Abstract
Objective
In dementia research, the Driving Scenes test from the Neuropsychological Assessment Battery has been shown to relate to memory, dementia diagnosis, and functional impairment. The aim of the current study was to examine Driving Scenes and its component scores, and their relationships with cognition and daily functioning, in a mixed dementia clinic sample.
Method
One hundred U.S. military veterans between the ages of 55 and 88 were administered a full neuropsychological protocol that included Driving Scenes.
Results
The Driving Scenes score and its subscores were strongly related to memory skills, and there were additional subscore associations with language and visuospatial functions. Driving Scenes uniquely predicted reported bill payment difficulties and tendency to get lost while driving, which were not predicted by other performances across cognitive domains.
Conclusion
Driving Scenes is a clinically and functionally relevant measure of memory. Although the Driving Scenes total score remains useful in dementia evaluations, component scores and error scores contribute additional practical information.
Keywords: Dementia, Everyday functioning, Learning and memory
The Driving Scenes test from the Neuropsychological Assessment Battery (NAB; Stern & White, 2003) was designed as a face-valid and ecologically valid measure for the assessment of visual attention as it pertains to driving safety. The individual first views a hand-drawn picture depicting a scene from the point of view of the driver of a car, and then they are shown five more scenes where each successive scene has some changes relative to the previous scene. The examinee is allotted 1 to 2 min per scene to identify as many changes as they can recall from one scene to the next. The NAB Psychometric and Technical Manual (White & Stern, 2003) described Driving Scenes as a measure of various attention skills including “working memory, visual scanning, attention to detail, and selective attention” (p. 18). That same manual also offered an exploratory factor analysis of the NAB core modules, in which Driving Scenes had its strongest factor loadings with measures of language and memory, not the other NAB Attention module tests. Therefore, the true function of this measure warrants some consideration.
Brown et al. (2005) evaluated Driving Scenes versus the gold-standard measure of driving safety, an on-road driving test, in a sample of older adults with no cognitive impairment or mild cognitive impairment (MCI), and found that safe drivers performed better than both unsafe and borderline-passing drivers. They found that Driving Scenes was 66% accurate at predicting road test results, though they did not report any further analysis of classification accuracy such as sensitivity and specificity values. Stern et al. (2016) evaluated a range of cognitive measures in healthy elderly individuals and those with MCI or dementia and found that Driving Scenes and Useful Field of View were the only two that were significant predictors of at-risk driving based on an on-road driving evaluation. However, in a study that compared scores for safe and unsafe drivers within diagnosis (Grace et al., 2005), there were no differences on Driving Scenes within Alzheimer’s disease or Parkinson’s disease groups, suggesting that the significant effects shown in other studies might have been driven by effects of diagnosis rather than by driving safety per se.
To that end, Gavett et al. (2012) found Driving Scenes to be reasonably predictive of Alzheimer’s disease diagnosis, with 65% sensitivity and 90% specificity when both were optimized, and >90% positive predictive power in settings where at least one of every three examinees has dementia. Similarly, Ashendorf et al. (2018) found that Driving Scenes was highly predictive of informant ratings of instrumental activities of daily living in a dementia research program sample, with up to 55% sensitivity and 96% specificity. On the other hand, in a study conducted with brain injury rehabilitation patients, Driving Scenes was found to be unrelated to clinician-rated functioning (Zgaljardic et al., 2011).
All investigations in the aforementioned studies that included Driving Scenes have used the Driving Scenes total score, which consists of all correct identifications of new, missing, or altered details in the pictures. Reliance on a total collapsed achievement score, though, may mask insight that can be gleaned from attending to component variables, and analyses of errors and finer-level scores often provide richer clinical detail (Libon et al., 2013). Recognition of newly added details, for example, might tap different memory processes than free recall of components that have been removed. Furthermore, those studies evaluated Driving Scenes relative to diagnosis or reported/observed functioning, but none compared it with a range of other cognitive skills to better tease out the cognitive substrates of the test. The present study presents Driving Scenes subscores and error scores and evaluates the relative implications and utility of each of these in a mixed dementia-clinic sample.
METHODS
Participants
One hundred consecutive older adults (ages 55–88) were referred for evaluation of memory concerns. All participants were U.S. military veterans tested by a neuropsychologist at a Department of Veterans Affairs hospital clinic following referral by primary care, neurology, or other specialists. As this is a mixed clinical sample, all referred individuals were included; the only exclusion criterion was inability to complete the Driving Scenes test, which was not seen in any of these participants, so analyses were conducted on the full sample of 100 older adults. Of those with cognitive impairment, the etiology was suspected to be Alzheimer’s disease in 19 cases (31%), vascular dementia or cerebrovascular disease in 24 cases (39%), Parkinson’s disease or Lewy body dementia in 6 cases (10%), a psychiatric disorder (posttraumatic stress disorder or depression) in 6 cases (10%), alcohol-related dementia in 3 cases (5%), frontotemporal dementia in 2 cases (3%), and other lesion/tumor in 2 cases (3%).
Measures
The Driving Scenes test is a “Daily Living” test from the NAB. The examinee is shown a series of six cartoon scenes drawn to look as if the examinee is viewing a road through the windshield of a car. Each scene is presented for at least 1 min but no more than 2 min. The objective is to identify, via pointing and/or verbal description, all of the changes that occur between each successive pair of scenes. Changes can be new additions (herein called “New”), deletions (“Missing”), or features that are still present but change in some way (such as lights or instrument readings on the dashboard; “Different”). The total raw score is the total number of changes identified across all scene pairs. Commission errors do not factor into the standard score, but for the purpose of this study, these are classified as “Intrusions.” Repetition errors within each scene were also tallied. In summary, the Driving Scenes variables explored in this study were total score, new, missing, different, intrusions, and repetitions.
