The United States Air Force (USAF) has an ever-increasing number of aircrew members requiring the use of spectacles for flying duties. 1–3 Spectacle wear can create significant compatibility issues with the unique life support systems that are essential for survival in the aerospace environment. 4 Soft contact lens (SCL) wear has been approved for aircrew for more than a decade to improve life support system compatibility, but not all ametropic aircrew can wear SCLs. 5 Photorefractive keratectomy (PRK) may be another alternative to spectacles and SCLs and may offer some distinct advantages in operational situations.
Although PRK has been widely accepted clinically, the USAF has concerns about its aeromedical and operational effects. Other types of refractive surgery, including surgical keratoplasty procedures such as radial keratotomy (RK), have resulted in corneal scars, diurnal refractive instability, excessive ocular glare, and change in prescription after prolonged exposure to altitude. 6–9 It is possible that PRK could increase a patient's susceptibility to these undesirable conditions. Although PRK laser ablation is, in principle, less invasive than RK, PRK is still a surgical manipulation of tissue and therefore initiates a progressive course of injury and healing. The associated tissue transformations can include disruption of the configuration of collagen molecules in the cornea, scar formation, and corneal haze. 10, 11 These conditions might impede vision even in an eye whose refractive error has been corrected perfectly, because they can exacerbate the effects of glare. Glare can be defined as excessive light in the visual field, which can scatter, degrade vision, and cause discomfort as long as the light persists. 12 Because glare can degrade the retinal image, it can interfere with performance in a variety of visual tasks and conditions. Glare elevates discrimination thresholds throughout the human contrast sensitivity (CS) function 13, 14 and can also impede motion perception and the discrimination of objects (including vehicles), particularly at night.
The potential threat from corneal haze and ocular glare may be underestimated or completely overlooked by standard high-contrast visual acuity (VA) tests, 15–17 which still comprise the primary basis for accession and retention in the military. This indicates a need for new tests to evaluate the potential duty impact on USAF personnel. These tests should determine whether PRK could be used to obtain refractive correction without impeding visual processing in low-contrast and glare conditions. Because haze and glare effects may evolve over time, 18 a meaningful test should accommodate a longitudinal evaluation.
In October 1998, the USAF endorsed a longitudinal clinical evaluation of the long-term effects of PRK on visual performance. The USAF chose PRK to investigate rather than laser in situ keratomileusis (LASIK) because of the possibility of flap displacement in the latter from blunt trauma occurring during an operational incident, such as ejection from the aircraft. 19–21 The overall PRK study included 5 groups of 20 subjects each, including a control group and a PRK-only group. All 100 subjects underwent detailed visual testing in the Refractive Surgery Center at Brooks Air Force Base, TX. Three groups participated in additional specialized tests representing the effect of different flight conditions on visual function. These groups included a PRK altitude study group, a group that rode the centrifuge to assess g effects on PRK-treated eyes, and a group participating in visual tests in simulated cockpit environments. This last group participated in the experiment described below in which an adaptive test battery, the Freiburg Acuity and Contrast Test (FrACT), was used to assess VA and CS.
Twenty nonflying, active duty USAF personnel (16 male and 4 female) ranging in age from 26 to 47 years volunteered as subjects. Each subject's informed consent was obtained as required by AF Instruction 40-403. All subjects completed the VA and CS phases of the experiment, to determine a baseline, before undergoing PRK at Wilford Hall Medical Center on Lackland Air Force Base, TX. Data were then collected at 6, 12, and 24-month post-PRK intervals. The mean refractive error for 40 eyes before PRK was −3.34 D, with myopia ranging from −1.25 to −6.00 D. Any post-PRK refractive error was determined before data collection at each interval, and all subjects wore corrective spectacles, fabricated by the USAF School of Aerospace Medicine's Optical Fabrication Laboratory, during data sessions. To maintain experimental uniformity, spectacles were included for all subjects, even if they required no correction. The study was successful in retaining 16 subjects throughout its 2-year duration, although only 14 completed all four data collection intervals.
