ABBA Letter Alternation: A Telehealth Inspired Measure of Executive Functioning/Inhibitory Control

Abstract

Objective:

To introduce ABBA Letter Alternation (ABBA) as a computerized measure of response inhibition/response alternation developed for telehealth following restrictions of in-person testing due to COVID-19. ABBA consists of two PowerPoint-administered trials: Letter Reading of 25 capital As or Bs individually presented, and Letter Alternation with instructions to say the opposite letter to what is presented.

Method:

We obtained initial normative ABBA performance from 899 healthy research volunteers participating in the Emory Healthy Brain Study (EHBS) with Montreal Cognitive Assessment (MoCA) scores 24/30 and higher. Cutpoints derived from the EHBS sample were applied to a series of 32 Parkinson disease (PD) patients being evaluated for deep brain stimulation to provide preliminary clinical validation. Trail Making B (TMT B) was also examined in both groups.

Results:

775 (86.2%) EHBS participants made 0-1 ABBA Letter Alternation errors, 58 (6.5%) EHBS participants had 2 ABBA alternation errors, and 66 (7.3%) made 3+ errors. Applying these thresholds to PD patients, 22 (68.8%) made 0-1 alternation errors, 3 PD (9.4%) patients made 2 errors, and 7 PD subjects (21.8%) made 3+ errors, which significantly differed in frequency from the EHBS group (χ²=9.8, p=.007). EHBS vs. PD differed on MoCA, a medium effect (p<.00001; η²=.076), and on TMT B (p<.00001; η²=.158), which is considered a large effect.

Conclusion:

These results provide initial support for ABBA Letter Alternation as a response inhibition/response alternation. Application in larger clinical samples, in both telehealth and face-to-face settings, will be needed to more fully establish ABBA’s clinical utility.

Introduction

Executive function testing is a critical component of clinical neuropsychology, assessing higher-order cognitive processes that enable goal-directed behavior, decision-making, problem-solving, and emotional regulation. While precise definitions of executive function are debated, three distinct types of executive function include inhibitory control, working memory, and cognitive flexibility (Diamond, 2013).

Due to restrictions designed to minimize virus transmission, the COVID-19 pandemic created the need to develop robust assessment methods that could be administered remotely (Bilder et al., 2020; Hewitt, Rodgin, Loring, Pritchard, & Jacobson, 2020). While many cognitive measures were readily adaptable to telehealth administration (e.g., language, verbal memory, cognitive screening), tests requiring a motor component for successful task completion (e.g., Trail Making Test, Block Design) did not easily lend themselves to a telehealth environment (Hewitt & Loring, 2020).

The Trail Making Test Part B (TMT B) is a well-recognized executive function measure assessing set-shifting and cognitive flexibility. TMT B is a critical component for participant characterization in our Emory Healthy Brain Study (EHBS), a large NIH-funded Alzheimer Disease (AD) biomarker discovery project designed to capture early conversion from normal age-related cognitive performance. During the early stages of the COVID-19 pandemic, we faced the challenge of assessing cognitive abilities without being able to administer TMT B remotely, creating the need to explore alternative methods for assessing response alternation and response inhibition that could be delivered remotely. We considered the Oral Trail Making Test (Mrazik, Perra, Brooks, & Naidu, 2014) as a potential telehealth alternative to the TMT-B. However, we had concerns that, without a visual cue, the task might place an additional load on working memory, beyond just assessing response alternation and inhibition. Additionally, we were concerned about its correlation of only r=.6 with the written TMT B. In our initial piloting of Oral TMT, participants often obtained z-scores less than −3.0, calling into question the representativeness of their limited normative dataset of only n=81 subjects.

We chose to modify a paper-and-pencil test that we had previously developed to assess response inhibition (Brikman, Drane, Prosje, DeFilippis, & Horne-Moyer, 2006; Drane et al., 2001). The original version of the ABBA letter alternation task consists of two trials where completion time serves as the primary dependent measure. The first trial establishes a baseline for letter reading speed (reading alternating capital As and Bs), while the second trial introduces a demand for response inhibition by requiring participants to say the opposite letter from the one presented. ABBA is conceptually similar to the Stroop Color and Word Test (Scarpina & Tagini, 2017), as both assess response inhibition of a prepotent automatic response. However, instead of inhibiting word reading, ABBA involves letter reading as the automatic response. Successful performance on the conflict task requires top-down control to override the dominant reading response, which represents a core aspect of executive functioning. This makes it conceptually similar to other response inhibition tasks such as go/no-go and antisaccade tasks.

