INTRODUCTION

Statistics revealing the aging of the population and higher-than-average motor vehicle crash (MVC) rates per mile driven for older drivers have focused attention on the assessment of the fitness of older drivers.1–5 Physicians play a major role in assessing fitness to drive and in reporting unsafe drivers to the authorities.1,6,7 When a physician deems a patient unfit to drive, it can undermine an older person’s sense of independence, contributing to depressive symptoms, social isolation, and a diminished quality of life,8–10 and can strain the physician–patient relationship.11

Legislation regarding a physician’s responsibility to report unsafe drivers varies by province in Canada6 and by state in the United States.12 In British Columbia (BC), physicians are mandated to report an unfit driver only if the driver continues to drive after they have been informed that they are unsafe to do so. In Quebec, Alberta, and Nova Scotia, reporting is discretionary. In all other Canadian provinces/territories, physicians are required to report to the licensing authorities all patients who, in their opinion, may be medically unfit to drive. Failure to report has resulted in physicians being found partially liable for MVCs in several civil suits, both in Canada and the United States.12–15

The Canadian Medical Association publishes a handbook titled Determining Medical Fitness to Drive: A Guide For Physicians.6 The handbook is organized into different sections corresponding to conditions that can affect driving (e.g., nervous system diseases, cardiovascular diseases, the aging driver). However, the guide does not outline a standard set of areas that should be assessed, and the recommendations are mainly empirical.

In Canada, primary care is provided predominantly by family physicians. No prior study has examined the attitudes and practices of Canadian family physicians regarding the assessment of fitness to drive and the reporting of unsafe older drivers. We conducted a national survey to obtain this information.

METHODS

We carried out a cross-sectional mailed survey in August and September of 2003. The intended population under study was English-speaking Canadian family physicians in active practices including patients 65 years and older. The 2003 Canadian Medical Directory was used to identify all physicians, sorted by province, with practice types listed as “family medicine” or “physician/general practice” and a preferred language of English because the survey was in English.16 Physicians whose preferred language was French, who make up about 25% of active family physicians,16 were excluded.

A stratified, simple random sampling strategy was used, with 5 strata corresponding to 5 regions of the country (BC, the Prairie provinces, Ontario, Quebec, and the Atlantic provinces). Nonproportional sampling fractions were applied such that 200 physicians were sampled within each stratum, for a total sample of 1,000 physicians. Sampling fractions ranged from 2% (Ontario) through 9% (Atlantic provinces) to 20% (Quebec).

The content of the questionnaire was developed based on a review of the literature and with consultations from family physicians, occupational therapists, geriatricians, and rehabilitation specialists with expertise in driving assessments. The questionnaire was pilot tested with 5 family physicians and revised based on their feedback. The survey started with a question to establish eligibility: “Are you in an active family practice that includes patients 65 years and older?” Those answering “no” were instructed to return the survey blank. The survey contained sections on (a) attitudes and practices towards driving assessments and the reporting of patients considered medically unsafe to drive; (b) components of assessments of older patients’ fitness to drive; (c) awareness of provincial reporting policies and procedures; and (d) information about the respondents (age, sex, years in practice), their medical practices, and their experience in performing fitness-to-drive assessments and reporting patients to the licensing authorities. The responses for most questions were on a 5-point ordinal scale (e.g., from “strongly agree” to “strongly disagree” or “always” to “never”).

We adopted several techniques to maximize the response rate: personalization (a cover letter using the respondent’s name and ink signatures), self-addressed and prestamped return envelopes using first-class stamps, a token of appreciation not conditional on response (a Canadian $2 bill, now out of circulation), and up to 5 contacts for each subject (a prenotice letter, the initial survey mailing, a reminder postcard, and up to 2 additional survey mailings).17,18 To ensure confidentiality, the surveys were administered anonymously, with no identifiers on the questionnaire or return envelope. Physicians were instructed to return an enclosed postcard with their identification information separately from their completed survey, enabling the identification of respondents but not their responses. The study was approved by the Research Ethics Board of the Toronto Rehabilitation Institute.