VA and CS were assessed by using a gapped Landolt ring (a “tumbling C”) target, presented in a visual performance battery called the FrACT. 22 In the FrACT, the subject uses a directional keypad to indicate the orientation (among 8 possible directions) of the Landolt target, which is presented as a black or gray symbol on a lighter background. The FrACT includes two components, one measuring VA and one measuring CS. This experiment was thus actually two experiments comprising both the acuity and contrast test components. Both tests use Best PEST (parameter estimation by sequential testing) adaptive threshold estimation to adjust the difficulty of the task across trials according to the subject's performance. In both the VA and the CS tests, subjects were allowed to respond to each trial at their own pace.
The FrACT VA test adjusts the size of the target (and hence, the width of the gap) across trials. If the subject responds correctly, the test shrinks the target for the next trial, whereas an incorrect response would cause the software to expand the target in the next trial (Fig. 1). In this way, the test “homes in” on the subject's VA threshold, which is then recorded in decimal form at the end of each test run (which comprised 30 trials in this study). Decimal acuity scores were converted to logMAR (log of the minimum angle of resolution) scores for data analysis.
The FrACT CS test estimates the contrast threshold similarly (Fig. 2). The diameter of the target is kept constant, and its contrast is adjusted across trials by modulating the symbol luminance level. When the subject responds correctly, contrast is reduced in the next trial by increasing the symbol luminance. When the subject responds incorrectly, contrast is increased in the next trial by decreasing the symbol luminance. At the end of each 30-trial test run, the program records the subject's Michelson contrast sensitivity threshold. This represents the reciprocal of the Michelson contrast ratio [(Lmax + Lmin)/(Lmax − Lmin)], corresponding to the sum of the target and the background luminances divided by their difference. In this experiment, the CS target diameter was 84 arc-min, with a corresponding gap size of 17 arc-min.
The subject viewed the stimuli monocularly from one end of an optical table (Fig. 3), and a chin rest was used to anchor the viewing position. Stimuli were displayed on a 21-inch Radius Precision View color monitor with a background luminance measuring approximately 125 cd/m2 driven by an Apple Power Macintosh G3. The luminance and contrast from this screen were measured and verified before each post-PRK data collection period. The monitor was placed to the side and was viewed in an optical mirror to yield a total viewing distance that approximated the 20 feet, or 6 m, typically used in clinical eye lanes to correspond to optical infinity.
Three within-subject parameters were included in the experimental design for both the VA and the CS tests. These comprised date in the subject's longitudinal PRK history (i.e., pre-PRK and 6-, 12-, and 24-month post-PRK); presence and type of glare in the visual field (no glare vs. laser glare vs. broadband white glare); and presence or absence of aircraft windshield material in front of the subject. At each of the four test dates, a 3 (glare conditions) × 2 (windscreen vs. no windscreen) design was conducted for both left and right eyes, yielding a total of 12 test runs for each visual test. Scores were averaged between eyes for each subject, and this mean score was recorded to represent each possible combination of the different levels of glare and windshield manipulations. VA and CS testing protocols were completed on separate days. The order in which the conditions were presented was counterbalanced across subjects.
The presence and characteristics of disability glare were varied among three levels. The first included no glare source. In the second, the laser glare condition, the Landolt target was surrounded by a ring-shaped, green (532-nm) laser glare source, which was superimposed on the stimulus display by using a 60-40 beam splitter. The beam splitter was mounted upright, behind the windscreen mounting, at a 45° slant relative to the sight line from the viewing position to the mirror in which the display monitor was viewed. The 40% reflection surface of the beam splitter faced the viewer, and the remaining 60% of the laser light was transmitted through the splitter, where it was sampled with a laser exposure monitoring system. The laser glare ring was created by rear-projecting a diverging laser beam on a diffusing screen mounted to the side of the beam splitter. A mask with a circular aperture and a matched inner disk were attached to the diffusing screen to yield an annular glare source whose inner and outer diameters subtended 3.3° and 4.8°, respectively, in the subject's visual field. The mean luminance of the annulus was approximately 6090 cd/m2 as viewed in the reflected image with no windscreen. This fell to approximately 4520 cd/m2 when the windscreen was added (see description below and Figs. 3 and 4). See the ”Laser System“ section below for a description of the laser apparatus.
