In this study, more than 20% of drivers were treated with at least one PDI drug (n = 373). The most frequent PDI drugs prescribed to drivers were Z-drug hypnotics (Zolpidem and Zopiclone), one anxiolytic (Bromazepam) and analgesics (Tramadol and Pregabalin). Compared to drivers without any PDI drug in their treatment, PDI+ drivers had more disability and more often a history of depressive disorder and were receiving more often polypharmacy. They had more frequently chronic pain and less frequently T2DM and AF.
The percentage of PDI+ drivers in this study was lower than in previous research. A study conducted in U.S. among 2990 older drivers aged 65–79 years old found that 70% were under central nervous system agent . Another one found 68.9% of PDI drug consumers among 225 drivers (mean age: 68 (12.8) years old) , according to the PDI medication list defined by Leroy and Morse in 2008, which includes not only central nervous system agents but also numerous other system agents (hematologic, cardiovascular-renal, gastrointestinal, metabolic, hormonal, neurologic, ophthalmic, otologic, antiparasitic, respiratory and analgesic ones) . These discrepancies may be explained by the fact that not all central nervous system agents were taken into account in the present study, but only those impairing driving performance according to the French classification  (see Additional file 1). Furthermore, the studies of Hetland et al. and Hill et al. included medically impaired drivers referred to an occupational therapy-based driving evaluation clinic inflating in all likelihood the rate of PDI-drug consumption [17, 18]. Also, self-medication was not assessed in this study which can have minimized the prevalence of PDI-drug consumption, whereas self-medication is frequent in the older population , especially analgesics , and in France, benzodiazepines , both of them being PDI drugs.
In this study, the most frequent PDI drugs were Z-drug hypnotics (Zolpidem (10.7%) and Zopiclone (8%)), followed by one benzodiazepine (Bromazepam (7.8%), and analgesics (Tramadol (6.9%) and Pregabalin (5.5%)). These results are consistent with previous research. One study found that Zopiclone and Diazepam were the two most frequent single legal drugs found in blood samples of older drivers suspected of driving under the influence of drugs. In this study, ethanol was detected in 81% of blood samples, Zopiclone in 9.8% and Diazepam in 9.3% . In the study of Hetland et al. in 2014, benzodiazepines, opioids and non-benzodiazepine hypnotics accounted respectively for 6.6, 4.5 and 4% of PDI drugs used among the 225 drivers . Benzodiazepines are overprescribed in older people [21, 23], particularly in the treatment of insomnia and anxiety disorders, which are both frequent in older adults [24, 25]. Non-benzodiazepine hypnotics, known as Z-drugs (Zopiclone and Zolpidem for example) are also commonly used to treat insomnia because they are often perceived as safer than benzodiazepines, while they are associated with an increased risk of falls and fractures . Moreover, older subjects frequently suffer from chronic pain [27, 28], which can explain the high rates of opioid use in this population (6–9%) [27, 29]. Opioid misuse is also frequent in older people (1–3%) , perhaps because of mood modifying effects of opioid agents, which are known to increase serotonin levels and cause addiction .
Our study found several factors associated with PDI-drug consumption. First, taking at least one PDI drug was associated with polypharmacy, which is frequently observed in older people [31,32,33]. Second, PDI-drug consumption was associated with history of depressive disorder and chronic pain, which is consistent with the 5 most frequent PDI drugs prescribed in this study. Indeed, insomnia and anxiety are frequently observed in depressed older people , and also described as a risk factor for late life depression . Moreover, previous research found that chronic pain is a risk factor for depression in older adults, and similarly, that depressed older people were more likely to suffer from chronic pain, corroborating the hypothesis, that neuroinflammation could be a common pathogenic factor of chronic pain and depression . Surprisingly, no antidepressant was found in the 5 most frequent PDI drugs, which may be explained by the fact that only 7.5% of PDI+ drivers had probable clinical depression (according to their GDS-15 score at inclusion), suggesting that insomnia, anxiety and chronic pain would be residual or prodromal symptoms of depression in these patients. Conversely, PDI+ drivers in this study suffered less frequently from T2DM and AF. This may be explained by a better management of patients suffering from these chronic diseases, with regular revision of prescriptions. Finally, PDI+ drivers had more disability than PDI- ones, which might be driven by the association of symptoms treated with PDI medications (like insomnia, anxiety or chronic pain) and medical comorbidity [24, 27, 37].
