OBJECTIVES:To evaluate child restraint use during transport to healthcare facilities and to identify age-related patterns and determinants of inappropriate restraint use. METHODS:A cross-sectional study, it included 502 caregivers of children aged 0-17 years presenting to a pediatric emergency department or outpatient clinic in Portugal between May 2025 and January 2026. Data were collected using a structured questionnaire. Restraint appropriateness was classified using a predefined algorithm aligned with United Nations Economic Commission for Europe (UNECE) Regulations No. 44 and No. 129. Associations were assessed using univariable analyses and multivariable logistic regression. RESULTS:Overall, 72.9% of transports were classified as appropriate and 27.1% as inappropriate. Appropriate child restraint use declined significantly with age, reaching its lowest levels among children aged 9-12 years, in whom more than half of transports were inappropriate (p < 0.001). Self-reported familiarity with child transport legislation was high (85.1%), although objective knowledge of specific safety criteria was substantially lower and was not independently associated with transport appropriateness. Absence of prior professional guidance (OR 1.94, 95% CI 1.20-3.11) and uncertainty regarding prior guidance (OR 3.85, 95% CI 1.60-9.25) were independently associated with inappropriate child restraint use. CONCLUSIONS:Inappropriate child restraint use during transport to healthcare facilities is common and shows a marked age-related decline. These findings identify middle childhood as a vulnerable period and support the integration of structured child passenger safety counseling into routine healthcare.
OBJECTIVE:Pedestrian fatality risks in low- and middle-income countries remain high due to mixed traffic conditions, roadside settlement activities, and inadequate pedestrian infrastructure. This study examines the associations between micro-level roadside infrastructure characteristics and pedestrian fatalities and identifies the spatial patterns and temporal trends of pedestrian fatalities along a high-speed multilane highway in India. METHODS:This study analyzed 1,687 pedestrian fatalities recorded between 2017 and 2022 using police-reported crash data along National Highway 44 in Haryana, India. Temporal heatmaps were used to identify high-risk periods, while kernel density estimation (KDE) was used to visualize the spatial concentration of pedestrian fatalities. A negative binomial regression model was employed to examine the associations between roadside infrastructure, land-use characteristics, and pedestrian fatalities. RESULTS:The model results indicated that the presence of earthen shoulders, service roads, and street lighting was associated with lower pedestrian fatalities, whereas junctions and bus stop locations were associated with higher fatalities. Higher concentrations of pedestrian fatalities were observed in commercial and mixed-use areas. Temporal analysis further showed that approximately 32% of pedestrian fatalities occurred during evening peak hours (18:00-21:00), with higher fatalities during weekends and nighttime periods. CONCLUSIONS:The findings highlight important associations between roadside infrastructure, land-use activity, and pedestrian fatality patterns along multilane highways. These results can support the identification of high-risk locations and guide targeted infrastructure improvements, including safer bus stop placement, improved lighting, pedestrian crossing facilities, and junction management. The findings can also assist planners and policymakers in developing effective strategies to improve pedestrian safety in India and similar low- and middle-income countries.
OBJECTIVE:Reliably determining collision risk in pedestrian-vehicle interactions remains a critical challenge for autonomous vehicles (AVs). Experienced human drivers are able to quickly evaluate the overall trend of a scene and predict potential risks based on their experience. In contrast, existing risk assessment models often lack experience-based judgment. This study develops and validates an expert-guided framework to quantify pedestrian-vehicle interaction risk by learning the takeover behaviors of safety drivers in AVs. METHODS:We extracted 113 real-world, high-risk pedestrian-vehicle interaction cases from an autonomous driving database based on extensive road-testing data. We recorded instances of takeover events during high-risk situations. To thoroughly analyze the limited yet factual data from these events, we developed an XGBoost model to infer drivers' judgments leading to takeovers. Key risk factors were then identified from the perspective of safety drivers. Finally, we validated the model's performance by assessing its consistency with actual human takeover behaviors. RESULTS:Through an analysis of high-risk takeover events in real emergency situations, the developed model was able to predict the takeover decisions of safety drivers with an accuracy of 92.2% and an F1 score of 91.3% using 765 time-sliced samples extracted from 113 high-risk cases. Interpretability analysis revealed that Time-to-Collision (TTC) and lateral distance are the main factors influencing the safety driver's risk perception and subsequent takeover behaviors. Furthermore, a comparison between the feature distributions at model-identified high-risk moments and those at actual driver takeovers revealed strong consistency in key indicators such as TTC and lateral distance. CONCLUSIONS:Based on real-world data from high-risk pedestrian-vehicle interactions, this study models interaction risk based on experienced drivers' risk perception and highlights key factors influencing human risk assessment. These findings provide insights for developing more human-aligned and interpretable risk assessment frameworks in autonomous driving.
