Prescribed anti-glaucoma medication consumption and road traffic crash.

BACKGROUND: Glaucoma is one of the leading causes of visual impairment worldwide. Influence of visual defects associated with this condition, as well as potential side effects of anti-glaucoma medications on driving may be a relevant traffic safety concern. This study therefore aimed to investigate whether and/or to what extent prescribed anti-glaucoma medicine consumption is associated with increased likelihood of crash risk, and traffic crash responsibility among drivers involved in road traffic crashes. METHODS: Data from three French national databases were extracted and matched as part of the CESIR (a combination of studies on health and road safety) project. The sample included 201 497 drivers involved in an injurious road crash in France from July 1, 2005 to December 31, 2015, and an age- and sex-matched control group (113 357 drivers) that was randomly drawn from the general population. Exposure to anti-glaucoma medications were compared between responsible and non-responsible drivers involved in a crash and between drivers involved in a crash and people from the control group. RESULTS: The proportion of drivers with prescribed anti-glaucoma medicine markedly increased with age. One type (OR = 0.79, 95% CI: 0.72-0.86) and two or more types (OR = 0.82, 95% CI: 0.68-0.98) anti-glaucoma medicine prescriptions were less frequent in crash-involved drivers than in controls. One type (OR = 0.99, 95% CI: 0.88-1.12) and two or more types (OR = 1.04, 95% CI: 0.82-1.33) anti-glaucoma medicine prescriptions were not associated with crash responsibility. CONCLUSION: Our findings are reassuring as regard to existing guidelines for safe driving for individuals using anti-glaucoma medications. Our results also suggest that driving behavior adaptation is effective mitigating potential traffic crash risks for people diagnosed with glaucoma.

Artificial Intelligence in Emergency Medicine: Viewpoint of Current Applications and Foreseeable Opportunities and Challenges.

Emergency medicine and its services have reached a breaking point during the COVID-19 pandemic. This pandemic has highlighted the failures of a system that needs to be reconsidered, and novel approaches need to be considered. Artificial intelligence (AI) has matured to the point where it is poised to fundamentally transform health care, and applications within the emergency field are particularly promising. In this viewpoint, we first attempt to depict the landscape of AI-based applications currently in use in the daily emergency field. We review the existing AI systems; their algorithms; and their derivation, validation, and impact studies. We also propose future directions and perspectives. Second, we examine the ethics and risk specificities of the use of AI in the emergency field.

Post-injury benzodiazepine and opioid use among older adults involved in road traffic crashes: A Swedish register-based longitudinal study.

AIM: Psychotropic drugs like opioids and benzodiazepines are prescribed for traumas resulting from road traffic crashes and the risk of developing an addiction deserves consideration. This study aims to shed light on how the consumption of those drugs evolves over time among older road traffic injury (RTI) victims. METHODS: We conducted a nationwide Swedish register-based longitudinal study to identify trajectories of post-RTI psychotropic drug use. All individuals aged 50 years and older who had a hospital visit for an RTI from 2007 to 2015 were followed up during a 2-year period; those who used the drugs prior to the RTI were excluded. Trajectories were identified by performing latent class trajectory analysis on drug dispensation data for opioids and benzodiazepines separately (66 034 and 66 859 adults, respectively, in total). RESULTS: Three trajectories were identified for opioids and four for benzodiazepines. The largest group in both instances included people with no-use/minimal use throughout the follow-up (81.3% and 92.8%). "Sporadic users" were more frequent among users of opioids (16.7%) than benzodiazepines (4.3%), whereas "chronic users" were found in similar proportions (2.0% and 1.8%). "Delayed chronic use" characterized the fourth group of benzodiazepine users (1.0%). CONCLUSION: Several trajectories of psychotropic drug use were identified after RTI, from limited to chronic. Although chronic use was uncommon, a better understanding of the factors likely to increase that risk is warranted given the seriousness of the problem.

Health conditions and the risk of home injury in French adults: results from a prospective study of the MAVIE cohort.