These results are a matter of great concern since the increased risk of car accident in drivers taking PDI drugs is widely documented [8, 18, 38, 39], especially for benzodiazepines [8, 40], Z-drug hypnotics [8, 41, 42] and opioids [8, 39, 43, 44]. Furthermore, most of the factors found here associated with PDI-drug consumption in older drivers are also associated with a higher risk of car accident, in particular depression [6, 7], chronic pain known to alter performance on attention tasks , and polypharmacy [17, 18]. In the study of Hetland et al., the average medically impaired driver was taking 5.9 (3.7) total routine medications , while half of older drivers interviewed for the AAA Foundation for Traffic Safety study were taking seven or more medications . Furthermore, PDI drugs tolerance is more likely to be poorer in older subjects than younger ones because of pharmacokinetic changes associated with ageing, like altered hepatic and renal functions that increase plasma elimination half-life, leading to increase of drug side effects like impaired attention, increased reaction time, hypersomnolence and confusion .
The concerning rate of aged drivers taking PDI drug might raise the question of searching for ways to decrease it. If no recommendation can be formally given to GPs based on this study because of several limitations, some aspects of drug prescription in aged population should be noted. First, the results of this study show the importance of questioning patients about their driving habits and considering them before prescribing. They might also highlight the importance of informing the patients and their families about the high risk of car accident while driving under the influence of PDI drugs, either prescribed or used for self-medication. Second, these results should also call for strategies to reduce the prescription of PDI drugs in older drivers and even avoid them when possible, keeping in mind that benzodiazepine and Z-drug hypnotic withdrawal can be achieved in the older population . Whenever possible, alternative therapies should be considered: cognitive behavioral therapy is known to efficiently treat chronic pain , depression , anxiety  and insomnia (with stimulus control, sleep restriction, sleep hygiene and relaxation) , also acupuncture, hypnotherapy, physical exercise and relaxation have demonstrated some efficacy in the treatment of chronic pain and depression . Third, since reducing polypharmacy should always be targeted, prescriptions should be regularly revised with the use of geriatrics tools like the Screening Tool of Older Person’s Prescriptions (STOPP) .
This study has some limitations. First, the results of this study have to be interpreted within the limit of its design. The fact that participants of the S.AGES cohort were included according to three specific diseases (T2DM, AF and chronic pain) might induce an inclusion bias, even though these diseases are highly prevalent in the older population [27, 49, 50]. Besides, Tramadol prescription in this study was probably higher than in the whole French population because one third of the participants were recruited in the chronic pain sub-cohort. Moreover, this study has a cross-sectional design: PDI status of participants may have changed during the follow-up. Another limitation is that the inclusion of the S.AGES data occurred about 10 years ago. The French list of PDI drugs has evolved during the last decade, which may have led to misestimate the number and distribution of such drugs among drivers. Nevertheless, according to the ANSM, in the whole French population in 2015, the rate of subjects taking benzodiazepines was still high (13.4% of the French population was prescribed a benzodiazepine at least once) . Regarding the use of opioids, Tramadol consumption (alone or in association) has highly increased (+ 68%) between 2006 and 2017 . Nevertheless, these most recent analyses of benzodiazepine (2015) and opioid analgesic (2017) consumptions by the ANSM were calculated from the whole French general population whatever the age, the driving status, and the state of health. They are therefore not easily comparable to descriptive analyses of this study population, which only comprises aged drivers with specific diseases.
Nevertheless, this study has several strengths: it was conducted in real life condition, with a large sample size, and data were recorded by patients’ GPs, with extensive socio-demographical, clinical and therapeutic information which gave a precise and detailed description of older drivers’ use of PDI drugs.