Normative Scores on the Clock Drawing Test Among Older Adults from a Large Population Survey in Norway: The HUNT Study

Abstract

Background:

The Clock Drawing Test (CDT) is used to screen for Alzheimer’s disease and other dementia disorders. Normative scores on the version from the Montreal Cognitive Assessment (MoCA) do not exist in the Nordic countries.

Objective:

To examine the normative scores of the CDT among adults aged 70 years and older.

Methods:

We included 4,023 cognitively healthy persons aged 70–97 years from a population survey in Norway. They were examined with the CDT, which has a total score between zero and three. A multiple multinominal regression model was applied with a CDT score as the dependent categorical variable and estimated the probabilities of scoring a particular score, stratified by age, sex, and education. These probabilities correspond to an expected proportion of the normative population scoring at, or below a given percentile.

Results:

None scored zero, 2.1% scored one, 14.9% scored two, and 83% scored three. Higher age, female sex and fewer years of schooling were associated with poorer performance. Scores of zero and one deviated from the normative score regardless of age, sex and education. A score of two was within the norm for a female older than 81 and a male older than 85.

Conclusions:

The majority (83%) of people 70 years and older had a score of three on the CDT. Lower age, male sex, and higher education were associated with a better performance. Scores of zero and one were below the normative score. Except for the very old, a score of two was also well below the normative score.

Keywords

Alzheimer’s disease Clock Drawing Test cognitive impairment dementia normative score

INTRODUCTION

It is well known that dementia, mild cognitive impairment (MCI) and age-dependent cognitive decline is common among people aged 70 years and older. Age-dependent cognitive decline can be difficult to separate from the beginning stage of a pathological process. 1 The use of a biomarker, either in spinal fluid or blood, or by use of imaging techniques would be strategies to separate mentally healthy older adults with subjective cognitive decline from those with early Alzheimer’s disease. However, lumbar puncture is not well accepted and is difficult to carry out except in specialist health care services. Brain imaging like positron emission tomography is costly and rarely available outside academic hospitals, while blood biomarkers are yet not well validated. A neuropsychological examination could be helpful, but such an examination is available only for a small number of older adults in most countries, if available at all. A third strategy that is cheap and often applied in primary health care is to use short and simple cognitive tests or questionnaires to identify people with cognitive impairment that should be offered full diagnostic assessment. Many tests exist for this purpose, such as the Mini-Mental State Examination (MMSE) and the Montreal Cognitive Assessment (MoCA), to mention the two most used.2,3, 2,3 Both are easy to administer, but many primary care physicians are often reluctant to use these tests either because they are not trained in them, are uncomfortable to ask the simple questions of the tests, or do not have the time to do such tests. Two other barriers exist; the use of MMSE is associated with a fee and to use MoCA one should be certified.

The Clock Drawing Test (CDT) could be an alternative. A systematic review reported that the CDT is as sensitive for detecting dementia as the MMSE and the MoCA. 4 Further, in a systematic review, Park and coworkers in 2018 reported pooled sensitivity of 82.0% and specificity of 75.7% using the scoring system of Shulman to screen for dementia, whereas it was 72.6% and 87.9% using the Sunderland scoring system. 5 A recent systematic review by Chan and coworkers of 2022 reported similar high sensitivity and specificity in screening for dementia using paper-and-pencil CDT. Among studies applying a brief scoring method pooled sensitivity and specificity was 83% and 80%, whereas it was 80% and 81% among studies using more detailed scoring systems. In comparison pooled sensitivity and specificity using digital CDT to screen for dementia was 83% and 87%. 6 In addition, the CDT is easy and quick to carry out, and is well received by older adults. According to Critchley, the CDT was thought to examine parietal lobe dysfunction, and Mendes et al. have argued that the test is a sensitive measure of constructional apraxia.7,8, 7,8 To perform a perfect CDT, several cognitive abilities, such as verbal comprehension, memory, executive function, attention, concentration, abstract thinking, and motor and perceptual function, must be intact.9,10, 9,10

Several complicated and time-consuming procedures exist on how to evaluate the performance on the CDT, but these require training and knowledge in neuropsychology.9 –12 The use of comprehensive instructions for rating will not be useful for a busy clinician who would rather have simple instructions such as the ones found in the MoCA. As far as we know, only one previous normative study used the MoCA version of the CDT. 13 As that study was from China, we believe that there is a need for normative studies of this CDT version from other parts of the world.

Most physicians would prefer to have a single cutoff value on a test, but as the performance on cognitive tests in most cases is dependent on age, sex, and education, a single cutoff value does not fit all. A way to overcome this problem is to define normative scores for people according to their age, sex, and education. Previously, we have adapted this method for the MMSE, MoCA, and the Word List Memory Test of the Consortium to Establish a Registry for Alzheimer’s Disease’s (CERAD) battery, and we believe that it would be valuable to have normative scores for the CDT as well.14 –16 With a few exceptions, previous normative CDT studies have shown that performance on the CDT is significantly influenced by age and education but not sex.13,17–30 , 13,17–30 However, most of the normative CDT studies have included few participants, especially those of a very high age (70+ years), and few studies are population based.13,17–30 , 13,17–30 In addition, most previous studies have applied complicated evaluation instructions.17 –30 Thus, there is a need for further normative studies, which preferably should include mentally healthy older adults from nonselective population surveys.

