LED Roadway Lighting: Impact on Driver Sleep Health and Alertness (2021)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

36 Corneal Illuminance Dosage Experiment— Naturalistic Driving Exposure Smart Road Exposure The results of the corneal illuminance dosage and irradiance for a 2-hour exposure in the lighted section of the roadway are shown in Table 8. The headlamps were always on, and the roadway luminance does not account for the additional luminance from the headlamps. The 2200 K LED could only achieve 0.8 cd/m2 at full brightness and, as a result, it had a lower dosage and irradiance than the other light sources. Public Road Exposure Table 9 shows the results of the light levels measured for different roadway function classes. Overall, the highest average roadway luminance was measured for the major collector/local func- tion class at 1.4 cd/m2, and the lowest average across all divisions was for interstates at 0.8 cd/m2. Interstates had the highest average vertical illuminance (4.1 lux), and the major collector/local function class had the lowest (2.8 lux). For the average horizontal illuminance, minor arterials (9.6 lux) had the highest, and interstates had the lowest (5 lux). These light levels are in-situ mea- surements and often exceed or fall under the light levels specified in IES RP-8-18. Light levels could exceed those specified in the RP when there are additional sources of light adjacent to the roadway light, stray light from commercial establishments, parking facilities, and the like. Light levels could fall under the specified levels due to luminaire failures or dirt accumulation. Consumer Device Exposure Average illuminance and total illuminance dosages from the consumer devices for a 2-hour exposure are shown in Table 10. In the full brightness mode, the Apple iPad had the highest average illuminance and dose, and Kindle Paperwhite® had the lowest. In dark mode, all the devices had lower average illuminances and doses, with the Apple iPad Pro having the lowest levels and the Apple iPad having the highest. Daily Exposure Measurements Table 11 shows the irradiances and illuminance dosages from the daily exposure measure- ments of 10 typical office workers at VTTI in 24 hours. Overall, participants experienced between 4 million and 31 million lux-s with an average of around 15 million lux-s of light. The average values in Table 11 denote the average irradiance and illuminance dosage experienced per second by the office workers during the 24 hours. C H A P T E R   4 Results

Results 37   Light Condition Road Luminance (cd/m2) Total Dose (lux-s) Total Irradiance (µW/cm2) 2200 K LED 0.8 9,448.0 2,310.7 3000 K LED 1.0 12,518.9 3,396.5 4000 K LED 1.0 12,663.0 3,191.1 5000 K LED 1.0 12,745.1 3,333.3 2100 K HPS 1.0 11,397.9 3,065.2 No-light <0.05 6,150.0 1,541.3 Table 8. Corneal illuminance dose and total irradiance under different light sources on the Smart Road for a 2-hour period. Division Functional Classification Avg. Luminance (cd/m2) Avg. Vertical Illuminance (lux) Avg. Horizontal Illuminance (lux) Bristol Interstate 1.8 4.2 16.2 Major Collector/Local 1.3 2.0 8.5 Minor Arterial 0.5 6.2 Principal Arterial 0.7 1.0 0.6 Culpeper Interstate 0.2 0.8 0.2 Major Collector/Local 1.5 1.2 0.6 Minor Arterial 0.4 1.2 1.2 Principal Arterial 1.7 1.7 1.4 Fredericksburg Interstate 3.1 8.4 Major Collector/Local 2.3 6.5 Minor Arterial 2.6 12.0 Principal Arterial 1.4 2.4 Hampton Roads Interstate 0.7 12.4 6.3 Major Collector/Local 1.7 6.1 13.3 Minor Arterial 1.4 6.0 8.2 Principal Arterial 5.3 14.1 Lynchburg Major Collector 0.8 4.7 15.5 Minor Arterial 3.0 7.7 Principal Arterial 1.7 3.8 14.2 Northern Virginia Interstate 0.3 2.4 3.6 Major Collector/Local 1.1 2.2 4.5 Minor Arterial 1.4 3.5 9.1 Principal Arterial 1.3 4.1 14.1 Richmond Major Collector/Local 2.0 2.0 10.3 Minor Arterial 2.0 10.3 Principal Arterial 2.0 10.3 Table 9. Road luminance and average vertical and horizontal illuminances for different function classes in each VDOT division. (continued on next page)

