Contributions of Cardiovascular Burden,Peripheral Inflammation,and Brain Integrity on Digital Clock Drawing Performance in Non-Demented Older Adults

Abstract

Background:

Greater cardiovascular burden and peripheral inflammation are associated with dysexecutive neuropsychological profiles and a higher likelihood of conversion to vascular dementia. The digital clock drawing test (dCDT) is useful in identifying neuropsychological dysfunction related to vascular etiology. However, the specific cognitive implications of the combination of cardiovascular risk, peripheral inflammation, and brain integrity remain unknown.

Objective:

We aimed to examine the role of cardiovascular burden, inflammation, and MRI-defined brain integrity on dCDT latency and graphomotor metrics in older adults.

Methods:

184 non-demented older adults (age 69±6, 16±3 education years, 46% female, 94% white) completed dCDT, vascular assessment, blood draw, and brain MRI. dCDT variables of interest: total completion time (TCT), pre-first hand latency, digit misplacement, hour hand distance from center, and clock face area. Cardiovascular burden was calculated using the Framingham Stroke Risk Profile (FSRP-10). Peripheral inflammation markers included interleukin (IL)-6, IL-8, IL-10, tumor necrosis factor-alpha, and high sensitivity C-reactive protein. Brain integrity included bilateral entorhinal cortex volume, lateral ventricular volume, and whole brain leukoaraiosis.

Results:

FSRP-10, peripheral inflammation, and brain integrity explained an additional 14.6% of the variance in command TCT, where FSRP-10 was the main predictor. FSRP-10, inflammatory markers, and brain integrity explained an additional 17.0% in command digit misplacement variance, with findings largely driven by FSRP-10.

Conclusion:

Subtle graphomotor behavior operationalized using dCDT metrics (i.e., TCT and digit misplacement) is partly explained by cardiovascular burden, peripheral inflammation, and brain integrity and may indicate vulnerability to a disease process.

Keywords

Alzheimer’s disease cardiovascular diseases cognition heart disease risk factors inflammation

INTRODUCTION

The US census estimates an exponential increase in our older adult populous by 2034 [1], which is accompanied by expected increases in neurocognitive disorders [2], including possible dementia, and higher rates of cardiovascular disease [3]. Cardiovascular and cerebrovascular disease contribute to over 77–86% of dementia diagnoses including Alzheimer’s disease (AD) [4]. By 2025, there will be an expected 7.5% increase in AD diagnoses throughout the US. For these reasons, there is a national push to identify prodromal AD characteristics in the community using neuropsychological tests. One such test is the paper and pencil version of the clock drawing test, which is a rich cognitive screening tool able to characterize and dissociate between patients with mild cognitive impairment [5] and dementia subtypes including AD from vascular dementia and Parkinson’s disease dementia [6–10]. Assessing suspected cognitive decline using digital assessment technology has emerged as a parsimonious and reliable means to flag emergent dementia and dementia related problems (see [11] for review) [12]. The clock drawing test has been modernized using digital technology (digital clock drawing test; dCDT) and shows promise for identifying subtle behavioral features that may be indicative of disease risk [12]. Clock drawing graphomotor output (i.e., variables measuring the production of individual pen strokes) has been useful for characterizing cognitively-well older adults [9, 13, 14], while specific latencies (i.e., time elapsed between pen stroke production and the time necessary to produce individual pen strokes) have been most useful for identifying those with dementia [9, 11]. Recent digital clock drawing research also suggests these metrics are correlated with specific cognitive abilities including processing speed, language, working memory, and declarative memory [13]. However, research has yet to examine the contributions of cardiovascular disease risk factors as related to subtle dCDT metrics in non-demented adults.

Cardiovascular disease is well known to potentiate cognitive decline and has been shown to be associated with dysexecutive difficulty including slower time to completion on tests measuring information processing speed and reduced working memory test performance [1, 15, 16]. Individuals with cardiovascular disease also show brain alterations including thinner frontal and more posterior cortical gray matter [17], reduced parietal gray matter volumes [18], and greater volume of white matter changes or leukoaraiosis (LA) seen on magnetic resonance imaging (MRI) [19–21].

Research associating the emergence of dysexecutive difficulty along with compromised MRI regions of the brain is consistent with additional findings linking peripheral markers of inflammation (e.g., interleukin 6, C-reactive protein) [22, 23] and LA [24]. Individuals with chronic inflammatory conditions perform worse on measures of attention, memory, and executive functioning [23, 25, 26]. To our knowledge, only one study investigated the relationship between inflammation and clock drawing, and showed that on the paper and pencil version of the test, older adults with higher levels of serum C-reactive protein (CRP) make more errors [27].

There are many reasons to hypothesize why dCDT latency metrics may associate with an increased cardiovascular and cerebrovascular burden. Slower time-based metrics are well-known to be associated with increasing age [28, 29] and greater MRI white matter disease burden [30–32]. For example, in the Seattle Longitudinal Study, Borghesani and colleagues found that MRI white matter fractional anisotropy is specifically associated with processing speed performance [33]. White matter disease is also closely related to cardiovascular burden and cognitive change. In fact, there is ample evidence suggesting that different aspects of cardiovascular health, such as elevated blood pressure and/or hypertension, smoking, and diabetes, adversely impact different cognitive domains, including executive abilities [34].

Similar to other time-based metrics, latencies captured by the dCDT change as a function of age, such that older individuals take longer to complete their drawing and demonstrate slower higher-order decision-making latencies in both conditions [14]. Similarly, clock drawings produced by those with greater LA are often accompanied with greater total completion time [35, 36] and slower decision-making latencies [37].

dCDT graphomotor variables represent the final graphical output produced from the initiation of pen-paper contact by the participant. Graphical output requires planning and intentional motor control. Individuals with frontal disorders or subcortical grey matter distortion tend to have atypically sized clocks; their graphical output is altered compared to cognitively-well peers [38, 39]. Also, individuals with frontal cortical grey matter disease are most often stimulus-bound and may produce larger drawings [40], whereas participants with a classic Parkinson’s disease (PD) presentation that involves subcortical basal ganglia disorder may produce smaller clocks, particularly PD peers with predominantly executive functioning deficits [38]. Smaller clock drawings are hypothesized to reflect reduced planning and graphomotor output associated with compromised frontostriatal functions [41]. Accuracy of digit placement associates with functional connectivity between the Basal Nucleus of Meynert (BNM) and the anterior cingulate cortex (ACC) [42].

Clock drawing also provides an opportunity for contrasting command and copy conditions, and improvements in performance from command to copy (i.e., making fewer errors in the copy than in the command condition), or lack thereof, are useful in distinguishing underlying cognitive deficits [10]. For example, individuals with a predominantly dysexecutive syndrome, as seen in individuals with more vascular comorbidities, typically do not improve in the copy condition, whereas those with AD usually do [7, 8]. The degree to which one may adjust their performance from command to copy may be an early phenotypic presentation of cerebrovascular health.