Aim of the study

The aim of the present study was to examine the normative scores of the CDT using the instructions and evaluation criteria from the MoCA. We hypothesized that the majority of cognitively healthy people aged 70+ years would achieve a full score on the CDT. However, in line with previous studies, we further anticipated that higher age is associated with poorer performance. As it has been shown that women do worse than men on visuospatial/constructive tasks, we assumed that older women would perform poorer than men on the CDT. But, in contrast to most previous results, we assumed that education would not be significantly associated with the score on the CDT, as the vast majority of our population have used a clock with numbers and hands on a daily basis since they began attending compulsory school.17 –30

Participants

Data were extracted from a population-based survey, the Trøndelag Health Study fourth wave of people aged 70 years and above (HUNT4 70+), which was carried out between September 2017 and February 2019. The purpose of the survey was to examine the prevalence of major (dementia) and minor cognitive (MCI) disorder according to the DSM-5 criteria among people aged 70 years and older. See paragraph “Diagnosis dementia and MCI” below how diagnoses were made. The population is considered to be representative of the population of Norway. Detailed information of the HUNT4 70+ study can be found elsewhere.31,32, 31,32 In all, 19,403 adults aged 70+ years were invited to participate, with 9,930 accepting, 5.9% of whom lived in a nursing home. Information on education was retrieved from Norwegian Statistics. The CDT was completed by 8,171 participants. To obtain a sample of healthy participants who were expected to have a normal CDT, we excluded 766 persons with dementia, 2,956 with MCI, and another 37 with a cognitive impairment that was not possible to classify, see paragraph Diagnosing dementia and MCI how these participants were identified. Of the remaining 4,412 participants, we examined whether any predefined clinical groups had a significantly lower score on the CDT. A group of persons with a score above seven on the anxiety subscale of the Hospital Anxiety and Depression Scale (HADS-A, n = 389) scored significantly lower (p 0.003) on CDT and were excluded. However, groups of persons with self-reported stroke (n = 284), diabetes (n = 421), any mental health complaints (n = 451), and a score of seven on the depression subscale of HADS-D, n = 306) were included as they did not score lower on CDT”. The cut point of above seven was chosen according to the findings of the systematic review of Bjelland et al. that included 747 validity studies. A score above seven on both subscales best balanced between sensitivity and specificity for a case of depression and anxiety, respectively, also in samples of the general population.33,34, 33,34 As judged by health personnel conducting the CDT test, none of the remaining 4,023 participants had impaired vision, hearing, or motor impairment that could influence negatively on the performance to complete the CDT. Using this strategy, similar to what was applied in previous normative studies, we included the CDT results from 4,023 participants.14,15,35,36 , 14,15,35,36

Assessments

To secure that the CDT was conducted and rated in the same way health personnel and nurse students that administered and evaluated the performance on CDT attended a two days training. It consisted of lectures on how to use the CDT, a video that demonstrated how to administer and conduct the test and to rate the result. In addition, practical training under supervision was carried out. The assessors could contact the supervisors at any time during the study period. The instructions were as follows: “Please draw a clock. Put in all the numbers, and set the time to 10 past 11.” The performance was scored as follows: Contour (1 pt.): the clock face must be a circle with only minor distortion acceptable. Numbers (1 pt.): all clock numbers must be present with no additional numbers; numbers must be in the correct order and placed in the approximate quadrants on the clock face. Hands (1 pt.): there must be two hands jointly indicating the correct time; the hour hand must be clearly shorter than the minute hand; hands must be centered within the clock face with their junction close to the clock center. According to this scoring system, the total score can vary between zero (worst) and three (best). To evaluate depression and anxiety, HADS was used by self-report. People scoring eight and above on both subscales are likely to be suffering from depression or anxiety.33,34, 33,34

Diagnosing dementia and MCI

Based on information from several sources including a structured interview with a family carer, rating of activities of daily living, information from medical records in nursing homes, assessments using MoCA, the Word List Memory Test (WLMT) from CERAD, and the short version of the Severe Impairment Battery (SIB, nursing home patients), diagnoses of dementia (major neurocognitive disorder) and MCI (minor neurocognitive disorder) were made according to the DSM-5 criteria.37 –39 For each participant, two experienced physicians from a pool of nine neurologists, psychiatrists, and geriatricians, who had participated in a training session and a pilot study prior to the survey, independently made the diagnoses. In case of disagreement, a third specialist was consulted. 40

Statistics

Characteristics of the participants were presented as means and standard deviations (SDs) or frequencies and percentages for the entire sample as well as stratified by sex, CDT score, and not being able to draw a circle or locate the numbers or hands correctly. The groups of participants were compared by χ²-test or ANOVA, as appropriate. The outcome variable, CDT, is a categorical variable with three ordered categories. To assess the association between age (measured in years), sex, and education (categorized as compulsory (10 years), secondary (11–13 years), and tertiary (≥14 years), an ordinal regression model was considered first. The parallel odds assumption was, however, not met and therefore a multinomial regression model was estimated instead. A potential nonlinear association between age and outcome was assessed by including age as first-, second-, and third-order component in the model. In addition, the model included all two- and three-way interactions between the covariates. The Bayesian information criterion (BIC), where smaller values imply a better model, was applied to reduce the model for excessive interactions. Marital status was considered as a potential covariate as well; however, it did not improve the model fit according to BIC and was therefore not included in the final model. The probabilities with corresponding 95% confidence intervals (CIs) for CDT total scores of one, two and three were then derived for each age value stratified by gender and education and illustrated graphically. These probabilities correspond to an expected proportion of the normative population scoring at, or below, a given score, i.e., are equivalent to percentiles. Clearly, it is improper to apply z-score-based cutoffs as in the case of normally distributed data for categorical data. Nevertheless, it is reasonable to assume that the probabilities for a certain score at or below approximately 16%, 7%, and 2.5% could be interpreted in the same way as z-scores of 1, 1.5, and 2 in the case of normally distributed data, i.e., indication of small, moderate, and large deviations from normative data.