38 LED Roadway Lighting: Impact on Driver Sleep Health and Alertness Division Functional Classification Avg. Luminance (cd/m2) Avg. Vertical Illuminance (lux) Avg. Horizontal Illuminance (lux) Salem Major Collector/Local 2.7 3.6 11.2 Minor Arterial 2.3 2.9 9.9 Principal Arterial 2.0 3.9 13.7 Staunton Interstate 1.1 1.9 4.0 Major Collector/Local 1.2 2.7 7.2 Minor Arterial 1.3 4.7 16.1 Principal Arterial 0.9 4.6 7.9 Overall Interstate 0.8 4.3 5.0 Major Collector/Local 1.4 2.8 7.9 Minor Arterial 1.1 3.2 9.6 Principal Arterial 1.1 3.3 8.6 Table 9. (Continued). Avg. Illuminance (lux) 2-Hour Dose (lux-s) Dark Mode Full Brightness Mode Dark Mode Full Brightness Mode Monitor (Dell E156FP—LCD monitor—15") 33.3 50.2 239,616.0 361,656.0 Television (LG 70UM7370PUA 70") 10.6 69.9 76,608.0 502,920.0 Kindle Paperwhite® 1.7 9.8 12,456.0 70,776.0 Apple iPad Pro (1st generation) 0.5 76.4 3,456.0 550,296.0 Apple iPad (2nd generation) 20.8 83.9 149,832.0 604,368.0 Table 10. Average illuminance and total dosage from consumer devices used in this experiment. Table 11. Irradiances and illuminance dosages from the daily exposure measurements of office workers. Irradiance (µW/cm2) 24-Hour Dose (lux-s) Participant Total Avg. Max SD Total Avg. Max SD 1 1,027,142.1 74.0 2,955.2 96.9 4,061,066.2 298.0 10,551.3 358.3 2 2,968,856.1 82.8 6,694.8 274.4 11,087,652.9 309.1 22,725.3 954.5 3 3,506,895.1 96.5 17,229.1 427.5 11,836,103.8 325.6 56,265.1 1,409.8 4 5,635,520.5 113.6 84,509.2 801.7 21,493,459.3 440.9 290,023.2 2,719.2 5 3,018,372.4 181.3 18,490.7 334.4 11,448,653.8 689.9 65,506.6 1,205.0 6 3,075,503.4 86.8 102,073.6 733.0 11,331,634.4 322.0 361,510.9 2,605.0 7 6,011,473.7 153.1 10,341.5 588.5 22,209,673.2 565.5 36,406.8 2,084.1 8 2,864,913.7 78.9 41,027.0 634.0 10,150,821.0 279.6 148,987.4 2,166.5 9 8,148,375.0 695.0 54,716.5 1,628.8 31,092,347.4 2,609.8 191,907.4 5,920.0 10 3,774,255.3 109.7 59,242.5 811.4 13,771,315.2 403.8 203,390.7 2,781.6 Aggregate Measures 4,003,130.7 167.2 102,073.6 188.6 14,848,272.7 624.4 361.510.9 710.1

