Research Papers

The Trail Making Test B: Minimizing the risk of false positives severely compromises the test's ability to identify unsafe drivers.

Version 1
Date added June 26, 2017
Downloaded 0 times/fois
Category 2017 CARSP XXVII Toronto
Tags Research and Evaluation, Session 2B
Author/Auteur Rose Wang
Stream/Volet Research and Evaluation

Slidedeck Presentation Only (no paper submitted)

2B_4_Wang

Abstract

"The Trail Making Test“ Part B(TMT-B) is among the most widely used screening measures to predict driving performance in a clinical setting for older drivers. However there is a lack of reported data describing the diagnostic properties of the TMT-B to identify unsafe drivers while minimizing the risk of inappropriately labeling safe drivers as unsafe. The primary objective was to determine the sensitivity of the TMT-B to identify unsafe drivers at high levels of specificity (90%, 95%, 100%). Secondary objectives were to examine the relationship between TMT-B completion time and sensitivity/specificity cut-offs and also assess the degree of evidence supporting TMT-B as a screening tool for unsafe driving. A systematic search process, based on subject headings and keywords words relating to the Trail Making Test B, sensitivity and specificity, and fitness-to-drive, was used to retrieve articles from the following databases: Ovid MEDLINE, Ovid EMBASE, CINAHL, and Web of Science. Retrieved studies were included in the analyses if they reported the TMT-B results separate from the TMT-A, and used a standardized on-road test that employed a pass/fail outcome, designating individuals as safe or unsafe to drive. Studies evaluating populations less than 55 years of age, or involving individuals who had sustained a traumatic brain injury were excluded.

Published ROC curves that assessed the discrimination of the TMT-B for driving performance were used to extract sensitivity values at high levels of specificity. Using a standardized approach, three investigators independently extracted values which were averaged to generate final sensitivity scores. Correlations between sensitivity/specificity and reported TMT-B completion time cut-offs were assessed using Pearson's r. To assess the degree of evidence supporting TMT-B as screening tool for unsafe driving, all studies initially retrieved were reviewed for key elements (e.g., gold standard employed, diagnostic properties reported). Forty-five articles were retrieved. Only six articles presented TMT-B sensitivity and specificity scores and utilized an on-road driving test as the gold standard. Four of these six articles reported ROC curves that allowed for estimation of sensitivity at high specificities (90%, 95%, and 100%).

Overall, the mean extracted sensitivities were 32.7%, 22.43% and 10.95% at specificities of 90%, 95%, and 100% respectively. Higher cut-offs in time to completion for TMT-B were associated with higher specificity (r=.852) and lower cut-off points generally associated with a higher level of sensitivity (r=.837). When cut-points that ensure high specificity are used, the sensitivity of TMB-B is very poor. For example, our analysis showed that for 95% specificity, the TMT-B failed to identify 77% of unsafe drivers. While failing to identify unsafe drivers is certainly a public safety concern, ensuring a high level of specificity is a priority to minimize the risk of misclassifying safe drivers as unsafe, thus, preventing seniors from suffering an overall reduction in quality of life as a result of inappropriately losing their driving privileges. Overall, when risk of mislabeling safe drivers as unsafe was minimized the TMT-B's ability to detect unsafe drivers was severely compromised."

Rose Wang