Introduction

While national-level clinical practice guidelines for physician assessments of fitness-to-drive for patients with a wide array of medical illnesses have been developed around the world, important limitations have been identified in the rigor of their development and in stakeholder involvement [1•, 2]. The Canadian Medical Association (CMA)’s Driver’s Guide: Determining Medical Fitness to Operate Motor Vehicles [3] is the primary tool used by physicians across Canada to guide decision making about assessing the impact of medical conditions on driving abilities, advising patients of the risks, and reporting patients to transportation authorities. The CMA Driver’s Guide also influences the content of national and provincial transportation policies pertaining to medical standards for drivers. The first edition of this Guide was published in 1974, and updates are provided every few years. Although the most recent publication of the CMA Guide is the 9th Edition in 2017, the present edition represented a minor opinion-based update of the 7th and 8th editions by two of our authors (F.M., M.J.R).

The term “dementia” encompasses a group of diseases (i.e., various types of dementia) that may have different effects on the functional skills required for safe driving [4]. In DSM-5, dementia was renamed major neurocognitive disorder [5]. It is known that patients with Alzheimer’s dementia show an inevitable decline in cognition, with an eventual decline in driving abilities over time [6]. While to date no longitudinal studies of declines in driving ability have been conducted for other forms of dementia, certain characteristics of these dementias likely have implications for fitness to drive. For example, vascular dementia can present with abrupt periods of worsening or stepwise declines in cognition associated with accumulation of cerebrovascular lesions [7]. Parkinson’s disease dementia and Lewy body dementia are often associated with motor and visuospatial dysfunction, as well as fluctuations in alertness and cognition that are inherently unpredictable and can be hazardous on the road [8•, 9•]. Furthermore, some frontotemporal dementias are associated with early executive dysfunction and behavioral changes (e.g., disinhibition, impulsivity, anger control issues) that may render driving hazardous [10•]. Insight may also be impaired in any of these illnesses, based on anatomical involvement, and thus patients’ own assessment of their driving ability may be unreliable [11]. Finally, all people with dementia are more prone to delirium, an acute change in mental status with unpredictable and sudden confusion [12]. Ultimately, then, progression to unsafe driving status is difficult to anticipate for patients with dementia.

We have previously conducted a systematic review of the literature from 2005 to 2015 pertaining to the risk of motor vehicle collisions and driving impairment associated with dementia [13••]. Briefly, in that work, we found two conflicting studies on motor vehicle collision (MVC) risk. One of these studies showed a 4-fold increase in MVCs among participants with dementia compared to healthy controls in a retrospective analysis but not in their prospective analysis [14]. The other study showed no statistically significant difference between those with and without dementia on retrospectively ascertained MVCs [15]. Among the seven recent studies that examined driving impairment, six revealed medium to large negative effects of dementia on driving performance (e.g., lane observance, speed control, turning). Importantly, we found that drivers with dementia had a 10-fold increased risk of failing a performance-based on-road assessment compared to healthy controls (RR 10.77, 95% CI 3.00–38.62, z = 3.65, p < 0.001). Therefore, we concluded that older adults with even very mild or mild stages of dementia are substantially more likely to exhibit impaired on-road driving performance and fail on-road tests, but the risk of actual crashes remained undetermined.

Although motor vehicle collisions are predicted by the World Health Organization to become the 3rd leading cause of disability-adjusted life years lost by 2020 [16], physicians are wary of advising their patients to stop driving because of potentially negative impacts on autonomy, health outcomes, quality of life, and the doctor-patient relationship [17,18,19,20,21]. This has led to under-reporting of patients with medical conditions that may impair driving to transportation authorities [20, 22], despite a demonstrated increased risk of collision [20, 22] that may be reduced by physician reporting [23]. A well-executed knowledge synthesis may increase confidence of clinicians using the guidelines [24], inform transportation policy, and provide a model for updating other international guidelines for medical illness and driving. In 2014, the Canadian Medical Association Journal published the Guidelines International Framework [25••] that outlines a systematic process of developing guidelines in a rigorous manner and involving stakeholders in the process. Our group sought to use this framework to update the CMA Driver’s Guide on the topic of dementia.

