A Pilot Study to Develop a Brief Question‐Based Screening Tool to Identify Higher‐Risk Older Drivers

To the Editor: Many older drivers are at higher risk of traffic crashes because of diseases, medications, and aging-related impairments, but no single clinical or demographic attribute accurately predicts future crashes.1 Existing screening tools require special examinations,2 equipment,3 or family presence;4 apply only to cognitively impaired drivers;5 or are too lengthy for busy settings.1, 6 Clinical decision rules provide a model for brief older driver screening by clinicians because they are developed from research rather than expert consensus7 and typically have simple scoring systems. To inform development of a decision rule for older driver screening, the current study sought to identify health- and driving-related questions associated with self-reported adverse driving events (ADEs) using only simple questions. Consecutive individuals aged 65 and older visiting an outpatient geriatric clinic or emergency department (ED) were invited to complete a confidential, pilot-tested survey. Eligible participants drove at least occasionally, spoke English, and lacked significant cognitive impairment (Six-Item Screener8 score ≥ 4). The survey included demographic characteristics and questions related to health, daily activities, medications, falls, injuries, and driving. Variables at least moderately positively associated (odds ratio ≥ 2.0) with an ADE (≥1 motor vehicle crashes or police stops as a driver within the previous 12 months) were included in a multivariable logistic regression to estimate their independent associations with ADEs. Two hundred forty-six participants were enrolled in equal proportions from the two sites (participation 43%; median age 76; 50% female) (Table 1). Clinic and ED enrollees were similar, except that clinic enrollees were older (median 79 vs 72; P < .001) and taking more medications (median 10 vs 6; P < .001). Thirty-six participants (15%) reported at least one ADE. The multivariable model included five variables (Hosmer-Lemeshow goodness-of-fit 0.59; area under receiver operator characteristic curve 0.72) (Table 1). The “CRASH” mnemonic was derived from this model: Ever feel confused or disoriented while driving? Regular [daily or almost daily] driver? Avoid driving alone? Ever have difficulty seeing the license plate in front while stopped? Asked by someone to hand over the keys in the past year? In this pilot study, five questions were identified that, if validated in future studies, could be used in clinical practice to screen older drivers for crash risk. Unlike other available screens,1-6 CRASH does not require specific examinations, computers or other equipment, or family participation. This study is novel in its focus on combining simple questions into a scored decision rule for older driver screening, and each CRASH variable is supported by prior research. Confusion while driving is consistent with research highlighting cognitive impairment as a crash risk factor.1 Cognitively impaired drivers may not accurately recall or report their experiences; this model was developed in cognitively intact drivers, and use in other populations requires future validation. Regular driving reflects the fact that greater road exposure affects crash risk. Avoiding driving alone is consistent with the “copilot” phenomenon, whereby drivers rely on passengers to assist with navigation or obstacle avoidance,9 perhaps because of underlying anxieties about impairments that affect driving confidence. Seeing trouble while driving is also logical; although there is insufficient evidence supporting vision screening alone to prevent older driver crashes,10 the visual domain remains important given its obvious connection to driving ability and the potential for interventions to enhance safety. Having been asked to hand over the keys had the strongest association with ADEs; other studies have suggested that input from family or friends, who have ridden with the older driver, can be useful in identifying at-risk drivers.1, 4 Limitations include using self-reported, unverified ADEs, which are an imperfect measure of driving ability, as the outcome. Furthermore, certain variables (e.g., recommendation to stop driving) may have emerged after, not before, ADEs. Study power was limited, and some variables in the model have nonsignificant estimates. With few outcomes, the final model is at risk of overfitting, but it was decided not to compare multiple models to avoid problems associated with post hoc analyses. Finally, the questions may perform differently in other settings or populations; this sample from two outpatient sites included generally healthy, community-dwelling older adults without significant cognitive impairment or acute illness. These results support efforts to develop a question-based decision rule for identification of higher-risk older drivers who need comprehensive testing. Necessary next steps, informed by these results, include prospective validation and possible refinement of a scored decision rule in an independent sample, using on-road driving evaluations as the outcome. If validated, this instrument may prove useful as the first step in tiered older driver assessment, an approach that could prioritize on-road testing for those at greater risk and avoid unfairly restricting safe drivers.2 Accepted for oral presentation at the 2012 American Geriatrics Society annual meeting, May 2012. The authors thank Erin Rodgers, BS, for her assistance with patient enrollment. The contents are solely the responsibility of the authors and do not necessarily represent the official views of the funding agencies. Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper. This work was supported by the Emergency Medicine Foundation (2009–2011 Research Fellowship), the John A. Hartford Center/University of Colorado Denver Center of Excellence, and Grant 5 R49 CE001168 from the Centers for Disease Control and Prevention. Author Contributions: Betz M. E.: Takes responsibility for the manuscript as a whole. Betz M. E.: Study concept and design, data acquisition, analysis and interpretation, and manuscript preparation. Haukoos J. S.: Data analysis and interpretation and manuscript preparation. Valley M. and Johnson R.: Data acquisition and management and manuscript preparation. Schwartz R., DiGuiseppi C., Lowenstein S. R.: Study concept and design, interpretation of data, and preparation of manuscript. Sponsor's Role: No sponsor had any direct involvement in study design, methods, subject recruitment, data collection, analysis, or manuscript preparation.