BACKGROUND: Home injury (HI) is a significant cause of mortality and morbidity in adults of all ages. Health conditions significantly impact HI among old adults, but little is known for other adults. STUDY DESIGN: Prospective cohort study. OBJECTIVE: We assessed the associations between health-related factors and HI's risk in a French study, the MAVIE (Mutualistes pour la recherche contre les Accidents de la VIE courante) cohort. METHODS: Poisson mixed models were fitted using health-related data information (diseases, treatments and disabilities) at baseline and the number of injuries prospectively recorded during the follow-up, adjusting for significant sociodemographics and exposure to a range of home activities. Attributable fractions were estimated based on risk ratio (RR) estimations measured in the fully adjusted models. RESULTS: A total of 6146 dwelling adults aged 15 or older were followed up for 5.1 years on average. Vertigo or dizziness (RR=2.36, 95% CI 1.06 to 5.01) and sciatica or back pain (RR=1.49, 95% CI 1.08 to 2.05) were independently associated with an increased risk of HI. These two groups of conditions showed the most significant associations among people aged 15-49, whereas musculoskeletal diseases other than rachialgias and arthropathies were the most significant health-related risk factor in people aged 50 and older. Sciatica or back pain represented the highest burden of HIs in overall adults (8%) and among people aged 15-49 (12%). CONCLUSION: Our results suggest that adults with musculoskeletal disorders and vertigo or dizziness symptoms have a higher risk of HI, regardless of age.

Implementation of a model of awareness-raising for taxi motorcyclists in Benin in relation to helmet use: a quasi-experimental study.

BACKGROUND: This study aims to test the effectiveness of an awareness-raising model designed based on the theory of planned behaviour regarding helmet use for motorcycle taxi drivers. METHODS: This quasi-experimental study took place in the cities of Parakou (intervention group) and Porto Novo (control group). Over a three-month period, a package of awareness-raising activities, based on the theory of planned behaviour, have been implemented in the intervention area. Data relate to knowledge, attitudes and practices regarding helmet use was collected prospectively before the intervention, at the end, and 6 months later. Stata 15 was used for data analysis. Chi-square or Fisher, Student's or Kruskal-Wallis tests was carried out. The difference-in-difference method was used to determine the specific effect of the awareness activities. RESULTS: After the intervention, there was an improvement in the total score in both groups compared to baseline. The total score increased by 0.2 (0.06-0.3) in the experimental group when the number of sessions attended increased by one (p = 0.005). The difference-in-difference estimator measured among subjects who attended at least one awareness session, controlling for socio-demographic variables, showed a significantly higher difference in the total score of subjects in the experimental group compared to those in the control group both at the end of the interactive sessions and 6 months later. CONCLUSION: This model improves the helmet-wearing behaviour of motorbike taxi drivers in the experimental area. It could be adapted and applied to other socio-professional groups and other types of users.

Factors associated with serious vehicular accidents: A cross-sectional study in hospital emergency rooms.

AIMS: Pictograms on medicine boxes warn of potential drug-related driving hazard; we studied their association with serious accidents. METHODS: Prospective study in emergency departments of the hospitals in Bordeaux and Périgueux (France), of drivers with serious (admitted at least 24 hours) or nonserious vehicular accidents. Minors, passengers, pedestrians or subjects incapable of answering an interview were excluded. Interviews ascertained driver and accident characteristics, use of drugs with or without pictograms, use of alcohol and abuse substances, sleepiness, distractions, and mind wandering at the time of the accident, RESULTS: Between 18 October 2016 and 26 December 2018, 1200 of the 6212 drivers admitted to the hospital emergency rooms, 741 nonserious, 459 serious, were interviewed. Serious accidents were associated with male sex (odds ratio 1.89, 95% confidence interval [1.36-2.64]), age above 60 years (3.64 [2.21-6.00]), driving on local roads (3.34 [2.34-4.76]), driving a motorcycle (3.39 [2.29-5.00]), having drunk alcohol within 6 hours (2.89 [1.85-4.51]) and using a drug with a pictogram during the 24 hours previous to the accident (1.57 [1.06-2.32]). From 207 police reports, 101 drivers were not responsible, and 106 were responsible, associated with age below 40 years, driving in overcast or rainy weather (2.62 [1.29-5.33]), on local roads (3.89 [1.90-7.95]), and use of at least 1 pictogram drug in the previous week (3.12 [1.31-7.41]). CONCLUSION: The known risks of alcohol and pictogram drugs, of riding motorcycles and using local roads were confirmed. As measured, behavioural sleepiness did not predict accidents.

A longitudinal study of mental health before and during COVID-19 lockdown in the French population.

