Association of alcohol use with helmet use in cyclists: analysis of the National Trauma Data Bank Authors: Daniel C Brock, Gaibo Yan, Chad T Wilson, Emanuel Narcis Husu Published: 2026 ABSTRACT AbstractBackgroundHelmet use significantly reduces the risk of traumatic brain injury among cyclists; however, the influence of alcohol use on helmet-wearing adherence remains understudied in large national data sets. The goal of this study was to investigate the relationship between alcohol use and helmet use in injured cyclists.MethodsThis study is a retrospective cross-sectional analysis of 155 766 cyclists with traumatic injuries from the National Trauma Data Bank spanning 8 years (2017–2023). Cyclists were categorized by blood alcohol concentration (BAC) as follows: not screened, sober (BAC <0.02), impaired (BAC 0.02–0.08), intoxicated (BAC ≥0.08), and multidrug intoxicated (BAC ≥0.08 plus ≥1 positive drug screen). The primary outcome was helmet usage. Secondary outcomes included mortality, hospital and intensive care unit length of stay, Injury Severity Score, and adverse events. Comorbid substance use disorders were considered as covariates.ResultsAlcohol use was associated with a dose-response decrease in helmet use. Helmet adherence was 45.9% in sober cyclists (n=51 566), compared with 19.9% in impaired (n=1592), 9.8% in intoxicated (n=8327), and 6.2% in multidrug intoxicated cyclists (n=3823; p<0.001). Alcohol use was associated with higher mortality rates (OR=2.3; p<0.001) and a greater proportion of injuries in the head and neck region (OR=1.4; p<0.001). Alcohol use when cycling was significantly associated with comorbid alcohol, tobacco, and substance use disorders (p<0.001). An interaction effect was observed between alcohol use and comorbid alcohol use disorder, where individuals with a history of alcohol use disorder exhibited a lower baseline helmet use that further declined with increasing BAC.ConclusionsAlcohol use demonstrated a significant dose-response relationship with helmet non-usage among cyclists, particularly among those with substance use disorders. This relationship represents a critical public safety concern requiring further investigation to address its public health implications.Level of EvidenceRetrospective Epidemiologic/Prognostic study, Level III.Keywords: helmet; alcoholism; Brain Injuries, Traumatic; Head Injuries, Closed WHAT IS ALREADY KNOWN ON THIS TOPIC Single-institution studies have shown that alcohol use is associated with reduced helmet use among cyclists. Single-institution studies have shown that alcohol use is associated with reduced helmet use among cyclists. However, national trends in the USA and the interaction between alcohol use and substance use disorders in relation to helmet use remain poorly understood. However, national trends in the USA and the interaction between alcohol use and substance use disorders in relation to helmet use remain poorly understood. WHAT THIS STUDY ADDS Using data from the National Trauma Data Bank, this study demonstrates that alcohol use and its interaction with substance use disorders are significantly associated with lower rates of helmet use in injured cyclists. Using data from the National Trauma Data Bank, this study demonstrates that alcohol use and its interaction with substance use disorders are significantly associated with lower rates of helmet use in injured cyclists. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY These findings highlight the need for public health campaigns and policy interventions that promote helmet use and discourage cycling under the influence of alcohol. These findings highlight the need for public health campaigns and policy interventions that promote helmet use and discourage cycling under the influence of alcohol. Introduction Bicycling is a fun, inexpensive, and eco-friendly form of transportation and leisure that has been rapidly growing in popularity.1 However, growing along with its popularity is the incidence of bicycle-related injuries. Injury severity ranges from minor soft tissue damage to major trauma resulting in permanent disability or death. In the USA, close to 130 000 bicycle-related injuries are reported annually, with nearly 1000 injuries being fatal.2 These injuries have a bimodal demographic distribution, with the greatest injury incidence occurring in younger and older men. The most common injuries are abrasions and lacerations (40% to 60%), followed by head injuries, accounting for 15% of injuries.3 Among