The domains assessed via the neuropsychological protocol included Processing Speed (Wechsler Adult Intelligence Scale 4th Edition (WAIS-IV) Coding (Wechsler, 2008), Salthouse Perceptual Comparisons Test (Salthouse & Babcock, 1991), and Trail Making Test Part A (Reitan, 1955)), Executive Functions (Trail Making Test Part B, WAIS-IV Digit Span longest backward span, and Lexical Fluency (FAS; Tombaugh et al., 1999)), Memory (California Verbal Learning Test 3rd Edition (CVLT-3; Delis et al., 2017), Wechsler Memory Scale 3rd Edition Logical Memory (Wechsler, 1997b), and Rey Complex Figure Test (RCFT; Meyers & Meyers, 1995) Recall), Language (Boston Naming Test (Kaplan et al., 2001), semantic (animal) fluency (Tombaugh et al., 1999), and Wechsler Adult Intelligence Scale 3rd Edition (Wechsler, 1997a) Similarities), and Visuospatial Functions (Clock Drawing (Freedman et al., 1994), RCFT Copy, and Hooper Visual Organization Test (Hooper, 1983)). No individuals’ results raised concern about performance validity based on embedded measures—most commonly Reliable Digit Span (Schroeder et al., 2012) and CVLT forced choice (Clark et al., 2012; Grewal et al., in press); there were some performances that fell below established younger-adult cutoffs, but in each case, these results were consistent with suspected or established neurodegenerative syndromes.
Functional capacities examined in this study were assessed via unstructured interview with the patient and collateral informant. Questions evaluated change and/or impairment in driving skills, financial management, and medication management, and responses resulted in binary ratings for each question. Where this could not be assessed (i.e., if there was no reliable collateral informant, or if the functional behavior was not applicable to the patient, such as if their spouse has always managed the finances or if they did not previously have a driver’s license), the variables were omitted. Specific variables derived from this interview were driving-risky errors (i.e., accidents or improper driving behaviors), driving-wayfinding errors (i.e., getting lost), bill payment errors, and medication management errors.
Data analysis
The examining clinician derived an estimated Clinical Dementia Rating (Morris, 1997) Sum of Boxes (CDR-SB; O’Bryant et al., 2008) score for each individual based on unstructured clinical interview data. Domain index scores were calculated using the mean of age-calibrated z-scores derived from published normative data for all cognitive measures in the study other than Driving Scenes. Descriptive statistics and correlations among all study variables were calculated, and t-tests were conducted to assess differences between groups. For the primary analyses, linear regression equations were derived by entering demographic variables (age and education) and the domain z-scores as predictor variables and each Driving Scenes score (total score and subscores) as the dependent variables. For domains that emerged as sharing significant relationships with Driving Scenes, it was of additional interest to determine which (if any) specific Driving Scenes subscore variables would emerge as predictive of each domain score, after also considering the shared variance among the scores. Therefore, a post hoc set of linear regression analyses consisted of Driving Scenes variables entered as predictors and domain z-scores as the dependent variables.
Diagnostic group classification (intact/MCI/dementia) was conducted using Jak/Bondi psychometric criteria for MCI (Jak et al., 2009) and clinical interview-based assessment of functional capacities to determine impairment in activities of daily living, which would inform diagnosis of dementia. Individuals demonstrating multiple within-domain standardized scores of z ≤ −1.0, in any domain, were classified as “cognitively impaired,” and everyone else was classified as “intact.” Those in the cognitively impaired group who also had functional impairment were classified in the “dementia” group, and those without apparent functional impairment were labeled as “MCI.” Diagnoses were psychometrically determined, and subjective concern was not used toward diagnosis (as every participant had some degree of subjective concern reported by either themselves or a collateral informant).
Secondary logistic regression analyses were conducted to assess the ability of cognitive skills, including most cognitive domains in addition to Driving Scenes total score, to predict report of limitations in various instrumental activities of daily living (driving-risky errors, driving-wayfinding errors, bill payment errors, medication management errors). Variables with multicollinearity were excluded. These analyses were conducted only with individuals who were impaired in at least one domain based on Jak/Bondi criteria (and thus were in the MCI or dementia groups) so that cognitive impairment would not be a confound for these analyses; in other words, all participants in these analyses had some degree of cognitive impairment.
The default p-value cutoff for significance was set to p <.05. However, Holm’s modified Bonferroni procedure was applied to each set of regression models in order to minimize the risk of type I error.
RESULTS
Using CDR-SB descriptors outlined by O’Bryant et al. (2008), of those with cognitive impairment, 30 (48%) were in the range of MCI, 18 (29%) were consistent with very mild dementia, 12 (19%) were classified as having mild dementia, and the remaining 2 (3%) had moderate dementia. Although cognitively impaired participants were initially divided, based on presence/absence of reported functional change, into MCI (n = 29) and dementia (n = 33) groups, there were no differences between these groups on any of the demographic or cognitive variables in this study, so they were collapsed back into a single cognitive impairment (n = 62) group. Table 1 offers demographic and descriptive summaries of the entire sample as well as each group independently.
Table 1.