The present study was designed to examine whether a metric assessing cardiovascular burden, peripheral inflammation, and an MRI measure of brain integrity could predict performance on the dCDT in non-demented older adults. Based on cognitive theories of normal aging involving the foundational changes in processing speed, we hypothesized that greater cardiovascular burden, greater peripheral inflammation, and poorer brain integrity (as measured by bilateral entorhinal thickness, total LA volume, and total lateral ventricular volume) would predict slower dCDT latencies and poorer performance on graphomotor variables in a sample of non-demented older adults. We also explored whether our predictors of interest contributed to differences in performance from command to copy. Since age influences the duration of latencies and cognitive performance (i.e., motor performance, processing speed performance, reasoning and memory abilities) [28], we included it as a covariate of interest. Cognitive reserve, which refers to the brain’s ability to offset pathological attacks by relying on previously acquired skills to maintain cognitive functioning [43], was also identified as a covariate based on its relationship with dCDT metrics [13].

METHODS

Participants

The University of Florida’s Institutional Review Board approved the current research investigation. All participants provided written informed consent and all aspects of the study followed principles from the Declaration of Helsinki. Participants were recruited via brochures and community mailings, memory screenings, and locally posted fliers.

Inclusion criteria: age≥55, English as their primary spoken language, intact instrumental activities of daily living (IADLs) per Lawton & Brody’s Activity of Daily Living Scale completed during the interview with participants [44], and willingness and eligibility to complete neuropsychological testing and neuroimaging. Exclusion criteria: evidence of a major neurocognitive disorder at baseline per the Diagnostic and Statistical Manual of Mental Disorders – Fifth Edition [45], significant medical illness potentially limiting lifespan, major psychiatric disorders, history of head trauma/neurodegenerative illness, documented learning disorder, seizure disorder or other significant neurological illness,<6^th grade education, substance abuse in the last year, major cardiac disease, and chronic medical illness thought to induce encephalopathy. All participants were screened for dementia over the phone using the Telephone Interview for Cognitive Status [46], and during an in-person interview with a neuropsychologist and trained research coordinator assessing comorbidity rating [46], anxiety, depression, IADLs, neuropsychological functioning, and digital clock drawing [47]. The same examiner administered all test items, and trained raters scored and double-entered behavioral data. Two neuropsychologists reviewed all baseline data to confirm the absence of any dementias. These participant data were studied as part of prior investigations [9, 13, 48].

Predictor variables of interest

Cardiovascular risk burden

Cardiovascular assessment and blood pressure protocol were performed by trained research staff on the same day as the cognitive testing. We operationalized cardiovascular risk using the Framingham Stroke Risk Profile (FSRP), which considers the compounding effects of various risk factors allowing for risk classification and prediction of stroke probability in the next 10 years, rather than separately considering individual risk factors [49]. The FSRP was recently revised to reflect current temporal trends, and the updated version was shown to better predict stroke risk in three large representative community samples [50]. For more details regarding the computation of the FSRP-10, please see the original publication by Dufouil and colleagues [50].

Peripheral inflammation

Peripheral inflammation was quantitated by plasma cytokine concentrations acquired from a fasting blood draw completed in the morning prior to cognitive testing. Plasma IL-6, IL-8, IL-10, Tumor Necrosis Factor-alpha (TNF-α), and high sensitivity C-reactive protein (hsCRP) were measured in duplicate from venous plasma using the Luminex MAGPIXtrademark platform (Thermo-Fisher Scientific, USA), according to the manufacturer’s instructions.

MRI assessment for brain integrity

We operationalized brain integrity by a composite of structural MRI metrics associated with prodromal neurodegenerative changes: bilateral entorhinal thickness, whole brain LA volume, and total lateral ventricular volume. Those markers were carefully selected based on theoretical rationale and previous publications [21, 51, 52]. All MRI metrics were acquired on Siemens 3T Verio scanner with an 8-channel head coil. We acquired T1-weighted (176 contiguous slices, 1 mm³ voxels, TR– TE=2500–3.77 ms) and Fluid Attenuated Inversion Recovery (FLAIR; 176 contiguous slices, 1 mm³ voxels, TR– TE=6000–395 ms) scans. Total intracranial volume (TICV) was estimated by FreeSurfer maskvol algorithm [53]. To form the brain integrity composite, we averaged the standardized scores from the left and right entorhinal thickness in mm³, total LA volume in mm³ (controlled for TICV), and total lateral ventricular volume in mm³ (controlled for TICV). The directionality of the z-scores was standardized such that positive scores indicated better brain integrity.

Bilateral entorhinal cortex. A known marker of medial temporal lobe disease and a major recipient of cholinergic input [54]. We used bilateral entorhinal cortices, given that clock drawing relies on verbally and spatially mediated memory abilities represented in both hemispheres. T1 images were processed using the FreeSurfer version 6.0 pipeline with measures calculated from the automatic subcortical segmentation and Desikan– Killiany– Tourville atlas cortical parcellation [55].

Whole brain LA volume. LA [19] represents white matter abnormalities seen on MRI or CT scans and is a risk factor for subcortical vascular dementia [56]. A reliable rater measured all scans for LA using FLAIR sequences with in-house macros using previously published methods [57].

Lateral ventricular volume. Larger volume of the lateral ventricles is a risk factor for dementia [58].

Digital clock drawing

A trained psychometrist administered the dCDT to command using the instructions to “draw the face of a clock, put in all the numbers, and set the hands to ten after eleven” followed by a copy condition where participants must copy a model clock. Clock drawings were completed using digital pen technology from Anoto, Inc. and corresponding smart paper [12]. The digital pen captures and measures pen positioning on smart paper 75 times/s. A Massachusetts Institute of Technology in-house software system (ClockSketch, a dCDT classification assist tool) classifies each pen stroke with at least 84% accuracy. A trained rater then reliably classified dCDT pen strokes (93–99% accuracy) and manually checked for scoring accuracy [12]. dCDT variables of interest were selected based on prior dCDT research used in memory clinic and perioperative settings. We considered latencies and graphomotor variables deemed important based on a recent publication operationalizing the production of a normal-appearing clock [9]. dCDT variables of interest are as follows.

Latency variables

Total completion time (TCT). Total time taken in seconds to complete all elements of the clock drawing, from first pen-paper contact until completion of the last pen stroke. Command TCT positively correlates with traditional neuropsychological measures of processing speed, language, working memory, and declarative memory, while copy TCT mainly correlates with processing speed and working memory [13]. Previously published normative reference: mean command TCT = 35.02±12.52 s; mean copy TCT = 27.40±8.93 s [9].