cycling injuries presenting to the emergency department (ED), head injuries, including traumatic brain injury, are the most common injury in all age groups, accounting for nearly a third of all injuries. Head injury is also associated with greater severity of injury and worse long-term sequela.4 Wearing a helmet can lower the incidence of head injury, reducing all types of head injury by 48%, with an even greater reduction among severe and fatal head injuries.5 Helmet usage rates within the USA vary among age groups and with respect to local helmet laws, with only 29% of individuals estimated to wear a helmet consistently when riding a bicycle.6 A major, controllable factor that decreases helmet usage is alcohol use.7,9 Previous research at a single US institution found that alcohol use in cyclists was inversely associated with helmet use, while being positively associated with injury severity, hospital resource use, and mortality.10 Cyclists who consumed alcohol were less likely to wear helmets, with only 16.5% of alcohol-consuming cyclists wearing a helmet compared with 43.2% of non-alcohol-consuming cyclists. However, there remains a gap in the literature with no large-scale, nationally representative studies in the USA to examine the association between alcohol use and helmet use among injured cyclists. Large representative databases, such as the National Trauma Data Bank (NTDB),11 provide an opportunity to address this relationship, thereby informing targeted public safety interventions to promote helmet use and address alcohol-related cycling risks. In this study, we analyze the relationship between alcohol use and helmet use in cyclists. We observed a significant negative correlation between increased blood alcohol concentration (BAC) and decreased rates of helmet use, alongside a decrease in helmet use among groups exhibiting comorbid substance use disorders. Thus, our findings demonstrate an actionable association between alcohol use and helmet use, providing a basis for additional research on preventing injuries and improving safety for cyclists. Methods Data source and patient selection This study used data from the NTDB Trauma Quality Programs Participant Use File (TQP PUF).11 The TQP PUF is maintained by the American College of Surgeons and is the largest repository of trauma registries in the USA, with over 11 million individual records from over 700 trauma centers. Records for the years 2017–2023 were filtered by International Classification of Diseases 10 (ICD-10) external cause of injury codes to select patients with pedal-cyclist injuries (E-codes V10–V19). Motor vehicle involvement was analyzed by stratifying ICD-10 codes into incidents involving collisions with cars, pick-up trucks, vans, busses, and other motorized vehicles. Our unfiltered cohort included a total of 191 695 patients with pedal-cyclist injuries (figure 1A). Helmet use trends in pedal-cyclists. (A) Study population diagram representing how incidents were filtered from the TQP PUF. (B) Rates of helmet use are significantly decreased in impaired, intoxicated, and multidrug-intoxicated cyclist groups. (C) Percentage of pedal-cyclists with ICD-10 codes indicating motor vehicle involvement per alcohol use group. *** indicates p<0.001. ICD-10, International Classification of Diseases 10; TQP PUF, Trauma Quality Programs Participant Use File. Inclusion criteria for injured pedal-cyclists included documented helmet use status at the time of injury in the “PROTDEV_HELMET” variable of the NTDB. Patients were excluded if they had missing demographic information (sex, age, race, and ethnicity), missing data on helmet use, or missing BAC screening results. A numeric BAC value or a report of “not screened” was required for downstream analysis. Missing data for primary and secondary outcomes were minimal, usually <1%, and summarized in online supplemental tables 1–4. We used helmet wearing status to stratify patients into helmeted and unhelmeted groups. Alcohol use was classified based on BAC thresholds defined by US federal guidelines. In the USA, the legal limit for alcohol use using BAC screening is 0.08%.12 The National Highway Systems Designation Act of 1995 further mandates that states should consider a “zero tolerance” policy of BAC screening of less than 0.02% for certain demographics, such as adults under the age of 21 years.13 Forty-three states currently use zero tolerance policies for monitoring BAC. We used these thresholds to classify patients into five distinct