Descriptive statistics
| Total sample | Intact | MCI/dementia | |
|---|---|---|---|
| (n = 100) | (n = 38) | (n = 62) | |
| Age | 73.9 (6.1) | 74.4 (4.9) | 73.7 (6.8) |
| Education | 12.8 (2.8) | 13.2 (2.6) | 12.5 (2.9) |
| Sex: female (%) | 3 (3%) | 0 (0%) | 3 (5%) |
| Race: White (%) | 93 (93%) | 36 (95%) | 57 (92%) |
| NAB Driving Scenes | |||
| Total raw | 34.5 (8.5) | 40.9 (4.4) | 30.6 (8.0) |
| T-Score | 39.5 (10.1) | 47.4 (6.4) | 34.7 (8.8) |
| New | 25.4 (5.4) | 28.7 (4.2) | 23.3 (5.1) |
| Different | 5.3 (3.1) | 6.7 (2.8) | 4.5 (3.0) |
| Missing | 3.9 (2.7) | 5.5 (2.3) | 2.9 (2.3) |
| Intrusions | 5.5 (4.5) | 4.5 (2.9) | 6.1 (5.2) |
| Repetitions | 1.4 (2.0) | 0.9 (1.6) | 1.7 (2.2) |
| Domain scores: z (SD) | |||
| Processing speed | −0.71 (1.01) | −0.11 (0.78) | −1.08 (0.96) |
| Executive functions | −1.09 (1.19) | −0.30 (0.84) | −1.58 (1.12) |
| Memory | −1.17 (0.96) | −0.35 (0.74) | −1.68 (0.69) |
| Language | −0.37 (0.77) | 0.10 (0.56) | −0.65 (0.74) |
| Visuospatial | −0.57 (1.05) | −0.05 (0.78) | −0.89 (1.07) |
| WTAR FSIQ | 99.8 (11.6) | 103.7 (9.8) | 97.5 (12.1) |
| Digit Span Backward | 3.8 (0.8) | 4.2 (0.8) | 3.6 (0.8) |
| Trail Making Test | |||
| Part A | 51.4 (24.8) | 39.6 (12.2) | 58.6 (27.8) |
| Part B | 189.0 (90.4) | 121.6 (55.6) | 230.4 (82.5) |
| Coding | 36.5 (12.3) | 42.5 (12.5) | 32.8 (10.7) |
| Salthouse PCT | 40.8 (13.7) | 48.6 (12.2) | 36.0 (12.4) |
| Clock Drawing Test | |||
| Command | 3.9 (3.6) | 2.5 (2.7) | 4.8 (3.9) |
| Copy | 1.7 (1.8) | 1.4 (1.4) | 1.9 (1.9) |
| Verbal Fluency | |||
| FAS | 28.7 (11.0) | 34.3 (9.9) | 25.2 (10.1) |
| Animals | 14.4 (5.4) | 16.7 (5.2) | 12.9 (5.1) |
| CVLT-3 | (n = 69) | ||
| T1–5 | 27.6 (9.6) | 32.8 (6.4) | 23.9 (9.8) |
| LDFR | 4.0 (3.3) | 5.7 (2.8) | 2.7 (3.0) |
| CVLT-3 Brief Form | (n = 31) | ||
| T1–4 | 16.8 (5.7) | 22.0 (3.3) | 14.7 (5.1) |
| LDFR | 3.1 (2.5) | 5.4 (2.2) | 2.2 (2.0) |
| WMS-III | |||
| Logical Memory I | 24.0 (10.9) | 32.3 (8.3) | 18.8 (9.0) |
| Logical Memory II | 11.7 (7.8) | 17.9 (6.0) | 7.9 (6.3) |
| RCFT | |||
| Copy | 23.4 (7.0) | 26.7 (5.4) | 21.3 (7.2) |
| Immediate | 8.0 (7.1) | 13.3 (7.0) | 4.7 (4.8) |
| Delayed | 8.0 (6.9) | 13.3 (6.4) | 4.6 (4.8) |
| Boston Naming Test | 49.5 (7.5) | 53.3 (4.2) | 47.2 (8.2) |
| WAIS-III Similarities | 16.7 (5.8) | 19.7 (4.6) | 14.8 (5.8) |
| Hooper | 18.0 (5.6) | 20.7 (3.9) | 16.3 (5.9) |
| Functional capacities | N/A | % impaired: | |
| Driving errors | 8/50 (16%) | ||
| Driving-wayfinding | 10/50 (20%) | ||
| Med management | 27/62 (44%) | ||
| Bill payment | 20/47 (43%) |
Note: MCI = mild cognitive impairment; N/A = not available; NAB = Neuropsychological Assessment Battery; WTAR FSIQ = Wechsler Test of Adult Reading estimated full scale intelligence quotient; PCT = Perceptual Comparisons Test; CVLT-3 = California Verbal Learning Test 3rd Edition; LDFR = long delay free recall; WMS-III = Wechsler Memory Scale 3rd Edition; RCFT = Rey Complex Figure Test; WAIS-III = Wechsler Adult Intelligence Scale 3rd Edition.
Table 2 reports the correlations of Driving Scenes variables with demographic variables and cognitive domain scores. Neither the total Driving Scenes score nor any of the component scores correlated with age. Higher education level was statistically related to higher total scores as well as more identified new details and missing details. Speed, language, memory, and executive functions all shared significant correlations with the Driving Scenes total score as well as almost all component scores. The visuospatial skills domain had comparatively weaker correlations with the Driving Scenes total score and all of its components, but was also the only domain that was significantly correlated with intrusion errors.
Table 2.