BACKGROUND: The impact of general population lockdown implemented in the face of the COVID-19 epidemic needs to be evaluated. We describe here a longitudinal study on the mental health of adults in France. METHODS: We did a secondary analysis of a web-based cohort, initially set up to study home and leisure injuries, in order to measure the consequences of the national lockdown implemented in France from 17 March 2020 to 11 May 2020, and to assess potential vulnerability and resilience factors. Eligible participants were invited to answer an online questionnaire designed to assess their living conditions and health during lockdown. Comparisons were done with answers provided 4.8 years earlier on average. RESULTS: On 15th April 2020, we sent email invitations to 9598 participants recruited between November 2014 and December 2019 and 1237 volunteers took part in the study by completing the online questionnaire. The proportion of those with anxiety symptoms markedly increased from 17.3 to 20.1%. The average self-rated level of mental health decreased from 7.77 to 7.58. Women, the elderly and the youngest appeared to be more vulnerable. A small living space (less than 30 m2) was associated with an increase in depression symptoms (PHQ-9 score), and poorer self-rated physical health at recruitment was associated with an increase in anxiety symptoms (GAD-7 score). On the contrary, the average self-rated level of physical health markedly increased from 7.44 to 7.94 between recruitment and lockdown, and the proportion of those who reported a level of 9 or 10 jumped from 25.7% at recruitment to 43.1% during lockdown. CONCLUSIONS: Mental health deteriorated during lockdown in France during the 2020 COVID-19 crisis. Overall, self-rated physical health improved but those who experienced a worse physical health were more likely to report anxiety symptoms.

Chronic medical conditions and their association with crash risk and changes in driving habits: a prospective study of the GAZEL cohort.

OBJECTIVES: To assess crash risk and driving habits associated with chronic medical conditions among drivers entering old age. DESIGN: Prospective cohort study. SETTING: French cohort GAZEL. PARTICIPANTS: 12 460 drivers in the analysis of road traffic crash, among whom 11 670 completed the follow-up period (2007-2014). We assessed driving cessation among 11 633 participants over the same period, and mileage and driving avoidance among the 4973 participants who returned a road safety questionnaire in 2015. MAIN OUTCOME MEASURES: Yearly occurrence of at least one road crash as a driver; time to driving cessation; mileage; driving avoidance: at night, with bad weather, in heavy traffic, with glare conditions, over long distances. RESULTS: Several potentially risky conditions (angina, myocardial infarction, coronary disease; stroke; nephritic colic, urinary stones; glaucoma) were associated with lower mileage and/or driving avoidance and did not increase crash risk. Neither driving avoidance nor lower mileage was found for other conditions associated with an increased crash risk: hearing difficulties (adjusted OR 1.19, 95% CI 1.06 to 1.34); joint disorders (1.17, 95% CI 1.06 to 1.30). Depression, anxiety and stress was associated with an increased crash risk (1.23, 95% CI 1.01 to 1.49) despite increased driving avoidance. Parkinson's disease was associated with driving cessation (adjusted HR 32.61, 95% CI 14.21 to 65.17). CONCLUSIONS: Depending on their condition, and probably on the associated risk perception, drivers entering old age report diverse driving habits. For example, hearing difficulties is a frequent condition, rarely considered a threat to road safety, and nonetheless associated with an increased crash risk.

Examining the Association Between Alcohol Consumption and Health Conditions in Community Dwelling Older Adults.

The Canadian low-risk drinking guidelines were developed for adults, however, the applicability to older adults was not considered in the development of the guidelines. The objectives of this study were to: (1) to examine alcohol use in community dwelling older males and females using the current Canadian guidelines; (2) to test lower limits of alcohol use on various health factors; and (3) to determine health factors associated with high-risk drinking in older males and females. Data on community dwelling older adults (aged 65 +) was used from the Canadian Injury Prevention Survey (n = 2274). Descriptive statistics and comparative analysis were used to compare alcohol consumption categories. Logistic regressions were performed to examine the relationships between health factors and alcohol consumption categories. 70% of the sample reported having at least one drink per week; 4.5% of males and 6.8% of females were high-risk drinkers according to the current guidelines. There were no significant associations between the current alcohol categories with demographics, behavioral risk factors or health conditions. Using the new guidelines, 21% of males and females were classified as high-risk drinkers, respectively. Diabetes and having an illness or disability before retirement was protective of high-risk drinking in males while having diabetes and poorer physical health was protective of high-risk drinking in females. The prevalence of high-risk drinkers is dependent on what alcohol classifications are used. Further studies are needed to determine the causal relationships between health-related factors and alcohol using standardized definitions of alcohol consumption.