Ethical considerations and data security

The HUNT4 70+ data collection has been approved by the Regional Committee for Medical and Health Research Ethics in Norway, the Norwegian Data Protection Authority, and the Norwegian Centre for Research Data and was carried out in accordance with the General Data Protection Regulations (GDPR). The participants gave informed consent; a proxy gave informed consent when a participant was judged to lack capacity. The current study was approved by the Regional Committee for Medical and Health Research Ethics (REK South East D 82985) and the Data Protection Authority (791342).

RESULTS

Table 1 shows the demographic and cognitive characteristics of the participants. In all, 2,162 (53.7%) were women and 1,861 (46.3%) were men. Their mean age was 76.2 (SD 5.0) years for all, 76.4 (SD 5.1) for women and 76.0 (SD 4.9) for men. Of the entire sample, we had information on the marital status of 3,983. Of them, 2,629 (66%) were married. The mean score on the Montreal Cognitive Assessment for all participants was 25.4 (SD 2.3), including 25.6 (SD 2.4) for women and 25.2 (SD 2.3) formen.

Table 1

Demographic and cognitive characteristics of the participants (n = 4,023)

Characteristics	All n = 4,023	Women n = 2,162	Men n = 1,861
Age, mean (SD)	76.2 (5.0)	76.4 (5.1)	76.0 (4.9)
Age groups, n (%)
70–74 y	1,982 (49.3)	1,011 (46.8)	971 (52.2)
75–79 y	1,160 (28.8)	666 (30.8)	494 (26.5)
80–84 y	590 (14.7)	311 14.4)	279 (15.0)
85 + y	291 (7.2)	174 (8.0)	117 (6.3)
Education groups, n (%)
Compulsory school, ≤10 y	700 (17.4)	454 (21.0)	246 (13.2)
Secondary school, 11–13 y	2,238 (55.6)	1,186 (54.9)	1,052 (56.5)
Tertiary education, ≥14 y	1,085 (27.0)	522 (24.1)	563 (30.3)
Marital status (n = 3,983), Yes (%)	2,629 (66.0)	1,197 (56.0)	1,432 (77.7)
MoCA score, mean (SD)	25.4 (2.3)	25.6 (2.4)	25.2 (2.3)

Of the 4,023 participants, 3,341 scored three on the CDT (83%), 599 (14.9%) scored two, 83 (2.1%) scored one, and no one scored zero. Table 2 shows the scores on the CDT across sex, age groups, educational level, and marital status. Higher age, female sex, fewer years of schooling, and being unmarried were all associated with low performance on the CDT. We further assessed the proportion of the participants who failed to draw a circle, set numbers, and/or locate the hands correctly. As shown in Table 3, higher age, female sex, and less education were associated with more problems in locating the hands correctly, whereas setting the numbers incorrectly was only associated with age. In the multiple multinomial regression model, higher age was associated with higher odds for a CDT score of one or two, compared to a CDT score of three (Table 4). Being a woman compared to a man was associated with higher odds for scoring two as compared to scoring three. In addition, the participants with tertiary education, as compared to those with secondary education, had lower odds for scoring one as compared to scoring three (see Table 4). The probabilities with corresponding 95% confidence intervals (CIs) for scoring one, two and three, separately for women and men aged 70 and older for three different educational levels, derived from the multiple multinomial regression model, are shown in Fig. 1. A score of zero or one should clearly be regarded as a deviation from a normative score regardless of the older adult’s age, sex, and education, since these correspond to percentiles of 7% or lower. Whether a score of two implies a deviation depends on the person’s age, sex and education. For example, an 82-year-old woman with a secondary education and a score of two on the CDT will belong to the proportion of the general population above the 15th percentile with the probability of 95% since the corresponding probability for this woman is 0.21 with 95% CI of 0.16–0.27. If the same woman scored one on the CDT, the related probability is 0.04 with 95% CI of 0.03–0.05, which implies that the woman belongs to a group of the 5% of women in the population with the same or lower score (zero) on the CDT.

Table 2

Score on the Clock Drawing test (CDT) across sex, age, educational and marital status.

Characteristics	CDT score 1	CDT score 2	CDT score 3	p ^a
	n = 83	n = 599	n = 3,341
Sex
Women, n = 2,162	53 (2.5)	354 (16.4)	1,755 (81.2)	0.002
Men, n = 1, 861	30 (1.6)	245 (13.2)	1,586 (85.2)
Age groups, n (%)
70–74 years, n = 1,982	35 (1.8)	257 (13.0)	1,690 (85.3)	<0.001
75–79 years, n = 1,160	22 (1.9)	159 (13.7)	979 (84.4)
80–84 years, n = 590	15 (2.5)	111 (18.8)	464 (78.6)
85+ years, n = 291	11 (3.8)	72 (24.7)	208 (71.5)
Education groups
Compulsory, n = 700	14 (2.0)	111 (15.9)	575 (82.1)	0.045
Secondary, n = 2,238	55 (2.5)	347 (15.5)	1,836 (82.0)
Tertiary, n = 1,085	14 (1.3)	141 (13.0)	930 (85.7)
Marital status
Married, n = 2,629	49 (1.9)	361 (13.7)	2,219 (84.4)	0.01
Not married, n = 1,354	33 (2.4)	228 (16.8)	1,093 (80.7)

^aPearson χ²-test.