The response rate was calculated using the proportional allocation method, which assumes that the proportion eligible among those who did not reply and whose mailing was returned undelivered is the same as the proportion eligible among responders.19 This method results in a conservative estimate of the response rate because it tends to overestimate the eligibility of nonresponders.

Because the survey was administered anonymously, we could not directly compare the characteristics of respondents and nonrespondents and could not use the same source of information to compare the characteristics of the respondents and the target sample of 1,000 physicians. The characteristics of the respondents were obtained from responses to the mailed questionnaire, while the characteristics of the target sample of 1,000 physicians were derived from data from the Canadian Medical Directory.16 For the target sample of physicians, the number of years of practice was estimated from their year of graduation from medical school plus 2 years of postgraduate training, and the community size for their city of practice was obtained from 2001 census information from Statistics Canada (http://www.statscan.ca).

Sampling weights were used to adjust the response frequencies for the nonproportional sampling scheme. The adjusted frequencies, which are estimates of the total Canadian population frequencies, are presented here. A weighted Pearson’s χ 2 test was used to compare response profiles between physicians in provinces with mandatory versus discretionary reporting of unsafe drivers.20 Physicians in BC were excluded from these comparisons because of the unique reporting legislation in that province, which falls somewhere between full mandatory and discretionary reporting of unsafe drivers.

Survey-weighted regression models were used to explore the effect on responses to key questions of 4 factors chosen a priori: (a) physicians’ years in practice, (b) the size of the community they practiced in, (c) the region they worked in, and (d) the proportion of their practice aged 65 years or older. For questions with ordinal responses (e.g., number of fitness-to-drive assessments performed, confidence in ability to evaluate driving fitness, whether physicians should be legally required to report unsafe drivers, whether cognitive assessment is included in the driving assessment), the continuation ratio regression model was used.21 For the question collapsed into a binary outcome (number of patients reported to the driving authorities; 0 or >1), binary logistic regression was used. Statistical significance was set at P < 0.05. No adjustment was made for multiple comparisons.22 For the analysis investigating the association between the time taken to complete a driving assessment and the number of components assessed, a survey-weighted Poisson regression model using polynomial contrasts for time of assessment was fitted. A response of “often” or “always” was treated as having assessed the component. Analyses were performed using the R Project for Statistical Computing (http://www.r-project.org/) using the survey package, version 3.6-1 (http://faculty.washington.edu/tlumley/survey/).

The sample size was selected by evaluating the expected statistical properties of a range of sample sizes and response rates. Assuming a 50% response rate, a sample of 200 physicians from each region was considered adequate because it met the following criteria: (a) the standard error of the estimate of the proportion of physicians who endorsed a given response would be, at most, 2.7 percentage points, and (b) for comparisons across provinces with mandatory and discretionary reporting legislation, with alpha = 0.05 and power = 80%, if the percentage of respondents endorsing a response in provinces without legislation was 15%, 25%, 35%, or 50%, the minimum detectable differences would be 9.1%, 11.1%, 12.2%, and 12.8%, respectively.

RESULTS

The outcome for each of the 1,000 mailed questionnaires is listed in Table 1, according to the guidelines of the American Association for Public Opinion Research.23 The response rate was 66.9% using the proportional allocation method.19

Table 1 Outcomes of the 1,000 Mailed Questionnaires
Full size table

Table 2 summarizes the characteristics of the eligible respondents who returned their questionnaires and of the target sample of 1,000 physicians. In general, for the characteristics available for comparison, the characteristics of the respondents and the target sample of physicians appear quite similar.

Table 2 Characteristics of Respondents and Target Sample of 1,000 Physicians
Full size table

Table 3 presents the weighted response frequencies for selected survey questions. Most physicians consider the assessment of the fitness to drive of older persons to be an important issue in their practice, but 45.8% do not feel confident in assessing the driving fitness of their patients and 46.7% do not feel that physicians are the most qualified professionals to identify unsafe older drivers. Despite a majority view that reporting unsafe drivers poses a conflict of interest and carries negative consequences for patients and the physician–patient relationship, 72.4% agree that physicians should be legally required to report unsafe drivers to the authorities. A higher proportion of respondents from provinces with mandatory reporting agree with mandatory reporting legislation, but this did not reach statistical significance (73.3 vs 62.1%, P = .07).