Correlations with Neuropsychological Assessment Battery Driving Scenes variables
| Total | New | Different | Missing | Intrusions | |
|---|---|---|---|---|---|
| Age | −0.15 | −0.10 | −0.06 | −0.19 | 0.16 |
| Education | 0.23* | 0.21* | 0.08 | 0.22* | 0.15 |
| Speed | 0.43*** | 0.38*** | 0.34*** | 0.19 | 0.07 |
| Memory | 0.66*** | 0.57*** | 0.46*** | 0.42*** | −0.13 |
| Language | 0.51*** | 0.38*** | 0.39*** | 0.42*** | 0.09 |
| Visuospatial | 0.24* | 0.16 | 0.20* | 0.19 | −0.27** |
| Executive | 0.48*** | 0.38*** | 0.31** | 0.38*** | −0.07 |
Note:
* p < .05,
** p < .01,
*** p < .001.
The regression models in Table 3 show that memory emerged as a consistent predictor of all three types of correct Driving Scenes responses as well as the total Driving Scenes score. Language performance also contributed, to a lesser extent, to Different and Missing responses and to the total Driving Scenes score. Intrusion errors were predicted by visuospatial skills; however, this model did not survive adjustment for multiple comparisons, so the omnibus model for prediction of intrusion errors did not emerge as significant. Repetition errors were not related to any of the cognitive domains; as 92% of the sample committed no more than three repetition errors, the restricted range of this variable may have been a factor.
Table 3.
Linear regressions of cognitive domains on Neuropsychological Assessment Battery Driving Scenes variables
| B | SE | β | t | p | |
|---|---|---|---|---|---|
| Total raw | R 2 = .590; F = 18.90; p < .001 | ||||
| Constant | 72.13 | 8.57 | 8.42 | <.001 | |
| Age | −0.39 | 0.1 | −0.28 | −3.95*** | <.001 |
| Education | −0.12 | 0.26 | −0.04 | −0.44 | .66 |
| Speed | 0.57 | 0.81 | 0.07 | 0.71 | .48 |
| Memory | 4.75 | 0.69 | 0.54 | 6.93*** | <.001 |
| Language | 3.37 | 1.22 | 0.3 | 2.77** | .007 |
| Visuospatial | −0.94 | 0.71 | −0.12 | −1.32 | .19 |
| Executive | 0.68 | 0.82 | 0.1 | 0.82 | .41 |
| New | R 2 = .404; F = 8.91; p < .001 | ||||
| Constant | 42.52 | 6.6 | 6.44 | <.001 | |
| Age | −0.19 | 0.08 | −0.21 | −2.45* | .016 |
| Education | 0.05 | 0.2 | 0.03 | 0.24 | .81 |
| Speed | 0.66 | 0.63 | 0.12 | 1.05 | .3 |
| Memory | 2.69 | 0.53 | 0.47 | 5.08*** | <.001 |
| Language | 0.81 | 0.94 | 0.11 | 0.86 | .39 |
| Visuospatial | −0.64 | 0.55 | −0.12 | −1.16 | .25 |
| Executive | 0.48 | 0.63 | 0.1 | 0.75 | .45 |
| Different | R 2 = .309; F = 5.89; p < .001 | ||||
| Constant | 16.52 | 4.08 | 4.05 | <.001 | |
| Age | −0.09 | 0.05 | −0.18 | −1.98 | .05 |
| Education | −0.2 | 0.12 | −0.18 | −1.62 | .11 |
| Speed | 0.43 | 0.39 | 0.14 | 1.12 | .27 |
| Memory | 1.17 | 0.33 | 0.36 | 3.57*** | <.001 |
| Language | 1.54 | 0.58 | 0.38 | 2.66** | .009 |
| Visuospatial | −0.19 | 0.34 | −0.07 | −0.57 | .57 |
| Executive | −0.29 | 0.39 | −0.11 | −0.74 | .46 |
| Missing | R 2 = .354; F = 7.20; p < .001 | ||||
| Constant | 13.22 | 3.35 | 3.94 | <.001 | |
| Age | −0.11 | 0.32 | −0.26 | −1.72** | .005 |
| Education | 0.03 | 0.27 | 0.03 | 3.32 | .75 |
| Speed | −0.55 | 0.48 | −0.21 | 2.17 | .09 |
| Memory | 0.89 | 0.28 | 0.32 | −0.44** | .001 |
| Language | 1.04 | 0.32 | 0.3 | 1.61* | .03 |
| Visuospatial | −0.12 | 0.04 | −0.05 | −2.89 | .07 |
| Executive | 0.52 | 0.1 | 0.23 | 0.32 | .11 |
| Intrusions | R 2 = .147; F = 2.26; p = .04 | ||||
| Constant | −0.15 | 6.58 | −0.02 | .98 | |
| Age | 0.07 | 0.08 | 0.09 | 0.86 | .39 |
| Education | −0.01 | 0.2 | −0.01 | −0.06 | .95 |
| Speed | 0.94 | 0.62 | 0.21 | 1.5 | .14 |
| Memory | −0.74 | 0.53 | −0.16 | −1.41 | .16 |
| Language | 1.13 | 0.94 | 0.19 | 1.2 | .23 |
| Visuospatial | −1.17 | 0.55 | −0.27 | −2.14* | .04 |
| Executive | −0.54 | 0.63 | −0.14 | −0.85 | .4 |
| Repetitions | R 2 = .109; F = 1.60; p = .15 | ||||
| Constant | −3.98 | 2.95 | −1.35 | .18 | |
| Age | 0.08 | 0.03 | 0.24 | 2.31* | .02 |
| Education | −0.06 | 0.09 | −0.08 | −0.66 | .51 |
| Speed | 0.02 | 0.28 | 0.01 | 0.07 | .94 |
| Memory | −0.27 | 0.24 | −0.13 | −1.14 | .26 |
| Language | −0.07 | 0.42 | −0.03 | −0.17 | .87 |
| Visuospatial | −0.36 | 0.25 | −0.19 | −1.46 | .15 |
| Executive | 0.21 | 0.28 | 0.12 | 0.74 | .46 |
| % Accuracy | R 2 = .225; F = 3.82; p = .001 | ||||
| Constant | 107.66 | 12.52 | 8.6 | <.001 | |
| Age | −0.24 | 0.14 | −0.16 | −1.66 | .1 |
| Education | −0.02 | 0.38 | −0.01 | −0.04 | .97 |
| Speed | −1.26 | 1.19 | −0.14 | −1.06 | .29 |
| Memory | 3.3 | 1 | 0.35 | 3.29** | .001 |
| Language | −1.08 | 1.78 | −0.09 | −0.6 | .55 |
| Visuospatial | 1.98 | 1.04 | 0.23 | 1.9 | .06 |
| Executive | 0.64 | 1.2 | 0.08 | 0.53 | .6 |
Note:
* p < .05,
** p < .01,
*** p < .001.
As Driving Scenes performance appears to be very closely related to memory skill, an additional linear regression was conducted to evaluate the role of each Driving Scenes score in predicting summary memory performance. Table 4 shows that each correct-response score contributes roughly equivalent information to memory ability, whereas each intrusion error contributes about half as much. Language scores were also related to multiple Driving Scenes scores, and in a Table 4 regression it was found to be significantly related to Missing responses, and to a lesser extent to New and Different responses. Consistent with the correlational analyses and the previous linear regression analyses, the only Driving Scenes score that was statistically significantly related to visuospatial skills was intrusion errors.
Table 4.
Linear regression of Neuropsychological Assessment Battery Driving Scenes variables on domain scores
| B | SE | β | t | p | |
|---|---|---|---|---|---|
| Memory | R 2 = .465; F = 16.34; p < .001 | ||||
| Constant | −3.55 | 0.36 | −9.85*** | <.001 | |
| New | 0.07 | 0.02 | 0.42 | 5.00*** | <.001 |
| Different | 0.08 | 0.03 | 0.27 | 3.16** | .002 |
| Missing | 0.07 | 0.03 | 0.19 | 2.31* | .02 |
| Intrusions | −0.03 | 0.02 | −0.16 | −2.06* | .04 |
| Repetitions | −0.01 | 0.04 | −0.02 | −0.23 | .82 |
| Language | R 2 = .294; F = 7.84; p < .001 | ||||
| Constant | −1.8 | 0.33 | −5.44*** | <.001 | |
| New | 0.03 | 0.01 | 0.2 | 2.06* | .04 |
| Different | 0.05 | 0.02 | 0.21 | 2.19* | .03 |
| Missing | 0.09 | 0.03 | 0.31 | 3.28** | .001 |
| Intrusions | 0.02 | 0.02 | 0.1 | 1.08 | .29 |
| Repetitions | 0.01 | 0.03 | 0.02 | 0.2 | .84 |
| Visuospatial | R 2 = .145; F = 3.18; p = .011 | ||||
| Constant | −1.03 | 0.5 | −2.06* | .04 | |
| New | 0.02 | 0.02 | 0.09 | 0.88 | .38 |
| Different | 0.06 | 0.04 | 0.17 | 1.55 | .12 |
| Missing | 0.03 | 0.04 | 0.07 | 0.63 | .53 |
| Intrusions | −0.06 | 0.02 | −0.27 | −2.69* | .01 |
| Repetitions | −0.04 | 0.05 | −0.08 | −0.82 | .41 |
Note:
* p < .05,
** p < .01,
*** p < .001.
Secondary analyses examined the relationship between cognitive scores and aspects of reported functional status. Cognitive scores in this evaluation did not predict reported driving errors (present in 8 of the 50 cognitively impaired participants who were currently or recently driving) or reported medication management errors (present in 27 of 62 participants). For reported financial skills/bill payment errors (present in 20 of 47 participants who currently or recently had at least some responsibility for personal financial management), Driving Scenes was the sole significant cognitive predictor (see Table 5). Likewise, for reported wayfinding difficulty while driving (present in 10 of 50 participants), Driving Scenes was again the only predictor to emerge as significant (see Table 6). The Memory domain was excluded from both of these analyses due to its high correlation with Driving Scenes and concern for multicollinearity, and Executive Functions was removed for similar reasons (high correlations with Speed and Language); when Driving Scenes was substituted with Memory score in both of these analyses, Memory did not become a significant contributor, so Driving Scenes did not simply act as a proxy for memory functions in these analyses.
Table 5.
Logistic regression for functional bill payment complaints among individuals with psychometric impairment (MCI or dementia)
| AUC = 0.813, Nagelkerke R2 = .337 | ||||
|---|---|---|---|---|
| Variable | B | SE | Wald Z | p |
| Constant | 5.55 | 6.77 | 0.67 | .413 |
| Age | −0.006 | 0.06 | 0.01 | .910 |
| Education | 0.11 | 0.17 | 0.41 | .520 |
| Driving Scenes | −0.16 | 0.08 | 4.54 | .033* |
| Speed | 0.19 | 0.51 | 0.14 | .708 |
| Executive Functions | 0.06 | 0.47 | 0.02 | .891 |
| Memory | 0.60 | 0.68 | 0.79 | .374 |
| Language | −0.38 | 0.47 | 0.02 | .891 |
| Visuospatial | 0.83 | 0.54 | 2.37 | .124 |
Note: AUC = area under the curve; MCI = mild cognitive impairment.