In the broadband condition, a broadband (white) light was used as a glare source and was situated behind the beam splitter in direct view, 99.7 cm from the viewer (Fig. 3). The broadband source was a ring-shaped fluorescent bulb (Stocker & Yale SteadyLite model 13 Plus) whose inner and outer diameters measured 2.25 inches (5.7 cm) and 3.25 inches (8.3 cm), respectively. This yielded an annular source that closely matched the angular dimensions of the laser source. The stimulus was viewed through the annulus and the mounting for the broadband light (Fig. 4). The emission peak for the broadband source was at 544 nm. The emission spectrum was broad and included additional peaks at 620 nm (red) and 440 nm (blue), yielding a metameric white light with a measured color temperature of 5453 K. A power controller was used to match the luminance of the white-light glare source to that of the laser source (~6090 cd/m2), and this match was verified with a Photo Research PR650 spectra colorimeter before each data session. As with the laser source, the luminance of the white glare ring fell to approximately 4520 cd/m2 when viewed through the windscreen material. This fluorescent light source exposed the subjects to a minimal fraction of the threshold limit values 23 for exposure to a broadband source.
To test the visual performance effects of viewing through an aircraft window (whose scratches, internal diffusion, and imperfections can degrade the optical quality of a visual stimulus), a section of polycarbonate windscreen material was mounted directly in front of the viewing position, between the subject and the monitor. The windscreen's measured transmission values averaged 74.6% for laser light and 73.0% for white light, and transmission did not vary significantly throughout the experiment. This experimental manipulation was accomplished by including the windscreen in half of the trial runs.
A 5-W Coherent Verdi laser (Santa Clara, CA;Fig. 3) was used to produce laser glare. The Verdi system is a compact, solid-state, diode-pumped, frequency-doubled neodymiun:vanadate (Nd: YVO4) laser that provides a single-frequency green (532-nm) output. The laser was operated at reduced power (~1.5 W) for the experiment, and all safety precautions for a class 4 laboratory laser 24 were strictly adhered to.
The subject's exposure to laser light was monitored by using a collection device comprising a biconvex lens, a Newport 818-SL detector head, a Newport 4832-C optical meter, and a LabView Virtual Instrument data monitoring application. A 532-nm notch filter (transmission value 48%) was attached to the detector head to restrict the spectrum of the measured light. This detector head was mounted on the opposite side of the beam splitter from the projection surface on which the annulus was projected, so that it viewed the direct (60%) image transmitted through the splitter instead of the reflected (40%) image seen by the subjects. The lens had an effective diameter of 49 mm and an effective focal length of 38 mm and was used to steer an image of the laser annulus into the detector. The measured irradiance at the eye was approximately 0.165 μW/cm2, with a typical run time of 100 s (16.5 μJ/cm2). The size and position of the collection device were used to derive correction values from which exposure at the viewing position could be measured and recorded throughout each trial run.a
A cumulative laser-exposure spreadsheet was maintained for each subject during each of the four data session periods, i.e., at baseline, 6 months after PRK, 1 year after PRK, and 2 years after PRK. The maximum permissible exposure (MPE) 24 for extended viewing for a 532-nm laser is 10 mJ/cm2, according to the 1993 American National Standards Institute standard for safe laser use, which was applied at the time this research was conducted.b All laser exposures during each data session period were added together, treated as one continuous exposure, and then compared with the MPE. The experimental goal was to expose each subject to no more than 40% of the MPE for each data session period. The cumulative exposure totals turned out to be much less than 40% of the continuous-exposure MPE. For all subjects, the cumulative exposure for this experiment at each data session period was less than 2% of the MPE.
VA data were collected in decimal form, transformed to log-MAR units, and averaged across eyes for experimental analysis. 25 Data were analyzed with a three-way repeated-measures analysis of variance (ANOVA), which included optical glare condition, months after PRK, and windscreen presence as experimental variables. A significant (p < 0.0001) main effect of optical glare condition was observed, with no-glare conditions yielding lower log-MAR means (superior performance) than broadband glare conditions, which in turn yielded lower logMAR values than did the laser glare conditions. The main effect of interposing the windscreen between the subject and the target and glare source was also significant (p < 0.0001), with higher logMAR means (inferior performance) when the screen was present. The main effect for months after PRK did not reach statistical significance (p = 0.055). Two two-way interactions were statistically significant: months after PRK and optical glare condition (p < 0.005) and optical glare condition and windscreen (p < 0.0005). The three-way interaction was not statistically significant. VA summary means are shown in Fig. 5.