Prescription medicine use by pedestrians and the risk of injurious road traffic crashes: A case-crossover study.

BACKGROUND: While some medicinal drugs have been found to affect driving ability, no study has investigated whether a relationship exists between these medicines and crashes involving pedestrians. The aim of this study was to explore the association between the use of medicinal drugs and the risk of being involved in a road traffic crash as a pedestrian. METHODS AND FINDINGS: Data from 3 French nationwide databases were matched. We used the case-crossover design to control for time-invariant factors by using each case as its own control. To perform multivariable analysis and limit false-positive results, we implemented a bootstrap version of Lasso. To avoid the effect of unmeasured time-varying factors, we varied the length of the washout period from 30 to 119 days before the crash. The matching procedure led to the inclusion of 16,458 pedestrians involved in an injurious road traffic crash from 1 July 2005 to 31 December 2011. We found 48 medicine classes with a positive association with the risk of crash, with median odds ratios ranging from 1.12 to 2.98. Among these, benzodiazepines and benzodiazepine-related drugs, antihistamines, and anti-inflammatory and antirheumatic drugs were among the 10 medicines most consumed by the 16,458 pedestrians. Study limitations included slight overrepresentation of pedestrians injured in more severe crashes, lack of information about self-medication and the use of over-the-counter drugs, and lack of data on amount of walking. CONCLUSIONS: Therapeutic classes already identified as impacting the ability to drive, such as benzodiazepines and antihistamines, are also associated with an increased risk of pedestrians being involved in a road traffic crash. This study on pedestrians highlights the necessity of improving awareness of the effect of these medicines on this category of road user.

Road traffic crash risk associated with benzodiazepine and z-hypnotic use after implementation of a colour-graded pictogram: a responsibility study.

AIMS: To assess potential change in medicine exposure and association with the risk of road traffic crash across a time period that started before the implementation of a grading system warning of the effect of medicine on driving performance. METHODS: Data from three French national databases were extracted and matched: the national health care insurance database, police reports and the national police database of injurious crashes. Drivers involved in such crashes in France, from July 2005 to December 2011 and identified by their national identifier, were included. Association with the risk of crash was estimated using a case-control analysis comparing benzodiazepine and z-hypnotic use among drivers responsible or not responsible for the crash. RESULTS: Totals of 69 353 responsible and 73 410 non-responsible drivers involved in an injurious crash were included. Exposure to benzodiazepine anxiolytics was associated with an increased risk of being responsible for a road traffic crash during the pre-intervention period (OR = 1.42 [1.24-1.62]). The association disappeared in the post-intervention period, but became significant again thereafter. The risk of being responsible for a crash increased in users of z-hypnotics across the study period. CONCLUSIONS: Our results question the efficacy of the measures implemented to promote awareness about the effects of medicines on driving abilities. Prevention policies relating to the general driving population, but also to healthcare professionals, should be reviewed.

Sparse conditional logistic regression for analyzing large-scale matched data from epidemiological studies: a simple algorithm.

This paper considers the problem of estimation and variable selection for large high-dimensional data (high number of predictors p and large sample size N, without excluding the possibility that N < p) resulting from an individually matched case-control study. We develop a simple algorithm for the adaptation of the Lasso and related methods to the conditional logistic regression model. Our proposal relies on the simplification of the calculations involved in the likelihood function. Then, the proposed algorithm iteratively solves reweighted Lasso problems using cyclical coordinate descent, computed along a regularization path. This method can handle large problems and deal with sparse features efficiently. We discuss benefits and drawbacks with respect to the existing available implementations. We also illustrate the interest and use of these techniques on a pharmacoepidemiological study of medication use and traffic safety.

New clinical decision instruments can and should reduce radiation exposure.

In this Perspective linked to Rodriguez and colleagues, Emmanuel Lagarde discusses the importance of decision instruments that can help physicians avoid subjecting patients to radiation exposure from unnecessary CT scans.

Variable selection on large case-crossover data: application to a registry-based study of prescription drugs and road traffic crashes.