Table 3

Proportion (percentages) of participants who could not draw a circle or set the number and hands correctly on the Clock Drawing Test (CDT)

Characteristics	Incorrect circle	Incorrect numbers	Incorrect hands
	n = 33	n = 237	n = 495
Sex
Women: Men, %	0.7 : 0.9	5.5 : 6.3	15.0 : 9.1
	p = 0.5	p = 0.3	p < 0.001
Age groups
70–74 : 75–79 : 80–84 : 85 + y, %	0.8 : 0.8 : 0.7 : 1.7	4.9 : 5.5 : 7.8 : 10.0	10.8 : 11.2 : 15.4 : 20.6
	p = 0.4	p = 0.001	p < 0.001
Educational status
Compulsory: Secondary: tertiary, %	0.9 : 1.0 : 0.5	4.7 : 5.9 : 6.7	14.3 : 13.6 : 8.4
	p = 0.3	p = 0.2	p < 0.001
Marital status
Married: Not Married, %	0.7 : 1.0	5.6 : 6.4	11.1 : 14.4
	p = 0.4	p = 0.4	p = 0.003

Table 4

Multinomial regression model (n = 4,023)

Variable	Bivariate models		Multiple model
	OR (95% CI)	p	OR (95% CI)	p
Age
CDT 1	1.06 (1.02; 1.11)	0.002	1.06 (1.02; 1.10)	0.003
CDT 2	1.05 (1.03; 1.06)	<0.001	1.05 (1.03; 1.06)	<0.001
CDT 3 – ref.	1		1
Sex, man (women -ref)
CDT 1	0.63 (0.40; 0.99)	0.043	0.65 (0.41; 1.02)	0.06
CDT 2	0.77 (0.64; 0.91)	0.003	0.78 (0.66; 0.94)	0.007
CDT 3 – ref.	1		1
Educational level
0–10 compulsory
CDT 1	0.81 (0.45; 1.47)	0.494	0.69 (0.38; 1.27)	0.234
CDT 2	0.50 (0.28; 0.91)	0.023	0.92 (0.72; 1.16)	0.475
CDT 3 – ref.	1		1
11–13 secondary – ref.	1		1
14+ tertiary
CDT 1	1.02 (0.81; 1.29)	0.859	0.54 (0.30; 0.97)	0.04
CDT 2	0.80 (0.65; 0.99)	0.041	0.84 (0.68; 1.04)	0.102
CDT 3 – ref.	1		1

Fig. 1

Probability (equivalent to percentile) for total score of one, two, and three on the CDT as a function of age stratified by gender and education, results of multinomial regression analysis. Black and dashed lines illustrate average probability and the corresponding 95% confidence interval, respectively. Vertical grid lines indicate age.

DISCUSSION

The main finding of the study was that 17% of the population of older, cognitively healthy participants did not have a full score on the CDT. None scored zero, 2.1% scored one, and 14.9% scored two. Higher age, female sex, and lower educational level were associated with having a lower performance. We further found that a score of one (and zero) implies a deviation from the normative score regardless of age, sex, and education, and that a score of two is within the range of normality among the very old of this sample. The results will be discussed in the order as reported.

As expected, the majority (83%) scored correctly on all CDT items. This result is difficult to compare with other studies, as we found only one study from China that used the same evaluation criteria as we did. 13 In that study 70.3% had a full score of three, whereas 25.6% scored two and 4.1% scored one. These figures are different compared to our results. The Chinese participants volunteered for the study and were not drawn from a population register. They were younger and had less education compared to the participants of the present study. As age is a strong predictor of performance on the CDT, it is not easy to explain the poorer performance among the participants from China. It may be that age, sex, and education have a different effect on CDT performance in the two countries, which could be related to culture or the habit of wearing a timepiece and checking a clock during their daily activities.

We expected that more than 83% would score three. One explanation for why 17% did not perform the CDT perfectly may be due to the large group of participants aged 80 and older included in the present study. In all, 21.9% belonged to this group and 55% of them were women. These oldest participants (80+ years) had the poorest performance. It is well known that performing complex cognitive tasks gets poorer with increasing age, due to the aging process. This decline is especially marked after age 80.41 –44 To perform CDT correctly, one needs intact cognitive abilities in various domains such as executive function, visuospatial abilities, abstract thinking, concentration, and verbal comprehension. For example, our results demonstrate that the oldest participants had problems with locating the numbers and setting the hands correctly, two complex cognitive tasks. Our finding that higher age was negatively associated with the performance on the CDT score fits well with results from several other studies using various instructions for use of the test and various ways of evaluating the results.13,17–30 , 13,17–30