ABBA was adapted for telehealth administration by utilizing a fixed pacing rate for stimulus presentation via PowerPoint, with letter alternation error frequency replacing self-paced completion time as the primary performance measure. In the original version, letters were presented in long horizontal rows, and completion time was the main metric. However, remote assessment, often conducted in more distracting environments, made it harder to gauge participants' engagement with the stimuli compared to in-person testing. Additionally, the self-paced nature of the original task reduced the likelihood of response errors. By requiring participants to respond to each individual letter presentation at a fixed rate, active engagement became more apparent. This fixed presentation pace also introduced response time pressure, increasing the likelihood of eliciting response errors.

In the present report, we contrast computerized ABBA Letter Alternation error frequency, TMT A, TMT B, and MoCA in PD patients undergoing evaluation for deep brain stimulation (DBS) with a large sample of cognitively normal research volunteers. Although MoCA was used as an exclusion criterion to characterize normal cognitive performance in research volunteers, group differences on MoCA are also described to contrast the cognitive profiles of both group. PD was selected for initial clinical ABBA characterization due to its association with an increased risk for executive function inefficiencies (Dirnberger & Jahanshahi, 2013; Kudlicka, Clare, & Hindle, 2011). Executive function difficulty in PD includes impaired response inhibition as reflected by the oculomotor antisaccade (Waldthaler et al., 2021) and go/no-go tasks (Georgiev, Dirnberger, Wilkinson, Limousin, & Jahanshahi, 2016). Although not designed specifically to investigate the cognitive profile of PD, we explored ABBAs clinical applicability using performance thresholds derived from the EHBS cohort applied to PD patients, predicting a higher frequency of ABBA alternation errors in PD patients compared to EHBS participants given PD’s increased rate of executive function/response inhibition inefficiency.

Methods

Participants

Research volunteers included 899 community-residing individuals (313 men, 34.8%; 586 women, 65.2%) who were enrolled in the Emory Healthy Brain Study (EHBS; (Goetz et al., 2019). All participants were self-characterized as cognitively normal, living independently in the community, and had MoCA scores ≥ 24/30. Participants with prior diagnoses of MCI or dementia of any type were excluded from EHBS enrollment. EHBS participant age ranged from 50-81 years (M= 65.3, SD=6.7); education ranged from 12-22 years (M=16.9 (SD=2.0).There were 779 White and 120 Black participants. This project was approved by the Emory University Institutional Review Board in accordance with the Declaration of Helsinki, and all participants provided written informed consent.

The clinical cohort consisted of 32 PD patients undergoing evaluation for DBS who were administered face-to-face neuropsychological evaluations in 2022-2023. PD age ranged from 42-72 years (M=62.1, SD=10.2); education ranged from 11-18 years (M=15.5, SD=2.2). There were 28 White (87.5%) and 4 (12.9%) Black PD participants consisting of 29 (90.6%) men and 3 (9.4%) women. One patient received a diagnosis of major neurocognitive disorder, 14 were diagnosed with minor neurocognitive disorder, and while the remaining patients exhibited cognitive inefficiencies likely attributable to PD, these were not severe or consistent enough to warrant a neurobehavioral diagnosis. There were no disease-related symptoms (e.g., dysarthria, hypophonia) of sufficient severity to potentially impact performance.

Procedures

ABBA Letter Alternation Test.

ABBA includes two trials using the same stimulus set of 25 capital letter As and Bs presented using PowerPoint. Stimuli are black letters (Ariel 115 font; bold) presented against a light grey background, chosen to reduce eyestrain. Letters are displayed for 1.5 seconds, with a jittered interstimulus interval averaging 3.2 seconds (R=2.6-6.5 seconds). Most EHBS participants (n=919) completed ABBA testing and MoCA testing via telehealth and then were administered TMT B during their subsequent in-person visit when they underwent lumbar puncture (LP), which occurred on average 40.6 days later (SD=35.1). Telehealth assessments were conducted at the participants’ home on computer laptop or desktop; no assessments were completed using smartphones. In-person cognitive testing was conducted before obtaining the lumbar puncture.

