Assessment of cognitive function pertaining to capacity for safe and independent living among elderly patients is a central responsibility of many geriatric medical clinics and service agencies. Specific concerns pertaining to judgments of driving capacity are also befalling upon the medical profession in primary care settings [1]. To aid in this task, a number of brief assessment screens are often employed to identify cognitive problems that may be indicative of a range of pragmatic concerns, including driving capacity [2]. Specifically, results on assessment instruments purported to assess attention, reaction time, and visuospatial abilities are often used to inform clinical judgment about driving capacity in such settings. Two such screening instruments typically used to gauge general cognitive function, and inform questions pertaining to driving capacity specifically, are the Folstein Mini Mental Status Exam (MMSE) [3] and the Clock Drawing Test (CDT).
The MMSE is a widely used cognitive screening tool, due to its brevity, ease of administration, and relative breadth [4]. Numerous studies over the past 40 years have supported its utility as a valid and reliable indicator of general cognitive function [5]. The MMSE consists of 30 items comprising subscales assessing orientation, word registration, attention (via a serial sevens or spelling task), word recall, and language. Additionally, a figure copy exercise is included to examine visuospatial abilities. The CDT is hypothesized to assess more specific aspects of planning, organization and visuospatial skill. Directions for completing the CDT involve asking a patient to draw the face of a clock, including the numbers, and then to place the hands to designate a certain time, such as "ten minutes after eleven." Although different scoring templates for the CDT exist, most often code for features such relative size, spacing and placement of numbers or hands, disorganization, perseveration, completeness, and other potential errors that are hypothesized to indicate cognitive impairment [6–8].
In addition to the MMSE, results on the CDT are often used in clinical settings to inform clinical impressions pertaining to whether or not patients are impaired to such an extent that they should not be driving [9]. Although empirical reviews note that performance on the CDT should only be examined in conjunction with other assessments in this regard, anecdotal evidence also suggests that the CDT is often used as a stand-alone instrument to inform judgments of driving capacity, in both medical and non-medical settings.
Despite this apparent widespread use, there appears to be a dearth of research addressing the validity of the CDT for detecting driving impairment. Although a small number of studies exist that suggest CDT scores may relate to driving problems, the size of this literature base coupled with methodological concerns indicate a need for further research. For example, one study [10] examined the effectiveness of the CDT and MMSE, in addition to the Trail Making Test, Part A [11] and a visual acuity test, in predicting driving ability as judged by a driving instructor after participants completed a road test. A discriminant function analysis indicated that the set of test scores and participant age correctly identified 80% of drivers judged to be impaired, and 85% of drivers judged not to be impaired, according to driving instructor assessments. The authors reported that the discriminant model did not include the MMSE, however, because it did not add significant discriminatory power. The authors then suggested that the overall battery may be useful as a screening instrument in primary health care settings for detecting potential problems in driving that would warrant further examination. Separate univariate data on the predictive power for each of the separate instruments, however, was not provided. Additionally, the authors incorporated a 4-point scoring system for the CDT that was created for the study and differs from scoring systems used in other studies. Furthermore, given that only the component instruments are typically used in practice as opposed to the more extensive batteries advocated, the unique predictive power of the CDT warrants further investigation.
Additional evidence for the potential utility of the CDT in predicting driving behaviors is provided in an examination of neurophysiologic phenomena related to caregiver reports of driving impairment in 79 individuals with Alzheimer's disease [12]. Single photon emission computerized tomography was incorporated to examine brain function. Additionally, scores on the MMSE, CDT, and caregiver ratings of driving ability were analyzed. CDT scoring was based upon a 5-point system that was constructed for the study. Results indicated that MMSE scores did not significantly differ between individuals based upon driving ability, but that CDT scores were predictive of driving impairment based upon level of impairment and whether participants were instructed to simply copy an existing clock, or construct their own according to specific directions. Furthermore, imaging also indicated that level of driving impairment related positively to changes in cortical function. These authors hypothesized that cognitive tests assessing visuospatial abilities and executive function may thus show greater discriminative power between driving impaired and non-impaired subjects than MMSE scores, which may be impacted to a greater extent by other non-relevant verbal tasks. The validity of the scoring system constructed for the CDT in comparison to other scoring systems, however, was not further explored.
A pilot study examining the comparability of simulated driving tests in predicting actual driving problems also suggested that CDT scores may be significant predictors [3]. A small sample of nine older adults was incorporated, four of whom were classified as cognitively impaired based in part on abnormal CDT and MMSE scores. It was found that simulated driving tasks correlated moderately with actual driving problems across the groups. No data was provided, however, on the extent to which the CDT or MMSE uniquely predicted impairment.
Given the typically low rate of follow-up for patients referred for more formal driving assessments, it would be beneficial to further investigate the relations between scores on the CDT and reports of actual driving problems. Furthermore, the predictive power of the CDT alone and in conjunction with other assessment tools in predicting reported driving problems has yet to be fully assessed. Additionally, previous studies examining CDT scores and driving behaviors have employed markedly small sample sizes, warranting future research with greater numbers of participants. Finally, previous studies differ in terms of what, if any, scoring systems were used to score the clock drawings. Thus, further investigation of the comparability of different scoring systems is needed.
To address these concerns, this study explored the relations of patient scores on the CDT and MMSE to patient or family reports of driving problems. In so doing, the utility and comparability of three scoring systems for the CDT that are commonly used by researchers and practitioners, namely the Shulman et al. [6], Sunderland et al. [7], and Wolf-Klein et al. [8]systems, were also examined. Specifically, the Shulman system incorporates a 1–6 rating scale, where higher scores indicate higher levels of impairment. Conversely, scores on the Wolf-Klein and Sunderland systems range from 1–10, with lower scores indicating greater levels of impairment. Although specific criteria differ, each system codes for elements pertaining to spacing, organization, and comprehension of the task, among other criteria.
Exploratory analyses also were conducted to examine the predictive utility of the CDT and MMSE in predicting whether driving problems, namely accidents or near misses, were reported. Further analyses examined whether linear relationships existed between CDT and MMSE scores and the reported number of such incidents. Finally, regression tests examined whether the CDT and MMSE uniquely predicted the number of reported incidents.