OBJECTIVE:The current study was conducted to assess the relationship between speeding behavior and handheld cellphone manipulation on different roadway classes. METHODS:Cambridge Mobile Telematics provided a dataset of trips in the United States made by drivers who used its smartphone-based platform from July through October 2024. The final sample comprised 35,000-40,000 trips per U.S. Census region each month (N = 593,454 trips). We used negative binomial regression models to predict the cellphone manipulation rate based on the speed limit and an indicator for how fast the vehicle was traveling relative to that limit during free-flow travel. An interaction term between speed limit and speeding behavior was included with covariates for time of day, day type, and area type. Models for limited-access and primary/collector roads were estimated separately due to little overlap in speed limits. RESULTS:A 5 mph increase in speed relative to the limit was associated with an increase in cellphone manipulation, although the magnitude of this increase was nearly 4 times as great on limited-access roads as on primary/collector roads. Statistically significant interactions between speeding and speed limit showed that the relationship between speeding magnitude and cellphone manipulation rates was exacerbated on both road types when the speed limit was higher. CONCLUSIONS:Both the reduced complexity of higher speed roads and the tendency for some drivers to take multiple risks likely help explain why speeding and cellphone manipulation occur together. These findings could aid traffic law enforcement in identifying locations where enforcing speeding and handheld cellphone laws together would be most effective. Countermeasures that raise perceived roadway complexity may also reduce the likelihood of both phone manipulation and speeding.
OBJECTIVE:The study addressed the multifaceted nature of severe curve-based head-on crashes by identifying hidden high-risk scenarios/combinations, stemming from higher-order interaction between driver, crash, environmental, traffic and roadway characteristics along with spatial relationship indicators. METHOD:Association rule mining (ARM), due to its greater flexibility in handling and quantifying interactions than conventional models, was used to extract the high-risk scenarios/combinations from 533 curve-based head-on crashes in the mountainous state of Himachal Pradesh, India. Higher-order interactions were manifested through 3 and 4-factor interactions after fixing severity as the consequent. A minimum support, confidence/severity-rate, and lift of 2%, 50%, and 1.3, respectively, were established for extracting the initial rule space. From this space, based on an absolute raw support, "key rules" with raw support >25 and "exceptions" with that under 25 but with lift ≥ 1.9, were further extracted to prioritize significant associations. These were then integrated after eliminating redundant rules and the final rule set constituted 20 key rules and 4 exceptions. The stability and generalizability of the rules were validated through Fisher's exact test and stratified bootstrap sampling-based stability analysis. RESULTS:Aggressive driving was a key trigger of many high-risk scenarios. Ineffective/inadequate risk communication and visibility restricted by narrow mountainsides (<1 m) increased the collision susceptibility on medium speed limit curves (30-50 kmph) by 1.6 times. Middle-aged heavy vehicle operators emerged as a vulnerable group on sections characterized opposing sequences/reverse curves, pavement width < 7 m and insufficient valley side clearance/buffer (<2.5 m). Severe collisions were 1.35 times more likely on longer curves (>90 m) with narrow mountainsides. Opposing sequences emerged as a significant hotspot of severe head-on crashes especially under conditions involving inadequate mountainside visibility (<1.5 m) and insufficient valley side buffer (<2.5 m). An important finding of the present study is that a relatively sharper curve in proximity (<240 m) increased the risk of a severe crash on a given subject curve. Specifically, very sharp curves (radius < 40m, length: 30-60 m) in succession emerged as one of the riskiest scenarios, doubling the collision risk. Collision risk increased significantly (1.986 times) with longer preceding curves, indicating potential hotspots. CONCLUSION:Curve-based head-on crashes exhibited increased severity and produced multiple hotspots under specific combinations/scenarios involving aggressive driver behavior, mountain side visibility constraints, insufficient valley side buffer, vehicle-specific dynamics and inadequate pavement width. The associated risks were particularly exacerbated on closely-spaced curve sequences featuring sharper or longer approach curves and opposing orientations. The consistent involvement of spatial relationship indicators in multiple high-risk scenarios, advocates shift toward a system-based curve safety assessment, especially in mountainous areas. The identified high-risk scenarios/locations/configurations facilitate comprehensive safety profiling based on factor combinations instead of individual features, for designing multiple targeted interventions.