The first trial (Letter Reading) requires participants to read each letter aloud as quickly as possible. Letter Reading is included not only to familiarize participants with the task, but also to reinforce the prepotent response set of reading/letter identification. Letter Alternation presents the same stimulus series but with instructions to state the opposite letter (e.g., say “A” when “B” is present and vice versa). “Opposite” appears at the top of each slide to minimize the likelihood of set loss error as a contributing factor to poor performance. In both conditions, the instructions are displayed on the screen and read aloud to the participants, who are asked to follow along. ABBA performance is characterized by Letter Alternation error count, including omission and commission errors (self-corrected errors are counted as errors).

Other Cognitive Measures.

Level of general cognitive function was assessed using the MoCA (Nasreddine et al., 2012). Details of our telehealth MoCA protocol are provided in our telehealth MoCA equivalence study (Loring, Lah, & Goldstein, 2023). MoCA scores included the single education correction point were appropriate. All participants were administered Trail Making Parts A and B (Army, 1944). TMT A is a measure of visuomotor speed requiring participants to connect numbers distributed in a spatial array as quickly as possible. TMT B is similar but contains numbers and letters, with the task to alternate between numbers and letters in ascending order. Prior testing using the same MoCA and Trails test versions was performed in 532 (59%) of the EHBS participants.

Analysis Plan.

Because most patients had 0-1 Letter Alternation errors reflecting an asymmetric and non-continuous performance distribution in both participant groups, we employed a staged approach to data analysis. ABBA errors represent count data, precluding normalization through data transformation. Consequently, we employed multiple approaches for data analysis including differences in ABBA Letter Alternation frequency reported using χ²; parametric/non-parametric correlations to characterize initial demographic performance influences on Letter Alternation errors, MoCA, and TMT B; ANOVAs to evaluate demographic MoCA, and TMT B group effects; and negative binomial regression to incorporate demographic influences on the relationship between ABBA Letter Alternation errors and MoCA and TMT B.

Results

EHBS Participants.

ABBA Letter Alternation error frequencies are presented in Table 1 and exhibit a highly asymmetric distribution. Most EHBS participants made either no ABBA Letter Alternation errors (n=630, 70.1%) or a single ABBA Letter Alternation error (n=145, 16.1%). There were 58 EHBS participants (6.5%) making 2 errors, and 66 (7.3%) made 3+ errors. Age was correlated with ABBA alternation error frequency (rho=.10, p=.002); the correlation of education with ABBA alternation errors was non-significant (rho=.02, p=.56). ABBA error frequency did not differ by sex (χ²=9.1, p=.61) or race (Black vs. White, χ²=9.8, p=.54).

Table 1.

ABBA Letter Alternation error frequencies for EHBS participants and PD patients

Errors	EHBS n	EHBS Percent	Cumulative	PD n	PD Percent	Cumulative
0	630	70.1	70.1	15	46.9	46.9
1	145	16.1	86.2	7	21.9	68.8
2	58	6.5	92.7	3	9.4	78.1
3	23	2.6	95.2	3	9.4	87.5
4	14	1.6	96.8	0	0	87.5
5	8	0.9	97.7	1	3.1	90.6
6	6	0.7	98.3	0	0	90.6
7	9	1.0	99.3	0	0	90.6
8	1	0.1	99.4	1	3.1	93.8
9	2	0.2	99.7	1	3.1	96.9
11	2	0.2	99.9	1	3.1	100.0
16	1	0.1	100.0	0	0	100.0
Total	899	100.0	100.0	32	100.0	100.0

Note: EHBS: Emory Healthy Brain Study, PD: Parkinson Disease

MoCA was significantly correlated with age (r=−.08, p=.017) and education (r=.14, p<.00001). Women obtained slightly higher MoCA scores than men (27.6, SD=1.8 vs. 27.0, SD=1.7, p<.00001) but with a small effect size (η²=.026). White participants obtained higher scores on the MoCA than Black participants (27.6, SD=1.7 vs. 26.5, SD=1.5, p<.00001, η²=.047).

Age and education were both statistically correlated with TMT B (age r=.30, p<.0001; education r=−.09, p=.009). There was no TMT B sex difference (women=65.2, SD=25.7; men=68.5, SD=28.8, p=.079, η² =.003). White participants had faster TMT B completion time than Black participants (64.3 seconds, SD=24.5 vs. 79.6 seconds, SD=36.4; p<.00001, η²=.038).

