School start time change and motor vehicle crashes in adolescent drivers

Abstract

Study Objectives:

The aim of this study was to examine the association between a 50-minute delay (7:20 am to 8:10 am) in high school start times in Fairfax County (FC) Virginia and changes in rates of adolescent motor vehicle crashes. Crash rates in FC were also compared to those in the rest of the state during the same time period.

Methods:

Virginia Department of Motor Vehicles crash data in drivers age 16 to 18 years old between September and June of each year in FC versus the rest of the state were compared in the combined 2-year periods preceding (2013–2014 and 2014–2015; T1) and following (2015–2016 and 2016–2017; T2) school start time change in the fall of 2015.

Results:

The crash rate per 1000 in 16- to 18-year-old licensed drivers in FC during T1 was significantly higher compared to T2, 31.63 versus 29.59 accidents per 1,000 (95% confidence interval, 1.0–1.14, odds ratio 1.07, P = .03). In contrast, adolescent crash rates in the rest of Virginia were not statistically significantly different at T1 versus T2. With regard to subtypes of crashes, there was a trend toward significance in distraction-related crashes per 1,000 in FC at T1 compared to T2 at 7.01 versus 6.13 (95% confidence interval, 0.99–1.31, odds ratio 1.14, P = .05), but were not significantly different in the remainder of the state.

Conclusions:

The results of this study suggest that school start time delay is associated with decreased adolescent motor vehicle crash risk, with significant implications for public health and safety.

Citation:

Bin-Hasan S, Kapur K, Rakesh K, Owens J. School start time change and motor vehicle crashes in adolescent drivers. J Clin Sleep Med. 2020;16(3):371–376.

BRIEF SUMMARY

Current Knowledge/Study Rationale: In order to assess the effect of school start time change on adolescent safety outcomes, this study examined the association between a 50-minute delay in high school star times in a large US school district and rates of adolescent motor vehicle crashes. Crash rates were compared in the 2-year period before and after the start time change in the district and in the rest of the state.

Study Impact: Results showed that both overall car crash and distracted driving crash rates decline significantly in conjunction with school start times delay. These findings support the implementation of healthy school start times as a potentially powerful intervention to reduce mortality and morbidity in adolescents.

INTRODUCTION

Unintentional or “accidental” injuries, defined as unplanned injuries that occur without any intention to cause harm to oneself or others and include falls, burns, sports- and recreational-related injuries and motor vehicle crashes, are the number one cause of death of adolescents in the United States,^1,2 and a major contributor to morbidity and health care costs such as hospitalization and emergency department visits.³ Thus, potentially remediable contributors to an increased risk of unintentional injuries in this population are an important target for public health interventions.⁴

Chronic sleep loss has been shown in a number of studies in adolescents to be associated with increased unintentional injuries^5–7 including risk of pedestrian injuries,⁸ sports-related injuries,⁹ and occupational injuries,^10,11 as well as injury-related risk behaviors (eg, infrequent seatbelt use, drinking and driving, texting while driving) in adolescents.¹² In particular, automobile crashes among adolescents have been reported in a number of studies to be related to insufficient sleep.^13–15 The term “drowsy driving” is intended to encompass the range of sleep-related impairments that could contribute to crash risk among drivers across the age spectrum; adolescents are at particularly high risk for drowsy driving as a consequence of deficient sleep (defined as insufficient sleep duration compared with sleep needs and circadian misalignment, ie, a mismatch between intrinsic biological circadian rhythms and extrinsic environmental demands). In the drowsy driving research specifically examining high school students (drivers in 10th through 12th grades), one study found that 51% of respondents surveyed indicated they had driven drowsy in the past year¹⁶; 5% of these drivers said they nodded off or fell asleep while driving, with 27% of those incidents reported to have resulted in a motor vehicle crash. Similarly, in an AAA Foundation study,¹⁷ 16% of drivers age 16 to 18 years reported driving while struggling to keep their eyes open at least once in the past 30 days. Finally, in an Italian study, 40% of students reported driving while sleepy, and of the individuals who reported having had a motor vehicle crash, 15% considered sleepiness to have been the main cause.¹⁸