Pre-first hand latency (PFHL). Time taken in seconds to set the first clock hand. This latency captures time from the previous stroke to the initiation of the first clock hand. Command PFHL specifically relates to working memory and the ability to disambiguate the syntactic proposition to set the hands for ‘10 after 11’. More effortful disambiguation of this syntactic proposition, as reflected by a longer PFHL, suggests greater working memory resources are needed to achieve accurate hand placement [13]. Previously published normative reference: mean command PFHL = 2.54±2.36 s; mean copy PFHL = 1.15±0.95 s [9].

Graphomotor variables

Digit misplacement. Sum of each digit’s distance, in degrees, from ideal placement around the clock face. Higher digit misplacement is associated with poorer performance on tasks of visual planning/reasoning abilities, and with reduced functional connectivity from the BNM to the ACC [42]. Previously published normative reference: mean command digit misplacement = 70.53±80.44 degrees; mean copy digit misplacement = 65.82±24.49 degrees [9].

Clock face area (CFA). Average of the horizontal and vertical radii in millimeters computed as the area of the clock circle (Area=π*r²). As previously noted, smaller clock drawings along with micrographia are often seen in older adults with subcortical diseases such as PD. Smaller clocks are hypothesized to reflect reduced planning and graphomotor output associated with compromised frontostriatal functions [41]. Previously published normative reference: mean command CFA = 5127.90±2379.39 mm²; mean copy CFA = 3103.35±1255.43 mm² [9].

Hour hand distance from center. Distance from the inner end of the hour hand to the center of the clock face in millimeters. Since age-related deterioration of the dorsal stream often leads to an upward attentional bias [59], older adults with subtle reductions in visuospatial and planning abilities may deviate upward when determining the center of the clock where the inner end of the hands should meet. Previously published normative reference: mean command hour hand distance from center = 4.40±4.06 mm; mean copy hour hand distance from center = 2.93±2.15 mm [9].

Statistical analyses

Statistical analyses were performed using SPSS v.25. We examined potential covariates using Spearman correlations between dCDT variables and participant demographics. Cognitive reserve was operationalized as a composite score including word reading ability using either the Wide Range Achievement Test or the Wechsler Test of Adult Reading, vocabulary knowledge using the Wechsler Abbreviated Scale Intelligence-II, and years of education; all considered estimates of premorbid intelligence. Word reading ability and vocabulary were converted to z-scores based on available published norms, while years of education were converted to z-scores based on the sample mean. All components were then averaged into a composite score. Normality of the distribution was achieved via natural log transformation for inflammatory markers (i.e., IL-6, L-8, IL-10, TNF-α, and hsCRP) and clock drawing latencies (TCT and PFHL), due to skewness of the data. FSRP-10, brain integrity, and other clock drawing variables were normally distributed and did not require transformation. Controlling for age and cognitive reserve, we used separate hierarchical regressions with FSRP-10, inflammatory markers (IL-6, IL-8, IL-10, TNF-α, and hsCRP), and brain integrity as independent variables, and clock drawing latencies (TCT, PFHL) and graphomotor elements (digit misplacement, clock face area, hour hand distance from center) as dependent variables. We corrected for multiple comparisons using a Sidak correction (p≤.005). Missing data were handled using pairwise deletion to maximize the sample size. Based on established criteria, the multicollinearity did not need to be accounted for, given that the variance inflation factor (VIF) was below 10 and tolerance was above 0.10 for all predictors [60].

RESULTS

Participants

From a well-characterized sample of 205 cognitively healthy older adult participants, two participants were excluded due to concerns about a learning disorder, and one participant was excluded due to having PD, thus retaining data for 202 participants. After excluding participants who did not complete cardiovascular assessment and blood draw, the sample included 184 participants. When calculating digit misplacement, participants who drew more or less than 12 digits were excluded from the analyses since the calculation of the digit misplacement variable requires such components (command: 18 excluded; copy: 5 excluded). See Fig. 1. In the retained sample, participants were, on average, 69±6 years old, highly educated (16±3 years), 46% female, predominantly white (94%), and endorsed minimal depressive symptoms (Beck Depression Inventory-II=4.56±5.04) and medical comorbidities (Charlson Comorbidity Index = 0.45±0.76). Table 1 illustrates the demographic characteristics of the sample. There were no sex differences in FSRP-10 and inflammatory markers. Table 2 displays frequency and descriptive statistics for cardiovascular burden markers, inflammatory markers, and brain integrity. Table 3 provides a description of the general function of each inflammatory marker of interest.

Fig. 1

Participant flow chart.

Table 1

Sample characteristics

	Mean (SD) or ratio	Min– Max
Age	68.80 (6.18)	55–85
Sex (Male: Female)	100: 84	–
Race (White: Non-White)	171: 13	–
Years of Education	15.92 (2.71)	9–24
Wide Range Achievement Test (n = 139)	52.06 (3.91)	39–57
Wechsler Test of Adult Reading (n = 44)	43.02 (6.28)	25–50
Cognitive reserve	0.50 (0.62)	– 1.22–1.91
Telephone Interview for Cognitive Status	37.87 (3.79)	26–47
Charlson Comorbidity Index	0.45 (0.70)	0–4
Beck Depression Inventory, 2^nd edition	4.36 (4.84)	0–30

Participants completed either the Wide Range Achievement Test (n = 139) or the Wechsler Test of Adult Reading (n = 44).

Table 2

FSRP Components, Inflammatory Markers, and Brain Integrity Markers

	Mean (SD) or ratio	Min– Max
FSRP-10 Composite	0.06 (0.04)	0.01–0.26
Prevalent current smoking	7/184 (3.80%)	–
Lipid high density lipoprotein (HDL)	56.30 (17.93)	17.00–107.00
Lipid low density lipoprotein (LDL)	102.81 (34.66)	39.00–252.00
Prevalent cardiovascular disease	3/184 (1.63%)	–
Prevalent atrial fibrillation	0/184 (0.00%)	–
Prevalent diabetes mellitus	24/184 (13.04%)	–
Prevalent high blood pressure meds	41/184 (22.28%)	–
Inflammatory Markers
Interleukin-6 (IL-6)	52.42 (87.61)	2.71–500.12
Interleukin-8 (IL-8)	18.08 (19.10)	4.05–99.33
Interleukin-10 (IL-10)	24.32 (19.50)	9.74–226.26
High Sensitivity C-Reactive Protein (hsCRP)	94.28 (103.29)	0.15–511.23
Tumor Necrosis Factor alpha (TNF-α)	19.91 (7.36)	7.15–63.27
Brain Integrity Composite	0.031 (0.63)	– 2.86–1.29
Left hemisphere entorhinal thickness	3.28 (0.25)	2.61–3.88
Right hemisphere entorhinal thickness	3.37 (0.33)	2.33–4.26
Leukoaraiosis raw volume	4092.18 (5181.66)	133.00–46261.00
Left lateral ventricular volume	15488.37 (9284.90)	3875.20–70311.60
Right lateral ventricular volume	14342.43 (8872.66)	2979.10–72149.90

FSRP-10, Framingham Stroke Risk Profile.