Methods

We followed the 2014 CMAJ Guidelines International Framework [25••] and the”ADAPTE” process for updating clinical practice guidelines [26]. The ADAPTE collaboration was “an international collaboration of researchers, guideline developers, and guideline implements who aimed to promote the development and use of clinical practice guidelines though the adaptation of existing guidelines.” [27].

Our team of authors includes researchers from Canada, Australia, Belgium, Ireland, England, and the USA, and representatives of geriatric psychiatry, geriatric medicine, neurology, family practice, occupational therapy, rehabilitation science, psychology, and pharmacology, as well as Canadian transportation administrators.

In April of 2016, the team met in Toronto, Canada, to complete the literature review [13••], and to review the existing clinical practice guideline recommendations on driving with dementia. Conflicts of interest were declared and authors were not allowed to make decisions pertaining to article inclusion, data extraction, or guideline recommendations for content areas on which they had contributed evidence. We reviewed 104 recommendations from 11 different national-level guidelines [3, 28,29,30,31,32,33,34,35,36,37]. We drafted a preliminary set of guidelines based on the review of the literature, the existing guidelines that were retrieved, and the clinical experience of the team. Each recommendation was assessed for level of evidence using the New Zealand Guideline Group Grading of Recommendations [38]. At the meeting, we voted on whether each recommendation (i.e., course of action) was supported by good evidence (Class A), fair evidence (Class B), expert opinion only (Class C).

The preliminary draft recommendations were further refined by two of the investigators (MJR and DC), and were then sent out for voting by the full team of investigators in July 2016. After an iterative process of electronic debate and discussion for those with less than 100% consensus, unanimous agreement was achieved for 23 guideline recommendations. For stakeholders, we focused on Geriatric Medicine, Geriatric Psychiatry, Neurology, Family Medicine, and Occupational Therapy. From November 2016 to March 2017, the revised guidelines were sent in an electronic survey to members of the Canadian Geriatrics Society (CGS), Canadian Academy of Geriatric Psychiatry (CAGP), and Canadian Association of Occupational Therapists (CAOT), the Canadian Neurological Sciences Federation (CNSF), and the Women’s College Family Practice (WCFP) for voting. We set an a priori definition of consensus as 90% agreement with the guideline recommendations.

Results

The research team generated 23 guideline recommendations with consensus (Table 1). We received 145 responses from stakeholders, which included 51 from CGS members, 51 from CAGP members, 25 from CAOT members, 1 from a CNSF member, 1 from a member of the WCFP, and 16 from other sources (e.g., colleague invitations). These responses included a total of 17 family physicians, 49 geriatricians, 35 geriatric psychiatrists, 8 general psychiatrists, 4 internal medicine specialists, 1 emergency medicine physician, and 31 occupational therapists.

Table 1 The proposed new evidence-informed recommendations on Driving with Dementia for consideration for the Canadian Medical Association Driver’s Guide as well as other national guidelines
Full size table

Responses from an additional 33 individuals were excluded because they consisted of missing data (e.g., all blank responses) or duplicate entries; or they were completed by professionals who either chose to withhold their area of practice (n = 6) or specialized in areas outside the scope of this analysis (i.e., residents, nurses, therapists, social workers, transportation stakeholders (n = 15)).

Of the 23 recommendations, more than 90% of respondents agreed with 14 (61%), and the remaining nine (39%) were endorsed by more than 80% but less than 90% of respondents.

In Table 1, we present the revised guideline recommendations for driving with dementia, along with the level of evidence and the percent of endorsement by the group. Recommendations with 80 to 89% endorsement, i.e., those falling below our a priori definition of consensus, include a summary of the comments of the respondents. Narrative comments from the research team follow some of the guideline recommendations as well for areas where the research team felt additional information was warranted, not necessarily based on the percentage of endorsement of the stakeholders.

Discussion

The development of these new guideline recommendations involved a much more rigorous process than prior editions of the CMA Driver’s Guide, which were rated poorly with respect to rigor of development and stakeholder involvement [1•]. Based on work on other clinical guidelines, we anticipated that enhancing the quality of the CMA guideline on driving with dementia may ultimately have a positive impact on patient outcomes [39], be more likely to be used in clinical practice [40], and increase the confidence of clinicians using the guidelines [41]. Rather than expert opinion of one or two authors (FM and MR authored the last 3 editions), we used a rigorous process of reviewing the literature, expert consensus, and stakeholder engagement for feedback. The stakeholders include multidisciplinary health care professionals who are involved with the clinical management of patients with dementia. Incorporating the perspectives of these professionals into the process of updating clinical guidelines on driving can [1] supplement the limited research by providing valuable opinion not available in the published literature [42](2) help to resolve conflicts over competing principles during the revision of specific recommendations from a multi-faceted approach [43, 44], and [3] enhance ownership of the recommendations and help to foster acceptance among clinicians [26, 43].