alcohol use categories: (1) Unscreened, (2) Sober with a BAC <0.02%, (3) Impaired with a BAC between 0.02% and 0.08%, (4) Intoxicated with a BAC ≥0.08%, and (5) Multidrug intoxicated with a BAC ≥0.08% plus a positive drug screen for at least one controlled/illicit substance. Individuals who were not screened were included for descriptive summaries to account for biases in BAC screening.14 To isolate the effect of alcohol use from confounding drug interactions, the sober, impaired, and intoxicated groups were restricted to patients with zero positive results on drug screening. Filtering patients according to our inclusion and exclusion criteria resulted in a final study population of 155 766. Study outcomes The primary outcome of this study was helmet wearing among different alcohol use groups. Covariate variables included comorbid conditions such as alcohol use disorder, tobacco use disorder, and other unspecified substance use disorder. Substance use disorder is an umbrella term that captures disorders such as smoking and alcohol use disorder, plus additional drug use disorders. Secondary outcomes included mortality, length of hospital stay and intensive care unit (ICU) stay, Injury Severity Score (ISS), and Abbreviated Injury Scale (AIS) Score for body regions. We also considered hospital adverse events as secondary outcome variables, including alcohol withdrawal syndrome, cardiac arrest, deep vein thrombosis, pulmonary embolism, acute kidney injury, acute respiratory distress syndrome, sepsis, stroke/cerebrovascular accident, hospital-acquired pneumonia, unplanned intubation, unplanned return to the ICU, and unplanned return to the operating room. Mortality was defined using the “disposition” entry in the NTDB, which was documented during the patient’s stay in the ED’s trauma center or on discharge. ISS were categorized into groups as described previously by Scott et al:15 minor (1–3), moderate (4–8), serious (9–15), severe (16–24), and critical (25–75). AIS severity scores ≥3 were categorized as severe and further broken down by ISS body region. Statistical methods Patient demographic information and outcome data were reported as frequencies and proportions for categorical variables and as median with IQR for continuous variables. Differences in patient demographics between groups were analyzed by Fisher’s exact test for categorical variables and Kruskal-Wallis test for continuous variables. Post hoc pairwise comparisons among alcohol use groups were conducted using Fisher’s exact tests for categorical outcomes and Wilcoxon rank-sum tests for continuous outcomes, with false discovery rate used to adjust for multiple hypothesis testing. To assess the rate of decline in helmet use across alcohol use groups with interacting comorbid conditions, we used logistic regression using a generalized linear model with a binomial family. The model was validated using the DHARMa package in R to test for statistical correctness.16 Univariable and multivariable logistic regression analysis was used to analyze factors associated with not wearing a helmet. The selection of variables for the multivariable models was conducted using a two-step process: first, relevant numerical features were standardized to ensure comparability of coefficients. Then a logistic regression model was fitted using GridSearchCV in scikit-learn with 10-fold cross-validation to optimize hyperparameters, including regularization strength and penalty type. This approach allowed for feature selection through the regularization process, and also considered the clinical importance of features. The final model included features with non-zero coefficients that were deemed clinically relevant. ORs derived from logistic regression were used to quantify the risk associated with each covariate. Patients were filtered from the TQP PUF using the pandas library in Python (V.3.9.21). Statistical analysis and plotting were performed using R V.4.4.2, using the tidyverse collection of packages and the finalfit package for statistical tables and logistic regression.17 A value of p<0.05 was considered statistically significant. The code used for this project can be found at the publicly available GitHub repository: https://github.com/toofastdan117/TQP_bike_alcohol_helmets. Results Application of the inclusion and exclusion criteria (see Methods) yielded 155 766 incidents labeled with pedal-cyclist external cause of injury ICD-10 codes (figure 1A). Of these, 60 593 (38.9%) wore a helmet at the time of injury, and 95 173 (61.1%) did not wear a helmet. 