Table 3

Description of inflammatory markers’ general function

Cytokine /Chemokine	Abbreviation	Function
Interleukin-6	IL-6	Pro-inflammatory cytokine; produced by T cells and macrophages; acute phase response; fever. Anti-inflammatory effects include inhibition of TNF-α and IL-10.
Interleukin-8	IL-8	Oxidative stress response pro-inflammatory chemokine; plays a role in innate immune response; mediates inflammatory responses and angiogenesis; secreted by macrophages and certain epithelial and endothelial cells.
Interleukin-10	IL-10	Anti-inflammatory cytokine; Secreted by monocytes. Inhibits macrophage and dendritic cell function; suppresses TNF-α.
Tumor Necrosis Factor-alpha	TNF-α	Pro-inflammatory cytokine. Mediates systemic inflammation. Produced by macrophages, monocytes, neutrophils, T cells, NK cells after stimulation with LPS. Also made by astrocytes, microglial cells, smooth muscle cells, and fibroblasts.
High sensitivity C-Reactive Protein	hsCRP	Pro-inflammatory cytokine; acute phase reactant; synthesized in liver hepatocytes, smooth muscle cells, macrophages, endothelial cells, lymphocytes, and adipocytes; promotes atherogenesis through effects on monocytes and endothelial cells.

Command condition

Latency variables

Total completion time. On average, participants took 35.06±12.94 s [range: 14.11–92.93] to complete the clock drawing. Hierarchical regression results are as follows. Block 1: age (B = 0.01, SE = 0.01, p≤0.001) and cognitive reserve (B=– 0.10, SE = 0.04, p = 0.01) significantly explained 10.9% of the variance in command TCT [F(2,143)=8.725, R² = 0.109, p < 0.001]. Block 2: FSRP-10, inflammatory markers, and brain integrity explained an additional 14.6% of the variance in command TCT above and beyond age and cognitive reserve [F change (7,136)=3.795, ΔR ² = 0.146, p≤0.001]. Significant individual predictor included FSRP-10 (B = 3.28, SE = 0.94, p≤0.001); IL-10 (B=– 0.20, SE = 0.08, p = 0.02), and hsCRP (B=– 0.05, SE = 0.02, p = 0.02) did not survive Sidak correction.

Pre-first hand latency. On average, participants took 3.00±2.83 s [range: 0.01–23.22] to set the first clock hand. Hierarchical regression models did not yield significant results for Block 1 (p = 0.44) or Block 2 (p = 0.93).

Graphomotor variables

Digit misplacement. On average, participants displayed 77.48±43.00° of total digit misplacement [range: 21.51–260.85]. Hierarchical regression results are as follows. Block 1: age (B = 0.54, SE = 0.53, p = 0.30) and cognitive reserve (B=– 19.80, SE = 5.27, p≤0.001) significantly explained 9.3% of the variance in command misplacement [F(2,144)=7.38, R² = 0.093, p≤0.001]. Block 2: Together, FSRP-10, inflammatory markers and brain integrity explained an additional 17.0% of the variance in command digit misplacement [F change(7,137)=4.53, ΔR² = 0.17, p≤0.001]. Significant individual predictor included FSRP-10 (B = 594.06, SE = 115.44, p≤0.001). See Table 4.

Table 4

Hierarchical regression summary for dCDT Latencies in the Command Condition

	dCDT Command
	TCT				PFHL
Predictor	R ²	ΔR²	B (SE)	t	R ²	ΔR²	B (SE)	t
Block 1	0.11^***	–			0.01	–
Age			0.01 (0.004)	3.47^***			0.003 (0.01)	0.23
Cognitive Reserve			– 0.10 (0.04)	– 2.50^*			– 0.18 (0.14)	– 1.27
Block 2	0.25^***	0.15^***	0.03	0.02
FSRP-10			3.28 (0.94)	3.49^***			– 2.05 (3.42)	– 0.60
IL-6			– 0.02 (0.05)	– 0.47			– 0.05 (0.19)	– 0.26
IL-8			0.02 (0.09)	0.24			0.14 (0.35)	0.38
IL-10			– 0.20 (0.08)	– 2.40^*			– 0.229 (0.31)	– 0.74
TNF-α			0.06 (0.10)	0.66			– 0.004 (0.36)	– 0.01
hsCRP			– 0.05 (0.02)	– 2.37^*			– 0.022 (0.08)	– 0.29
Brain Integrity			0.06 (0.06)	1.05			– 0.16 (0.23)	– 0.69

TCT, total completion time. PFHL, pre-first hand latency. ^***p≤0.001; ^**p≤0.01; ^*p≤0.05. After correcting for multiple comparisons, ^**p≤0.01; ^*p≤0.05 did not survive the correction.

Clock face area. On average, participants drew a clock face with an area of 4108.43±2263.77 mm² [range: 1159.25–6773.76]. Hierarchical regression results are as follows. Block 1: age (B = 20.59, SE = 28.57, p = 0.47) and cognitive reserve (B = 817.16, SE = 302.95, p = 0.01) explained 6.2% of the variance in command CFA [F(2,144)=4.75, R² = 0.062, p = 0.01]. Block 2: Together, FSRP-10, inflammatory markers, and brain integrity did not explain any significant additional variance in command CFA over and above age and cognitive reserve [F change(7,137)=0.59, ΔR² = 0.027, p = 0.76].

Hour hand distance from center. On average, participants drew the hour hand 3.98±3.93 mm [range: 0.09 – 31.09] away from the center of the clock face. Hierarchical regression models did not yield significant results for Block 1 (p = 0.64) or Block 2 (p = 0.19). See Table 5.

Table 5

Hierarchical regression summary for dCDT graphomotor variables in the command condition

	dCDT Command
	Misplacement				CFA				HH Dist Ctr
Predictor	R ²	ΔR²	B (SE)	t	R ²	ΔR²	B (SE)	t	R ²	ΔR²	B (SE)	t
Block 1	0.09^***	–			0.06^*	–			0.01	–
Age			0.54 (0.53)	1.03			20.59 (28.57)	0.72			– 0.03 (0.05)	– 0.57
Cognitive Reserve	– 19.80 (5.27)	– 3.76^***			843.68 (286.26)	2.95^**			– 0.37 (0.52)	– 0.71
Block 2	0.26^***	0.17^***			0.09	0.03			0.09	0.08
FSRP-10			594.06 (115.44)	5.15^***			5666.00 (6852.10)	0.83			13.25 (11.96)	1.11
IL-6			– 1.95 (6.32)	– 0.31			509.35 (375.02)	1.36			1.24 (0.66)	1.88
IL-8			– 2.72 (11.80)	– 0.23			– 754.52 (700.17)	– 1.08			– 2.67 (0.1.23)	– 2.17^*
IL-10			– 7.46 (10.52)	– 0.71			– 525.07 (624.65)	– 0.84			– 0.28 (1.09)	– 0.26
TNF-α			6.01 (12.10)	0.5			– 102.90 (718.07)	– 0.14			2.01 (1.25)	1.61
hsCRP			0.04 (2.59)	0.02			23.33 (153.77)	0.15			– 0.32 (0.27)	– 1.20
Brain Integrity			– 3.42 (7.66)	– 0.45			– 423.71 (454.92)	– 0.93			– 1.66 (0.81)	– 2.05

CFA, clock face area. HH Dist Ctr, hour hand distance from center. ^***p≤0.001; ^**p≤0.01; ^*p≤0.05. After correcting for multiple comparisons, ^**p≤0.01; ^*p≤0.05 did not survive the correction.