Several post hoc comparisons were performed to characterize the effects of PRK on acuity in combination with various visual conditions. These comparisons were made with Tukey tests for honest significant differences. The post hoc contrasts included Spjotvoll-Stoline adjustment to correct for the unequal subject numbers that resulted from the loss of several subjects over the course of the study. 26 VA performance was compared between each post-PRK date and the corresponding baseline pre-PRK test, for each combination of optical glare (no glare, laser, and broadband) and windscreen (present vs. absent) conditions. With the no-glare condition, there were no statistically significant changes in VA over time, either with or without the screen in place. Under the laser-glare condition, there were no significant across-period changes without the screen, but the changes at 6 months (p < 0.01) and 12 months (p < 0.05) after PRK were statistically significant with the screen in place. The baseline logMAR VA for the laser condition with the screen in place was 0.29 (equivalent to 20/39 Snellen), whereas the 6- and 12-month acuities were both 0.22 (20/33). This represents an improvement in VA at the 6- and 12-month periods when compared with baseline. The only time period showing a statistically significant change in the broadband optical glare condition versus baseline was the 24-month post-PRK period with the screen in place (p < 0.05). LogMAR VA at this baseline was 0.01 (20/21), whereas the 24-month post-PRK period was 0.07 (20/24), which represents a decrement in visual performance equivalent to three letters on a Snellen chart.
Because the laser- and broadband-glare sources were luminance matched and subtended equal visual angles, VA data from the laser-glare condition were compared with broadband-glare VA data at each data collection period. There was a significant difference (p < 0.0005) between the laser- and broadband-glare source conditions at every data collection period, both with and without the screen in place. LogMAR VA for the laser-glare source condition ranged from 0.15 (20/28) to 0.30 (20/40), whereas logMAR VA for the broadband-source condition ranged from −0.05 (20/18) to 0.07 (20/24). This relation is evident by examining the curves in Fig. 5 and represents a much better performance on the FrACT VA task with the broadband-glare source in place than with the laser-glare source.
No statistically significant differences were observed in VA between conditions with and without the windscreen in the absence of a glare source. In the presence of laser glare, VA did differ significantly between screen and no-screen conditions at all four data collection periods, with higher logMAR values observed when the screen was present (pre-PRK, p < 0.0005; 6 months, p < 0.05; 12 months, p < 0.001; and 24 months, p < 0.05). When laser glare was added, logMAR VA ranged from 0.22 (20/33) to 0.30 (20/40) with the screen in place and from 0.14 (20/28) to 0.23 (20/34) with no screen in place, indicating better performance in the latter condition. In broadband-glare conditions, logMAR means were higher with the screen present at all four data collection periods, but this difference was statistically significant only at the 6-month post-PRK period (p < 0.05), at which time logMAR VA was 0.02 (20/21) with the screen in place and −0.05 (20/18) with no screen.
A three-way repeated-measures ANOVA was performed on the CS data, including months after PRK, optical glare condition, and windscreen presence as experimental variables. CS data were analyzed after logarithmic transformation. The main effects for all three variables were statistically significant (p < 0.0001), as were the two-way interactions between months after PRK and optical glare condition (p < 0.0001) and between windscreen presence and optical glare condition (p < 0.0001). The three-way interaction was also significant (p < 0.0001). Log CS summary means are shown in Fig. 6.
Post hoc comparisons were performed to characterize the effects of PRK on CS in combination with various visual conditions. These comparisons were made with Tukey's honestly significant difference tests incorporating Spjotvoll-Stoline correction to adjust for unequal subject numbers. 26 CS was compared between each post-PRK date and the corresponding baseline pre-PRK test for each combination of optical glare (no glare, laser, and broadband) and windscreen (present vs. absent) conditions. In the absence of a glare source, log CS scores were significantly lower at all three post-PRK data collection periods, with and without the screen in place (at 6 months, p < 0.0005 with and without screen; at 12 months, p < 0.001 with and without screen; at 24 months, p < 0.05 without screen and p < 0.0005 with screen). The pre-PRK mean log CS without the screen in place was 1.77, whereas the 6-, 12-, and 24-month post-PRK means were 1.63, 1.64, and 1.67, respectively. With the screen in place, the pre-PRK baseline mean log CS was 1.82, whereas the 6-, 12-, and 24-month post-PRK means were 1.61, 1.60, and 1.62, respectively. With broadband glare, means from all three post-PRK data collection periods were also significantly lower except at the 12-month period with the screen in place (at 6 month, p < 0.0005 with and without screen; at 12 months, p < 0.005 without screen; at 24 months, p < 0.001 with and without screen). The pre-PRK mean log CS for the broadband-glare condition without the screen was 1.28, whereas the 6-, 12-, and 24-month means were 1.13, 1.16, and 0.99, respectively. With the screen in place, the pre-PRK mean log CS was 0.94, whereas the 6-, 12-, and 24-month means were 0.80, 0.85, and 0.81, respectively. With laser glare, only one post-PRK data collection period was significantly lower than baseline, namely, the 24-month period without the screen (p < 0.0005). The pre-PRK mean log CS for the laser condition without the screen was 0.92, whereas the 24-month mean was 0.78. Obviously, the statistically significant post-PRK CS scores measured in the no-glare and glare conditions represent a decrement in visual performance relative to the pre-PRK baseline.