* p < .05.
* * p < .01.
* * * p < .001.
Table 6.
Logistic regression for functional driving-wayfinding complaints among individuals with psychometric impairment (MCI or dementia)
| AUC = 0.800, Nagelkerke R2 = .266 | ||||
|---|---|---|---|---|
| Variable | B | SE | Wald Z | p |
| Constant | 11.90 | 9.40 | 1.60 | .205 |
| Age | −0.08 | 0.08 | 0.89 | .347 |
| Education | −0.07 | 0.20 | 0.12 | .728 |
| Driving Scenes | −0.20 | 0.09 | 4.44 | .035* |
| Speed | 0.80 | 0.64 | 1.54 | .214 |
| Executive Functions | −0.25 | 0.59 | 0.18 | .673 |
| Memory | 0.59 | 0.79 | 0.55 | .460 |
| Language | −0.11 | 0.80 | 0.02 | .888 |
| Visuospatial | −0.02 | 0.58 | 0.001 | .972 |
Note: AUC = area under the curve; MCI = mild cognitive impairment.
* p < .05.
* * p < .01.
* * * p < .001.
DISCUSSION
This study explored the cognitive and functional correlates of the NAB Driving Scenes test and its components. Given the correlates and implications observed in this study, Driving Scenes appears to be most closely associated with memory abilities, with potential implications for statistical prediction of instrumental activities of daily living (ADL) performance.
Unsurprisingly, the three component scores that comprise the total Driving Scenes score all contributed unique information. They all contribute equally to the test’s ability to assess memory skills, suggesting that the total score as designed is appropriate and sufficient. The score might be improved as a memory measure by also penalizing for intrusion errors—possibly a half point per intrusion error, according to the current analyses, though this would need to be investigated further before it could be implemented. If one instead wanted to use information from Driving Scenes to inform other domains, they would need to consider the component scores differently. For example, Missing items are the ones that are most related to language abilities—perhaps because they often require verbal description given that they are not on the stimulus page, whereas other items can be adequately indicated by pointing and do not necessitate similarly verbal responses. If Driving Scenes were to be used as a language test, it would call for heavier weighting of these responses. Driving Scenes is weaker as a measure of visuospatial skills, as only intrusion errors were a significant predictor of this domain, so the traditional total score would not be contributory. This unique relationship between intrusion errors and visuospatial skills was unexpected, as intrusions on memory tests are commonly associated with memory impairment. It may be that “intrusion errors” in this study are functionally different from those seen on episodic memory tests (Gaines et al., 2008); for example, here they may be a result of inappropriate attention to perceptual errors within individuals with weaker visuospatial skills. The most consistently significant predictors of Driving Scenes performance were memory and language, which is supportive of a hypothesized relationship between Driving Scenes and areas in and around the medial and anterior temporal lobe, reflecting function of regions including the hippocampus and the entorhinal cortex (Berron et al., 2020; Rodrigue & Raz, 2004), as is true for other measures requiring verbal episodic memory (Keith et al., 2023) and semantic retrieval (Bookheimer, 2002).
Although Driving Scenes is described as a face-valid and ecologically valid measure of visual attention by the measure’s developers, other cognitive skills may be essential for the successful completion of this task. Attention is a heterogeneous domain including a variety of cognitive skills such as simple focused attention, divided attention, selective attention, sustained attention, and working memory (Lezak et al., 2012), or discrete attention processes according to other models (e.g., Mirsky et al., 1991). The assessment of attention is further complicated in that completion of cognitive measures requires the use of multiple cognitive skills. Initial psychometric evaluation of Driving Scenes revealed greater associations of the test with measures that are thought to reflect learning and memory (White & Stern, 2003). Similar findings emerged in this study. Analyses indicated that at least working memory measures were not significant predictors of Driving Scenes performance. The completion of Driving Scenes should require ~6–10 min and may therefore not adequately capture aspects of vigilance or sustained attention that would be observable over longer time periods. Future studies may benefit from incorporating additional measures of attention to fully explore this measure’s association with various aspects of attention.
Driving Scenes was a strong predictor—and, in fact, the only predictor in this study—of reported bill payment problems and reported tendency to get lost while driving. It (and all other measures) did not predict other reported functional capacities in people with cognitive impairment. There are caveats to these analyses, importantly including the fact that some functional deficit reporting subgroups were very small and the fact that there was no structured format for obtaining this information, as well as the fact that all were reported by the participant and/or their collateral informant and were not objectively observed. Regardless, the fact that it uniquely predicted some functional concerns supported previous similar findings from a different sample (Ashendorf et al., 2018). This finding was not seen in a brain injury sample (Zgaljardic et al., 2011), which likely reflects the fact that different clinical populations experience functional difficulties for different reasons. In a dementia sample like the current one, the proximal cause of functional decline is most likely to be memory impairment, so a test like Driving Scenes that is most sensitive to memory decline may be more likely to predict functional decline.