65 308 (41.9%) patients had undergone BAC screening, and 90 458 (58.1%) patients were not screened. Patients without BAC screening were included to account for previously reported selection bias in patient screening across racial and ethnic groups.14 Among patients who underwent BAC screening, 51 566 (78.9%) were sober with a BAC ˂0.02%. A subset of 1592 patients (2.4%) was classified as impaired with 0.02% ≤BAC <0.08%. A further 8327 (12.8%) patients were classified as intoxicated with a BAC ≥0.08%. Lastly, 3823 (5.8%) patients were classified as multidrug intoxicated with a BAC ≥0.08% with a co-occurring positive drug screen for ≥1 controlled/illicit substance. The helmeted group had a greater proportion of sober patients on BAC screen (sober: 39%, impaired: 0.5%, intoxicated: 1.4%, multidrug: 0.4%) compared with the non-helmeted group (sober: 29.3%, impaired: 1.3%, intoxicated: 7.9%, multidrug: 3.8%) (table 1). Unscreened patients were similarly represented in the helmeted (58.7%) and non-helmeted (57.7%) groups. After stratifying ICD-10 injury codes by motor vehicle involvement, 39.0% of non-helmeted cyclists had injuries related to motor vehicle collisions, compared with 20.9% of helmeted cyclists (p<0.001). Compared with the helmeted group, the non-helmeted group was associated with male sex, younger age groups, black race, and Hispanic ethnicity (p<0.001). Table 1Demographics of pedal-cyclists by helmet useHelmet (n=60 237)No helmet (n=94 436)P valueAlcohol UseNot screened35 323 (58.6)54 332 (57.5)<2.2E-16Sober23 548 (39.1)27 790 (29.4) Impaired315 (0.5)1268 (1.3) Intoxicated813 (1.3)7475 (7.9) Multidrug238 (0.4)3571 (3.8) Motor vehicle involvementYes12 635 (21.0)36 987 (39.2)<2.2E-16No45 578 (75.7)54 239 (57.4) Unspecified2024 (3.4)3210 (3.4) SexMale47 995 (79.7)76 297 (80.8)7.66E-08Female12 242 (20.3)18 139 (19.2) Age (years)0–92955 (4.9)9177 (9.7)<2.2E-1610–196026 (10.0)20 043 (21.2) 20–294307 (7.2)9424 (10.0) 30–396175 (10.3)10 420 (11.0) 40–497855 (13.0)10 804 (11.4) 50–5912 038 (20.0)15 038 (15.9) 60–6912 864 (21.4)12 686 (13.4) 70+8017 (13.3)6844 (7.2) RaceWhite52 244 (86.7)67 519 (71.5)<2.2E-16Black1722 (2.9)12 701 (13.4) Asian2450 (4.1)2633 (2.8) Native American173 (0.3)907 (1.0) Pacific Islander135 (0.2)291 (0.3) Other3513 (5.8)10 385 (11.0) EthnicityHispanic3803 (6.3)15 873 (16.8)<2.2E-16Not Hispanic56 434 (93.7)78 563 (83.2) Categorical variables are represented in raw counts with respective percent-of-whole enclosed within parentheses. Demographics of pedal-cyclists by helmet use Categorical variables are represented in raw counts with respective percent-of-whole enclosed within parentheses. The proportion of sober individuals wearing a helmet was 45.9%, compared with 19.9% for the impaired group (p<0.001), 9.8% for the intoxicated group (p<0.001), and 6.2% for the multidrug intoxicated group (p<0.001) (figure 1B). Patients who received a BAC test were roughly twice as likely than unscreened patients to be involved in a motor vehicle collision (p<0.001), with smaller differences observed among alcohol use groups (figure 1C). For clinical outcomes, mortality rates were 2.29 times higher in the group that did not wear a helmet (p<0.001) (table 2). This correlation was observed for both mortality in the trauma center’s ED and in-hospital mortality at discharge. The median length of hospital stay was slightly higher in the non-helmeted group compared with the helmeted (p<0.001). The non-helmeted group had higher lengths of stay in the ICU (p<0.001). However, helmeted cyclists had slightly higher median ISS compared with the non-helmeted group (p<0.001). The non-helmeted group had significantly higher proportions of both minor (1.82-fold, p<0.001) and critical (1.68-fold, p<0.001) injuries relative to the helmeted group. Table 2Clinical outcomes and Abbreviated Injury Scale Scores for pedal-cyclists by helmet useHelmet (n=60 237)No helmet (n=94 436)P valueTotal mortality% Deceased639 (1.1)2265 (2.4)<2.2E-16ED mortality% Deceased179 (0.3)750 (0.8)<2.2E-16Hospital mortality% Deceased460 (0.8)1515 (1.6)<2.2E-16Length of hospital stay (days)Median (IQR)3.0 (2.0 to 4.0)3.0 (2.0 to 5.0)<2.2E-16Length of ICU stay (days)Median (IQR)3.0 (2.0 to 4.0)3.0 (2.0 to 5.0)<2.2E-16ISSMedian (IQR)9.0 (5.0 to 14.0)6.0 (4.0 to 11.0)<2.2E-16ISS categoriesMinor (0–3)5069 (8.4)13 984 (14.8)<2.2E-16Moderate (4–8)20 554 (34.1)34 903 (37.0) Serious (9–15)25 033 (41.6)30 216 (32.0) Severe (16–24)7175 (11.9)9557 (10.1) Critical (>25)2362 (3.9)5632 (6.0) Total ISS count (%) 615 206 (40.2)916 623 (59.8) Total ISS regionsHead and neck80 652 (13.1)161 746 (17.6)<2.2E-16Chest37 553 (6.1)83 381 (9.1) Abdominal and pelvic105 733 (17.2)88 690 (9.7) Extremities26 552 (4.3)40 968 (4.5) Face102 501 (16.7)143 474 (15.7) External262 215 (42.6)398 364 (43.5) Severe ISS count (%) 87 896 (37.4)146 885 (62.6) Severe ISS regionsHead and neck25 702 (29.2)78 640 (53.5)<2.2E-16Chest538 (0.6)1093 (0.7) Abdominal and pelvic39 384 (44.8)32 670 (22.2) Extremities7707 (8.8)11 239 (7.7) Face14 506 (16.5)23 025 (15.7) External59 (0.1)218 (0.1) Continuous variables are represented as median with IQR and categorical variables are depicted as percentages.ED, emergency department; ICU, intensive care unit; ISS, Injury Severity Score. Clinical outcomes and Abbreviated Injury Scale Scores for pedal-cyclists by helmet use Continuous variables are represented as median with IQR and categorical variables are depicted as percentages. ED, emergency department; ICU, intensive care unit; ISS, Injury Severity Score. To analyze if alcohol use groups differed in injury severity, we analyzed AIS scores for six different body regions: (1) Head and neck, (2) Chest, (3) Abdominal and pelvic, (4) Extremities, (5) Face, and (6) External injuries. There were higher proportions of head and neck and chest injuries in the non-helmeted group (p<0.001) (table 2). The non-helmeted group also showed lesser proportions of abdominal and pelvic injuries. Among patients with severe injuries (AIS >2), the non-helmeted group had a higher proportion of injuries classified as severe. Head and neck injuries comprised the majority of severe injuries in the non-helmeted group, accounting for 53.5% of patients, compared with 29.2% in the helmeted group (p<0.001). The non-helmeted group again had a lower proportion of severe abdominal and pelvic injuries. There was no difference in reported incidence of “polytrauma,” defined as trauma events affecting multiple body regions. We next examined clinical outcomes during hospitalization. The most pronounced difference between helmeted and non-helmeted cyclists was alcohol withdrawal syndrome, which occurred 7.3-fold more frequently in non-helmeted patients (p=2.1×10−78) (table 3). Non-helmeted cyclists consistently experienced higher proportions of adverse outcomes, with the next three largest effect sizes observed for unplanned intubation (OR=2.38; p=2.16×10−24), acute respiratory distress syndrome (OR=2.31; p=4.08×10−7), and acute kidney injury (OR=2.22; p=2.87×10−9). Other significant clinical outcomes higher in the non-helmeted group included sepsis, cardiac arrest, hospital-acquired pneumonia, hypotension (systolic blood pressure <90 mm Hg), and unplanned admission to the operating room and ICU. Table 3Adverse outcomes for pedal-cyclists by helmet useAdverse outcomesHelmet (n=60 237)No helmet (n=94 436)P valueOR (95% CI)Hypotension (SBP <90 mm Hg)721 (1.2)1841 (2.0)1.43E-311.65 (1.51 to 1.8)Alcohol withdrawal syndrome57 (0.1)646 (0.7)7.81E-787.27 (5.54 to 9.7)Cardiac arrest146 (0.2)446 (0.5)2.50E-131.95 (1.62 to 2.37)Deep vein thrombosis219 (0.4)349 (0.4)0.8631.02 (0.86 to 1.21)Pulmonary embolism100 (0.2)164 (0.2)0.7521.05 (0.81 to 1.36)Acute kidney injury64 (0.1)221 (0.2)3.81E-092.2 (1.66 to 2.96)Acute respiratory distress syndrome42 (0.1)151 (0.2)5.63E-072.29 (1.62 to 3.31)Sepsis44 (0.1)136 (0.1)4.58E-051.97 (1.39 to 2.84)Stroke/CVA61 (0.1)124 (0.1)0.0961.3 (0.95 to 1.79)Hospital-acquired pneumonia125 (0.2)321 (0.3)1.71E-061.64 (1.33 to 2.03)Unplanned intubation154 (0.3)572 (0.6)2.05E-242.38 (1.98 to 2.86)Unplanned admission to ICU374 (0.6)700 (0.7)5.70E-031.19 (1.05 to 1.36)Unplanned return to operating room186 (0.3)380 (0.4)1.63E-031.32((1.11 to 1.59)Numbers represent patient counts for the presence of adverse outcomes.CVA, cerebrovascular accident; ICU, intensive care unit; SBP, systolic blood pressure. Adverse outcomes for pedal-cyclists by helmet use Numbers represent patient counts for the presence of adverse outcomes. CVA, cerebrovascular accident; ICU, intensive care unit; SBP, systolic blood pressure. The higher proportion of alcohol withdrawal syndrome in the non-helmeted group prompted our hypothesis that factors such as substance use disorders and concurrent comorbidities may influence individuals’ decision to wear a helmet. The non-helmeted group showed higher proportions of patients with substance use disorder (OR=6.24; p<1×10−100), tobacco use disorder (OR=5.89; p<1×10−100), and alcohol use disorder (OR=5.88; p<1×10−100) (table 4). Also notable was the higher proportion of non-helmeted cyclists with cirrhosis (OR=5.27; p=2.50×10−63) and chronic obstructive pulmonary disease (COPD) (OR=2.83; p<1×10−100). In a logistic regression model with interactions, recorded alcohol use in the trauma center (OR=1.23, p<0.001), alcohol use disorder (OR=2.05, p<0.001), tobacco use disorder (OR=4.46, p<0.001), and substance use disorder (OR=2.95, p<0.001) were independently associated with increased odds of not wearing a helmet (figure 2A). A significant interaction was observed between alcohol use categories and alcohol use disorder (interaction OR=1.15, p<0.001), indicating a synergistic effect associated with not wearing a helmet. No significant interactions were found between alcohol use categories and tobacco or substance use disorders. Table 4Comorbid conditions for pedal-cyclists by helmet useComorbid conditionsHelmet (n=60 237)No helmet (n=94 436)P valueOR (95% CI)Alcohol use disorder730 (1.2)6227 (6.6)05.89 (5.45 to 6.38)Tobacco use disorder2207 (3.7)17 313 (18.3)05.92 (5.66 to 6.2)Substance use disorder729 (1.2)6566 (7.0)06.25 (5.78 to 6.76)Congestive heart failure249 (0.4)678 (0.7)7.62E-151.75 (1.51 to 2.03)End-stage renal disease95 (0.2)261 (0.3)1.16E-061.76 (1.38 to 2.25)Diabetes2139 (3.6)4159 (4.4)3.66E-171.25 (1.19 to 1.32)Hypertension9582 (15.9)13 064 (13.8)3.63E-280.85 (0.83 to 0.88)Chronic obstructive pulmonary disease505 (0.8)2169 (2.3)9.91E-1182.84 (2.58 to 3.14)Bleeding disorder304 (0.5)359 (0.4)3.25E-040.75 (0.65 to 0.88)Cirrhosis77 (0.1)629 (0.7)9.13E-635.25 (4.14 to 6.75)Cerebrovascular accident240 (0.4)484 (0.5)1.16E-031.29 (1.1 to 1.51)Numbers represent patient counts for the presence of comorbid conditions. Comorbid conditions for pedal-cyclists by helmet use Numbers represent patient counts for the presence of comorbid conditions. Helmet use, alcohol use, and comorbid conditions. (A) Interaction between alcohol use and substance-related comorbidities, illustrating their combined association with reduced helmet use. (B) Multivariate logistic regression showing ORs with 95% CIs for covariates associated with helmet non-use. BAC, blood alcohol concentration; CI, confidence interval; OR, odds ratio. Given the significant association of comorbidities, including alcohol, tobacco, and substance use disorders with lower helmet usage, we performed univariate and multivariate logistic regression analysis to evaluate the association of risk factors on rates of helmet use. Using multivariate logistic regression, multidrug intoxication status had the greatest association with not wearing a helmet out of all covariates tested (OR=4.81, p<0.001) (figure 2B, online supplemental table 5). This was followed by tobacco use disorder (OR=4.38, p<0.001), intoxication status (OR=4.17, p<0.001), and black race (OR=3.96, p<0.001). Collectively, alcohol use, substance-related behaviors, and specific demographic characteristics were associated with decreased likelihood of helmet use. Discussion This study used a large, nationally representative trauma database to investigate the relationship between alcohol use and helmet use among injured pedal-cyclists in the USA. Our findings demonstrate a significant negative correlation between alcohol use and helmet use. Intoxicated pedal-cyclists, defined as having a BAC ≥0.08%, exhibited a 36.1% lower rate of helmet use compared with sober pedal-cyclists, and impaired pedal-cyclists (BAC between 0.02% and 0.08%) had a 26% lower rate. These results were further compounded if pedal-cyclists tested positive for ≥1 illicit/controlled substance when intoxicated, having a helmet wearing rate of just 6.2%. These findings are consistent with previous research in other countries, such as Finland7 and Germany,8 which also found that alcohol use is negatively associated with helmet use in pedal-cyclists. Our study confirms previously established associations between non-helmeted cyclists and increased risks of mortality and head and neck injuries.18 Our 2.29 increased mortality rate in non-helmeted cyclists agrees with prior studies, which found increased odds of mortality ranging from 1.7 to 3.1.18,20 We attribute the increased mortality rate observed in non-helmeted cyclists to the greater proportion of severe head and neck injuries, as reflected by higher regional ISS. The non-helmeted group exhibited higher rates of alcohol-related complications, including alcohol withdrawal