OBJECTIVES:Secondary crashes on freeways pose significant safety risks and are often preventable with timely intervention. This study aims to develop a real-time prediction framework for secondary-crash risk using traffic flow precursor characteristics, enabling proactive traffic safety management. METHODS:A novel secondary-crash identification method based on a crash buffer and speed contour map was first proposed to accurately determine the spatiotemporal influence range of primary crashes. Using historical crash data from Interstate I-405, traffic flow features (including flow rate, speed, and occupancy) were extracted. Four machine learning algorithms-Support Vector Machine (SVM), Random Forest (RF), XGBoost, and CatBoost-were employed to construct prediction models. A two-tier prediction framework was developed, consisting of a primary-crash risk prediction sub-model and a secondary-crash risk prediction sub-model. The outputs of the two sub-models were integrated using a voting strategy. Model performance was evaluated using Accuracy, Precision, Recall, F1-score, and AUC. RESULTS:The proposed identification method effectively distinguished secondary crashes from historical data. Among the tested algorithms, CatBoost demonstrated the best overall performance in both sub-models. The two-tier prediction framework outperformed single‑tier models, achieving higher recall and AUC values for secondary‑crash detection. Although the integrated model slightly reduced precision, it significantly improved the identification of secondary-crash events, which is critical for safety-sensitive applications. The model also provided probabilistic risk outputs, enhancing interpretability for traffic management decision-making. CONCLUSIONS:This study presents an effective and practical framework for real-time prediction of freeway secondary-crash risk. The proposed two-tier prediction framework, combined with traffic flow precursor features and machine learning techniques, improves prediction accuracy and enhances early warning capabilities. The findings provide theoretical and methodological support for proactive traffic safety management and secondary-crash prevention. Future work should incorporate more diverse datasets and external factors (e.g., weather and road conditions) to further improve model generalizability.
OBJECTIVE:Long railway tunnels expose train drivers to abrupt luminance transitions, spatial confinement, and monotonous visual environments that may elevate physiological load and compromise operational safety. This study aims to characterize the dynamic evolution of train drivers' physiological load across distinct sections of a long railway tunnel and to establish a quantitative multi-indicator evaluation framework to support tunnel safety design and driver-state monitoring. METHODS:Naturalistic driving experiments were conducted in the Heishan Railway Tunnel (16 km) in western China, with drivers traversing the tunnel at a constant speed of 155 km/h. Drivers' oculomotor signals (pupil diameter, mean fixation duration, standard deviation of fixation-point distribution), electrodermal activity (EDA), and heart rate variability (SDNN) were synchronously recorded using a Tobii Pro Glasses 3 eye tracker and a BIOPAC MP160 system. The evaluation path (17 km) was divided into entrance, transition, middle, and exit sections. A fuzzy comprehensive evaluation (FCE) model integrated with the entropy weight method (EWM) was developed for quantitative load assessment. RESULTS:All five section-discriminative indicators showed statistically significant between-section differences with medium-to-large effect sizes. The entrance section yielded the highest composite physiological load score (76.74), reflecting an overt visual-sympathetic dual-activation stress response. The exit section ranked second (74.08), corresponding to a reactivation response. The middle section produced the lowest composite score (69.41), but multi-indicator joint analysis revealed that this low score reflects reduced stress-response intensity rather than reduced risk, corresponding to a covert hypovigilant state rather than a true safety zone. The transition section showed intermediate values (72.43). CONCLUSIONS:Train drivers in long railway tunnels exhibit a distinctive three-zone risk structure-overt high-load at the entrance, covert hypovigilance in the middle, and reactivation at the exit-differing qualitatively from patterns reported for highway tunnels. The EWM-FCE framework established here provides a quantitative basis for railway tunnel safety design and driver-state monitoring.
OBJECTIVE:To compare the head injury protection offered by rearward-facing child restraint system (CRS) models without a base attached using either the United States (US) or European belt path during frontal-oblique impacts. METHODS:The test buck comprised the Consumer Reports simulated test bench, Federal Motor Vehicle Safety Standards (FMVSS) 213a door surrogate, post-mounted shoulder belt and high-speed camera array. The test buck did not include front-row structures and was rotated relative to the sled deck to represent frontal-oblique far-side (30°) and near-side (30° and 60°) impacts. The 12-month-old Child Restraint/Air Bag Interaction (CRABI-12) anthropomorphic test device (ATD) was seated in one of two rearward-facing infant CRS models without a base attached using either the US or European belt path. A total of 12 sled tests were performed using the FMVSS 213 frontal impact crash pulse (48 km/h, 23 g). Peak forward excursion of the CRS, peak lateral excursion of the ATD head, head injury criterion 15 ms (HIC15) and peak linear accelerations 3 ms clip (PLA3) of the head and chest of the ATD were compared across tests and related to belt path. Head and chest injury metrics were compared to injury assessment reference value (IARVs): HIC15, 390; head PLA3, 80 g; chest PLA3, 60 g. RESULTS:For both CRS models in 30° far- and near-side impact tests, the US belt path had higher head injury metrics compared to the European belt path. For both CRS models in 60° near-side impacts, head and chest injury metrics were greater for the European belt path compared to the US belt path. Head and chest injury metrics were typically below IARVs in 30° far- and near-side impact tests, but IARVs were exceeded in 60° near-side impact tests. For all impacts, CRS models attached using the US belt path underwent substantially greater peak forward excursions compared to the European belt path. CRS models attached using the European belt path in the in 60° impacts had the lowest peak forward excursions. For 30° far-side impacts, peak lateral excursion of the ATD head was greater for CRS models attached using the European belt path compared to the US belt path. CONCLUSIONS:The belt path attachment method of rearward-facing infant CRS models installed without a base influenced pediatric occupant kinematics and injury metrics in frontal-oblique impacts. The European belt path reduced head injury metrics relative to the US belt path for 30° impacts, likely due to reduced forward excursion, but resulted in greater head and chest injury metrics in 60° impacts where increased lateral loading likely exacerbated outboard yaw motion of the CRS into the door surrogate. Conversely, the US belt path produced substantially greater forward excursions that may potentially lead to interaction with front-row vehicle structures, suggesting that head injury metrics were likely underestimated in the current study.
