velocity. A gain of 1.0 represented perfect compensation of eye rotation evoked by the head rotation.
VR Headset, Visual Stimulus and Trial Design
A Sony PlayStation VR headset (Sony Interactive Entertainment LLC, San Mateo, CA) was worn ( Figure 1 ) for 2 sequential trial block play sessions (of 30 minutes each) of a first-person 3D flying game (Eagle Flight, Ubisoft Entertainment, San Francisco, California, USA) requiring head movement ( Figure 2 ) to control flight direction (pitch, yaw, and roll axes). Baseline testing preceded VR exposure, and each VR session was followed by post-VR testing (approximately 10 minutes in duration) of the measures described above ( Figure 3 ). The PlayStation VR headset was chosen because it is the most popular current VR headset, comparable technically to other commercial VR headsets, and the Sony PlayStation platform has an array of video games appropriate for children. The headset (1.3 lb) is equipped with a binocular organic light-emitting diode (OLED) display (90-120 Hz, 1080 × 960 resolution) providing a field-of-view surround of 100°. The headset contains an accelerometer-gyroscope magnetometer to precisely track head motion. The headset is equipped with adjustment for interpupillary distance and allows spectacle wear for children with glasses. It blocks out all visual orientation to the real world, providing an immersive VR experience. The Eagle Flight game required the child to fly above and through architectural obstacles in the city of Paris, France, with other birds. The child used a hand-controller to modify the speed of flight and emit sonic blasts (shrieks) at competing birds. A video component—showing a representative segment of the Eagle Flight game—is available and accompanies the online version of this manuscript. To access the video component, simply click on the image ( Figure 4 ) visible below (online version only).
Data Analysis
Differences between means of baseline measures and measures obtained after 30 and 60 minutes of VR exposure were compared using a 1-tailed paired samples t test for binocular logMAR CDVA, SEQRE (in diopters), stereopsis (converted to logarithm of seconds of arc ), binocular eye alignment (in prism diopters), postural stability-sway score, and V-VOR gain. Both a 2-tailed t test and Mann-Whitney test were used to compare Peds SSQ responses. Because none of the measures revealed substantial differences between the results obtained at 30 and 60 minutes, the statistical results reported are differences between those measured at baseline and those measured after 60 minutes of VR exposure. Significance was defined as P ≤.05.
Results
Forty-seven of the 50 children (94%) completed both of the 2 successive 30-minute VR play sessions. All of the children had some exposure to 2D video game play on computer video displays or handheld devices, but none had used a VR headset for immersive 3D game play. Their visuomotor measures are shown in the graphs of the results and reported as visual acuity, refractive error, binocular eye alignment, stereoacuity, balance, and simulator sickness symptoms. The visual vestibulo-ocular reflex was tested in 5 of the children (chosen at random) before and after VR exposure.
Corrected Distance Visual Acuity
To determine whether VR play caused a degradation of distance visual acuity in the children, we measured binocular CDVA before and immediately after each viewing session. As shown in Figure 5 A, CDVA for the group at baseline was an average 0.041 ± 0.03 logMAR (Snellen 20/22). After 30 minutes of VR exposure, CDVA averaged 0.040 ± 0.03 logMAR and after 60 minutes, 0.021 ± 0.03 logMAR (Snellen 20/21). CDVA remained unaltered, equivalent to baseline after an hour of VR play ( P = .14).
Refractive Error
Display screens in VR headsets are positioned at an average vertex distance of approximately 40 mm from each eye. Though special optics are employed to project the display to a viewing distance of approximately 2 m, awareness of the near-eye display might be expected to induce accommodative spasm (transient myopia) in some children. To assess the presence or absence of this phenomenon, in addition to CDVA, refractive error (SEQRE) was measured for each eye. At baseline ( Figure 5 B), SEQRE in the right eye measured a mean +0.60 ± 0.90 D and after 60 minutes of VR exposure +0.48 ± 0.41 D. For the left eye, SEQRE was a mean +0.53 ± 0.90 D and after the 60 minutes of VR play, +0.61 ± 0.57 D. SEQRE did not alter from baseline after 60 minutes of VR exposure for the right ( P = .36) or left eye ( P = .41).
Binocular Eye Alignment
To assess for any heterophoria or heterotropia evoked by VR play, binocular eye alignment was measured viewing a distance accommodative target. Any esodeviation was plotted as negative values and exodeviations as positive values. As shown in Figure 5 C, alignment at baseline was an average esophoria of –0.51 ± 0.81 PD. After 60 minutes of VR play, the average esophoria measured –0.38 ± 0.70 PD ( P = .96).
Stereoscopic Vision
Stereoscopic vision is dependent critically on proper binocular eye alignment. As an additional measure for any disruption of binocularity caused by VR play, we measured stereoacuity in each child. Stereoacuity at baseline was an average 80 ± 14 seconds of an arc ( Figure 5 D). After 60 minutes of VR exposure, stereoacuity measured a mean 95 ± 24 seconds of an arc, which did not differ significantly from baseline ( P = .45).
Balance/Postural Stability
A concern of VR viewing in normal adults is transient imbalance (postural instability or sway) induced by spatial disorientation, notably for VR scenes that entail motion through space and changes in head orientation, for example, simulated flying or driving. Balance testing for postural stability was conducted in each child, as shown in the graph of Figure 6 . Children were required to stand heel-to-toe or on one foot—a modified Romberg test—while holding an electronic sway meter ( Figure 7 ). The average stability score at baseline was 68 ± 4, with a score of 100 denoting perfect stability, that is, no sway. After 60 minutes of VR play, the stability score reduced to 62 ± 8. The reduction (9%) was not significantly different from baseline ( P = .18).