One limitation of this study is the homogeneity of this mostly male military veteran sample. Although this is a strength in the sense that findings are less likely to be attributable to differences between participants, it also renders the findings questionably generalizable to other groups and populations. Future studies should evaluate these analyses in other samples. Although the population is demographically homogeneous, another limitation is that the use of a dementia clinic sample of convenience may be too diagnostically heterogeneous, as relationships between cognitive skills and underlying mechanisms may be clouded by diagnostic discrepancies and/or shared variance that is wholly or largely due to cognitive impairment itself. Future research might look at diagnosis-specific outcomes, such as exploring the test’s utility in Alzheimer’s disease versus vascular dementia. This would also allow for analyses clarifying whether the extent or pattern of cognitive deficits in a diagnostic group affects the relationship between Driving Scenes scores and specific functional capacities. As discussed previously, this study also did not examine Driving Scenes performance relative to formal measures of attention, so although it clearly has a very strong relationship with memory functions in this population, its relationship with attention skills remains unclear.
Strengths of this study include the novelty of investigating component scores and the use of a clinical sample rather than research volunteers. Future directions should include comparing NAB Driving Scenes with other Daily Living tests and other specific memory measures for this purpose, as well as investigating lateralizing component scores, ideally in a stroke sample, for use in assessing spatial attention. In conclusion, the NAB Driving Scenes test is a valid measure of memory, and its total score and components can be used to better understand cognitive skills and potential functional challenges.
Funding
No direct funding was acquired or used for this study. This work was supported by and conducted at the VA Central Western Massachusetts Healthcare System.
Conflict of Interest
None declared.
Disclaimer
The contents of this article do not represent the views of the U.S. Department of Veterans Affairs or the United States Government.
Author Contributions
Lee Ashendorf (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Writing—original draft, Writing—review & editing), Brandon Gavett (Methodology, Writing—review & editing), and Susanne Withrow (Conceptualization, Methodology, Writing—review & editing)
Contributor Information
Lee Ashendorf, Mental Health Service Line, VA Central Western Massachusetts Health Care System, Worcester, MA, USA; Department of Psychiatry, University of Massachusetts Chan Medical School, Worcester, MA, USA.
Susanne Withrow, Behavioral Health Service Line, VA Pittsburgh Health Care System, Pittsburgh, PA, USA.
Brandon E Gavett, Department of Neurology, UC Davis School of Medicine, Davis, CA, USA.
References
- Ashendorf, L., Alosco, M. L., Bing-Canar, H., Chapman, K. R., Martin, B., Chaisson, C. E., et al. (2018). Clinical utility of select neuropsychological assessment battery tests in predicting functional abilities in dementia. Archives of Clinical Neuropsychology, 33(5), 530–540. 10.1093/arclin/acx100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Berron, D., van Westen, D., Ossenkoppele, R., Strandberg, O., & Hansson, O. (2020). Medial temporal lobe connectivity and its associations with cognition in early Alzheimer’s disease. Brain, 143(4), 1233–1248. 10.1093/brain/awaa068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bookheimer, S. (2002). Functional MRI of language: New approaches to understanding the cortical organization of semantic processing. Annual Review of Neuroscience, 25(1), 151–188. 10.1146/annurev.neuro.25.112701.142946. [DOI] [PubMed] [Google Scholar]
- Brown, L. B., Stern, R. A., Cahn-Weiner, D. A., Rogers, B., Messer, M. A., Lannon, M. C., et al. (2005). Driving Scenes test of the Neuropsychological Assessment Battery (NAB) and on-road driving performance in aging and very mild dementia. Archives of Clinical Neuropsychology, 20(2), 209–215. 10.1016/j.acn.2004.06.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Clark, L. R., Stricker, N. H., Libon, D. J., Delano-Wood, L., Salmon, D. P., Delis, D. C., et al. (2012). Yes/no versus forced-choice recognition memory in mild cognitive impairment and Alzheimer’s disease: Patterns of impairment and associations with dementia severity. The Clinical Neuropsychologist, 26(7), 1201–1216. 10.1080/13854046.2012.728626. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Delis, D. C., Kramer, J. H., Kaplan, E., & Ober, B. A. (2017). California Verbal Learning Test (3rd ed.). NCS Pearson. [Google Scholar]
- Freedman, M., Leach, L., Kaplan, E., Winocur, G., Shulman, K. I., & Delis, D. C. (1994). Clock drawing: A neuropsychological analysis. Oxford University Press. [Google Scholar]
- Gaines, J. J., Gavett, R. A., Lynch, J. J., Bakshi, R., & Benedict, R. H. (2008). New error type and recall consistency indices for the brief visuospatial memory test–revised: Performance in healthy adults and multiple sclerosis patients. The Clinical Neuropsychologist, 22(5), 851–863. 10.1080/13854040701565216. [DOI] [PubMed] [Google Scholar]