With respect to factors contributing to deficient sleep, drowsy driving, and adolescent motor vehicle crash risk, there is mounting evidence to support an important role for early school start times. Several studies have found an association between earlier high school start times and increased crash rates.^{14,15,19–21} Furthermore, a report published by the University of Minnesota Center for Applied Research and Educational Improvement in 2014 examining the effect of school start time delay in a number of districts across the country documented a substantial decrease in adolescent car crash rates in the year before and after school start time change.²¹ The reduction in crash rates was as high as 65% to 70% in several communities, although the rates in other communities ranged from a decrease of 6% to an increase of 9%. The wide range of changes in crash rates was theorized to be attributable to geographical and driving distance differences, traffic patterns (eg, suburban vs rural) and the relative pre-post start time difference as well as the absolute start time after the change. However, these data included only 1 year before-and-after change (failing to account for possible year-to-year variability unrelated to start time change), largely assessed smaller school districts, and did not compare crash rate data in these districts with the rest of the state.

The aim of this study was to examine motor vehicle crash rates in adolescent drivers in one of the largest school districts in the United States, Fairfax County Public Schools (FCPS) Virginia in the 2-year period prior to and following a 50-minute delay in high school start times. These data were compared to corresponding crash rates in the rest of the state. Crash data were also further examined with respect to crashes involving injuries, distraction, speed, and alcohol-related crashes.

METHODS

To study the effect of the school start time change in FCPS (2015–2016), we analyzed the crash rates in 16- to 18-year-old drivers in Fairfax County (FC) versus the rest of state, in the combined 2 years before (2013–2014 and 2014–2015; T1) and after (2015–2016 and 2016–2017; T2) a 50-minute change in high school start times (7:20 am to 8:10 am) in the fall of 2015.

Crash data

The Virginia Department of Motor Vehicles (DMV) provided de-identified data on licensed drivers and driving accidents for 16- to 18-year-olds from September through June of each year (T1 and T2). Crash rates, including overall crash rates, crashes associated with injuries, distraction-related, speeding-related, and alcohol-related crashes were compared within FC for the given time periods. Based on the DMV guidelines, reportable traffic crash was defined as a “crash on a public highway which involves death, injury or property damage in excess of $1,500.” Crashes associated with injuries are those involving “a person who is injured within 30 days as a result of a traffic crash involving at least one vehicle.” Distraction-related crashes were defined as “crashes that involve at least one distracted driver” and speeding-related crashes are those “that involve a driver exceeding the posted speed limit or driving too fast for conditions.” An alcohol-related crash is defined as “a crash where BAC data (≥ 0.1) is used in addition to police reports to determine alcohol-related status.” Although there are a number of other categories of crashes included in the Virginia DMV data base, including unrestrained persons, pedestrian involved, cell phone involved, bicycle involved, etc, we chose to report on those categories that are most common and/or involve a significant percentage of injuries/fatalities and that we believe are most relevant to the issue of school start times and resulting chronic sleep loss/circadian misalignment. Further comparison was made between FC and the rest of the state of Virginia (minus FC) during the same period.

Statistical analysis

We conducted the following statistical tests to compare crash rates in Virginia with crash rates in Fairfax. (1) Chi-square tests of homogeneity were done to compare the crash rate for 2013–2015 Virginia versus crash rate for 2013–2015 Fairfax. P value was highlighted along with odds ratios and their 95% confidence intervals. (2) Similarly, chi-square tests were also done for 2015–2017 crash rates of Virginia versus FC. The main parameter of interest was interaction between region (FC and rest of Virginia) and time to assess whether or not rate of accidents changed (decreased) at a faster rate in FC as opposed to the rest of Virginia in conjunction with the intervention of delaying start time in high schools in FCPS.

The rate of subcategories of crashes (injuries, distraction, speed and alcohol-related crashes) were also similarly compared separately for FC and rest of Virginia to determine whether there was a significant change in the rate of various subcategories of crashes from T1 to T2 within each of the two regions.