We hypothesized that performance on the CDT would be independent of schooling, while Norwegian persons born just before, during, and shortly after the Second World War have used an old-fashioned clock with numbers and hands every day since they started school. In addition, illiteracy is almost absent in the Norwegian population. Some experts in the field have supported the assumption that the CDT is educationally unbiased.10,45,46 , 10,45,46 Were we mistaken? According to the multinomial regression analysis, years of schooling influenced the performance on the CDT, but there was no difference between those with compulsory and secondary education. The group with tertiary education scored better. The differences were particularly pronounced across the three educational levels in setting the hands correctly. Older persons with compulsory (14.3%) and secondary (13.6%) educations, compared to those with tertiary education (8.4%), failed to set the hands correctly. Setting the hands correctly requires a good function in abstract thinking, an ability that is necessary for passing higher education and one that develops further by studying at a college or a university. In comparison, studies where the educational level was dichotomized using compulsory school as the threshold value reported that education had an influence on the score, at least for a subgroup.8,47, 8,47 Other studies did not find an association between education and the CDT score using the same threshold value for education. One study, in line with our research, reported an association using a cutoff point at the college level (no college versus college). 20 The association between education and performance on the CDT seems to be dependent on which threshold values for education have been applied, and maybe as well on the educational system of the population in which the CDT is examined. This could be one of the more persuasive arguments for why normative scores on cognitive tests like the CDT should be examined among people with the same culture and educational system.

In contrast to most other studies, we found that a larger proportion of men compared to women had a full score on the CDT and that more men placed the hands correctly on the clock. We were not surprised as visuo-construction is an important ability for performing the CDT correctly. Numerous studies have reported that visuo-construction is sex-dependent, and that men at any age perform visuo-constructive tasks better than women, whereas women do better on verbal memory tasks.14,15,48–51 , 14,15,48–51

Regardless of age and education, the probability of scoring one on the CDT was between 0.5% and 12% among women and between 0.3% and 10% among men, when including the 95% CI values, whereas the probability of scoring two was between 8% and 40% among women and 6% and 38% among men. The question arises as to how we can interpret these findings and make them useful for clinical practice. Having results on a normally distributed continuous scale, a z-score of zero (average score) would correspond to the 50th percentile, whereas a z-score of 1 would correspond to the 15.87th percentile, 1.5 to the 6.68th percentile and 2 to the 2.28th percentile. As we have categorical data, the z-score is not applicable. However, taking all reservations into consideration, we can assume that a performance on the CDT corresponding to the 15.87th percentile and above could be considered as a normal result. A score corresponding to between the 6.68th and the 15.87th percentile would be a mild deviation from the normative score, a score between the 2.88th and the 6.68th percentile a moderate deviation, and below the 2.28th percentile a severe deviation from the normative score. Under this assumption, a score of zero or one on CDT would, regardless of sex, age, and educational level, but including the 95% CI values, be a deviation from the normative score. Among older women, a score of two would lie above the 15.87th percentile, with a probability of 95% for women aged 84 years with primary school, 82 years for those with secondary school, and 86 years for those with a tertiary education. The corresponding ages for older men would be 92, 86 and 92 years, respectively. The drop in age for older adults with secondary education is difficult to understand. However, as we cannot directly transfer the understanding of percentiles corresponding to performances on a continuous scale with performances on categorical data, we suggest use of a pragmatic approach. By simplifying the findings without taking notice of educational level we conclude that a score of zero and one on the CDT is not normal for any person and a score of two should be considered normal for women aged 82 and older and for men aged 86 and older. Otherwise, a score of two is not normal for persons below those ages.

The main result of the present study indicates that except for very old people approaching the age of 90 years, cognitive healthy older people succeeded in performing the CDT version of MoCA in a perfect way. For screening purposes or to establish a baseline for later follow-up this version may be sufficient. A systematic review of various scoring systems for evaluation of performance on CDT supports this view, as the authors could not find any scoring system that was superior in terms of predictive value. The conclusion of the review says: “simpler is better”. 52

Limitations and strengths

This study has some limitations. For one, we cannot be sure that the included sample is absolute representative of older people of the catchment area and the older population of Norway as the participation rate was relatively low. We have some information of the non-participants. In a study of 2023 using data from about half of the invited participants to HUNT4 70+ it was found that non-participation was less common among younger women (>75 years) having tertiary education. 53 From the Norwegian national primary health care registry non-participants of HUNT4 above the age of 80 years visited a general practitioner less frequent than participants, and home nurse service and nursing home care were more often provided to non-participants. In addition, dementia diagnosis was less recorded in the registry among the participants. These differences were small but could indicate that frailty and dementia were more common among the non-participants compared to the participants. 32 Thus, we assume that the differences are not a significant concern for the representativeness of our study as we wanted to include cognitively healthy older people. A strength of the study is the large number of participants, especially participants above the age of 80 years, a group where the cognitive aging process is advanced. Another strength is that the study is based on a nonselective group that took part in a population-based survey, which ensured that older adults with various health problems and with various demographic characteristics were included. Although the attrition rate is relatively low, we will argue that due to the design of the study (non-selective community survey) and the large number of participants there is a low concern for selection bias. Second, we excluded not only those with a diagnosis of dementia, MCI, and severe cognitive impairment of unknown etiology but also a group of 389 people with a high score on HADS-A, indicating anxiety. This group had a significantly lower score on the CDT. Some of the persons in the anxiety group could have been included, but we believe that this would not have changed the results, because the group is small compared to the sample of 4,023. Why older people with anxiety scored poorer on the CDT is another question. Another retrospective study among memory clinic patients did not find any association between anxiety and performance on the CDT. 54 We assume that anxious people have reduced self-confidence, and for this reason they might have been in a stressed situation when performing the CDT in front of a test administrator. Anxiety is also associated with loss of concentration. Third, the dementia and MCI diagnoses were not made by physicians examining the participants in person but by using all available information given by the participants and their next of kin, from medical records, and from the cognitive testing results. Experienced physicians made diagnoses based on consensus by use of standardized criteria. We assume that this procedure was sufficient to make correct diagnoses. Fourth, only one health professional scored the performance on the CDT. Preferably two or more should have done this evaluation. Preferably an inter- rater agreement study should have been conducted prior to the main study. This was unfortunately not done. Searching the literature we only found one inter-rater study of acceptable quality that reported inter-rater-reliability of the CDT version of MoCA among cognitively healthy individuals. In the Chinese study by Shao et al, five raters evaluated the performance on the CDT version of MoCA and found an inter-rater correlation coefficient (ICC) of 0.71, which is acceptable. 13 Two other studies from the same research group, one including patients with various dementia disorders and a smaller control group and one including orthopedic patients with unknown cognitive status reported lower inter-rater agreement.55,56, 55,56 Price et al. reported ICC of 0.28–0.58 across three raters and Frei et al. an average weighted kappa of 0.60. In both studies novice (untrained) assessors rated the performances.55,56, 55,56 Cumming et al. reported kappa statistics for each of the three items of the CDT among 1,119 stroke patients with unknown cognitive status from 102 assessors in 54 centers. For contour, numbers and hands, kappa was 0.49, 0.67 and 0.46, respectively. 57 Both Price and Cumming recommended that raters should be trained. An Italian study reported inter-rater agreement of the total score on MoCA of ICC 0.99 and a Spanish study reported inter- rater agreement of Spearman correlation of 0.85.58,59, 58,59 Based on the comprehensive training of the assessors in the recent study and findings of inter-rater agreement in other studies, especially the Chinese study of cognitively healthy persons, we assume that the agreement between the raters in the present study was acceptable.