PURPOSE: In exploratory analyses of pharmacoepidemiological data from large populations with large number of exposures, both a conceptual and computational problem is how to screen hypotheses using probabilistic reasoning, selecting drug classes or individual drugs that most warrant further hypothesis testing. METHODS: We report the use of a shrinkage technique, the Lasso, in the exploratory analysis of the data on prescription drugs and road traffic crashes, resulting from the case-crossover matched-pair interval approach described by Orriols and colleagues (PLoS Med 2010; 7:e1000366). To prevent false-positive results, we consider a bootstrap-enhanced version of the Lasso. To highlight the most stable results, we extensively examine sensitivity to the choice of referent window. RESULTS: Antiepileptics, benzodiazepine hypnotics, anxiolytics, antidepressants, antithrombotic agents, mineral supplements, drugs used in diabetes, antiparkinsonian treatment, and several cardiovascular drugs showed suspected associations with road traffic accident involvement or accident responsibility. CONCLUSION: These results, in relation to other findings in the literature, provide new insight and may generate new hypotheses on the association between prescription drugs use and impaired driving ability.

Comparing responsible and non-responsible drivers to assess determinants of road traffic collisions: time to standardise and revisit.

OBJECTIVES: In 1994, Robertson and Drummer formalised the responsibility analysis as an alternative to case-control studies in the study of road traffic crashes. Our objective was to assess whether published responsibility studies respected standards defining adequate case-control studies. SETTING: Using Web of Science, Scopus and Google Scholar, we identified responsibility studies indexed until October 2013, which cited Robertson and Drummer's paper. After reviewing titles, abstracts and reference lists, we selected original studies comparing responsible and non-responsible drivers, published in peer-reviewed journals or proceedings. DATA AND ANALYSIS: We applied a grid to judge the conformity of the responsibility assessment to the original definition, and whether methods addressing representativeness of selection, accuracy of measures of exposures, confounding variables, and power met standards defining adequate case-control studies. RESULTS: Of 205 titles, we identified 20 papers. Methods of responsibility assessment were the original in three papers. Variations across studies concerned the number of mitigating factors included, or the use of questionnaires rather than police data (n=3). The ratio responsible/non-responsible drivers varied from 0.90 to 5.31, due to major variations in sampling methods, threshold selection, and data completeness. Selection or measurement issues were discussed in 13 papers, but seldom addressed in the design. A comparison of confounding factors in responsible and non-responsible drivers was presented in five papers. CONCLUSIONS: Basic requirements of case-control studies are often not or not clearly met in responsibility studies. There is a need to revisit the method and to adapt existing publication standards to the way responsibility studies are reported.

Specificity of postconcussion symptoms at 3 months after mild traumatic brain injury: results from a comparative cohort study.

OBJECTIVE: To assess the specificity of symptoms included in various symptom lists used to identify postconcussion syndrome (PCS), by using follow-up data comparing patients with mild traumatic brain injury (MTBI) and control patients during the month prior to injury and 3 months later. SETTING: The adult emergency department of a teaching hospital in Bordeaux, France. PARTICIPANTS: A cohort of patients with MTBI (n = 536) and a comparison group with nonhead injuries (n = 946). MAIN MEASURES: Postconcussion symptoms listed in the Rivermead Postconcussion Symptoms Questionnaire (RPQ), the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV), and the 10th International Classification of Diseases (ICD-10). RESULTS: Analyses were performed comparing symptom occurrence in patients with MTBI and controls, before and 3 months after injury. Eight symptoms were identified as being specific to MTBI: headache, dizziness, intolerance of stress, forgetfulness, poor concentration, taking longer to think, blurred vision, and personality change. CONCLUSION: The relevance of symptoms proposed to constitute PCS should be reviewed. A more specific definition of PCS would make diagnosis easier and facilitate prevention as well as treatment of patients with MTBI.

The risk of road traffic crashes for occupational drivers: A responsibility study with comparison to the general population.