OBJECTIVES:Existing autonomous vehicle (AV) crash-injury studies are predominantly variable-centered and often separate scenario classification from injury modeling. This study developed a scenario-first framework to determine whether injury probability conditional on an observed crash is organized primarily by broad pre-crash interaction structures or by localized operational design domain (ODD) conditions. METHODS:A validated database of 2,946 AV crashes recorded from January 2015 to July 2024 was assembled from regulatory reports and verified public sources. Without using injury outcomes, crashes were mapped to five scenario skeletons and clustered within skeletons using Gower distance and k-medoids to derive 10 scenario-ODD phenotypes. Dual-baseline injury enrichment, Bayesian hierarchical models, model comparisons, bootstrap stability checks, and sensitivity analyses assessed adjusted injury differences and robustness. RESULTS:No phenotype met the strong dual-baseline enrichment criterion. S05-P01 had the highest observed injury proportion (51.9%) and global enrichment (RRg = 1.41) but no within-skeleton enrichment (RRss = 1.00). The two vulnerable-road-user phenotypes showed elevated global injury tendencies (RRg = 1.31-1.33), whereas the three rear-end/longitudinal phenotypes had lower injury proportions (RRg = 0.55-0.70). Seven of nine estimable phenotype deviations had 95% credible intervals excluding zero, although their magnitudes were modest. M2 had the lowest Brier score (0.227) and BIC (3939.5) and the highest AUROC (0.719); M5 had the highest AUPRC (0.431) and most favorable calibration slope (0.689). Seven of 10 phenotype classifications were unchanged across all threshold bundles. CONCLUSIONS:Pre-crash interaction structure provided the primary organization of conditional injury probability, while the phenotype layer identified localized within-skeleton variation and translated broad scenarios into operationally interpretable ODD micro-contexts. The two levels support complementary tasks in scenario testing, ODD review, and targeted safety validation.
OBJECTIVE:In the U.S., motor vehicle crashes are the leading cause of death for adolescents ages 15 - 18, and alcohol use increases the risk of crashes. There is a need for greater clarity regarding the factors most effective for behavior change in preventing teen driving after drinking. Given the centrality of teen-parent relationships in influencing teen behavior, understanding the relational dynamics in this context is vital to prevention. To inform the development of more effective prevention programs for teens and parents, this study examines teen drivers' and parents of teen drivers' perspectives on drinking and driving behaviors, prevention, and parental involvement. METHODS:Surveys and focus groups were conducted among 41 teens (ages 15 - 18) and 27 parents (of teens ages 15 - 18) residing in 11 cities in Orange County, California. Participants completed an online survey assessing demographic information, driving behaviors, family relationships, physical and mental health, and substance use history. Participants then attended a one-hour in-person focus group, held separately for teens (5 focus groups; 7-13 participants per group) and parents (4 focus groups; 5-8 participants per group). During these sessions, participants engaged in open discussions about imagined scenarios involving riding with or driving while impaired, as well as imaginal scenarios designed to explore perspectives, decision-making, and prevention strategies. Focus group transcripts were coded by a multidisciplinary team using a deductive approach to explore and summarize themes. RESULTS:Teens reported varying experiences with alcohol, yet overall described a strong safety orientation, awareness of consequences, openness to parental limits, and interest in having explicit safety conversations related to driving after drinking. Although teens differed in their preferences for the timing and duration of these discussions, most expressed a desire for genuine parental understanding. Parents similarly described differing perceptions of teen alcohol use, with some expressing trust in their teens not using alcohol and others acknowledging alcohol use as common and driving after drinking as a notable concern. Parenting approaches regarding drinking ranged from expectations of strict alcohol abstinence from their teens to more permissive practices, including permitting supervised and social drinking. Parents reported a preference that their teens first turn to them for support and, when necessary, seek help from other trusted, practical sources, including family, friends, or rideshare services. CONCLUSIONS:Findings emphasized the need for proactive prevention strategies targeting driving after drinking. Teens suggested that prevention efforts should collaboratively engage both teens and parents to improve safety outcomes. Parents highlighted that comprehensive prevention efforts should extend beyond the home to multiple contexts and be adaptable to diverse family structures. These study findings present directions for future research, prevention activities, and programs.