Conclusions

Using the instructions from the Montreal Cognitive Assessment to carry out and evaluate the Clock Drawing Test (CDT), we found that 83% of 4,023 cognitively healthy adults aged 70 years and older, recruited from a nonselective population survey, scored a maximum score of three, whereas none scored zero, 2.1% scored one, and 14.9% scored two. According to a multiple multinominal regression model, higher age, female sex, and fewer years of schooling negatively influenced the categorical score. Failure to set the numbers and hands were the major reasons for not scoring three on the CDT. Drawing a circle incorrectly was a rare finding among the older participants. Regardless of age, sex, and education, a score below two should be considered as a deviation from the normative score in adults aged 70 years and older. A score of two is within the norm for women of age 82 and older and men aged 86 and older. This is a pragmatic recommendation as it is impossible to directly transfer the understanding of percentiles of continuous data to categorical data. It should be noticed that the normative scores found in the present work do not apply for migrants whose limited knowledge of the spoken language and culture could be a barrier. 60

AUTHOR CONTRIBUTIONS

Knut Engedal (Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Visualization; Writing – original draft; Writing – review & editing); Jurate Saltyte Benth (Conceptualization; Data curation; Formal analysis; Writing – original draft; Writing – review & editing); Jørgen Wagle (Conceptualization; Formal analysis; Methodology; Writing – original draft; Writing – review & editing); Linda Gjøra (Conceptualization; Data curation; Methodology; Writing – original draft; Writing – review & editing); Geir Selbæk (Conceptualization; Formal analysis; Methodology; Project administration; Writing – original draft; Writing – review & editing); Karin Persson (Conceptualization; Formal analysis; Investigation; Methodology; Writing – original draft; Writing – review & editing).

Footnotes

ACKNOWLEDGMENTS

The data collection took place as part of the fourth wave of the Trøndelag Health Study (HUNT). The HUNT is a collaboration between the HUNT Research Centre (Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology NTNU), the Trøndelag County Council, the Central Norway Regional Health Authority, and the Norwegian Institute of Public Health.

FUNDING

The authors have no funding to report.

CONFLICT OF INTEREST

KE, the first author, is Editorial Board Member of this journal, but was not involved in the peer-review process nor had access to any information regarding its peer-review. GS reports participation in an advisory board of companies manufacturing monoclonal antibody drugs for Alzheimer’s disease outside the submitted work, and KP reports work with Novo Nordic and Roche in clinical trials outside the submitted work. JSB, JW, and LG have no conflicts of interest to declare.

DATA AVAILABILITY

The data supporting the findings in this study are not publicity available due to restrictions of the HUNT databank.

References

Jessen

, Amariglio

, Buckley

, et al. The characterisation of subjective cognitive decline. Lancet Neurol 2020; 19: 271–278.

Folstein

, Folstein

and McHugh

. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12: 189–198.

Nasreddine

, Phillips

, Bédirian

, et al. The Montreal Cognitive Assessment, MoCA: A brief screening tool for mild cognitive impairment. J Am Geriatr Soc 2005; 53: 695–699.

Tsoi

, Chan

, Hirai

, et al. Cognitive tests to detect dementia: A systematic review and meta-analysis. JAMA Intern Med 2015; 75: 1450–1458.

Park

, Jeong

and Seomun

. The clock drawing test: A systematic review and meta-analysis of diagnostic accuracy. J Adv Nurs 2018; 74: 2742–2754.

Chan

JYC

, Bat

BKK

, Wong

, et al. Evaluation of digital drawing tests and paper-and-pencil drawing tests for the screening of mild cognitive impairment and dementia: a systematic review and meta-analysis of diagnostic studies. Neuropsychol Rev 2020; 32: 566–576.

Critchley

. The parietal lobes. London: Edward Arnold & Co, 1953.