Because the laser and broadband optical glare sources were matched daily for luminance and subtended equal visual angles, CS data were compared between the laser and broadband conditions at each data collection period, both with and without the windscreen in place. A statistically significant difference between the glare conditions was observed at all data collection periods with and without the windscreen (p < 0.0005 for all comparisons except for 6 months with screen, for which p < 0.005). The mean log CS values for broadband glare ranged from 1.28 to 0.80, whereas values for laser glare ranged from 0.92 to 0.66. These findings are consistent with the VA comparisons indicating better visual performance in the broadband-glare condition than in the laser-glare condition.
As with VA, there were no statistically significant differences observed in CS between conditions with and without the windscreen in the absence of a glare source. However, there was a definite ”windscreen“ effect in the presence of both glare sources. CS was significantly lower with the windscreen in place (indicating poorer visual performance) for both glare sources at all four data collection periods (p < 0.0005, except for the laser condition at 24 months, for which p < 0.005). For broadband glare without the screen, log CS means ranged from 1.28 to 0.99, whereas means ranged from 0.94 to 0.80 with the screen in place. For laser glare without the screen, log CS means ranged from 0.92 to 0.78, whereas means ranged from 0.70 to 0.66 with the screen in place.
In this discussion, we highlight findings that indicate meaningful changes in visual performance resulting from our three experimental manipulations and distinguish these from minor or local effects that can emerge when considering all the possible interactions and contrasts embodied in a 4 × 3 × 2 three-way design. Our indices of visual performance comprised the VA and CS scores obtained in the FrACT. In each of the visual tasks, we consider the effects of PRK across the 2-year study period, then we consider the disruptive effect of laser and broadband glare, and then we consider the effect of interposing windscreen material between the viewer and the task.
We identified few significant changes in FrACT VA, relative to baseline performance, across the 24-month period after PRK. Consistent with reports of improved performance in some other USAF PRK experiments (J. Bruce Baldwin, personal communication), significant improvement in VA was observed at the 6- and 12-month data collection periods under conditions including the laser-glare source and the windscreen. However, we observed no other improvements in visual performance under any of the other conditions in this experiment, including the laser-glare condition without the windscreen. Although statistically significant, this improvement under laser-plus-windscreen conditions represents only about one-half of a Snellen VA line. Some of this minor improvement in VA might have resulted from a change in retinal image magnification related to transferring the refractive correction from the spectacle plane to the corneal plane. 17, 27 VA values were not corrected for magnification differences in this study, and the mean myopic decrease of approximately 3 D (15-mm vertex distance) would result in an approximate 4% increase in retinal image size. This change, however, would equate to a less-than-one-letter gain in VA, even more modest than the observed half-line improvement. The only other statistically significant difference across time was a decrement in VA at the 24-month data collection period with the broadband-glare source and windscreen in place. This represents a loss of less than one-half of a Snellen VA line and is consistent with a previous finding of a small mean loss (1.5 letters) of high-contrast VA in glare conditions 1 year after PRK. 28 In sum, it does not appear that PRK aggravated the subjects' glare susceptibility meaningfully in the VA task.