RESULTS

A total of 62,968 16-year-old drivers and 62,950 16- to 18-year-old drivers registered in FC in T1 and T2 periods respectively were compared to 358,315 and 372,474 in the rest of Virginia in T1 and T2 respectively. In the T1 period, there were 1,992 crashes, a rate of 31.63 per 1,000 licensed teen drivers (age 16 to 18 years) reported in FC (53.5% males) (Table 1), of whom 5.09 per 1,000 resulted in injuries, 7.01 per 1,000 were related to distraction while driving, 5.27 per 1,000 were speed related and 0.68 were deemed alcohol related. In the T2 period in FC, the crash rate per 1,000 licensed drivers significantly dropped compared to T1, from 31.63 to 29.59 accidents per 1,000 (95% CI, 1.0–1.14, P = .03) (49.2% male). Looking at the different subtypes of the accidents, distraction-related crashes was the only domain that approached statistical significance, decreasing from 7.01 in T1 to 6.13 in T2 (95% CI, 0.99–1.31, P = .05).

Table 1.

Comparison between adolescent driver car crash rates in Fairfax County and the rest of Virginia in 2013–2015 and 2015–2017.

In the rest of Virginia, there was a slight, nonstatistically significant increase in the crash rate during the two measurement periods from 56.25 per 1000 in T1 (50.5% male) to 57.06 per 1,000 in T2 (95% CI, 0.96–1.0, P = .13) (52.7% male). Furthermore, there were no differences in the four subtypes of crashes at T1 and T2.

DISCUSSION

The results of this study in one of the largest school districts in the United States suggest that there is an association between a delay in high school start times and reduced rates of motor vehicle crashes among adolescent drivers. This study address some of the shortcomings of previous research by including a 2-year before-and-after assessment period, comparing crash rates versus the rest of the state, and examining subcategories of crashes potentially relevant to the effects of chronic sleep loss and circadian misalignment in one of the largest school districts in the United States. Although a causal mechanism cannot be inferred from these cross-sectional data, the lack of a significant decrease in adolescent driver crash rates in the rest of the state during the same time period is supportive of the association. In addition, several previous studies have found a similar association between early school start times and adolescent crash rates^8,14,19 and a decrease in crash rates in the immediate period following start time change in smaller school districts.

As is true in many states, the Virginia DMV data used in this study did not include drowsy driving as a separate category for car crashes, so total crash data were by default the primary crash outcome analyzed. However, the analysis of subtypes of crashes showed that distraction-related driving decreased in FC at T2 compared to T1 but remained stable in the rest of the state. A recent systematic review reported that behaviors related to distracted driving (including texting, reading text messages, and talking on the phone) are associated with an increased risk of car crashes and are particularly common in young drivers.^22–24 The potential link between chronic sleep loss commonly experienced by adolescent drivers, especially those attending high schools with earlier start times,^25,26 and distracted driving behaviors has been previously demonstrated.¹² Although the mechanisms for this association remain somewhat speculative, there is clearly an association between insufficient sleep and deficits in self-regulation/executive functioning²⁷ which in turn can be associated with poor decision-making and increased risk-taking behavior in adolescents.^28–36 Although it should be noted that we did not see a change in the rates of other subtypes of crashes in FC associated with start time delay that might also be postulated to be related to poor self-regulation (speeding and alcohol related), the finding that distracted driving did decrease suggests that there may be multiple crash risk factors that improve with school start time delay.

Although we do not have sleep data on the specific sample described in this study, data from several other relevant sources in FC suggest that on average, public high school student drivers in the district experienced chronic sleep loss prior to the start time change and that school start time delay was associated with an increase in sleep duration and decrease in self-reported daytime sleepiness. A quasi-experimental survey study examining school start time data before and after the change showed that before start time changes, high school students slept a mean (standard deviation) of 7 hours, 25 minutes (70 minutes) on school nights and had a prevalence of daytime sleepiness of 78.4%.³⁷ Adjusted for potential confounders, following the 50-minute delay students slept 30.1 minutes longer (95% CI 24.3–36.0) on school nights and had less daytime sleepiness (−4.8%; 95% CI −8.5% to −1.1%).³⁷ In addition, a survey conducted in the spring of 2015 of 431 student drivers in FCPS found that prior to start time change, 42% of the students reported getting less than 7 hours of sleep on average on school nights, and 48% reported drowsy driving (defined as having ever in the last year “driven a car or motor vehicle while feeling drowsy”); the covariate-adjusted prevalence of drowsy driving was 13.9% (95% CI 3.0% to 24.9%) higher in students who slept less than 7 hours on school nights than in those who slept 8 or more hours.³⁸ Moreover, before-and-after analyses of health and safety outcomes associated with start time change in FCPS found that the percent of student drivers self-reporting drowsy driving declined significantly (−8.4%; 95% CI −15.9% to −0.9%; P < .05³⁹).

