Here, we present a protocol to show how to perform two types of cognitive assessment tools derived from the paper-pencil version of the Trail Making Test.
The Trail Making Test (TMT) is a well-accepted tool for evaluating executive function. The standard TMT was invented more than 60 years ago and has been modified into many versions. With the development of digital technologies, TMT is now modified to a digitized version. The present study demonstrated digital TMT (dTMT) performed on a computer, and Walking TMT (WTMT) on the floor. Both revealed more information compared with the traditional version of TMT.
With a rapidly aging population, dementia is considered to be a major public health concern. The number of elderly patients with dementia worldwide is about 47 million according to the World Health Organization1. Executive function impairment is not only a common type of cognitive dysfunction in aged individuals, but has been reported as a predictor of progression from mild cognitive impairment (MCI) to clinical Alzheimer’s disease (AD)2,3. As the third most widely used test in neuropsychology4, the Trail Making Test (TMT) is employed as a well-accepted tool to evaluate executive functions, especially sustained attention and set-shifting5, even in elderly patients6.
The standard TMT is a paper-pencil test consisting of two parts: tMT-A and TMT-B5. The former calls for the test-taker to draw lines connecting randomly distributed numbers (1–25) on a test paper in ascending order (1->2->3…), whereas the latter requires the test-taker to set numbers and letters (1->A->2->B…) alternatively. The performance of TMT is generally scored in the time taken to complete each part correctly7. TMT has been translated into different languages. The Chinese version of TMT was developed in 20068. Since Chinese characters are quite distinct from English letters, the Chinese version of TMT was used in our procedure.
Apart from the standard version, TMT has been modified in different ways by researchers (e.g., oral TMT9, driving TMT10, walking TMT (WTMT)11) to assess specific populations or find details under different conditions, such as driving and walking. Of note, some studies conferring different numbers compared with the standard TMT are also reported to be of high validity and reliability. For example, THINC-Integrated Tool (THINC-it) developed by the McIntyre group used 9 numbers and letters for TMT-B12; WTMT reported by Schott and colleagues used 15 numbers for TMT-A13. In the same way, many evaluating systems of TMT have been built beyond the complete time scoring, which are reported to be helpful in finding more items besides executive dysfunction, or to be accessible for participants who are not suitable to complete the standard TMT. For example, some researchers investigated the errors in TMT and found that errors in TMT-B were associated with mental tracking and working memory in patients with psychiatric disorder14. Another group from Greece suggested derived TMT scores [TMT-(B−A) or TMT(B/A)] as indices to detect impairment in cognitive flexibility across the adult life span15. Generally, alternative evaluating systems of TMT can be summarized as follows: (1) completion time analysis—TMT completion time is calculated in seconds16; (2) error analysis—different types of TMT errors are classified and quantified14; (3) intermanual differences—different abilities of completing TMT between the dominant hand and the nondominant hand are compared17; and (4) derived Trail Making Test indices—different characterizations between completing TMT-A and TMT-B are analyzed15. The alternative scoring methods provide additional information. For example, the utility of TMT error analysis could reveal cognitive deficits not traditionally captured using completion time as the sole outcome variable in patients with schizophrenia and depression14. The lack of any significant intermanual difference helped to discriminate the cognitive dysfunction from the influence of the motor disorder17. Derived TMT indices could detect impairment in cognitive flexibility across the adult life span and minimize the effect of demographics and other cognitive background variables15.
With advances in modern technology, computer-based digital applications have been increasingly integrated into traditional cognitive interventions, most of which are designed as similar to the original test as possible, rather than created as new tools. Digital or computerized TMT (dTMT) has been proven to have the potential to capture additional information, with the structure of the existing test mainly unchanged in recent years18,19.
This study aimed to introduce a computer-based Chinese version of dTMT-A and dTMT-B, as well as a WTMT. Both are modified TMTs and have been confirmed to have high sensitivity and specificity to screen patients with MCI, Parkinson’s Disease, Alzheimer’s Disease, and so forth, based on the movement of upper and lower limbs20,21. Detailed scoring methods were also presented because digital technologies incorporated into dTMT and WTMT might help capture more information compared to the paper-pencil version of TMT.