It is notable that only 8 of the 23 guideline statements were thought to have fair or good evidence to support them, while the remainder were still at the level of expert opinion. This gap in evidence is shared by many other areas of medical fitness to drive, such as various psychiatric conditions [45] and reflects the professional neglect of the healthcare professions of the importance of transport, and in particular the private automobile, to health and social inclusion. One barrier to enhancing understanding of these matters is that the effects of illnesses on driving crosses many different professional domains, and is not seen to specifically fall within any particular academic or clinical specialty. This problem is in turn is matched by a lack of education in traffic medicine at undergraduate [46], graduate, and postgraduate levels of medicine [47]. This paper may help promote a stronger evidence base for guidance on medical fitness to drive, one which is belatedly being recognized by funding bodies through the funding of studies that link databases of health records with police and crash data [48]. A particularly important focus for future research will be to gain a sense of the relative impact of combinations of co-morbidity, as exemplified by the additive effect of co-existing vision and hearing impairment [49].

The rigorous approach we used should be used to inform other national efforts in different countries, with international collaboration. Other national dementia and driving guideline development groups can use these findings to enhance the evidence base upon which their own guidelines are built. This will lend all guidelines greater validity and credibility to the developed guideline. This project has already fostered international collaborations, for example with researchers and clinicians in the UK who are developing similar Guidelines through a Delphi process. By working from a common internationally derived evidence-based foundation, it is hoped that the various national guidelines will share many common elements while still retaining the ability to adapt to unique local circumstances. There are many benefits to such an approach. First, this research produces guideline revisions that are not jurisdiction-specific, as they are based on the international scientific evidence. Secondly, it entails the establishment of a multinational collaboration of researchers and clinicians (e.g., the USA, Canada, Ireland, Belgium, and the UK), which functions to (a) incorporate a wide range of perspectives into the newly-revised recommendations and (b) facilitate the dissemination and adoption of the updated recommendations into clinical practice and public policy, utilizing the existing networks of the research team within each of the nations involved in the guideline revision process.

It is worth noting some limitations to this work. First, the guideline development process is limited by the quality of the evidence reviewed. Guidelines include not only evidence-informed recommendations but, given the weak or lacking evidence in some areas, also include authors’ guidance based on clinical practice, a form of evidence albeit less rigorous and more prone to bias. Second, our process was also limited by a lack of broader input of family physicians, and the input of those most affected by the guidelines, i.e., drivers with mild dementia and their caregivers, and we did not seek input from representatives of driving authorities, and organizations such as the Alzheimer Society. The time frame of the study did not permit us to get permission to send the survey to the national organization of Family Physicians, and we relied on a small convenience sample of local family physicians only. Third, lower levels of agreement are not necessarily indicative of the lower levels of evidence. In fact, low agreement may pertain to practical concerns about how the recommendations would be implemented or concerns with impact on the relationship between the clinicians and their patients.

As further evidence emerges, the most effective manner to incorporate the new evidence into guidelines in a timely manner would be via the development of a standing working group that continuously updates the evidence informing driving guidelines. Unfortunately, most countries cannot afford to fund such groups especially when all other medical conditions are considered. The present study, funded by a research granting agency for a limited period of time, illustrates that point. A funded international consortium of researchers and guideline developers to continuously update the evidence for driving guidelines would have international benefits, including harmonizing an approach to these problems, despite different legislative structures.

Conclusion

The proposed guidelines listed in this paper are not meant to be prescriptive. Rather, they serve as a list of evidence-based elements that dementia and driving guideline developers should consider for inclusion in their national guidelines. This research represents the next step in the evolution of evidence-based guidelines. The adoption of a rigorous scientific approach to guideline development will enhance the credibility of future national guidelines on fitness to drive among patients with various medical conditions.