Results 39   Experimental Conditions—Corneal Illuminance Dosage Measurements Table 12 shows the total and average illuminance dosages from all the experimental condi- tions in the DSHA Experiment. The α-opic equivalent daylight illuminance values, calculated based on CIE S 026/E:2018, for each experimental condition are shown in Table 13. Driver Sleep Health and Alertness Experiment Assay Performance The inter-assay coefficient of variation from the four salivary assays run for this experiment performed by SolidPhase, Inc. (Portland, ME) was between 3% and 7%. Intra-assay coefficients of variation calculated from control samples of 3.6 and 24.4 picograms per milliliter (pg/ml) assayed were 4.1% and 4.4%, respectively. The minimum detection limit of the salivary assay is 0.5 pg/ml melatonin. The inter-assay coefficient of variation from the two plasma assays run for this experiment performed by SolidPhase, Inc. was between 7% and 10%. Intra-assay coefficients of variation calculated from control samples of 4.6 and 14.3 pg/ml were 6.7% and 7.5%, respectively. The minimum detection limit of the plasma assay is 0.5 pg/ml melatonin. Comparison of Plasma and Salivary Melatonin in Positive Control Experiment The LMM results comparing salivary melatonin with plasma melatonin are shown in Table 14. The main effects of the type of melatonin and time were significant. The difference in the mela- tonin between saliva and plasma was statistically significant, with the amplitude in the plasma being higher than saliva (see Figure 12). Light Condition Avg. of Total Dose (lux-s) 2-Hour SD of Total Dose (lux-s) Avg. Dose (lux/s) SD of Avg. Dose (lux/s) Pre-exposure 1,439,692.3 165,592.4 200.0 23.0 2100 K HPS–HIGH 13,162.1 4,272.2 1.8 0.6 4000 K LED–HIGH 13,459.8 4,405.0 1.9 0.6 4000 K LED–MED 9,858.3 3,683.3 1.4 0.5 4000 K LED–LOW 8,016.7 2,389.6 1.1 0.3 No-light 5,681.9 1,534.3 0.8 0.2 Table 12. Total and average illuminance dosages across all experimental conditions in the DSHA Experiment. α-opic Equivalent Daylight (D65) Illuminance, lux Light Condition S-cone-opic M-cone-opic L-cone-opic Rhodopic Melanopic Pre-exposure 66.4 173.0 194.5 112.4 87.1 2100 K HPS–HIGH 0.3 1.2 1.9 0.5 0.3 4000 K LED–HIGH 0.6 1.6 1.8 1.1 0.8 4000 K LED–MED 0.5 1.2 1.4 0.8 0.6 4000 K LED–LOW 0.4 1.0 1.1 0.6 0.5 Table 13. Equivalent daylight illuminance values from CIE S 026/E:2018 for each experimental light condition in the DSHA Experiment.

40 LED Roadway Lighting: Impact on Driver Sleep Health and Alertness Effect of Road Light Exposure on Melatonin The LMM results of the effects of light condition and time on salivary melatonin are presented in Table 15. Only the main effect of light condition was significant. The melatonin differences between the positive control and all road exposure conditions were statistically significant, with the positive control showing the most suppression (see Figure 13). However, there were no differences between any of the road exposure conditions. Effect of Road Light Exposure on Detection Distance Detection Distance The LMM results of the effects of light condition and object color on detection distance are presented in Table 16. All the main effects, along with the interaction involving light condition and object color, were significant. The interactive effects between lap and light conditions, and light conditions and object color, will be discussed further. The interaction between lighting condition and the covariate, lap, was significant, indicating that the slope of the detection distance was not equal across the light conditions with increas- ing the lap numbers. Further analysis of the interaction between lap and each of the light con- ditions showed that, under the HPS light condition, the detection distances decreased as the lap number increased (see Figure 14). This phenomenon was not observed across any of the other light conditions. Detection distance was the highest across all the light conditions in Effect Numerator (Num) DF Denominator (Den) DF F-Value P-Value Type (plasma versus saliva) 1 19 38.6 <.0001 Time 4 73.4 2.79 0.0325 Type*Time 4 70 1.77 0.1444 Note: DF = degrees of freedom. Table 14. Results of the LMM comparing salivary melatonin with plasma melatonin in the positive control. 0 5 10 15 20 25 12:30 AM 1:00 AM 1:30 AM 2:00 AM 2:30 AM 3:00 AM 3:30 AM M el at on in (p g/ m l) Time Plasma Saliva Figure 12. Effect of melatonin from plasma and saliva over time. Values are least-square means of melatonin. Error bars represent standard errors.

Results 41   Effect Num DF Den DF F-Value P-Value Light Condition (LC) 5 69 8.35 <.0001 Time (T) 4 237 1.07 0.3712 LC × T 20 262 0.61 0.9021 Table 15. Statistical results from LMM analysis of salivary melatonin. B B B B B A 0 4 8 12 16 20 Sa liv ar y M el at on in (p g/ m l) 2100 K HPS HIGH 4000 K LED HIGH 4000 K LED MED 4000 K LED LOW No Light Positive Control Light Condition Figure 13. Effect of light condition on melatonin. Values are least-square means of melatonin. Error bars denote standard errors. Uppercase letters indicate post hoc groupings (from pairwise comparisons). Effect Num DF Den DF F-Value P-Value Light Condition (LC) 4 1834 9.6 <.0001 Object Color (OC) 6 1834 190.23 <.0001 LC × OC 24 1834 6.64 <.0001 Lap (L) 1 1835 0.46 0.4996 L × LC 4 1834 2.94 0.0195 Table 16. Statistical results from LMM analysis of detection distance.