The experiment identified a notable difference in VA with the laser-glare source in place versus the broadband-glare source. This was not expected before the experiment (i.e., it was expected that the VA would be similar across the conditions), because both glare sources were matched for luminance, size, and spatial configuration, and this luminance match was verified daily. VA measured with the broadband optical glare source in place was the equivalent of two Snellen chart lines better than with the laser-glare source in place. This difference could be considered operationally relevant to a USAF aircrew member. Because the pupillomotor response under photopic conditions is typically determined by luminance for both broadband and monochromatic stimuli, pupil size should have been similar in the two glare conditions and not a factor in differing visual performances. 29 The visual performance disparity may best be explained by the optical qualities of the two different glare sources. When looking at the laser-glare source, it was obvious that there was some coherent spatial noise (laser speckle), whereas the broadband source was composed of incoherent light with no speckle. It has been reported that laser speckle can cause spatial masking that impedes visual performance when VA is measured with square-wave gratings made up of laser light. 30 Although the Freiburg stimuli are computer generated Landolt Cs and were viewed through the center of the optical glare sources, the laser speckle could have had a surround spatial masking effect on the stimulus, which may have impaired visual performance with the laser-glare source in place.
Because windscreen material can be expected to scatter light from its surface scratches and imperfections, it was expected that visual performance with the screen present would be degraded when compared with performance without the screen. The windscreen degraded VA appreciably more with the laser-glare source in place. The rough surface of the windscreen probably increased the laser speckle and created a more sizeable spatial masking effect. The windscreen also scattered the incoherent light from the broadband optical glare source, but it evidently did not mask the stimulus as profoundly as did the laser speckle.
The mean log CS scores recorded in the absence of a glare source and those recorded with broadband glare were significantly higher in the baseline pre-PRK data period than for all three post-PRK periods; this was true with or without the windscreen, except for the broadband-glare-plus-screen condition at 12 months. The log CS means in the absence of a glare source at the 6-, 12-, and 24-month data collection periods were very similar, indicating no further decrease in CS after the initial drop-off in visual performance. The post-PRK log CS means with broadband glare were also very similar, except for a continued drop-off at 24 months without the screen. This drop contrasts with the VA findings, where performance was, if anything, better at the postoperative periods.
This decrease in CS performance after PRK might indeed represent a real consequence of corneal change after PRK. Some researchers have reported that PRK can aggravate higher-order optical aberrations in the cornea, 31–33 and this might have been a factor in this study. Our finding does not yet constitute a conclusive demonstration or explanation of the nature of the decrease, and some alternative explanations can be invoked, but none explains the drop-off completely. It is not clear why only one significant change over time was identified in log CS mean scores that were recorded in the condition including laser glare. Given the sizeable decrease in CS mean scores recorded in the absence of glare and in the presence of broadband glare, we might expect some parallel drop-off in the laser-glare conditions. It remains possible that such a drop-off did occur, but the test instrument was not sufficiently sensitive to distinguish it from other sources of variance in the laser-glare conditions. Consistent with this is the observation, illustrated in Fig. 6, that log CS means in the laser-glare condition were mostly lower than baseline at 6, 12, and 24 months.
Part of the decrease in CS performance might be attributable to an inconsistency in the Freiburg CS data collection program, whereby some CS values >100 (at the high end of the FrACT continuum) can be overestimated (Michael Bach, personal communication). As an example, there were no CS values >100 for the 6-month data collection period when the no-glare condition was combined with the screen condition, but there were six CS values >100, ranging as high as 151, for the pre-PRK combination of no glare and screen conditions. These high values might have skewed the data somewhat, but this can only be a partial explanation, because the mean CS values are still considerably higher before PRK than the 6-month values, even if all values are artificially limited to a maximum of 100. This possible inconsistency in the Freiburg program also does not explain why no subject managed to score >100 on the CS scale after receiving PRK.
The laser- and broadband-glare sources disrupted the CS task in a manner similar to that observed in the VA task. Both CS and VA performances were significantly worse with laser glare than with broadband glare. This is a notable finding and one that merits replication and further investigation. Specifically, it will be worthwhile to determine whether laser light is indeed “privileged” in its capacity to disrupt visual stimuli, and if so, whether this results from coherent speckle effects. 34 It is also notable that the windscreen aggravated the disruption of CS under glare conditions. This is operationally relevant because modern aviators view the world through windscreens or canopies, and these age, become scratched and pitted by windblast, and generally provide an increasingly imperfect optical medium as time progresses.