Copy condition

Latency variables

Total completion time. On average, participants took 28.54±9.78 s [range: 13.31–72.23] to complete the clock drawing. Hierarchical regression results are as follows. Block 1: age (B = 0.02, SE = 0.004, p≤0.001) and cognitive reserve (B=– 0.05, SE = 0.04, p = 0.17) explained 11.8% of the variance in copy TCT [F(2,144)=9.61, R² = 0.118, p≤0.001]. Block 2: Together, FSRP-10, inflammatory markers, and brain integrity explained an additional 9.5% of the variance in copy TCT above and beyond age and cognitive reserve [F change(7,137)=2.37, ΔR² = 0.095, p = 0.03). IL-10 (B=– 0.17, SE = 0.08, p = 0.04) did not survive Sidak correction.

Pre-first hand latency. On average, participants took 1.48±1.22 s [range: 0.01–9.48] to set the first clock hand. Hierarchical regression models did not yield significant results for Block 1 (p = 0.72) or Block 2 (p = 0.40). See Table 6.

Table 6

Hierarchical regression summary for dCDT Latencies in the Copy Condition

	dCDT Copy
	TCT				PFHL
Predictor	R ²	ΔR²	B (SE)	t	R ²	ΔR²	B (SE)	t
Block 1	0.12^***	–			0.01	–
Age			0.20 (0.004)	4.24^***			0.01 (0.01)	0.54
Cognitive Reserve			– 0.05 (0.04)	– 1.38			0.08 (0.14)	0.59
Block 2	0.21^*	0.10^*			0.07	0.06
FSRP-10			1.42 (0.89)	1.60			– 2.06 (3.19)	– 0.65
IL-6			– 0.04 (0.05)	– 0.80			– 0.03 (0.18)	– 0.16
IL-8			0.04 (0.09)	0.39			0.17 (0.33)	0.53
IL-10			– 0.17 (0.08)	– 2.08^*			– 0.40 (0.29)	– 1.39
TNF-α			0.11 (0.09)	1.21			– 0.07 (0.34)	– 0.20
hsCRP			– 0.04 (0.02)	– 1.78			0.07 (0.07)	0.96
Brain Integrity			0.02 (0.06)	0.39			0.49 (0.22)	2.27^*

TCT, total completion time. PFHL, pre-first hand latency. ^***p≤0.001; ^**p≤0.01; ^*p≤0.05. After correcting for multiple comparisons, ^**p≤0.01; ^*p≤0.05 did not survive the correction.

Graphomotor variables

Digit misplacement. On average, participants displayed 69.18±27.51° of total digit misplacement [range: 21.51–260.85]. Hierarchical regression results are as follows. Block 1: age (B = 0.30, SE = 0.33, p = 0.37) and cognitive reserve (B=– 15.98, SE = 3.33, p≤0.001) explained 14.0% of variance in copy misplacement [F(2,144)=11.71, R² = 0.14, p≤0.001]. Block 2: Together, FSRP-10, inflammatory markers, and brain integrity did not explain any significant additional variance in copy misplacement above and beyond age and cognitive reserve [F change(7,137)=1.20, ΔR² = 0.05, p = 0.31).

Clock face area. On average, participants drew a clock face with an area measuring 2922.36±1180.82 mm² [range: 283.90–12782.68]. Hierarchical regression models did not yield significant results for Block 1 (p = 0.42) or Block 2 (p = 0.45).

Hour hand distance from center. On average, participants drew the hour hand 3.11±2.35 mm [range: 0.30–20.92] away from the center of the clock face. Hierarchical regression models did not yield significant results for Block 1 (p = 0.51) or Block 2 (p = 0.34). See Table 7.

Table 7

Hierarchical regression summary for dCDT Graphomotor Variables in the Copy Condition

	dCDT Copy
	Misplacement				CFA				HH Dist Ctr
Predictor	R ²	ΔR²	B (SE)	t	R ²	ΔR²	B (SE)	t	R ²	ΔR²	B (SE)	t
Block 1	0.14^***	–			0.01	–			0.01	–
Age			0.30 (0.33)	0.89			– 17.45 (15.51)	– 1.13			– 0.03 (0.03)	– 1.05
Cognitive Reserve			– 15.98 (3.33)	– 4.80^***			– 101.12 (153.67)	– 0.66			– 0.14 (0.31)	– 0.46
Block 2	0.19^***	0.05			0.06	0.05			0.07	0.06
FSRP-10			– 8.50 (78.48)	– 0.11			– 1219.44 (3623.06)	– 0.34			– 1.92 (7.18)	– 0.27
IL-6			– 2.02 (4.30)	– 0.47			134.79 (200.02)	0.67			– 0.24 (0.39)	– 0.60
IL-8			5.02 (8.02)	0.63			90.89 (371.28)	0.25			0.55 (0.73)	0.75
IL-10			– 16.55 (7.16)	– 2.31^*			19.15 (329.25)	0.06			– 1.13 (0.65)	– 1.73
TNF-α			3.19 (8.23)	0.39			– 217.43 (378.68)	– 0.57			1.38 (0.75)	1.83
hsCRP			2.27 (1.76)	1.29			45.06 (81.23)	0.56			– 0.17 (0.16)	– 1.04
Brain Integrity			– 0.05 (5.21)	– 0.01			– 156.31 (245.74)	– 0.64			– 0.62 (0.49)	– 1.28

Differences in performance from the command to copy condition

On average, participants improved from the command to the copy condition on all clock drawing measures.

Digit misplacement. Block 1 did not explain significant variance in performance difference from command to copy in digit misplacement [F(2,144)=1.51, R²=0.002, p = 0.86]. Block 2 (i.e., FSRP-10, inflammatory markers, brain integrity) explained 14.5% of the variance in performance difference from command to copy in digit misplacement, [F change(7,125)=3.03, ΔR² = 0.15, p = 0.01]; however, it did not survive Sidak correction.

FSRP-10, inflammatory markers, and brain integrity did not predict performance difference from command to copy, over and above age and cognitive reserve, for total completion time, pre-first hand latency, clock face area, and hour hand distance from center.