The development of the dTMT and initial application was approved by The Seventh Medical Center of PLA Army General Hospital Review Board. Subjects signed approved informed consent documents prior to testing TMT.
1. General Method Development
2. Design and Testing of the dTMT
NOTE: As mentioned earlier, dTMT has two parts: dTMT-A and dTMT-B. These two tests should be performed sequentially (dTMT-A proceeding dTMT-B), without being reversed.
3. Direct Data Collection and Definitions in dTMT
4. Design and Testing of the WTMT
NOTE: Similar to dTMT, WTMT also has two parts: WTMT-A and WTMT-B. These two tests should be performed sequentially (WTMT-A proceding WTMT-B), without being reversed.
5. Direct Data Collection and Meaning Explanation in WTMT
NOTE: As shown in Figure 5, the human gait cycle has been divided into different subphases. In detail, spatial and temporal parameters are defined and calculated as follows.
6. Data Collection and Statistics
Seven aged patients with Mild Cognitive Impairment (Elderly with MCI), seven aged subjects with Parkinson’s Disease (Elderly with PD), and seven aged healthy individuals (Healthy Elderly) were recruited, and dTMT-A, dTMT-B, WTMT-A, and WTMT-B, were performed. After the tests, data were collected and analyzed using SPSS software.
As a whole, the demographical data of participants showed that all groups were matched well in terms of age, gender, educational level, dominant hand, Clinical Dementia Rating (CDR) score, Global Deterioration Scale (GDS) score, TUG: timed Up and Go Test (TUG), and so forth (p > 0.05).
As shown in Table 2, most of the data of dTMT-A between Healthy Elderly, Elderly with MCI, and Elderly with PD were similar, such as Total time to completion (18.15 ± 5.12 s vs. 19.67 ± 7.12 s vs. 19.85 ± 3.89, P = 0.812), Number of Errors (0.14 ± 0.38 vs. 0.29 ± 0.49 vs. 0.29 ± 0.49, P = 0.796), and so forth. This means all the participants had similar scores if they are assessed by traditional TMT-A. However, there existed some different variables captured by dTMT-A. As shown in Table 2, Elderly with PD exhibited a larger total pathway deviation of each step (Pb = 0.017, Pc = 0.048), a larger variability of pathway deviation (Pb = 0.000, Pc = 0.000), and a lower velocity of drawing of each step (Pb = 0.001, Pc = 0.025) compared with Elderly with MCI and Healthy Elderly, respectively.
As shown in Table 3, the differences in completing dTMT-B were reflected in more aspects relative to dTMT-A. Aged patients with MCI needed a longer time of completion (P = 0.000) and had more errors (P = 0.000), more time inside the circle (P = 0.000) or tolerance circle (P = 0.000), more pathway deviation (P = 0.035), and lower velocity in drawing (P = 0.000) compared with healthy elderly. Meanwhile, Elderly with PD needed a longer time of completion (P = 0.000), and had more errors (P = 0.000), more time inside the circle (0.000) but less time inside the tolerance circle (P = 0.000), more pathway deviation (P = 0.032), larger variability of pathway deviation (P = 0.001), and obviously lower velocity of drawing of each step (P = 0.000) compared with aged healthy individuals. All the results indicated that dTMT can detected amount of significant differences between aged healthy participants and aged patients.
As shown in Table 4, gait data in WTMT-A could detect more differences between Elderly with PD in comparison with other individuals, especially in terms of speed (Pb = 0.000, Pc = 0.002), step length (Pb = 0.004, Pc = 0.016), stride length (Pb = 0.005, Pc = 0.019), and so forth. All these data implied that WTMT-A could capture obvious differences between aged PD patients and aged healthy participants.
As shown in Table 5, gait data in WTMT-B could find more differences between groups. Aged patients with MCI and PD needed a longer time (Pa = 0.001, Pb = 0.000) and more steps to complete the test (Pa = 0.000, Pb = 0.000). Their step and stride length seemed shorter relative to aged healthy participants. In addition, aged patients with PD showed even more severe trend in comparison with MCI subjects. The marked differences are step length (0.045 m ± 0.02 vs. 0.049 m ± 0.02, Pc = 0 .002), stride length (0.91 m ± 0.04 vs. 0.96 m ± 0.03, Pc = 0.012), and Gait variability of step length (0.112 ± 0.0030 vs. 0.120 ± 0.0034, Pc = 0.000).