- Gavett, B. E., Lou, K. R., Daneshvar, D. H., Green, R. C., Jefferson, A. L., & Stern, R. A. (2012). Diagnostic accuracy statistics for seven Neuropsychological Assessment Battery (NAB) test variables in the diagnosis of Alzheimer’s disease. Applied Neuropsychology: Adult, 19(2), 108–115. 10.1080/09084282.2011.643947. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Grace, J., Amick, M. M., D’abreu, A., Festa, E. K., Heindel, W. C., & Ott, B. R. (2005). Neuropsychological deficits associated with driving performance in Parkinson’s and Alzheimer’s disease. Journal of the International Neuropsychological Society, 11(06), 766–775. 10.1017/S1355617705050848. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Grewal, K. S., Trites, M., Kirk, A., MacDonald, S. W., Morgan, D., Gowda-Sookochoff, R., et al. (in press). CVLT-II short form forced choice recognition in a clinical dementia sample: Cautions for performance validity assessment. Applied Neuropsychology: Adult. 10.1080/23279095.2022.2079088. [DOI] [PubMed] [Google Scholar]
- Hooper, E. H. (1983). Hooper Visual Organization Test (VOT). Western Psychological Services. [Google Scholar]
- Jak, A. J., Bondi, M. W., Delano-Wood, L., Wierenga, C., Corey-Bloom, J., Salmon, D. P., et al. (2009). Quantification of five neuropsychological approaches to defining mild cognitive impairment. The American Journal of Geriatric Psychiatry, 17(5), 368–375. 10.1097/JGP.0b013e31819431d5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kaplan, E. F., Goodglass, H., & Weintraub, S. (2001). The Boston naming test (2nd ed.). Lippincott, Williams & Wilkins. [Google Scholar]
- Keith, C. M., Haut, M. W., Wilhelmsen, K., Mehta, R. I., Miller, M., Navia, R. O., et al. (2023). Frontal and temporal lobe correlates of verbal learning and memory in aMCI and suspected Alzheimer’s disease dementia. Aging, Neuropsychology, and Cognition, 30(6), 923–939. 10.1080/13825585.2022.2144618. [DOI] [PubMed] [Google Scholar]
- Lezak, M. D., Howieson, D. B., Bigler, E. D., & Tranel, D. (2012). Neuropsychological assessment (5th ed.). Oxford University Press. [Google Scholar]
- Libon, D. J., Swenson, R., Ashendorf, L., Bauer, R. M., & Bowers, D. (2013). Edith Kaplan and the Boston process approach. The Clinical Neuropsychologist, 27(8), 1223–1233. 10.1080/13854046.2013.833295. [DOI] [PubMed] [Google Scholar]
- Meyers, J. E., & Meyers, K. R. (1995). Rey Complex Figure Test and Recognition Trial: Professional manual. Psychological Assessment Resources. [Google Scholar]
- Mirsky, A. F., Anthony, B. J., Duncan, C. C., Ahearn, M. B., & Kellam, S. G. (1991). Analysis of the elements of attention: A neuropsychological approach. Neuropsychology Review, 2(2), 109–145. 10.1007/BF01109051. [DOI] [PubMed] [Google Scholar]
- Morris, J. C. (1997). Clinical dementia rating: A reliable and valid diagnostic and staging measure for dementia of the Alzheimer type. International Psychogeriatrics, 9(S1), 173–176. 10.1017/S1041610297004870. [DOI] [PubMed] [Google Scholar]
- O’Bryant, S. E., Waring, S. C., Cullum, C. M., Hall, J., Lacritz, L., Massman, P. J., et al. (2008). Staging dementia using clinical dementia rating scale sum of boxes scores: A Texas Alzheimer's Research Consortium study. Archives of Neurology, 65(8), 1091–1095. 10.1001/archneur.65.8.1091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Reitan, R. M. (1955). The relation of the trail making test to organic brain damage. Journal of Consulting Psychology, 19(5), 393–394. 10.1037/h0044509. [DOI] [PubMed] [Google Scholar]
- Rodrigue, K. M., & Raz, N. (2004). Shrinkage of the entorhinal cortex over five years predicts memory performance in healthy adults. Journal of Neuroscience, 24(4), 956–963. 10.1523/JNEUROSCI.4166-03.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Salthouse, T. A., & Babcock, R. L. (1991). Decomposing adult age differences in working memory. Developmental Psychology, 27(5), 763–776. 10.1037/0012-1649.27.5.763. [DOI] [Google Scholar]
- Schroeder, R. W., Twumasi-Ankrah, P., Baade, L. E., & Marshall, P. S. (2012). Reliable digit span: A systematic review and cross-validation study. Assessment, 19(1), 21–30. 10.1177/1073191111428764. [DOI] [PubMed] [Google Scholar]
- Stern, R. A., Abularach, L. M., Seichepine, D. R., Alosco, M. L., Gavett, B. E., & Tripodis, Y. (2016). Office-based assessment of at-risk driving in older adults with and without cognitive impairment. Journal of Geriatric Psychiatry and Neurology, 29(6), 352–360. 10.1177/0891988716666378. [DOI] [PubMed] [Google Scholar]
- Stern, R. A., & White, T. (2003). NAB, neuropsychological assessment battery: Administration, scoring, and interpretation manual. Psychological Assessment Resources. [Google Scholar]
- Tombaugh, T. N., Kozak, J., & Rees, L. (1999). Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming. Archives of Clinical Neuropsychology, 14(2), 167–177. 10.1016/S0887-6177(97)00095-4. [DOI] [PubMed] [Google Scholar]
- Wechsler, D. (1997a). Wechsler adult intelligence scale (3rd ed.). Psychological Corporation. [Google Scholar]
- Wechsler, D. (1997b). Wechsler memory scale (3rd ed.). Psychological Corporation. [Google Scholar]
- Wechsler, D. (2008). WAIS-IV administration and scoring manual. Pearson. [Google Scholar]
- White, T., & Stern, R. A. (2003). Neuropsychological assessment battery: Psychometric and technical manual. Psychological Assessment Resources, Inc. [Google Scholar]
- Zgaljardic, D. J., Yancy, S., Temple, R. O., Watford, M. F., & Miller, R. (2011). Ecological validity of the screening module and the daily living tests of the neuropsychological assessment battery using the Mayo-Portland adaptability Inventory-4 in postacute brain injury rehabilitation. Rehabilitation Psychology, 56(4), 359–365. 10.1037/a0025466. [DOI] [PubMed] [Google Scholar]