DISCUSSION

The present study aimed to examine the role of cardiovascular burden, peripheral inflammatory markers, and brain integrity on digitally acquired clock drawing latency and graphomotor metrics in older adults. Findings suggest that in non-demented older adults, cardiovascular burden, peripheral inflammation, and brain integrity together partly explain performance in digital clock drawing total completion time and digit misplacement (command condition) over and above age and estimates of cognitive reserve.

Cardiovascular burden, markers of peripheral inflammation, and disease markers, as seen on MRI (brain integrity), explained 14.6% of the variance in command TCT above and beyond age and cognitive reserve. The relationship was largely driven by FSRP-10. Command TCT relies on a multitude of cognitive domains, including processing speed, language, working memory, and declarative memory. Individuals with reduced brain integrity, such as those with mild cognitive impairment and PD, typically have longer TCT [13, 38]. Consistent with literature showing poorer attentional, executive functioning, and speed of information processing in individuals with higher cardiovascular burden [1] and inflammation [23, 26], our results suggest that cardiovascular burden and inflammation help predict command TCT performance. Cardiovascular burden, inflammatory markers, and brain integrity did not predict pre-first hand latency over and above age and cognitive reserve. This may partly be due to several factors, including our sample’s limited cardiovascular burden, the cognitively healthy nature of our sample, and the restricted range of data, thereby limiting our ability to detect more nuanced relationships. Our findings also highlight the significance of cognitive reserve on digitally acquired cognitive assessment. Cognitive reserve was more strongly associated with clock drawing variables in the command condition (total completion time and digit misplacement) than in the copy condition.

Next, over and above age and cognitive reserve, command digit misplacement appears sensitive to cardiovascular burden, peripheral inflammation, and brain integrity, with results largely driven by FSRP-10. These findings are consistent with previous research showing that older adults with higher cardiovascular burden, measured using the FSRP-10, perform more poorly on neuropsychological tests of reasoning, visual organization, attention, visual scanning, and motor speed [61]. A recent study shows digit misplacement is reliant upon semantics, visuospatial, visuoconstructional, and reasoning, and those data also suggest unique associations of digit misplacement with functional connectivity between the BNM and the ACC, two regions highly involved in attentional networks [42]. Therefore, command digit misplacement may be more sensitive to attentional network disturbances brought about by vascular comorbidities.

Individuals with severely compromised cardiovascular functions, such as those with vascular dementia [10], do not show the typical improvement from the more cognitively demanding command condition to the simpler copy condition. The present study explored changes from command to copy in cognitively well older adults. We did not find cardiovascular burden, peripheral inflammation, and brain integrity to be associated with more limited improvements from command to copy in digit placement. These results may, once again, reflect the cognitively healthy nature of our sample. In a sample of older adults with AD, we would expect a greater degree of improvement in digit misplacement from the command to copy condition. Notably, this degree of improvement in AD samples may depend upon cerebrovascular disease risk, given that vascular pathology is often seen in participants with AD [62]. However, more work is needed in this area to test this hypothesis.

We recognize study limitations. First, there is a clear need for improved racial, ethnic, and educational diversity in our sample, likely affecting the generalizability of our findings. CRP and cardiovascular burden are known to be elevated in individuals with chronic stress [63], which highlights the importance of expanding this research to groups that may be experiencing more social stressors, including perceived discrimination. Second, our sample was cognitively healthy; therefore, our findings may not be generalizable to more cognitively impaired samples. Third, given the nature of the digit misplacement variable, we could not compute this variable for participants who did not draw a clock face or who had±12 digits, thereby reducing our sample size. We further recognize that while our findings speak to the utility of a digitally-acquired version of the CDT, we did not use other, perhaps more traditional, neuropsychological measures. However, a previous investigation from our team demonstrates the cognitive correlates of latencies and graphomotor elements of digital clock drawing [13].

Despite these limitations, our sample was well characterized from a neuroanatomical and neuropsychological standpoint with carefully and theoretically selected measures of cognitive functioning, brain integrity, and cognitive reserve. We thoroughly examined vascular contributions to digital clock drawing performance from various angles, including cardiovascular health metrics, blood serum inflammatory markers, and neuroanatomical markers. Finally, we used objectively-quantified clock drawing latencies and graphomotor elements using available digital technology. We show that, in non-demented older adults, certain medically reported markers can predict subtle cognitive-behavioral features potentially indicative of a vulnerability. We also highlight the relevance of using digital screening tools, such as the digital clock drawing test, in populations with increased cardiovascular comorbidities. Future directions include expanding the current research to samples that are more compromised from a cardiovascular standpoint and conducting a deeper exploration of the unique role of each inflammatory marker on cognitive screening tools. Lastly, we hope to pose these questions in older adults with different types of cognitive impairment (i.e., vascular dementia, AD) and of varying levels of severity (i.e., mild cognitive impairment, dementia).

Footnotes

ACKNOWLEDGMENTS

We thank the participants who volunteered their time with this investigation. We thank Donna Weber, BS, for her expertise as a research coordinator. We additionally thank the Moldawer laboratory for the team’s expertise in inflammatory marker measurement.

FUNDING

This project was supported by the National Institute of Health (grant no. R01AG055337, CP; P50AG047266, CP; R01NR014181; R01NS082386, CP; UL1R001427, CP), by a scholarship from the American Psychological Foundation, and by a dissertation grant from the American Psychological Association.

CONFLICT OF INTEREST

Catherine Dion has no conflict of interest to report. Jared J. Tanner, David J. Libon, and Catherine C. Price are Associate Editors of this journal but were not involved in the peer-review process nor had access to any information regarding its peer-review. David J. Libon also receives royalties from Oxford University Press and Linus Health.

DATA AVAILABILITY

The data supporting the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

References

O’Brien

, Erkinjuntti

, Reisberg

, Roman

, Sawada

, Pantoni

, Bowler

, Ballard

, DeCarli

, Gorelick

(2003) Vascular cognitive impairment. Lancet Neurol 2, 89–98.

(2021) 2021 Alzheimer’s disease facts and figures. Alzheimers Dement 17, 327–406.

Tsao

, Aday

, Almarzooq

, Alonso

, Beaton

, Bittencourt

, Boehme

, Buxton

, Carson

, Commodore-Mensah

(2022) Heart disease and stroke statistics— 2022 update: A report from the American Heart Association, Circulation 145, e153–e639.

Azarpazhooh

, Avan

, Cipriano

, Munoz

, Sposato

, Hachinski

(2018) Concomitant vascular and neurodegenerative pathologies double the risk of dementia. Alzheimers Dement 14, 148–156.

Ahmed

, Brennan

, Eppig

, Price

, Lamar

, Delano-Wood

, Bangen

, Edmonds

, Clark

, Nation

, Jak

, Au

, Swenson

, Bondi

, Libon

(2016) Visuoconstructional impairment in subtypes of mild cognitive impairment. Appl Neuropsychol Adult 23, 43–52.

Cosentino

, Jefferson

, Chute

, Kaplan

, Libon

(2004) Clock drawing errors in dementia: Neuropsychological and neuroanatomical considerations. Cogn Behav Neurol 17, 74–84.