PD Subjects.

ABBA alternation error frequencies for PD patients are presented in Table 1, displaying a highly asymmetric distribution. Most PD patients (22, 68.8%) made 0-1 errors, 3 PD (9.4%) patients made 2 errors, and 7 PD subjects (21.8%) made 3+ errors. The PD patient who made 11 ABBA alternation errors was the patient with a major neurocognitive disorder diagnosis. Sex and race were not contrasted due to small cell numbers. There was no correlation between ABBA Letter Alternation with either age (rho=−.01, p=.98) or education (rho=−.17, p=.36).

Group Contrasts.

The average EHBS MoCA score of 27.4 (SD=1.8) was significantly higher than the PD MoCA average of 24.5 (SD=3.4; p<.00001; η²=.076) reflecting a medium effect size. PD MoCA scores ranged from 13-30 (19-30 when excluding major neurocognitive disorder patient). Significant group differences for TMT B were present (EHBS: 66.3 sec (SD=26.8); PD: 140.2 sec (SD=90.5; p<.00001) representing a large effect (η²=.158).

Because of overdispersion associated with the high frequency of zero ABBA Letter Alternation errors, we fitted both negative binomial and zero-inflated negative binomial regression models to control demographics when evaluating group differences characterizing the relationship of ABBA Letter Alternation to TMT B and MoCA. Because there is little change in the parameter estimates, we chose to use negative binomial regression due to its greater simplicity. Demographic factors included age (10-year increments), race (Black, White), sex at birth (Male, Female), and education (16 years or less, 17-18 years, doctorate; Figure 1). When developing our model, we excluded the single EHBS subject who had 16 errors, treating this score as a performance outlier. This score exceeds the highest number of errors in the PD group by 5 errors, suggesting that a factor other than ability was contributing to this score. Otherwise, cognitive difficulty of this magnitude would be reflected in TMT B and the MoCA, both of which were normal. Exclusion of this subject resulted in improved model fit and little change in coefficients.

Figure 1.

Median regression lines including demographics illustrating relationship between ABBA letter alternation errors with Trail Making Test Part B and with MoCA summary score

Note: TMT B: Trail Making Test Part B, MoCA: Montreal Cognitive Assessment, EHBS: Emory Healthy Brain Study, PD: Parkinson Disease

Negative binomial regression results confirmed the relationship between ABBA letter alternation errors with MoCA and TMT B when controlling for demographics (see Figure). The exponential incident rate estimate for MoCA demonstrated that for every one point increase in MoCA score, the incident rate of ABBA errors decreased by 11% controlling for other variables (p=0.0027). For TMT B, every additional 10-second TMT B increment, there was a 9% increase in the incident rate of ABBA errors (p<0.0001).

Group Classification Differences.

Groups differed in the frequency of error-free ABBA Letter Alternation responding (χ²=7.8, p=.005), with fewer PD patients having error-free ABBA alternation performance compared to EHBS study participants. However, for clinical characterization based error frequency, we chose to form 3 groups. Normal performance is inferred with 0-1 ABBA alternation errors, a pattern displayed by 86.4% of the EHBS normative group, a level closely approximating the −1.5 standard deviation threshold that is often used as a cutpoint to differentiate “normal” from “abnormal” performance in normal parametric distributions. Two errors (6.5% of the EHBS sample) are considered borderline, whereas 3+ errors (7.1%) are considered to be impaired. 802 (86.2%) EHBS participants made 0-1 alternation error, 58 (6.5%) EHBS participants had 2 ABBA alternation errors, and 66 (7.3%) made 3+ errors. For PD patients, 22 (68.8%) made 0-1 errors 3 PD (9.4%) patients made 2 alternation errors, and 7 PD subjects (21.8%) made 3+ errors, which significantly differed in frequency from the EHBS group (p=.0073).