This study identified no systematic changes in VA relative to baseline after PRK. Although some local interaction between the data period and glare conditions was observed, it does not appear that PRK aggravated (or alleviated) meaningfully the effects of disruptive glare in the acuity task. Notably, the visual disruption from laser glare was almost two full Snellen lines worse, throughout the study, than the disruption from a broadband-glare source with the same configuration and luminance. This greater effect might result from masking associated with coherent spatial noise (speckle) surrounding the laser stimulus. The windscreen exacerbated the disruptive effect of glare on acuity to a varying degree.
A significant drop-off in CS was observed at all data collection periods after PRK under viewing conditions with no glare source and under conditions with broadband glare. We do not rule out the possibility that this could represent a true decline in CS after PRK. Both the laser- and broadband-glare sources caused a substantial decrement in CS performance throughout the study; this effect was again significantly greater for the laser source. Considered in conjunction with the VA data, this is a notable finding. The windscreen was more of a disruptive factor under both glare conditions with the CS task.
The methods and experimental conditions in these experiments were limited and did not embrace all of the potential visual scenarios that aircrew might encounter following PRK. Further investigation with other glare sources and visual tasks may be necessary.
The authors would like to thank Michael Bach, PhD, Professor of Ophthalmology at the University of Freiburg, Freiburg, Germany, and the originator of the FrACT program, for his advice and clarification of the FrACT procedures during this experiment.
1. Provines WF, Woessner WM, Rahe AJ, Tredici TJ. The incidence of refractive anomalies in the USAF rated population. Aviat Space Environ Med 1983;54:622–7.
2. Miller RE 2nd, Woessner WM, Dennis RJ, O'Neal MR, Green RP Jr. Survey of spectacle wear and refractive error prevalence in USAF pilots and navigators. Optom Vis Sci 1990;67:833–9.
3. Baldwin JB, Dennis RJ, Ivan DJ, Miller RE 2nd, Belihar RP, Jackson WG, Tredici TJ, Datko LM, Hiers PL. The 1995 aircrew operational vision survey: results, analysis, and recommendations. Brooks AFB, TX: USAF School of Aerospace Medicine, 1999.
4. Dennis RJ, Miller RE 2nd, Peterson RD, Jackson WG Jr. Contact lens wear with the USAF protective integrated hood/mask chemical defense ensemble. Aviat Space Environ Med 1992;63:565–71.
5. Dennis RJ, Green RP Jr, Ketchum NS. The medical acceptability of soft contact lens wear by USAF tactical aircrews. Aviat Space Environ Med 1992;63:122–7.
6. Veraart HG, van den Berg TJ, Ijspeert JK, Cardozo OL. Stray light in radial keratotomy and the influence of pupil size and straylight angle. Am J Ophthalmol 1992;114:424–8.
7. Deitz MR, Sanders DR, Raanan MG, DeLuca M. Long-term (5- to 12-year) follow-up of metal-blade radial keratotomy procedures. Arch Ophthalmol 1994;112:614–20.
8. McDonnell PJ, Nizam A, Lynn MJ, Waring GO 3rd. Morning-to-evening change in refraction, corneal curvature, and visual acuity 11 years after radial keratotomy in the prospective evaluation of radial keratotomy study. The PERK Study Group. Ophthalmology 1996;103:233–9.
9. Mader TH, Blanton CL, Gilbert BN, Kubis KC, Schallhorn SC, White LJ, Parmley VC, Ng JD. Refractive changes during 72-hour exposure to high altitude after refractive surgery. Ophthalmology 1996;103:1188–95.
10. Lee YG, Chen WY, Petroll WM, Cavanagh HD, Jester JV. Corneal haze after photorefractive keratectomy using different epithelial removal techniques: mechanical debridement versus laser scrape. Ophthalmology 2001;108:112–20.
11. Corbett MC, Prydal JI, Verma S, Oliver KM, Pande M, Marshall J. An in vivo investigation of the structures responsible for corneal haze after photorefractive keratectomy and their effect on visual function. Ophthalmology 1996;103:1366–80.
12. Brennan DH. Glare in aviation. Health Phys 1989;56:665–9.
13. Finlay D, Wilkinson J. The effects of glare on the contrast sensitivity function. Hum Factors 1984;26:283–7.
14. Yates JT, Harrison JM, O'Connor PS, Ballentine C. Contrast sensitivity: characteristics of a large, young, adult population. Am J Optom Physiol Opt 1987;64:519–27.