Figure 1: Computer. Computer for dTMT-A and dTMT-B (upper panel), print screen of dTMT, subjects choose Part A to start dTMT-A, or Part B to start dTMT-B (lower panel). Please click here to view a larger version of this figure.
Figure 2: IDEEA. Device for WTMT-A and WTMT-B. Please click here to view a larger version of this figure.
Figure 3: Example of WTMT-A and WTMT-B. As shown in the figure, subjects need to begin from START and walk to the END. Please click here to view a larger version of this figure.
Figure 4: IDEEA accelerometers and the location. The figure showed how to wear the IDEEA accelerometers correctly. Please click here to view a larger version of this figure.
Figure 5: Human gait cycle divided into different subphases. Stand phase was about 60% of gait cycle, and Swing phase was about 40% of gait cycle. Please click here to view a larger version of this figure.
Healthy Elderly | Elderly with MCI | Elderly with PD | P Value | |
N = 7 | N = 7 | N = 7 | ||
Age | 67.14 ± 4.22 | 65.14 ± 3.39 | 66.29 ± 3.90 | 0.63 |
Gender(M:F) | 4:03 | 5:02 | 4:03 | 0.589 |
Dominant hand(R%) | 100 | 100 | 100 | |
Education (yrs) | 10.00 ± 1.91 | 11.43 ± 2.51 | 10.14 ± 1.36 | 0.353 |
MMSE | 29.00 ± 1.15 | 27.86 ± 1.35 | 28.43 ± 1.27 | 0.263 |
CDR | 0.14 ± 0.24 | 0.5 ± 0.00 | 0.29 ± 0.39 | 0.066 |
GDS | 2.28 ± 0.49 | 2.71 ± 0.76 | 2.29 ± 0.75 | 0.487 |
TUG (S) | 10.07 ± 1.51 | 11.02 ± 0.60 | 11.72 ± 1.24 | 0.052 |
Table 1: Demographic data of participants. Mean ± SD. M:F = Male: Female; R% = Right hand percentage; yrs = years; MMSE = Mini Mental State Examination.; MCI = Mild Cognitive Impairment; PD = Parkinson’s Disease; CDR = Clinical Dementia Rating; GDS = Global Deterioration Scale; TUG = timed Up and Go Test; S = Seconds
Healthy Elderly | Elderly with MCI | Elderly with PD | P Value | |
N = 7 | N = 7 | N = 7 | ||
Total time to Completion | 18.15 ± 5.12 | 19.67 ± 7.12 | 19.85 ± 3.89 | 0.821 |
Number of Errors | 0.14 ± 0.38 | 0.29 ± 0.49 | 0.29 ± 0.49 | 0.796 |
Total time inside each circle | 6.94 ± 1.99 | 6.91 ± 3.31 | 7.81 ± 2.46 | 0.773 |
Inside circle percentage | 39.13 ± 7.70 | 35.42 ± 10.25 | 40.02 ± 11.63 | 0.665 |
Total time inside each tolerance circle | 1.57 ± 0.80 | 2.09 ± 0.88 | 1.85 ± 0.49 | 0.442 |
Inside tolerance circle percentage | 8.74 ± 3.02 | 10.80 ± 3.07 | 9.61 ± 3.55 | 0.498 |
Total Line cancelling times | 0.14 ± 0.38 | 0.29 ± 0.49 | 0.14 ± 0.38 | 0.764 |
Total pathway deviation of each step | 38.41 ± 2.52 | 39.30 ± 3.07 | 42.99 ± 3.99b, c | 0.039 |
Variability of pathway deviation | 1.72 ± 0.24 | 2.36 ± 0.55a | 3.66 ± 0.46b, c | 0 |
Velocity of drawing of each step | 21.38 ± 2.59 | 19.00 ± 2.40 | 15.70 ± 2.55b, c | 0.002 |
Table 2: dTMT-A data of participants. Mean ± SD. MCI = Mild Cognitive Impairment; PD = Parkinson’s Disease. One-way-ANOVA and post hoc analysis with LSD. a = P < 0.05 Elderly with MCI relative to Healthy Elderly; b = P < 0.05 Elderly with PD relative to Healthy Elderly; c = P < 0.05 Elderly with PD relative to Elderly with MCI.