Mendez

, Ala

and Underwood

. Development of scoring criteria for the clock drawing task in Alzheimer’s disease. J Am Geriatr Soc 1992; 40: 1095–1099.

Freedman

, Leach

, Kaplan

, et al. Clock Drawing: A neuropsychological analysis. New York: Oxford University Press, 1994.

10.

Shulman

. Clock-drawing: Is it the ideal cognitive screening test? Int J Geriatr Psychiatry 2000; 15: 548–561.

11.

Sunderland

, Hill

, Mellow

, et al. Clock drawing in Alzheimer’s disease. A novel measure of dementia severity. J Am Geriatr Soc 1989; 37: 725–729.

12.

Wang

, Shi

, Zhao

, et al. Longitudinal changes in clock drawing test (CDT) performance before and after cognitive decline. PLoS One 2014; 9: e97873.

13.

Shao

, Dong

, Guo

, et al. Clock-drawing test: Normative data of three quantitative scoring methods for Chinese-speaking adults in Shijiazhuang City and clinical utility in patients with acute ischemic stroke. Brain Behav 2020; 10: e01806.

14.

Engedal

, Gjøra

, Benth

JŠ

, et al. The Montreal Cognitive Assessment: Normative data from a large, population-based sample of cognitive healthy older adults in Norway – The HUNT Study. J Alzheimers Dis 2022; 86: 589–599.

15.

Wagle

, Selbæk

, Benth

JŠ

, et al. The CERAD Word List Memory Test: Normative data based on a Norwegian population-based sample of healthy older adults 70 years and above. The HUNT Study. J Alzheimers Dis 2023; 91: 321–343.

16.

Engedal

, Benth

JŠ

, Gjøra

, et al. Normative scores on the Norwegian version of the Mini-Mental State Examination. J Alzheimers Dis 2023; 92: 831–842.

17.

Stewart

, Richards

, Brayne

, et al. Cognitive function in UK community-dwelling African Caribbean elders: Normative data for a test battery. Int J Geriatr Psychiatry 2001; 16: 518–527.

18.

Kim

and Chey

. Effects of education, literacy, and dementia on the Clock Drawing Test performance. J Int Neuropsychol Soc 2010; 16: 1138–1146.

19.

Carvalho

and Caramelli

. Normative data for middle-aged Brazilians in Verbal Fluency (animals and FAS), Trail Making Test (TMT) and Clock Drawing Test (CDT). Dement Neuropsychol 2020; 14: 14–23.

20.

von Gunten

, Ostos-Wiechetek

, Brull

, et al. Clock-drawing test performance in the normal elderly and its dependence on age and education. Eur Neurol 2008; 60: 3–8.

21.

Bozikas

, Giazkoulidou

, Hatzigeorgiadou

, et al. Do age and education contribute to performance on the clock drawing test? Normative data for the Greek population. J Clin Exp Neuropsychol 2008; 30: 199–203.

22.

Hubbard

, Santini

, Blankevoort

, et al. Clock drawing performance in cognitively normal elderly. Arch Clin Neuropsychol 2008; 23: 295–327.

23.

Siciliano

, Santangelo

, D’Iorio

, et al. Rouleau version of the Clock Drawing Test: Age- and education-adjusted normative data from a wide Italian sample. Clin Neuropsychol 2016; 30 (Suppl 1): 1501–1516.

24.

Caffarra

, Gardini

, Zonato

, et al. Italian norms for the Freedman version of the Clock Drawing Test. J Clin Exp Neuropsychol 2011; 33: 982–988.

25.

Nyborn

, Himali

, Beiser

, et al. The Framingham Heart Study clock drawing performance: Normative data from the offspring cohort. Exp Aging Res 2013; 39: 80–108.

26.

Santana

, Duro

, Freitas

, et al. The Clock Drawing Test: Portuguese norms, by age and education, for three different scoring systems. Arch Clin Neuropsychol 2013; 28: 375–387.

27.

Mazancova

, Nikolai

, Stepankova

, et al. The reliability of Clock Drawing Test scoring systems modeled on the normative data in healthy aging and nonamnestic mild cognitive impairment. Assessment 2017; 24: 945–957.

28.

Yassuda

, da Silva

, Lima-Silva

, et al. Normative data for the Brief Cognitive Screening Battery stratified by age and education. Dement Neuropsychol 2017; 11: 48–53.

29.

Shanhu

, Linhui

, Xiaoqing

, et al. Effects of age and education on clock-drawing performance by elderly adults in China. Clin Neuropsychol 2019; 33 (Suppl 1): 96–105.

30.

Turcotte

, Gagnon

, Joubert

, et al. Normative data for the Clock Drawing Test for French-Quebec mid- and older-aged healthy adults. Clin Neuropsychol 2018; 32 (Suppl 1): 91–101.

31.

Krokstad

, Langhammer

, Hveem

, et al. Cohort profile: The HUNT Study, Norway. Int J Epidemiol 2013; 42: 968–977.

32.

Åsvold

, Langhammer

, Rehn

, et al. Cohort profile update: The HUNT Study, Norway. Int J Epidemiol 2023; 8: e80–e91.

33.

Zigmond

and Snaith

. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983; 67: 361–370.

34.

Bjelland

, Dahl

, Haug

, et al. The validity of the Hospital Anxiety and Depression Scale: An updated literature review. J Psychosom Res 2002; 52: 69–77.

35.