syndrome, which may reflect underlying alcohol use disorder and associated comorbidities. Higher rates of comorbid conditions, such as cirrhosis and chronic obstructive pulmonary disease (COPD), are likely to contribute to higher mortality rates after traumatic injury. In addition to the lack of head protection, the higher incidence of head injuries in non-helmeted cyclists may also reflect contributing factors, such as impaired motor function when intoxicated. Potential confounding variables include the mechanism and time of injury. Compared with unscreened patients, those who underwent BAC testing, regardless of alcohol use level, had higher odds of involvement in motor vehicle-related injuries. This pattern likely reflects selection bias in trauma testing practices, whereby trauma centers are more likely to obtain alcohol and drug screens in patients involved in high-energy mechanisms such as motor vehicle crashes.21 Another contributing factor is that pedal-cyclists who consume alcohol are more likely to cycle at night when there is poorer visibility for both cyclists and motor vehicle drivers.22 Additionally, cyclists who consume alcohol have a greater probability of injury due to being less experienced cyclists.7 23 Given the protective effects of wearing helmets on head injuries, factors that increase helmet usage can decrease the incidence and severity of cycling-related head injuries.5 15 20 24 Therefore, reduction or abstinence from alcohol may lead to greater helmet usage among cyclists. Our study demonstrated that non-helmeted cyclists were associated with various comorbid conditions, the most significant being alcohol, tobacco, and substance use disorders. A prior NTDB study spanning 2010–2016 reported similar findings, demonstrating that non-helmeted cyclists were positively associated with cirrhosis (OR=9), tobacco use disorder (OR=5.6), and congestive heart failure (OR=3.5).24 Similar associations in comorbidities were also observed in non-helmeted motorcyclists.25 Studies using self-reported measures of alcohol use disorder support our findings that individuals with this condition have lower rates of seatbelt and helmet use.26 27 This highlights an underlying need to improve treatment of alcohol and substance use disorders. Failure to adequately treat the underlying disorders increases the likelihood of ED visits, placing additional strain on the healthcare system and impacting public safety. The number of substance use-related ED visits is increasing, along with repeat ED visits.28 Previous studies have shown that alcohol use disorder is significantly associated with increased risk of diabetes and cirrhosis, and smoking is linked with development of COPD.29,32 This suggests that alcohol use disorder is a significant driver of adverse health conditions, and our results suggest that alcohol and other substance use disorders play a prominent role in helmet-wearing decisions. The rise in substance use disorders presents a growing public health hazard in the general population; therefore, understanding systemic changes to addressing substance use disorders may help reduce risky behaviors among pedal-cyclists. In the USA, helmet laws and the legality of cycling when intoxicated are governed by state-level police powers, and a large proportion of pedal-cyclists are not fully aware of the risks.13 There have been suggestions that prevention campaigns may help lower the prevalence of alcohol use when cycling, similar to those for driving under the influence of alcohol. Previous studies showed that helmet subsidies, helmet education, and legislative laws are effective in increasing helmet usage and reducing head injury.33,35 To date, no such national campaigns for public awareness of cycling when intoxicated exist in the USA, despite evidence that education campaigns encouraging helmet use in pedal-cyclists are effective at increasing helmet usage and decreasing rates of traumatic brain injury. There is also little debate concerning the promotion of unsafe cycling practices from bike-sharing companies, such as positioning rental bicycles without helmets outside of urban bars. Further studies are needed to analyze the effectiveness of laws and public awareness campaigns on alcohol usage and helmet usage with respect to public behaviors among pedal-cyclists. Strengths of this study include its large sample size and its objective measure of alcohol use using BAC screening. This allowed us to categorize patients into alcohol use groups