42 LED Roadway Lighting: Impact on Driver Sleep Health and Alertness the initial laps. However, as the number of laps increased, the detection distances became shorter until and beyond lap number 10. After lap 10, the detection distances were lower than 4000 K LED–High (see Figure 14). For the remaining light conditions, the detection distances increased with increasing light levels (see Figure 14). The no-light condition had the shortest detection distances across all the light conditions. Post hoc analysis of the two-way interaction involving lighting condition and object color showed that the statistical differences between light conditions were only observed for red and yellow targets. For the red targets, the 2100 K HPS–High condition had the longest detec- tion distance compared to all other light conditions (see Figure 15a). There were no statis- tical differences between the detection distances of any other light conditions for the red target. For the yellow target, statistical differences between light conditions depended on the light levels. The detection distances were the highest for the 2100 K HPS and 4000 K LED at the highest light level (see Figure 15b). The detection distance differences between 4000 K 4000 K LED - HIGH 4000 K LED - MED 4000 K LED - LOW 2100 K HPS - HIGH No Light 0 20 40 60 80 100 120 0 2 4 6 8 10 12 14 16 18 20 22 D et ec tio n D is ta nc e (m ) Lap Figure 14. Effect of lap on detection distance for each light condition. Red Yellow(b)(a) Figure 15. Effect of light condition on detection distance for (a) red and (b) yellow targets. Values are least-square means. Error bars denote standard errors. Uppercase letters indicate post hoc groupings (from pairwise comparisons).

Results 43   LED–Medium and No-Light, and 4000 K LED–Low and No-Light, were also statistically sig- nificant (see Figure 15b). Color Recognition Distance Table 17 shows the LMM results of the effects of light condition and object color on color recognition distance. All the main and interaction effects are significant. The interactive effects between lap and light conditions, and light conditions and object color, will be discussed further. As observed in the detection distance LMM, the significant interaction between lighting con- dition and the covariate, lap, indicated that the slope of the color recognition distance was not equal across the light conditions with increasing lap numbers. Similar to the behavior of detec- tion distance for the 2100 K HPS–High condition, the color recognition also showed a steady decrease with an increase in the number of laps. Such behavior was not observed in any of the other light conditions. Color recognition distances in the 2100 K HPS–High light condition were the highest in initial laps, and as laps progressed distances decreased until they were lower than the 4000 K LED–High after lap number 14 (see Figure 16). Color recognition distances for the remaining light conditions depended on the light level and increased when light levels increased (see Figure 16). The no-light condition had the lowest color recognition distance. Effect Num DF Den DF F-Value P-Value Light Condition (LC) 4 1832 9.62 <.0001 Object Color (OC) 6 1831 193.66 <.0001 LC × OC 24 1831 7.14 <.0001 Lap (L) 1 1832 11.07 0.0009 L × LC 4 1832 3.76 0.0048 Table 17. Statistical results from LMM analysis of color recognition distance. 4000 K LED - HIGH 4000 K LED - MED 4000 K LED - LOW 2100 K HPS - HIGH No Light 0 20 40 60 80 100 120 0 2 4 6 8 10 12 14 16 18 20 22 Lap D et ec tio n D is ta nc e (m ) Figure 16. Effect of lap on color recognition distance for each light condition.