OBJECTIVE:Future autonomous vehicles increase the use of reclined seating configurations, driving the need for adapted occupant restraint systems that are increasingly developed using human body models (HBMs). This study provides biofidelic target data for positioning HBMs in reclined postures, addressing the current limitation of default upright posture. METHODS:An existing MRI dataset of 11 volunteers (5 male, 6 female; age 39.2 ± 12.7 years; height 174 ± 9.3 cm; body mass index (BMI) 24.8 ± 2.7) was analyzed to quantify spinal curvature and pelvic alignment. Participants were scanned while seated in a replica of a vehicle seat in upright (U) and reclined (R) seating configurations (20° and 50° backrest angle). For each posture, scan markers were placed on the endplates of sacrum and vertebral bodies from L5 to T6 in midsagittal plane, yielding 50 two-dimensional markers per volunteer and posture. These markers were utilized for calculation of intervertebral angles (IVA) and global angles including sacral slope (SS), lumbar lordosis (LL: L5-L1), and thoracolumbar angle (TLK: L2-T10). Additionally, pelvis angle (PA) was measured as the angle between the line connecting the pubic tubercle with the anterior superior iliac spine and the vertical axis. Combining PA with SS resulted in the composite angle PASS. To characterize global spinal curvature, third-order polynomial curve-fitting and principal component analysis (PCA) with marker coordinates normalized by arc length were analyzed. Polynomial curve-fitting was also applied in a preliminary manner to explore potential transformations into reclined dependent on upright posture. RESULTS:Spearman's rho test revealed significant correlations in reclined posture (* = p < 0.05; ** = p < 0.01) between LL and SS (ρR=-0.83∗∗), and between LL and PASS (ρR=-0.74∗). No significant relationship was found between LL and PA in both seating configurations (ρU=-0.38,ρR=-0.43). Relations between respective global angles were statistically significant positively related with larger seatback inclination (ρPA=0.67∗,ρPASS=0.76∗∗,ρSS=0.80∗∗,ρLL=0.94∗∗,ρTLK=0.67∗). Lumbar IVAs exceeded thoracic IVAs in both postures, with the largest posture-related changes in lumbar segments L5/L4 and L4/L3 as well as SS. PCA of normalized spinal curvature identified three principal components accounting for more than 97% of the total variance across seating configurations, while the first component (≥ 85%) demonstrates the greatest changes in SS and lumbar IVAs at ±2 SD. Fitting of a polynomial curve resulted in a good fit with a third order polynomial for both seating postures. Prediction of reclined posture using polynomial curves could be achieved with a mean RMSE of 13.82 mm. CONCLUSIONS:Created data revealed significant posture-dependent changes in sacral slope and lumbar IVAs. A preliminary methodology for predicting reclined spinal curvature from upright posture is presented. Findings require further validation with larger datasets and consideration of covariates to ensure robustness.
OBJECTIVE:Ankle injury is one of the most common AIS 2 motor vehicle crash injury in the lower extremity. Prior work suggests females may be at greater risk for ankle fracture than males. Accurately predicting such ankle injuries using human body models (HBMs), particularly in frontal collisions, presents a unique challenge due to the complex loading mechanisms and differences in local geometry. Injury predictors derived from tissue level testing may be able to help overcome these challenges through strain-based injury prediction. This study aimed to provide tissue-level strain data for injury prediction through matched simulations with past Postmortem Human Subjects (PMHS) studies and morphed, subject-specific Total Human Model for Safety (THUMS) V6.1 tibiae. METHODS:Simulations were created for 17 PMHS tibiae subjected to inferomedial dynamic loading to generate medial malleolar fractures (previously published). THUMS tibiae were morphed to maintain baseline THUMS mesh size while generating accurate local geometry for each PMHS. Boundary and displacement loading conditions were matched to each subject-specific experimental test. Values of maximum principal strain (MPS) and 95th percentile maximum percentile strain (MPS95) of the tibia cortical bone were extracted from each simulation at the vertical force, vertical displacement, and energy at fracture from the matched experimental tests. Local injury risk curves were then developed with three functions (Weibull, log-normal, log-logistic) to generate potential curves to assess injury based on MPS and MPS95. Age and sex were considered as possible covariates. RESULTS:All simulations ran to completion and exhibited force and strain behavior consistent with the experimental work. Force-displacement curves displayed a softer response for the HBM compared to PMHS tests. Axial force at fracture, calculated at the ankle joint center, ranged from 1900 to 7500 N and shear force from 1100 to 4200 N. Max principal strain at fracture ranged from 1.9% to 9.2% and from 0.7% to 6.2% for MPS95. Specimen age and sex were initially considered as covariates but lacked statistically significance and were ultimately removed from these prediction metrics. The Weibull function was the best fit for the MPS and MPS95 risk functions. The Weibull function without covariates was used to generate fits of the cumulative distribution for MPS and MPS95. CONCLUSIONS:This work used MPS and MPS95 to predict AIS 2 distal tibia injury using THUMS V6.1 simulations matched to past physical PMHS testing. These simulations demonstrated similar strain at fracture to typical failure values for cortical bone. Differences in HBM stiffness showed a potential challenge when applying untuned tissue-level injury metrics to the human body model. The curves generated in this study can be a useful foundation for approximating gross ankle injury based on MPS and MPS95 with THUMS V6.1.