Discussion

This report demonstrates that ABBA Letter Alternation error frequency is increased in PD patients compared to cognitively healthy research volunteers providing initial evidence of its clinical utility for telehealth assessment of response inhibition/response alternation. While further clinical validation is needed, its usefulness is supported by the large group difference with TMT B, although larger studies across a variety of cognitive measures will be needed to establish both convergent and discriminative support for task specificity. While different interpretative performance characterizations exist (Guilmette et al., 2020), we suggest that 0-1 errors on ABBA alternation be considered normal, 2 errors as borderline, and 3 or more errors as impaired. Borderline (2 ABBA Letter Alternation errors, 13.6^th percentile) and impaired cutpoints (3+ ABBA alternation errors, 6.5^th percentile). This roughly correspond to thresholds commonly used to identify atypical performance levels (i.e., −1.5 SD, −2 SD) and is appropriate where the goal of assessment is to identify basic ability impairment rather than testing aptitude (Calamia, Markon, Denburg, & Tranel, 2011). We draw a parallel to the interpretation of abnormal antisaccades and suggest that ABBA errors are better understood as soft neurological signs rather than psychometric performance metrics. In this context, distinctions of normal, borderline, vs. impaired are used to guide clinical decision-making. Given the small but statistically significant age effect on ABBA Letter Alternation error frequency, recommended cutpoints should be considered as guidelines, allowing for appropriate adjustment in clinical inference at the older age levels based upon clinical judgement and experience.

Although ABBA Letter Alternation errors obtained during are used to characterize impairment, ABBA Letter Reading serves an important role by reinforcing the prepotent reading response before assessing alternation/response inhibition. After the reading condition is completed, participants are told explicitly that the same series of letters will again be presented, but this time with different task demands. Being told that they will be seeing the same stimulus series provides reassurance that there will be no unanticipated pacing changes that might increase performance anxiety.

Although this study this is not a null hypothesis significance testing (NHST) experiment designed as a primary investigation of inhibitory control in PD, we considered it important to demonstrate the magnitude of group differences independent of ABBA Letter Alternation performance. After establishing demographic influences on both MoCA and TMT B in the EHBS cohort, we performed negative binomial regression to incorporate demographic influences on the relationship between ABBA Letter Alternation errors and MoCA and TMT B between groups. PD patients were included to characterize performance as initial support of ABBA Letter Alternation clinical sensitivity. However, future studies characterizing a relationship between PD disease factors such as those reflected by instruments such as the Unified Parkinson’s Disease Rating Scale (UPDRS) will be helpful in establishing the clinical meaningfulness of poor ABBA performance.

The large effect size TMT B group difference provides initial evidence of ABBA Letter Alternation’s clinical utility given their established risks of executive dysfunction and inhibitory difficulty. However, future studies designed explicitly to study PD inhibitory control will need to characterize PD subgroups more carefully since cognitive phenotypes likely will influence ABBA performance patterns (Barvas et al., 2021; Kudlicka, Clare, & Hindle, 2013).

Despite the large group TMT B difference, we do not suggest strict equivalence of ABBA Letter Alternation errors with TMT B completion time. While sequencing errors on TMT B and time to completion are both discriminative in Alzheimer’s disease identification, scores are often in disagreement when examined on the individual level (Ashendorf et al., 2008). Nevertheless, these data suggest that ABBA Letter Alternation performance provides an additional window into response alternation/response inhibition aspects of executive function that otherwise could not be easily assessed via telehealth, and should provide additional behavioral data for assessment in clinical case formulation.

Although ABBA was developed due to difficulty adapting TMT B to a telehealth environment, ABBA shares important features with the Stroop Color and Word test (Golden, 1978; Scarpina & Tagini, 2017)} by assessing inhibition of a pre-potent response. ABBA was chosen rather than the Stroop because of ABBA’s demonstrated clinical sensitivity in our prior studies using the paper and pencil version, and because the Stroop is not as adaptable to the telehealth environment (including smartphone administration) given concerns regarding the influence of display heterogeneity on stimulus presentation (e.g., color hue and size vs. simple black block letters). Future studies comparing traditional Stroop testing to ABBA will provide a valuable performance contrast.