Table 3 Weighted Frequency of Responses for Selected Survey Questions (%)
Full size table

Most physicians feel that they would benefit from further education about driving assessment and from a screening instrument to identify high-risk drivers. Nearly one quarter of physicians (23.5%) are not aware of the Canadian Medical Association handbook Determining Medical Fitness to Drive: A Guide for Physicians,6 and 31.2% of those who are aware of it rarely/never use it. There was no significant difference in the use of the handbook by physicians from provinces with mandatory versus discretionary reporting (P = .95).

Table 4 describes the weighted frequencies of performing driving assessments and of reporting patients who are unfit to drive to the licensing authorities. Physicians in provinces with mandatory reporting assess fewer patients (P < .001) but report more (P = .02). The majority who reported patients to the licensing authorities in the previous year only reported 1 to 2. Cross tabulations of number of assessments with number of drivers reported were calculated for provinces with mandatory versus discretionary reporting. For each category of number of assessments done (i.e., 1–2, 3–5, 6–9, >10), physicians in provinces with mandatory reporting reported a higher percentage of patients than physicians in provinces with discretionary reporting did (results not shown). Multiple regression analyses (see Table 5) revealed that having a higher proportion of older patients in one’s practice was associated with a higher odds of conducting driving assessments and of reporting unsafe drivers, while practicing in a province with mandatory reporting of unsafe drivers was associated with a lower odds of conducting assessments but a higher odds of reporting.

Table 4 Weighted Frequencies of Number of Fitness-to-Drive Assessments Performed and Number of Patients Reported to the Licensing Authorities in the Past Year by Region
Full size table
Table 5 Odds Ratios for Physicians Performing Fitness-to-Drive Assessments and Reporting Patients who are Unsafe to Drive to the Licensing Authorities
Full size table

The reporting practices of physicians were asked for 3 different scenarios, as outlined in Table 6. The only significant difference between physicians from provinces with mandatory versus discretionary reporting was that physicians from mandatory reporting provinces are more likely to report an unsafe driver who agrees to stop driving (P < .001).

Table 6 Weighted Frequencies of Reporting Patients to the Licensing Authorities for Specific Scenarios
Full size table

Table 7 shows the variability of the components included in physicians’ assessments of older patients’ fitness to drive. A patient history of driving crashes and infractions and cognitive testing are often/always obtained by less than half of physicians. Physicians from provinces with mandatory reporting are more likely to obtain a driving history (P = .02), do cognitive testing (P = .007) and an ECG (P = .005), and refer patients to a specialist (P < .001) and for a geriatric assessment (P = .02). Physicians from provinces with discretionary reporting are more likely to do cardiac (P < .001), hearing (P = .003), and joint exams (P = .05) and to refer patients for a road test (P = .04).

Table 7 Weighted Frequencies of Carrying Out Various Components of Fitness-to-Drive Assessments in Older Patients (%)
Full size table

Most driving assessments take between 10 and 30 minutes, with 39.3% taking 10–20 minutes and 40.8% taking 20–30 minutes (Table 3). Physicians from provinces with mandatory reporting spend a significantly greater amount of time conducting assessments (P = .04). A positive linear relationship was found between the time taken to conduct the assessment and the number of components included in the assessment (P < .001).

DISCUSSION

Many physicians do not feel confident or qualified in assessing the driving fitness of their patients. The majority feel that reporting patients who are unsafe to drive to the licensing authorities puts them in a conflict of interest and has negative consequences for patients, patients’ families, and the patient–doctor relationship. Despite these misgivings, almost three quarters agree that physicians should be legally required to report unsafe drivers.