15. Atwood DA. The effects of headlight glare on vehicle detection at dusk and dawn. Hum Factors 1979;21:35–45.
16. Bichao IC, Yager D, Meng J Disability glare: effects of temporal characteristics of the glare source and of the visual-field location of the test stimulus. J Opt Soc Am A 1995;12:2252–8.
17. Bullimore MA, Olson MD, Maloney RK. Visual performance after photorefractive keratectomy with a 6-mm ablation zone. Am J Ophthalmol 1999;128:1–7.
18. Hersh PS, Stulting RD, Steinert RF, Waring GO 3rd, Thompson KP, O'Connell M, Doney K, Schein OD. Results of phase III excimer laser photorefractive keratectomy for myopia. The Summit PRK Study Group. Ophthalmology 1997;104:1535–53.
19. Norden RA, Perry HD, Donnenfeld ED, Montoya C. Air bag-induced corneal flap folds after laser in situ keratomileusis. Am J Ophthalmol 2000;130:234–5.
20. Leung AT, Rao SK, Lam DS. Traumatic partial unfolding of laser in situ keratomileusis flap with severe epithelial ingrowth. J Cataract Refract Surg 2000;26:135–9.
21. Goodman RL, Johnson DA, Dillon H, Edelhauser HF, Waller SG. Laser in situ keratomileusis flap stability during simulated aircraft ejection in a rabbit model. Cornea 2003;22:142–5.
22. Bach M. The Freiburg visual acuity test: automatic measurement of visual acuity. Optom Vis Sci 1996;73:49–53.
23. American Conference of Governmental Industrial Hygienists. Threshold Limit Values for Chemical Substances and Physical Agents, and Biological Exposure Indices. Cincinnati, OH: American Conference of Governmental Industrial Hygienists, 1992.
24. American National Standards Institute. American National Standard for the Safe Use of Lasers. ANSI Z136.1–2000. Orlando, FL: The Laser Institute of America, 2000.
25. Holladay JT, Prager TC. Mean visual acuity. Am J Ophthalmol 1991;111:372–4.
26. Spjotvoll E, Stoline MR. An extension of the T-method of multiple comparison to include the cases with unequal sample sizes. J Am Stat Assoc 1973;68:976–8.
27. Applegate RA, Howland HC. Magnification and visual acuity in refractive surgery. Arch Ophthalmol 1993;111:1335–42.
28. Verdon W, Bullimore M, Maloney RK. Visual performance after photorefractive keratectomy: a prospective study. Arch Ophthalmol 1996;114:1465–72.
29. Adrian W. Spectral sensitivity of the pupillary system. Clin Exp Optom 2003;86:235–8.
30. Artigas JM, Felipe A. Effect of luminance on photopic visual acuity in the presence of laser speckle. J Opt Soc Am A 1988;5:1767–71.
31. Mierdel P, Kaemmerer M, Krinke HE, Seiler T. Effects of photorefractive keratectomy and cataract surgery on ocular optical errors of higher order. Graefes Arch Clin Exp Ophthalmol 1999;237:725–9.
32. Mrochen M, Kaemmerer M, Mierdel P, Seiler T. Increased higher-order optical aberrations after laser refractive surgery: a problem of subclinical decentration. J Cataract Refract Surg 2001;27:362–9.
33. Ninomiya S, Maeda N, Kuroda T, Fujikado T, Tano Y. Comparison of ocular higher-order aberrations and visual performance between photorefractive keratectomy and laser in situ keratomileusis for myopia. Semin Ophthalmol 2003;18:29–34.
34. Williams DR. Visibility of interference fringes near the resolution limit. J Opt Soc Am A 1985;2:1087–93.
a Conservative assumptions were adopted wherever possible in the monitoring of laser exposure and the application of correction values. For example, although the glare annulus was large enough to qualify as an extended source, its output was measured in units of irradiance corresponding to the more concentrated power deliver from a point source. In addition, in applying a correction factor to represent the difference between the energy gathered by a human pupil vs. a collection lens 49 mm in diameter; the selected pupil diameter was 7 mm, which is larger (and hence would lead to more exposure) than the pupil of an alert human subject in photopic viewing conditions.
b This standard has since been updated. The new American National Standards Institute 2000 standard specifies exposure guidelines and MPE levels that are similar to, but typically more liberal than, the earlier 1993 standard.