to analyze dose-response effects. The large sample size provided the necessary statistical power to disentangle the contributions of sex, age, race, and comorbid conditions using a multivariate logistic regression model. Alcohol use remained independently associated with reduced helmet use after adjustment for these factors. Covariate adjustment is recommended when using NTDB data to study trauma outcomes, as outlined by Haider et al.36 Lastly, unlike self-reported assessments, which are vulnerable to under-reporting due to social desirability bias,37 our study relied on objective biological measurements of alcohol use derived from BAC testing. This study is limited by the study population consisting of pedal-cyclists who presented to the ED with injuries, which naturally presents an enrichment for individuals who get seriously injured and says little about helmet use in the setting of alcohol use among those who do not get seriously injured. The demographics may be skewed, with certain cultures, ethnicities, and socioeconomic backgrounds being less likely to present to the ED. Furthermore, it is known that certain demographics, such as patients with black race and Hispanic ethnicity, receive BAC screenings at disproportionately high rates in the NTDB.14 Pedal-cyclists who present to the ED may have a higher bias towards being actively intoxicated, possibly due to the fact that intoxicated patients are more likely to be brought to the ED with concerns of head injury, given that the behavior of intoxicated patients can be difficult to distinguish from patients with head injuries.10 Large databases such as the NTDB also contain missing data and limited contextual information for how injuries occurred beyond ICD coding. Missing data should be addressed through careful filtering and reporting or imputation.36 Lastly, substantial variability exists in how trauma centers administer BAC and drug screening tests.38 If trauma center information is available, future studies should account for this heterogeneity by clustering at the trauma center level. Future work should dissect the effect of polysubstance use, where patients present with a positive BAC screen and drug test for an illicit/controlled substance. Lastly, our study did not differentiate between manually powered traditional bicycles and e-bikes. ICD-10 codes were updated in 2023 to include terms for e-bike injuries; however, there is still a limited sample size in the NTDB. Early work using the National Electronic Injury Surveillance System demonstrates that e-bike riders have significantly higher rates of head injury when compared with traditional bicycles.22 39 40 Future work using the NTDB should differentiate between traditional bicycles and e-bikes. Conclusions This study demonstrated a significant dose-response association between alcohol use and decreased helmet use among injured pedal-cyclists using a nationally representative trauma database. Alcohol use was associated with comorbidities such as alcohol, tobacco, and substance use disorders that interacted with the proportion of helmet users. Public health efforts promoting cycling safety should prioritize strategies addressing alcohol use, substance use disorder treatment, trauma prevention education, and equitable infrastructure to promote safe cycling. A holistic public health approach is needed to mitigate alcohol-related injuries for this mode of transportation and recreation. Supplementary material Acknowledgements The authors thank Duc Nguyen, Sukru Aras, and Kenneth Nobleza of the Texas Children’s Hospital Research Data Network for providing timely biostatistics support for this project. The authors also thank Dr. Nguyen for reviewing the article and offering feedback. Footnotes Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: This study involves human participants. This study was reviewed by the Institutional Review Board at Baylor College of Medicine and was determined to be exempt from ethics approval. All code, tables, and figures used in this study are deidentified and compliant with the American College of Surgeons’ guidelines for publishing research using the TQP PUF. Data availability free text: Data are available through the American College of Surgeons NTDB TQP PUF. Data availability statement Data may be obtained from a third party and are not publicly available. References Associated Data Supplementary Materials Data Availability Statement Data may be obtained from a third party and are not publicly available.