These findings are consistent with a study from Saskatchewan, Canada that found that while most of that province’s physicians (58.1%) felt that reporting unsafe drivers negatively affected the physician–patient relationship, an overwhelming majority (92.5%) indicated that the interests of the public should prevail over the needs of the individual driver.11 Similarly, a survey of 467 geriatricians in the United States found that 92.3% would report their unsafe drivers with dementia.24 However, our findings contrast with those from a survey of Canadian neurologists (N = 289) that found that only 44% of respondents were in favor of mandatory reporting by physicians.25 This difference may relate in part to the fact that neurologists, who provide care for patients with several of the leading conditions that affect driver safety (e.g., seizures, strokes, and dementias),26,27 may be concerned with the burden of having to report unsafe drivers, possibly making them more reluctant to take on the responsibility of mandatory reporting.

The high proportion of physicians who indicate that reporting a patient who is unfit to drive negatively impacts the physician–patient relationship underscores the need for physician training in how to communicate the need to stop driving to their patients. Our findings that many physicians lack confidence in their driving assessments and do not consider physicians to be the most qualified professionals to conduct these assessments highlights the need for further physician education about fitness-to-drive assessments. In fact, the vast majority of physicians indicate that they would benefit from further education in this area. This finding is consistent with a study of Finnish and Swedish general practitioners, which showed that only 21% of Finnish and 18% of Swedish physicians agreed with the statement “My training in traffic medicine is sufficient for assessing the driving fitness of older (65+) drivers.”27

Surprisingly, a large number of physicians in our survey were either not aware of, or did not use, the Canadian Medical Association handbook Determining Medical Fitness to Drive: A Guide For Physicians. This is in keeping with findings from an American study that found that 69% of the geriatricians were not aware of the American Medical Association guidelines regarding medical conditions affecting drivers.28 This suggests that physicians need to be made more aware of existing guidelines.

Our study found considerable variability in the reported components of the assessment of fitness to drive of older patients. Unexpectedly, less than half of the physicians often/always obtain a patient history of driving crashes and infractions or perform cognitive testing as part of their assessments. However, these findings are consistent with a study that showed that the majority of geriatricians did not keep a record of their patients’ driving status and that only 8% used mental status as a criterion for recommending driving cessation.28 It is possible that physicians focus exclusively on the presenting medical condition that is perceived to be the potential barrier to safe driving, or that physicians do not have a clear idea about what conditions are most important to screen for and what should be included in a thorough assessment for older drivers. Our findings are consistent with a Norwegian study that found that general practitioners had a variety of approaches to assessing fitness to drive29 and with an American study that found no clear consensus on the criteria used to recommend driving cessation.28 These findings suggest that greater standardization is needed on how physicians should assess their older patients’ medical fitness to drive.

Our study raises some interesting questions about the possible effects of mandatory reporting legislation on physician behavior. Paradoxically, physicians from provinces that require the reporting of unsafe drivers appear to perform fewer driving assessments than respondents from provinces with discretionary reporting do. The requirement to report, with its potential negative repercussions, may act as a disincentive for physicians to initiate a driving assessment.

Our study has some limitations. Although our response rate was not optimal, it was relatively high and consistent across regions. Because the survey was administered anonymously, we could not directly compare the characteristics of respondents with nonrespondents and we had to use different sources of information to compare the characteristics of respondents with the target sample of 1,000 physicians. Although it is reassuring that the respondent characteristics that we were able to compare with those of the target sample of physicians were quite similar, we cannot exclude the possibility of nonresponse bias. As is the case in all surveys, some of the responses may not reflect actual attitudes and practices. Social desirability bias may have influenced some of the responses, but the anonymity of the survey likely reduced this bias. Recall bias is a potential threat for the questions regarding physicians’ practices, but it is unlikely that physicians were very inaccurate about their practices. Lastly, because specialist physicians and physicians who primarily speak French were excluded from the survey, our results are not generalizable to these physicians.

In conclusion, while most Canadian family physicians do not believe that they are the most qualified to assess fitness to drive, and feel that this role puts them in a conflict of interest and negatively affects the patient–physician relationship, the majority feel that it should be the legal responsibility of physicians to report medically unsafe drivers to the licensing authorities. To aid physicians in this challenging role, our study suggests the need for additional physician education about how to conduct driving fitness assessments, greater standardization of the components of these assessments, and better communication strategies for physicians who need to inform their patients to stop driving.