Department of Epidemiology, College of Public Health, University of Iowa, Iowa City, IA, United States; Injury Prevention and Research Center, College of Public Health, University of Iowa, Iowa City, IA, United States.
Department of Occupational and Environmental Health, College of Public Health, University of Iowa, Iowa City, IA, United States; Injury Prevention and Research Center, College of Public Health, University of Iowa, Iowa City, IA, United States.
J Safety Res. 2019 Feb;68:215-222. doi: 10.1016/j.jsr.2018.10.014. Epub 2018 Nov 13.
Classifying risky driving among new teenage drivers is important for efficiently targeting driving interventions. We thoroughly investigated whether novice drivers can be clustered by their driving outcome profiles over time.
A sample of 51 newly licensed teen drivers was recruited and followed over a period of 20 weeks. An in-vehicle video recording system was used to gather data on dangerous driving events referred to as DDEs (elevated g-force, near-crash, and crash events), risky driving behaviors referred to as RDBs (e.g., running stop signs, cell phone use while driving), and miles traveled. The DDE and RDB weekly rates rate were determined by dividing the number of DDEs and RDBs in a week by the number of miles traveled in that week, respectively. Group-based trajectory modeling was used to map the clustering of DDE rate and RDB rate patterns over time and their associated covariates.
Two distinct DDE rate patterns were found. The first group (69.1% of the study population) had a lower DDE rate which was consistent over time. The second had a higher DDE rate pattern (30.9%) and characterized by a rising trend in DDE rate followed by a steady decrease (inverted U-shaped pattern). Two RDB rate patterns were also identified: a lower RDB rate pattern (83.4% of the study population) and a higher RDB rate pattern (16.6%). RDB and DDE rate patterns were positively related, and therefore, co-occurred. The results also showed that males were more likely than females to be in the higher DDE and RDB rate patterns.
The groups identified by trajectory models may be useful for targeting driving interventions to teens that would mostly benefit as the different trajectories may represent different crash risk levels. Practical applications: Parents using feedback devices to monitor the driving performance of their teens can use the initial weeks of independent driving to classify their teens as low or high-risk drivers. Teens making a very few DDEs during their early weeks of independent driving are likely to remain in the lower risk group over time and can be spared from monitoring and interventions. However, teens making many DDEs during their initial weeks of unsupervised driving are likely to continue to make even more DDEs and would require careful monitoring and targeted interventions.
对新青少年驾驶员的危险驾驶行为进行分类对于有效地针对驾驶干预措施非常重要。我们深入研究了新手驾驶员是否可以根据其随时间推移的驾驶结果分布进行聚类。
招募了 51 名新获得驾照的青少年驾驶员并对其进行了 20 周的随访。使用车载视频记录系统收集危险驾驶事件（高 g 力、近撞和撞车事件）和危险驾驶行为（例如，闯红灯、开车时使用手机）以及行驶里程的数据。每周 DDE 率和 RDB 率分别通过将一周内的 DDE 数量和 RDB 数量除以一周内行驶的英里数来确定。使用基于群组的轨迹建模来映射随时间推移的 DDE 率和 RDB 率模式的聚类及其相关协变量。
发现了两种不同的 DDE 率模式。第一组（研究人群的 69.1%）的 DDE 率较低且保持一致。第二组的 DDE 率较高（30.9%），特点是 DDE 率呈上升趋势，随后稳定下降（倒 U 型模式）。还确定了两种 RDB 率模式：较低的 RDB 率模式（研究人群的 83.4%）和较高的 RDB 率模式（16.6%）。RDB 和 DDE 率模式呈正相关，因此同时发生。结果还表明，男性比女性更有可能处于较高的 DDE 和 RDB 率模式。
轨迹模型识别的群体可能有助于针对青少年进行驾驶干预，因为不同的轨迹可能代表不同的碰撞风险水平，大多数青少年将从中受益。实际应用：父母使用反馈设备来监测青少年的驾驶表现，可以使用青少年独立驾驶的最初几周来将他们归类为低风险或高风险驾驶员。青少年在独立驾驶的最初几周内很少发生 DDE，则随着时间的推移，他们很可能继续处于低风险群体中，从而可以免于监控和干预。但是，在最初几周内未受监督的驾驶中发生许多 DDE 的青少年很可能会继续发生更多的 DDE，因此需要仔细监控和有针对性的干预。