OBJECTIVES:To describe the relative recorded contribution of commuting crashes to occupational road traffic injuries (ORTIs) in Spain between 2020 and 2024 using recent official surveillance reports, and to summarize available subgroup patterns by age, sex, occupation, and economic activity when comparable information was reported. METHODS:An integrated descriptive synthesis was conducted using publicly available annual reports published by the Spanish National Institute for Safety and Health at Work (Instituto Nacional de Seguridad y Salud en el Trabajo, INSST). Reports for 2020, 2021, 2022, 2023, and 2024 were reviewed in full. Extracted variables included total ORTIs, at-work crashes, commuting crashes, and the proportion of ORTIs among all occupational injuries with sick leave. Secondary variables were extracted when sufficiently comparable across reports, including age group, sex, occupation, economic activity division, contract type, and selected severity-related indicators. A structured extraction matrix was used. The analysis was descriptive, and annual commuting-to-at-work ratios were calculated to summarize the relative recorded burden of commuting crashes compared with at-work crashes. RESULTS:ORTIs in Spain increased from 52,248 cases in 2020 to 76,327 in 2024, while their proportion among all occupational injuries with sick leave rose from 10.3% to 11.8%. Commuting crashes accounted for more recorded ORTIs than at-work crashes in every year analyzed, with commuting-to-at-work ratios of 2.24 in 2020, 2.17 in 2021, 2.20 in 2022, 2.40 in 2023, and 2.58 in 2024. Across the available reports, younger workers appeared among the categories with higher recorded incidence for at-work ORTIs, while driving-related occupations, particularly drivers and mobile machinery operators, and postal/courier and transport-related activity divisions were frequently reported among the main at-work ORTI categories. Subgroup findings for commuting ORTIs were less uniformly reported across years and should therefore be interpreted cautiously. Later reports provided more detailed sex-disaggregated information, particularly for commuting crashes. CONCLUSIONS:Recent official Spanish surveillance reports show that commuting crashes account for a larger recorded share of ORTIs than at-work crashes. These findings support giving commuting more explicit attention in occupational road safety surveillance. However, because the study is based on aggregated annual reports, the results should be interpreted as evidence of burden distribution rather than individual-level risk or causal mechanisms. Future research using individual-level data should examine exposure, transport mode, commuting distance, shift schedules, fatigue, and work-organization factors.
OBJECTIVES:This study aimed to determine the prevalence of alcohol mixed with energy drink (AmED) usage among commercial drivers in Ghana and examine associated demographic, behavioral, physical, and psychosocial work factors. METHODS:This analytical cross-sectional study used a convenience sampling technique to recruit 7,540 commercial drivers from the Greater Accra Region. Bivariate analyses and hierarchical binary logistic regression were used to identify individual and organization factors associated with AmED use. Independent-samples t-tests were performed to compare users and non-users on safety- and mental health-related outcomes. Analysis was done using Jamovi statistical software version 2.6.23.0. RESULTS:About 29% of drivers reported the use of AmED during work in the 30 days prior to data collection. AmED use was more prevalent among younger drivers and those with shorter driving experience, longer work hours, and reported higher job demands and job insecurity. Drivers reporting AmED use showed statistically higher rates of risky driving indicators, sleep problems, and depressive symptoms. CONCLUSIONS:The findings indicate that AmED use is relatively common among commercial drivers in Ghana. One out of every four drivers sampled use AmED during work. It is recommended that road safety agencies in Ghana implement routine random roadside alcohol and stimulant screening (including AmED detection) among commercial drivers to enhance monitoring and early identification of potential driving impairment.