Libon

, Malamut

, Swenson

, Sands

, Cloud

(1996) Further analyses of clock drawings among demented and and nondemented older subjects. Arch Clin Neuropsychol 11, 193–205.

Libon

, Swenson

, Barnoski

, Sands

(1993) Clock drawing as an assessment tool in dementia. Arch Clin Neuropsychol 8, 405–416.

Davoudi

, Dion

, Formanski

, Frank

, Amini

, Matusz

, Wasserman

, Penney

, Davis

, Rashidi

(2021) Normative references for graphomotor and latency digital clock drawing metrics for adults age 55 and older: Operationalizing the production of a normal appearing clock. J Alzheimers Dis 82, 59–70.

10.

Price

, Cunningham

, Coronado

, Freedland

, Cosentino

, Penney

, Penisi

, Bowers

, Okun

, Libon

(2011) Clock drawing in the Montreal Cognitive Assessment: Recommendations for dementia assessment. Dement Geriatr Cogn Disord 31, 179–187.

11.

Libon

, Swenson

, Lamar

, Price

, Baliga

, Pascual-Leone

, Au

, Cosentino

, Andersen

(2022) The Boston Process Approach and Digital Neuropsychological Assessment: Past research and future directions. J Alzheimers Dis 87, 1419–1432.

12.

Souillard-Mandar

, Davis

, Rudin

, Au

, Libon

, Swenson

, Price

, Lamar

, Penney

(2016) Learning classification models of cognitive conditions from subtle behaviors in the digital clock drawing test. Mach Learn 102, 393–441.

13.

Dion

, Arias

, Amini

, Davis

, Penney

, Libon

, Price

(2020) Cognitive correlates of digital clock drawing metrics in older adults with and without mild cognitive impairment. J Alzheimers Dis 75, 73–83.

14.

Piers

, Devlin

, Ning

, Liu

, Wasserman

, Massaro

, Lamar

, Price

, Swenson

, Davis

, Penney

, Au

, Libon

(2017) Age and graphomotor decision making assessed with the digital clock drawing test: The Framingham Heart Study. J Alzheimers Dis 60, 1611–1620.

15.

Muller

, Grobbee

, Aleman

, Bots

, Van

der Schouw Y

(2007) Cardiovascular disease and cognitive performance in middle-aged and elderly men. Atherosclerosis 190, 143–149.

16.

Schievink

, van

Boxtel MP

, Deckers

, van

Oostenbrugge RJ

, Verhey

, Köhler

(2022) Cognitive changes in prevalent and incident cardiovascular disease: A 12-year follow-up in the Maastricht Aging Study (MAAS), Eur Heart J 43, e2–e9.

17.

Leritz

, McGlinchey

, Kellison

, Rudolph

, Milberg

(2011) Cardiovascular disease risk factors and cognition in the elderly. Curr Cardiovasc Risk Rep 5, 407–412.

18.

Cardenas

, Reed

, Chao

, Chui

, Sanossian

, DeCarli

, Mack

, Kramer

, Hodis

, Yan

(2012) Associations among vascular risk factors, carotid atherosclerosis, and cortical volume and thickness in older adults. Stroke 43, 2865–2870.

19.

Hachinski

, Potter

, Merskey

(1986) Leuko-araiosis: An ancient term for a new problem. Can J Neurol Sci 13, 533–534.

20.

Libon

, Price

, Giovannetti

, Swenson

, Bettcher

, Heilman

, Pennisi

(2008) Linking MRI hyperintensities with patterns of neuropsychological impairment: Evidence for a threshold effect. Stroke 39, 806–813.

21.

Price

, Mitchell

, Brumback

, Tanner

, Schmalfuss

, Lamar

, Giovannetti

, Heilman

, Libon

(2012) MRI-leukoaraiosis thresholds and the phenotypic expression of dementia. Neurology 79, 734–740.

22.

Harris

, Ferrucci

, Tracy

, Corti

, Wacholder

, Ettinger

Jr , Heimovitz

, Cohen

, Wallace

(1999) Associations of elevated interleukin-6 and C-reactive protein levels with mortality in the elderly. Am J Med 106, 506–512.

23.

Trollor

, Smith

, Agars

, Kuan

, Baune

, Campbell

, Samaras

, Crawford

, Lux

, Kochan

(2012) The association between systemic inflammation and cognitive performance in the elderly: The Sydney Memory and Ageing Study. Age 34, 1295–1308.

24.

Dichgans

, Wardlaw

, Smith

, Zietemann

, Seshadri

, Sachdev

, Biessels

, Fazekas

, Benavente

, Pantoni

(2016) METACOHORTS for the study of vascular disease and its contribution to cognitive decline and neurodegeneration: An initiative of the Joint Programme for Neurodegenerative Disease Research. Alzheimers Dement 12, 1235–1249.

25.

Marsland

, Petersen

, Sathanoori

, Muldoon

, Neumann

, Ryan

, Flory

, Manuck

(2006) Interleukin-6 covaries inversely with cognitive performance among middle-aged community volunteers. Psychosom Med 68, 895–903.

26.

Schram

, Euser

, De

Craen AJ

, Witteman

, Frölich

, Hofman

, Jolles

, Breteler

, Westendorp

(2007) Systemic markers of inflammation and cognitive decline in old age. J Am Geriatr Soc 55, 708–716.

27.

Ravaglia

, Forti

, Maioli

, Arnone

, Pantieri

, Cocci

, Nativio

, Muscari

, Pedone

, Mariani

(2003) The clock-drawing test in elderly Italian community dwellers: Associations with sociodemographic status and risk factors for vascular cognitive impairment. Dement Geriatr Cogn Disord 16, 287–295.

28.

Salthouse

(1996) The processing-speed theory of adult age differences in cognition. Psychol Rev 103, 403.

29.

Salthouse

(2010) Selective review of cognitive aging. J Int Neuropsychol Soc 16, 754–760.

30.

Van den Heuvel

, Ten

Dam V

, De Craen

, Admiraal-Behloul

, Olofsen

, Bollen

, Jolles

, Murray

, Blauw

, Westendorp

(2006) Increase in periventricular white matter hyperintensities parallels decline in mental processing speed in a non-demented elderly population. J Neurol Neurosurg Psychiatry 77, 149–153.

31.

Raz

, Rodrigue

, Acker

(2003) Hypertension and the brain: Vulnerability of therefrontal regions and executive functions. Behav Neurosci 117, 1169–1180.

32.

Rabbitt

, Scott

, Lunn

, Thacker

, Lowe

, Pendleton

, Horan

, Jackson

(2007) White matter lesions account for all age-related declines in speed but not in intelligence. Neuropsychology 21, 363.

33.

Borghesani

, Madhyastha

, Aylward

, Reiter

, Swarny

, Schaie

, Willis

(2013) The association between higher order abilities, processing speed, and age are variably mediated by white matter integrity during typical aging. Neuropsychologia 51, 1435–1444.