BACKGROUND: Road accidents are the leading type of work-related fatalities, but the impact of work-related travel on overall traffic safety has been scarcely studied. OBJECTIVE: The main objective of the present study was to assess drivers' relative road accident risk between work-related and personal journeys. METHODS: A responsible/non-responsible case-control study was performed on a sample of 7,051 road accidents in France from the VOIESUR project. Logistic regression determined odds-ratios according to work-related versus personal travel, and identified risk factors for responsibility, specific to each of the two sub-groups. RESULTS: Drivers traveling on duty or commuting home were significantly less often responsible for accidents than drivers on personal journeys: OR = 0.75 [0.63; 0.89] and 0.65 [0.53; 0.80] respectively. Responsibility was significantly more frequent in commuting to versus from work: OR = 1.38 [1.06; 1.78]. Among on-duty drivers, professional passenger-transport drivers had the lowest risk of responsibility (OR = 0.25 [0.11; 0.58]), while those on temporary or work/study contracts and professional light goods vehicle drivers had the highest risk (OR = 11.64 [2.15; 62.94] and OR = 29.83 [5.19; 171.38] respectively). When driving under the influence of alcohol, risk of responsibility was higher in commuting home than in personal journeys. CONCLUSION: On-duty drivers showed lower risk of responsibility for an accident than other drivers. However, on-duty drivers on temporary or work/study contracts, who are usually not subject to specific regulations, showed higher risk, and should be the subject of particular attention regarding occupational risk prevention.

Attention problems in childhood and adult substance use.

OBJECTIVE: To assess the link between childhood attention problems (AP) and substance use 18 years later. STUDY DESIGN: This cohort study was conducted in a community sample of 1103 French youths followed from 1991 to 2009. Exposures and covariates were childhood behavioral problems (based on parental report at baseline), early substance use, school difficulties, and family adversity. Outcome measures were regular tobacco smoking, alcohol problems, problematic cannabis use, and lifetime cocaine use (based on youth reports at follow-up). RESULTS: Individuals with high levels of childhood AP had higher rates of substance use (regular tobacco smoking, alcohol problems, problematic cannabis use, and lifetime cocaine use). However, when taking into account other childhood behavioral problems, early substance use, school difficulties, and family adversity, childhood AP were related only to regular tobacco smoking and lifetime cocaine use. Early cannabis exposure was the strongest risk factor for all substance use problems. CONCLUSION: This longitudinal community-based study shows that, except for tobacco and cocaine, the association between childhood AP and substance use is confounded by a range of early risk factors. Early cannabis exposure plays a central role in later substance use.

Prescription of antidepressants and the risk of road traffic crash in the elderly: a case-crossover study.

AIM: To investigate the impact of antidepressants on the risk of road traffic crash in the elderly. METHODS: Reports from the Universal Quebec Automobile Insurance Agency database were matched with data on antidepressant prescription from the Quebec Health Insurance Agency. The case-crossover analysis consisted in comparing exposure during a period immediately before the crash with exposure during earlier periods, for the same subject. RESULTS: One hundred and nine thousand four hundred and six drivers between 66 and 84 years of age involved in a traffic crash between 1988 and 2000 were included. Two thousand nine hundred and nineteen (2.7%) were exposed to an antidepressant on the day of the crash. Case-crossover analysis found an increased risk of crash in drivers with a prescription of antidepressants before their crash when compared with a prescription of antidepressants 4 to 8 months before the crash (OR = 1.19, 95% CI 1.08, 1.30 to 1.42. 95% CI 1.30, 1.55). With the most recent control periods, results were not significant. CONCLUSION: A patient's mental state is probably more similar between two periods that are close to each other than up to 8 months before. Consequently, the risk of crash is likely to be linked to symptoms of depression.

Deep Learning Transformer Models for Building a Comprehensive and Real-time Trauma Observatory: Development and Validation Study.

BACKGROUND: Public health surveillance relies on the collection of data, often in near-real time. Recent advances in natural language processing make it possible to envisage an automated system for extracting information from electronic health records. OBJECTIVE: To study the feasibility of setting up a national trauma observatory in France, we compared the performance of several automatic language processing methods in a multiclass classification task of unstructured clinical notes. METHODS: A total of 69,110 free-text clinical notes related to visits to the emergency departments of the University Hospital of Bordeaux, France, between 2012 and 2019 were manually annotated. Among these clinical notes, 32.5% (22,481/69,110) were traumas. We trained 4 transformer models (deep learning models that encompass attention mechanism) and compared them with the term frequency-inverse document frequency associated with the support vector machine method. RESULTS: The transformer models consistently performed better than the term frequency-inverse document frequency and a support vector machine. Among the transformers, the GPTanam model pretrained with a French corpus with an additional autosupervised learning step on 306,368 unlabeled clinical notes showed the best performance with a micro F1-score of 0.969. CONCLUSIONS: The transformers proved efficient at the multiclass classification of narrative and medical data. Further steps for improvement should focus on the expansion of abbreviations and multioutput multiclass classification.