An additional factor to consider in assessing the effect of school start times on car crash rates is that of circadian misalignment. A pronounced shift to a more evening chronotype has been shown to coincide with pubertal development in adolescents.^40,41 Given that evening types cannot easily fall asleep earlier at night and must still meet societal demands to wake up and function early in the morning, this often results in a sleep duration that is insufficient to meet sleep needs. In addition, timing of activity and alertness patterns during the day are also shifted later. Several studies in adults (which included young adults) have suggested that an evening chronotype is associated with impaired driving.^42–44 Moreover, an evening chronotype has been reported to be specifically associated with increased risk-taking in potentially dangerous traffic conditions⁴⁵ In the FC study of student drivers discussed previously³⁸ compared with those with a morning chronotype, the adjusted prevalence of drowsy driving was 15.2% (95% CI 4.5% to 25.9%) higher among those with an evening chronotype. Thus, later school start times (8:30 am or later⁴⁶) may positively affect sleep and risk-taking behavior both by increasing sleep time and better aligning sleep timing.

In addition to those previously discussed, limitations of this study include incomplete data regarding whether other school districts in Virginia may have also changed (delayed or advanced) high school start times. Unfortunately, there is no centralized repository of these data for school districts in the United States. However, after reviewing the websites for all of the remaining 151 Virginia Public School Districts by region (excluding FC) listed by the Virginia Department of Education for high school start times in each district, to the best of our knowledge, no other school district made an equivalent change in high school start times at the beginning of the school year during the study period (fall 2013 to fall 2016). Additional limitations include the potential inclusion in the FC and Virginia samples of 16- to 18-year-old drivers who were not enrolled in a public high school, and lack of more granular data regarding day of the week or location of crashes. Because of the categorical nature of the DMV data, we were not able to examine the effect of potentially important combined factors such as speed plus alcohol. Although our time-of-day data, as shown in Figure 1, mirror that of previous studies, including one in FC, that have shown an increase in crash rates in high school drivers specifically during school commuting periods,^47–49 interestingly the crash rates in the 12:00 pm to 2:59 pm period appears higher in T1 compared to T2, potentially reflecting the differences in dismissal times before and after school start time change.

Figure 1. Time of day of motor vehicle crashes in Fairfax County and the rest of Virginia at T1 and T2.

We also do not have information about other factors potentially affecting adolescent crash rates such as commuting distance and traffic patterns. We chose to use Virginia DMV data rather than those from the FC Police Department database, in order to compare equivalent information at the county versus state levels. It is also unclear why the adolescent crash rate in the rest of the state was significantly higher both at T1 and T2 compared to FC (at least twofold); however, that further highlights the finding that there was a significant decline only in FC crash rates despite the lower base rate. Future studies should include a direct comparison of changes in sleep duration and daytime sleepiness associated with school start time change, including more objective measures of sleep such as actigraphy, drowsy driving measures, and car crash risk in adolescent drivers. In addition, statewide car crash databases need to begin to include drowsy driving as a subcategory of impairment-related crashes, similar to distraction-related and alcohol-related categories. Although this will necessarily involve a change in approach to defining and identifying contributing factors, especially on the part of law-enforcement officials present at the crash scene, these data are critical in regard to developing preventive strategies that are effective in reducing motor vehicle crashes, particularly in the adolescent population.

The results of this study support healthy high school start times as a potential mitigator of car crashes in adolescents and supplies further evidence of the implications for public health and safety of delaying school start times. Pairing this operational intervention with educational efforts focusing on the dangers of drowsy driving in adolescents, including as a mandatory component of drivers’ education programs and in public service campaigns targeting parents of teen drivers, as well as with legislation restricting driving for junior operators,⁵⁰ are likely to further extend the benefits of school start time change.

ABBREVIATIONS

BAC

blood alcohol content

DMV

Department of Motor Vehicles

FC

Fairfax County

FCPS

Fairfax County Public Schools

T1

2013-2014 and 2014-2015 period

T2

2015-2016 and 2016-2017 period

DISCLOSURE STATEMENT

All authors have seen and approved the manuscript. Work for this study was conducted at Boston Children’s Hospital. The authors report no conflicts of interest.