34.

Liebel

, Jones

, Oshri

, Hallowell

, Jerskey

, Gunstad

, Sweet

(2017) Cognitive processing speed mediates the effects of cardiovascular disease on executive functioning. Neuropsychology 31, 44.

35.

Dion

, Davoudi

, Formanski

, Wiggins

, Davis

, Penney

, Tanner

, Price

(2020) Cerebrovascular contributions to the pre-post-operative cognitive slowing in clock drawing after total knee arthroplasty in older adults. Anesth Analg 130, 486.

36.

Libon

, Penney

, Lamar

, Swenson

, Garrett

, Davis

(2011) Digital Clock Drawing Test (dCDT)– V: Using total drawing time and latencies to assess decision making and capacity to shift mental set in patient with Alzheimer’s disease and subcortical vascular dementia. J Int Neuropsychol Soc 17, 271.

37.

Zhào

, Wei

, Do

EY-L

, Huang

(2019) Assessing performance on digital clock drawing test in aged patients with cerebral small vessel disease. Front Neurol 10, 1259.

38.

Dion

, Frank

, Crowley

, Hizel

, Rodriguez

, Tanner

, Libon

, Price

(2021) Parkinson’s disease cognitive phenotypes show unique clock drawing features when measured with digital technology. J Parkinsons Dis 11, 779–791.

39.

Freedman

, Leach

, Kaplan

, Winocur

, Shulman

, Delis

(1994) Clock drawing: A neuropsychological analysis, Oxford University Press, USA.

40.

Lee

, Kim

, Choi

, Sohn

(2009) Characteristics of clock drawing test (CDT) errors by the dementia type: Quantitative and qualitative analyses. Arch Gerontol Geriatr 48, 58–60.

41.

Eknoyan

, Hurley

, Taber

(2012) The clock drawing task: Common errors and functional neuroanatomy. J Neuropsychiatry Clin Neurosci 24, 260–265.

42.

Dion

, Tanner

, Formanski

, Davoudi

, Rodriguez

, Wiggins

, Amin

, Penney

, Davis

, Heilman

(2022) The functional connectivity and neuropsychology underlying mental planning operations: Data from the digital clock drawing test. Front Aging Neurosci 14, 868500.

43.

Stern

(2009) Cognitive reserve. Neuropsychologia 47, 2015–2028.

44.

Lawton

, Brody

(1969) Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontologist 9, 179–186.

45.

American Psychiatric Association (2013) Diagnostic and Statistical Manual of Mental Disorders (DSM-5^®). American Psychiatric Publishing, Washington, DC.

46.

Welsh

, Breitner

, Magruder-Habib

(1993) Detection of dementia in the elderly using telephone screening of cognitive status. Neuropsychiatry Neuropsychol Behav Neurol 6, 103–110.

47.

Davis

, Penney

, Pittman

, Libon

, Swenson

, Kaplan

(2011) The digital clock drawing test (dCDT) I: Development of a new computerized quantitative system. J Int Neuropsychol Soc 17, 273.

48.

Hizel

, Warner

, Wiggins

, Tanner

, Parvataneni

, Davis

, Penney

, Libon

, Tighe

, Garvan

, Price

(2019) Clock drawing performance slows for older adults after total knee replacement surgery. Anesth Analg 129, 212–219.

49.

Wolf

, D’Agostino

, Belanger

, Kannel

(1991) Probability of stroke: A risk profile from the Framingham Study. Stroke 22, 312–318.

50.

Dufouil

, Beiser

, McLure

, Wolf

, Tzourio

, Howard

, Westwood

, Himali

, Sullivan

, Aparicio

, Kelly-Hayes

, Ritchie

, Kase

, Pikula

, Romero

, D’Agostino

, Samieri

, Vasan

, Chene

, Howard

, Seshadri

(2017) Revised Framingham Stroke Risk profile to reflect temporal trends. Circulation 135, 1145–1159.

51.

Price

, Garrett

, Jefferson

, Cosentino

, Tanner

, Penney

, Swenson

, Giovannetti

, Bettcher

, Libon

(2009) Leukoaraiosis severity and list-learning in dementia. Clin Neuropsychol 23, 944–961.

52.

Price

, Tanner

, Schmalfuss

, Brumback

, Heilman

, Libon

(2015) Dissociating statistically-determined Alzheimer’s disease/vascular dementia neuropsychological syndromes using white and gray neuroradiological parameters. J Alzheimers Dis 48, 833–847.

53.

Crowley

, Huang

, Tanner

, Zhao

, Schwab

, Hizel

, Ramon

, Brumback

, Ding

, Price

(2018) Considering total intracranial volume and other nuisance variables in brain voxel based morphometry in idiopathic PD. Brain Imaging Behav 12, 1–12.

54.

Velayudhan

, Proitsi

, Westman

, Muehlboeck

J-s

, Mecocci

, Vellas

, Tsolaki

, Kłoszewska

, Soininen

, Spenger

(2013) Entorhinal cortex thickness predicts cognitive decline in Alzheimer’s disease. J Alzheimers Dis 33, 755–766.

55.

Fischl

(2012) FreeSurfer. Neuroimage 62, 774–781.

56.

Price

, Jefferson

, Merino

, Heilman

, Libon

(2005) Subcortical vascular dementia: Integrating neuropsychological and neuroradiologic data. Neurology 65, 376–382.

57.

Crowley

, Tanner

, Ramon

, Schwab

, Hizel

, Price

(2018) Reliability and utility of manual and automated estimates of total intracranial volume. J Int Neuropsychol Soc 24, 206.

58.

Carmichael

, Kuller

, Lopez

, Thompson

, Dutton

, Lu

, Lee

, Aizenstein

, Meltzer

(2007) Ventricular volume and dementia progression in the Cardiovascular Health Study. Neurobiol Aging 28, 389–397.

59.

Mańkowska

, Heilman

, Williamson

, Harciarek

(2018) Age-related changes in the allocation of vertical attention. J Int Neuropsychol Soc 24, 1121–1124.

60.

Tabachnick

, Fidell

, Ullman

(2013) Using multivariate statistics, Pearson Boston, MA.

61.

Elias

, Sullivan

, D’Agostino

, Elias

, Beiser

, Au

, Seshadri

, DeCarli

, Wolf

(2004) Framingham stroke risk profile and lowered cognitive performance. Stroke 35, 404–409.

62.

Kapasi

, DeCarli

, Schneider

(2017) Impact of multiple pathologies on the threshold for clinically overt dementia. Acta Neuropathol 134, 171–186.

63.

Beatty

Moody DL

, Brown

, Matthews

, Bromberger

(2014) Everyday discrimination prospectively predicts inflammation across 7-years in racially diverse midlife women: Study of women’s health across the nation. J Soc Issues 70, 298–314.