There are several important limitations associated with the report. As with all home telehealth testing, there is potential for distractions to impact task performance. However, evidence indicates that MoCA scores are comparable between in-person and telehealth administration (Loring, Lah, & Goldstein, 2023), suggesting that any such effect is likely negligible. Additionally, MoCA responses are verbal rather than reliant on response speed making variability in home computer displays and hardware unlikely to serve as significant confounds. Although differences in operating systems were not evaluated, all assessments in this study were conducted using computers rather than smartphones, which could be explored in future research. The EHBS sample included more women than men, whereas the PD sample was predominantly male, reflecting sex differences in PD epidemiology. Both groups were highly educated, with the EHBS participants averaging 16.9 years of education—a common bias in clinical research recruitment—and the PD group averaging 15.5 years. Consequently, generalization of these findings to individuals with fewer years of education should be made with appropriate caution. Since lower education is often associated with reduced cognitive reserve, however, it is possible that ABBA could be more sensitive to detecting inhibitory control issues in participants with more limited educational backgrounds. Similar caution should be employed when interpreting results from other demographic backgrounds. Although no performance validity measures (PVTs) were included, DBS candidates in our experience are highly motivated since neuropsychological findings are critical in determining their candidacy for the procedure. Moreover, we have no reason to expect that research volunteers would be motivated to exaggerate their performance. Even if a small number of participants are not fully motivated, this is mitigated by the large sample size of nearly 900 participants.

One advantage of ABBA is that individuals with limited education can perform the task since it makes few cognitive demands other than simple letter reading, and it is well tolerated by the participants. Thus, there may be less of an education performance influence (Adólfsdóttir, Wollschlaeger, Wehling, & Lundervold, 2017). Another important benefit of ABBA is its time efficiency, taking less than 5 minutes for task administration. In addition, task demands are not frustrating to participants, although they often comment with a smile about unusual task demands when the Letter Alternation instructions are given. ABBA is also adaptable for patients whose native language is not English, providing the opportunity for broad international use and clinical collaboration. Since ABBA was developed as part of a federally funded research project, the task is in the public domain and is freely available without test purchase fee or per-use charges (ABBA PPT download; ABBA Scoring Form). Development of a variety of tasks to measure cognitive constructs has been identified as an important future direction for the field (Bilder & Reise, 2019). We hope that ABBA can contribute to meeting that need.

Table 2.

Negative binomial regression reflecting the association between ABBA errors and MoCA in EHBS volunteers and PD patients adjusting for demographics

Parameter	Estimate	Exponential* Estimate	p value
Intercept	1.56 ± 1.25	4.74	0.21
MoCA Summary	−0.12 ± 0.04	0.89	0.0027
EHBS vs PD	−0.73 ± 0.37	0.48	0.0470
Age, 10 years	0.24 ± 0.10	1.28	0.0177
African American vs Caucasian	0.10 ± 0.20	1.10	0.63
Female vs male	0.25 ± 0.15	1.29	0.09
Master’s vs bachelor’s degree or below	0.19 ± 0.16	1.20	0.24
Doctorate vs bachelor’s degree or below	0.17 ± 0.20	1.18	0.40

Note: TMT B: Trail Making Test Part B, MoCA: Montreal Cognitive Assessment, EHBS: Emory Healthy Brain Study, PD: Parkinson Disease

^*Exponential estimate is the incident rate. For every one unit increase in MoCA summary score, the incident rate of ABBA errors decreases by 11% controlling for other variables and this decrease is significant (p=0.0027).

Table 3.

Negative binomial regression reflecting the association between ABBA errors and Trail Making Test Part B in EHBS volunteers and PD patients adjusting for demographics

Parameter	Estimate	Exponential Estimate*	p value
Intercept	−2.25 ± 0.74	0.11	0.0023
TMT B, 10 sec.	0.09 ± 0.02	1.09*	<.0001
EHBS vs PD	−0.23 ± 0.41	0.79	0.56
Age, 10 years	0.18 ± 0.11	1.19	0.10
African American vs Caucasian	0.07 ± 0.20	1.07	0.73
Female vs male	0.21 ± 0.15	1.24	0.15
Master’s vs bachelor’s degree or below	0.25 ± 0.16	1.28	0.11
Doctorate vs bachelor’s degree or below	0.19 ± 0.19	1.21	0.32

Note: TMT B: Trail Making Test Part B, MoCA: Montreal Cognitive Assessment, EHBS: Emory Healthy Brain Study, PD: Parkinson Disease

^*Exponential estimate is the incident rate. For every additional 10-second spent in trail B, there is 9% increase in the incident rate of ABBA errors and this increase is significant (p<0.0001) holding the variables constant. The incident rate of ABBA errors for EHBS participants is 0.79 times the incident rate for PD patients controlling for demographics and TMT B (p=0.56).