OBJECTIVES:To estimate the association between moderate-to-severe generalized anxiety and past-year drink-driving in Korean adults and test whether this association differs by sex. The study addresses a road-safety question that is not captured by alcohol-use measures alone: whether a standardized anxiety signal is visible in population-level drink-driving surveillance. METHODS:We conducted a cross-sectional complex-survey analysis of pooled 2021-2024 Korea National Health and Nutrition Examination Survey microdata. The analytic sample included 20,859 adults aged 19 years or older. The exposure was a GAD-7 score of 10 or higher, with sensitivity analyses using the 8 or higher cut point and continuous GAD-7 score. The outcome was self-reported past-12-month drink-driving, recoded as binary. Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) were estimated using KNHANES strata, primary sampling units, and interview weights. Models adjusted for sex, age, urban or rural residence, income, education, occupation, marital status, and family-relationship status; additional analyses examined sex interaction, depressive-symptom co-adjustment, alcohol-use adjustment, multiple imputation for missing data, and restriction to past-year drinkers. RESULTS:Weighted drink-driving prevalence was 1.6%, and GAD-7 score of 10 or higher prevalence was 4.6%. Moderate-to-severe anxiety was associated with higher adjusted odds of drink-driving after demographic, socioeconomic, and family-context adjustment (aOR, 2.20; 95% CI, 1.08-4.45; p = 0.03). The GAD-7 by sex interaction was not statistically significant (p = 0.13). The association was directionally consistent using a GAD-7 cut point of 8 or higher (aOR, 2.06; 95% CI, 1.17-3.62; p = 0.01), a continuous GAD-7 score per 5-point increase (aOR, 1.58; 95% CI, 1.35-1.84; p < 0.001), multiple imputation of missing data (aOR, 2.23; 95% CI, 1.14-4.35; p = 0.02), and depressive-symptom co-adjustment (aOR, 3.30; 95% CI, 1.50-7.25; p = 0.003). Additional alcohol-use adjustment attenuated the estimate (aOR, 1.95; 95% CI, 0.93-4.10; p = 0.08), and the past-year-drinker analysis produced a similar point estimate. Thus, the anxiety signal weakened after alcohol-use adjustment but remained near twofold in magnitude. CONCLUSIONS:Moderate-to-severe anxiety was associated with higher odds of self-reported drink-driving in Korean adults, but the association weakened after alcohol-use adjustment. These findings should be interpreted as hypothesis-generating rather than causal. They do not imply that anxiety alone predicts drink-driving or that anxiety screening should be used for enforcement. Instead, they suggest that drink-driving may be one point where alcohol use, anxiety, and risk behavior meet. Future drink-driving prevention research should measure anxiety, alcohol use, and driving exposure together rather than treating drink-driving only as an alcohol-use problem.
OBJECTIVES:The emergence of autonomous vehicles is expected to introduce nonstandard seating configurations, including reclined postures, that fall outside the design basis of current restraint systems. A growing body of research has investigated occupant kinematics in these configurations, but most of this work relies on anthropomorphic test devices or postmortem human subjects. Human volunteer data, which captures active musculoskeletal responses and demographic variability, remains limited and has not been synthesized across studies. METHODS:This scoping review identified 10 human volunteer studies encompassing 127 volunteers and 1,327 trials that examined occupant kinematics in reclined seating during pre-impact braking, lane change, and low-severity impact events. Studies were included if they used human volunteers in reclined positions ≥40˚. Studies were excluded if they exclusively used ATDs, PMHS, or computational modeling. Studies were classified by B-pillar mounted or seat-integrated belt types. A weighted meta-regression was performed across 20 group-level observations from 6 studies to test whether D-ring type influences the relationship between seatback angle and forward head excursion while controlling for braking severity and sex composition. RESULTS:The meta-regression identified a significant interaction between seatback angle and D-ring type (β = -9.49, p < 0.0001, R2 = 0.868). B-pillar belts produced increasing forward head excursion with recline (+7.20 mm/deg), while integrated belts showed the opposite trend (-2.29 mm/deg). This interaction remained significant after controlling for peak deceleration and percent male (p < 0.0001, R2 = 0.917), and after conducting a leave-one-study-out sensitivity analysis (p ≤ 0.015). The same pattern was observed at the torso. Males showed more excursion under B-pillar restraint at reclined angles, while females showed more excursion under integrated restraint. Belt forces decreased with recline regardless of D-ring type, indicating a load path shift from belt to seat structure. CONCLUSIONS:These results identify D-ring anchor location as the primary identifiable moderator of the inconsistencies previously reported across studies, rather than platform setup (sled vs. vehicle) or pulse severity. Other factors, such as seat pan angle, anthropometry, and pulse characteristics could also contribute but could not be fully controlled for in this study. Because forward excursion depends on D-ring anchor location, kinematic data from one belt configuration should not be used to validate computational models designed for the other. The lack of impact-severity data for females, older adults, and high-BMI occupants represents a critical gap in the current volunteer dataset that should be addressed before computational models can be considered broadly validated. Future volunteer testing should include these populations to establish response corridors that ensure restraint systems in highly autonomous vehicles are designed and validated across the full range of real-world occupant anthropometry.