Healthy Elderly | Elderly with MCI | Elderly with PD | P Value | |
N = 7 | N = 7 | N = 7 | ||
Total time to Completion | 32.07 ± 10.93 | 67.56 ± 9.87a | 89.95 ± 12.12b,c | 0 |
Number of Errors | 0.14 ± 0.38 | 2.86 ± 1.07a | 1.29 ± 0.49b,c | 0 |
Total time inside each circle | 6.03 ± 1.72 | 27.83 ± 5.05a | 7.81 ± 2.46b,c | 0 |
Inside circle percentage(%) | 19.16 ± 3.86 | 41.47 ± 6.76a | 22.46 ± 3.35c | 0 |
Total time inside each tolerance circle | 3.51 ± 0.91 | 9.73 ± 1.46a | 3.93 ± 2.21c | 0 |
Inside tolerance circle percentage(%) | 11.26 ± 2.20 | 14.47 ± 1.62a | 4.57 ± 2.86b,c | 0 |
Total Line cancelling times | 0.29 ± 0.38 | 0.86 ± 1.07 | 0.43 ± 0.53 | 0.35 |
Total pathway deviation of each step | 86.02 ± 7.36 | 95.36 ± 6.76a | 95.56 ± 8.78b | 0.051 |
Variability of pathway deviation | 2.158 ± 0.173 | 2.024 ± 0125 | 2.659 ± 0.332b,c | 0 |
Velocity of drawing of each step | 16.85 ± 1.79 | 8.41 ± 1.09a | 4.91 ± 0.91b, c | 0 |
Table 3: dTMT-B data of participants. Mean ± SD. MCI = Mild Cognitive Impairment; PD = Parkinson’s Disease. One-way-ANOVA and post hoc analysis with LSD. a = P < 0.05 Elderly with MCI relative to Healthy Elderly; b = P < 0.05 Elderly with PD relative to Healthy Elderly; c = P < 0.05 Elderly with PD relative to Elderly with MCI.
Healthy Elderly | Elderly with MCI | Elderly with PD | P Value | |
N = 7 | N = 7 | N = 7 | ||
Total time to Completion | 68.43 ± 4.86 | 76.57 ± 7.66 | 98.29 ± 9.36b,c | 0 |
Number of Errors | 0.29 ± 0.49 | 0.29 ± 0.49 | 0.57 ± 0.53 | 0.487 |
Steps (n) | 80.86 ± 2.34 | 81.29 ± 3.30 | 81.71 ± 3.90 | 0.886 |
Swing duration (%) | 36.86 ± 1.32 | 35.03 ± 0.84a | 35.48 ± 1.25b | 0.022 |
Step duration (%) | 63.00 ± 1.35 | 64.97 ± 0.84 a | 64.52 ± 1.25b | 0.014 |
Speed (m/s) | 1.01 ± 0.10 | 0.82 ± 0.57a | 0.68 ± 0.04b,c | 0 |
Step length (m) | 0.51 ± 0.02 | 0.50 ± 0.01 | 0.49 ± 0.02b,c | 0.01 |
Stride length (m) | 1.02 ± 0.04 | 1.00 ± 0.02 | 0.96 ± 0.04b,c | 0.011 |
Gait variability of step length | 0.111 ± 0.0011 | 0.112 ± 0.0011 | 0.113 ± 0.0014 | 0.156 |
Table 4: WTMT-A data of participants. Mean ± SD. MCI = Mild Cognitive Impairment; PD = Parkinson’s Disease. One-way-ANOVA and post hoc analysis with LSD. a = P < 0.05 Elderly with MCI relative to Healthy Elderly; b = P < 0.05 Elderly with PD relative to Healthy Elderly; c = P < 0.05 Elderly with PD relative to Elderly with MCI.