44 LED Roadway Lighting: Impact on Driver Sleep Health and Alertness Similar to the detection distances LMM, the post hoc pairwise comparisons of the two-way interaction between light condition and object color showed that differences between color recognition distances were statistically significant only for the red and yellow targets. For red targets, color recognition distances were statistically significant between 2100 K HPS–High and 4000 K LED–Low, and 2100 K HPS–High and no-light conditions (see Figure 17a). For yellow targets, both 2100  K HPS–High and 4000  K LED–High had the longest color recognition distances, which were significantly higher than 4000 K LED–Medium and Low and no-light conditions (see Figure 17b). There were no differences between the Medium- and Low-light conditions for the 4000 K LED. In general, the color recognition distances for yellow targets were longer than the red targets. Effect of Road Light Exposure on PERCLOS The LMM results of the effect of light condition and time on PERCLOS are shown in Table 18. None of the main effects or the interaction effect is significant. As a result, there are no major differences in the PERCLOS in the times measured across all the light conditions (see Figure 18). Effect of Light Exposure on SDLP Table 19 shows the LMM results of the effect of light condition and time on SDLP. None of the main effects or the interaction effect is significant. While the SDLP under the 2100 K HPS–High condition for the first two times was higher, this difference was not statistically significant (see Figure 19). Overall, there were no major SDLP differences between lighting conditions and times (see Figure 19). Red Yellow(b)(a) Figure 17. Effect of light condition on color recognition distance for (a) red and (b) yellow targets. Values are least-square means. Error bars denote standard errors. Uppercase letters indicate post hoc groupings (from pairwise comparisons). Effect Num DF Den DF F-Value P-Value LC 4 75.2 0.62 0.6529 T 3 75 0.23 0.8773 LC × T 12 75 0.3 0.9874 Table 18. Statistical results from LMM analysis of PERCLOS.

Results 45   0% 10% 20% 30% 40% 50% 1:00 AM 1:30 AM 2:00 AM 2:30 AM 3:00 AM Pe rc en ta ge o f E ye C lo su re Time 4000K LED–HIGH 2100K HPS–HIGH 4000K LED–MED 4000K LED–LOW No Light Figure 18. Effect of light condition on PERCLOS. Values are the least-square means of PERCLOS. Error bars denote standard errors. Effect Num DF Den DF F-Value P-Value LC 4 121 1.11 0.3536 T 2 43.5 0.96 0.3897 LC × T 8 120 0.5 0.851 Table 19. Statistical results from LMM analysis of SDLP. 0 300 600 900 1,200 1,500 1:30 to 2:00 AM 2:00 to 2:30 AM 2:30 to 3:00 AM St an da rd D ev ia tio n of L at er al Po si tio n (m m ) Time 4000K LED–HIGH 2100K HPS–HIGH 4000K LED–MED 4000K LED–LOW No Light Figure 19. Effect of light condition on SDLP. Values are the least-square means of SDLP. Error bars denote standard errors. Higher SDLP is correlated with more drowsiness.

46 LED Roadway Lighting: Impact on Driver Sleep Health and Alertness Effect of Road Light Exposure on Subjective Self-Reported Measure of Drowsiness (KSS) The LMM results of the effect of light condition and time on KSS are shown in Table 20. The main effects of light condition and time are significant. KSS scores were the highest for the 4000 K LED–High condition and the lowest for the positive control (see Figure 20). The KSS score differences between positive control and 4000 K LED–High, positive control and 4000 K LED–Medium, and positive control and the no-light condition were statistically significant. An increase in the time increased the KSS score (see Figure 21). The KSS scores between 1:30 AM, 2:00 AM, and 2:30 AM were significantly different from one another. Effect Num DF Den DF F-Value P-Value LC 5 111 5.04 0.0003 T 4 239 47.23 <.0001 LC × T 20 259 1.27 0.1977 Table 20. Statistical results from LMM analysis of KSS. A, B B B A, B B A 1 2 3 4 5 6 7 8 9 Ka ro lin sk a Sl ee pi ne ss S ca le 2100 K HPS HIGH 4000 K LED HIGH 4000 K LED MED 4000 K LED LOW No Light Positive Control Light Condition Figure 20. Effect of light condition on KSS. Values are the least-square means of KSS. Error bars denote standard errors. Uppercase letters indicate post hoc groupings (from pairwise comparisons). Higher KSS scores indicate higher drowsiness.

Results 47   A A B C C 1 2 3 4 5 6 7 8 9 12:30 AM 1:00 AM 1:30 AM 2:00 AM 2:30 AM 3:00 AM 3:30 AM Ka ro lin sk a Sl ee pi ne ss S ca le Time Figure 21. Effect of time on KSS. Values are the least-square means of KSS scores. Error bars denote standard errors. Uppercase letters indicate post hoc groupings (from pairwise comparisons). Higher KSS scores indicate higher drowsiness.