OBJECTIVE:Exit ramps are the most crash-prone locations on motorways due to frequent conflicts between exiting vehicles and through traffic, often resulting in rear-end and angle crashes. In South Korea, Color Guidance Markings (CGMs) were introduced as an innovative countermeasure to mitigate such conflicts by visually guiding drivers to appropriate exit ramps in advance. This study estimates crash modification factors of CGMs on Korean motorways. METHOD:This study estimates crash modification factors of CGMs using a robust analytical approach. The Comparison Group method was applied to before-and-after crash data from 20 treatment and 87 comparison sites on Korean motorways. In addition, Regression-To-The-Mean (RTM) violations were evaluated using an established RTM assessment method to determine whether the observed crash reductions exceeded the expected RTM effects. RESULTS:The results show that CGMs installed at exit ramps with low traffic volumes of less than 20,000 vehicles per lane per day reduced crashes by 47.5% and fatalities and serious injuries by 74.4%. However, reductions were lower, at 31.0% and 54.7%, respectively, at exit ramps with higher traffic volumes exceeding 20,000 vehicles per lane per day. The analysis showed that observed crash reductions exceeded expected RTM effects at up to 75% of treatment sites. CONCLUSIONS:These findings indicate that CGMs effectively reduce the likelihood and severity of crashes at exit ramps. The extensive implementation of CGMs is plausible for reducing crashes and crash severity on Korean motorways. However, because the Comparison Group method cannot inherently account for RTM effects and may over- or underestimate treatment effects, the results need to be confirmed using the Empirical Bayes (EB) method. The application of the EB method requires developing safety performance functions for Korean motorways, which should be established in future research.
OBJECTIVES:The utility of the Abbreviated Injury Scale (AIS) for injury benchmarking and epidemiologic research depends on accurate injury severity designations. The objective of this study was to assess whether aggregated independent clinician assessments of injury severity differ from the official consensus-based AIS severity designations, focusing on common thoracic injuries. METHODS:Twenty-one thoracic injuries were selected, including rib fractures, sternal fractures, pneumothorax, and hemothorax. Radiographic images from patients with these injuries were obtained from the Crash Injury Research and Engineering Network (CIREN) database for 2017-2023. Clinicians were recruited to participate in an anonymous survey and independently rated each injury across four domains: threat to life, time to recovery, long-term morbidity, and overall injury severity. Overall severity ratings were recorded on an ordinal scale (1-6) anchored to AIS severity levels for injuries in other body regions. Agreement between aggregated clinician ratings and official AIS consensus severity designations was evaluated using ordinal regression. RESULTS:There were 18 physicians who completed the survey, most of whom were trauma surgeons practicing at Level I trauma centers. Overall severity ratings exactly matched the AIS designation in 43% of responses, and 89% were within one severity level. Greater years of clinical practice and training in trauma surgery or emergency medicine were associated with higher agreement with AIS designations. Clinician-assigned overall severity ratings tended to be lower than AIS designations for pneumothorax and hemothorax at higher severity levels. Among the surveyed domains, perceived threat to life contributed most strongly to overall severity ratings. CONCLUSIONS:Aggregated clinician assessments of injury severity generally aligned with AIS severity levels but frequently rated certain thoracic injuries as less severe at higher AIS levels. These findings suggest that clinician perceptions of severity may differ from consensus-based AIS designations, particularly for pneumothorax and hemothorax.
OBJECTIVE:To deconstruct the microscopic dynamic games and risk evolution mechanisms between right-turning heavy vehicles (HVs) and non-motorized vehicles (NMVs) at signalized intersections. METHODS:High-fidelity unmanned aerial vehicle (UAV) video data from three intersections in Nanjing were used to extract 527 conflict events and reconstruct five interaction patterns. Using time-to-collision (TTC) and post-encroachment time (PET) as surrogate safety measures, K-means clustering categorized conflict severity into potential, moderate, and severe levels. An ordered logit model (OLM) with interaction terms was developed to quantify the non-linear impacts of vehicle kinematics, behaviors, and environmental factors. RESULTS:HV speed and jerk (deceleration rate) are primary catalysts for severe conflicts. Microscopic behaviors exhibit a "double-edged sword" effect: proactive HV yielding reduces risk, but being "forced to stop" by NMVs during yielding increases severe conflict probability by 11.8%. A "safety in numbers" effect (>4 NMVs) decreased risk by 6.0%, while comprehensive safety facilities reduced severe conflict probability by 7.7%. CONCLUSIONS:This study reveals dynamic risk mechanisms for right-turning HVs, providing an empirical basis for optimizing intersection facility deployment and advanced driver assistance system (ADAS) warning strategies to enhance traffic safety.