Healthy Elderly | Elderly with MCI | Elderly with PD | P Value | |
N = 7 | N = 7 | N = 7 | ||
Total time to Completion | 78.57 ± 4.86 | 92.29 ± 7.72a | 109.00 ± 5.66b,c | 0 |
Number of Errors | 0.57 ± 0.79 | 1.14 ± 1.07 | 0.86 ± 0.69 | 0.479 |
Steps (n) | 89.71 ± 2.63 | 96.71 ± 2.29a | 100.57 ± 3.74b,c | 0 |
Swing duration (%) | 37.20 ± 1.21 | 36.56 ± 1.23 | 36.47 ± 1.15 | 0.476 |
Step duration (%) | 62.80 ± 1.21 | 63.44 ± 1.23 | 63.53 ± 1.15 | 0.476 |
Speed (m/s) | 0.98 ± 0.06 | 0.83 ± 0.08a | 0.73 ± 0.03b,c | 0 |
Step length (m) | 0.51 ± 0.02 | 0.49 ± 0.02 | 0.45 ± 0.02b,c | 0 |
Stride length (m) | 1.01 ± 0.04 | 0.96 ± 0.03a | 0.91 ± 0.04b,c | 0 |
Gait variability of step length | 0.114 ± 0.0033 | 0.120 ± 0.0034a | 0.112 ± 0.0030c | 0.001 |
Table 5: WTMT-B data of participants. Mean ± SD. MCI = Mild Cognitive Impairment; PD = Parkinson’s Disease. One-way-ANOVA and post hoc analysis with LSD. a = P < 0.05 Elderly with MCI relative to Healthy Elderly; b = P < 0.05 Elderly with PD relative to Healthy Elderly; c = P < 0.05 Elderly with PD relative to Elderly with MCI.
Traditional paper-pencil TMT has been well used worldwide for more than 50 years. However, digital TMT is advantageous. First, traditional TMT is considered as an executive function tool, while both dTMT and WTMT have aspects reflecting motor ability besides cognitive function. Considering that the cognitive-motor dual task has gained great attention in recent years26, digital technologies can provide researchers with more information on this integrated task compared with the traditional TMT27. Second, digital TMT is a sensitive tool compared with the traditional version. Digital TMT does not need additional time relative to traditional ones, which has enough compliance of subjects.
A critical step in the protocol is to perform dTMT and WTMT with no interruption, because both tests collected time variables. Subjects need to complete the tests fluently. Any delay induced by examiners, or misunderstanding, distraction, etc., should be minimized or eliminated.
There are two modifications to be mentioned. First, for dTMT, the real-time pressure of the stylus onto the screen is a sensitive variable for drawing, which has been confirmed in a digital Clock Drawing Test28. With more development, software that could detect the stylus pressure onto the screen during dTMT will give physicians more information in future. Second, for WTMT, a new device that can detect and analyze trunk sway might be helpful to find more evidence in movement disorder patients29,30, because IDEEA only provides gait data. However, as far as we know, IDEEA is the first digital accelerometry used in WTMT.
The current study introduced two types of TMTs in a digitized version. These new types of TMTs were derived, rather than being an exact copy of the traditional TMT. Robert P. Fellows found that the computerized TMT needed fewer circles compared to the traditional TMT, in case the circles were too crowded31. However, this difference cannot impede the wide use of the digital TMT in the future.
Since digital technology is becoming more and more popular in our daily life, digital devices should be used in early diagnosis of cognitive disorders and movement disorders32. dTMT and WTMT are both derived from traditional TMT but can capture more variables than the paper-based TMT. Both new modified TMTs could be used to screen patients with cognitive disorders and movement disorders. Particularly for those patients with upper limb disability, WTMT is particularly useful.
A limitation of the present study was its small sample size. Consequently, the sensitivity and specificity of digital TMT could be demonstrated. However, dTMT and WTMT could find additional information for the physicians to determine the cognitive function and motor ability of the participants. However, more studies are needed to validate the findings.
The authors have nothing to disclose.
The authors thank Xiaode Chen for digital technology support.
Minisun | LLC | Intelligent Device for Energy Expenditure and Activity (IDEEA) | |
Surface Pro 2 | Microsoft | computer |