Borland

, Nägga

, Nilsson

, et al. The Montreal Cognitive Assessment: Normative data from a large Swedish population-based cohort. J Alzheimers Dis 2017; 59: 893–901.

36.

Thomann

, Goettel

, Monsch

, et al. The Montreal Cognitive Assessment: normative data from a German-speaking cohort and comparison with international normative samples. J Alzheimers Dis 2018; 64: 643–655.

37.

Morris

, Mohs

, Rogers

, et al. Consortium to establish a registry for Alzheimer’s disease (CERAD) clinical and neuropsychological assessment of Alzheimer’s disease. Psychopharmacol Bull 1988; 24: 641–652.

38.

Schmitt

, Saxton

, Ferris

, et al. Evaluation of an 8-item Severe Impairment Battery (SIB-8) versus the full SIB in moderate to severe Alzheimer’s disease patients participating in a donepezil study. Int J Clin Pract 2013; 67: 1050–1056.

39.

American Psychiatric Association. Diagnostic and statistical manual of mental disorders: DSM-5. Washington, DC: American Psychiatric Association, 2013.

40.

Gjøra

, Strand

, Bergh

, et al. Current and future prevalence estimates of mild cognitive impairment, dementia, and its subtypes in a population-based sample of people 70 years and older in Norway: The HUNT Study. J Alzheimers Dis 2021; 79: 1213–1226.

41.

McCarrey

, An

, Kitner-Triolo

, et al. Sex differences in cognitive trajectories in clinically normal older adults. Psychol Aging 2016; 31: 166–175.

42.

Finkel

, Reynolds

, McArdle

, et al. Latent growth curve analyses of accelerating decline in cognitive abilities in late adulthood. Dev Psychol 2003; 39: 535–550.

43.

Ferreira

, Ferreira Santos-Galduróz

, Ferri

, et al. Rate of cognitive decline in relation to sex after 60 years-of-age: A systematic review. Geriatr Gerontol Int 2014; 14: 23–31.

44.

Lam

, Chiu

, Ng

, et al. Clock-face drawing, reading and setting tests in the screening of dementia in Chinese elderly adults. J Gerontol B Psychol Sci Soc Sci 1998; 53: 353–357.

45.

Ratcliff

, Dodge

, Birzescu

, et al. Tracking cognitive functioning over time: Ten-year longitudinal data from a community-based study. Appl Neuropsychol 2003; 10: 76–88.

46.

Yamamoto

, Mogi

, Umegaki

, et al. The clock drawing test as a valid screening method for mild cognitive impairment. Dement Geriatr Cogn Disord 2004; 18: 172–179.

47.

Sugawara

, Yasui-Furukori

, Umeda

, et al. Clock drawing performance in a community-dwelling population: Normative data for Japanese subjects. Aging Ment Health 2010; 14: 587–592.

48.

Aartsen

, Martin

and Zimprich

. Gender differences in level and change in cognitive functioning. Results from the Longitudinal Aging Study Amsterdam. Gerontology 2004; 50(1), 35–38.

49.

de Frias

, Nilsson

and Herlitz

. Sex differences in cognition are stable over a 10-year period in adulthood and old age. Neuropsychol Dev Cogn B Aging Neuropsychol 2006; 13: 574–587.

50.

Sundermann

, Biegon

, Rubin

, et al. Does the female advantage in verbal memory contribute to underestimating Alzheimer’s disease pathology in women versus men? J Alzheimers Dis 2017; 56: 947–957.

51.

Engedal

, Gjøra

, Bredholt

, et al. Sex differences on Montreal Cognitive Assessment and Mini-Mental State Examination scores and the value of self-report of memory problems among community-dwelling people 70 years and above: The HUNT Study. Dement Geriatr Cogn Disord 2021; 50: 74–84.

52.

Mainland

, Amodeo

and Shulman

. Multiple clock drawing scoring systems: Simpler is better. Int J Geriatr Psychiatry 2014; 29: 127–136.

53.

Skirbekk

, Bowen

, Håberg

, et al. Marital histories and associations with later-life dementia and mild cognitive impairment risk in the HUNT4 70+ Study in Norway. J Aging Health 2023; 35: 543–555.

54.

Khachan

, Kanjolia

and Nair

. Does anxiety affect the clock-drawing task in patients in a memory clinic? A clinical correlation study. Innov Aging 2020; 16 (Suppl 1): 284.

55.

Price

, Cunningham

, Coronado

, et al. Clock drawing in the Montreal Cognitive Assessment: recommendations for dementia assessment. Dement Geriatr Cogn Disord 2011; 31: 179–187.

56.

Frei

, Woodward

, Zhang

, et al. Considerations for clock drawing scoring systems in perioperative anesthesia settings. Anesth Analg 2019; 128: e61–e64.

57.

Cumming

, Lowe

, Linden

, et al. The AVERT MoCA Data: scoring reliability in a large multicenter trial. Assessment 2020; 27: 976–981.

58.

Pirani

, Nasreddine

, Neviani

, et al. MOCA 7.1: Multicenter validation of the first Italian Version of Montreal Cognitive Assessment. J Alzheimers Dis Rep 2022; 6: 509–520.

59.

Delgado

, Araneda

, Behrens

. Validation of the Spanish-language version of the Montreal Cognitive Assessment test in adults older than 60 years. Neurologia (Engl Ed) 2019; 34: 376–385.

60.

Maestri

, Nicotra

, Pomati

, et al. Cultural influence on clock drawing test: A systematic review. J Int Neuropsychol Soc 2023; 29: 704–714.