CAMCOG: Detailed description, population data and psychometric properties CAMCOG: Detailed description, population data and psychometric properties The Cambridge Cognitive Examination (CAMCOG) is a concise neuropsychological test for the assessment of cognitive impairment in elderly people. It was designed specifically to assist in the diagnosis of dementia at an early stage. CAMCOG assesses a broad range of cognitive functions, as is required for the diagnosis of dementia, and it minimises floor and ceiling effects by covering a range of item difficulty. Aims q assessed both in an oral and in a written form. Some standard neuropsychological items which are included in the CAMCOG are verbal fluency, similarities and the identification of objects photographed from unusual views. A summary of CAMCOG items and subscales is provided in Table 1. All the MMSE items are included in the cognitive examination but the following are not used in calculating the CAMCOG score; naming two objects (pencil, wristwatch), registration and recall of three words, writing a sentence and paper folding. The processes involved in these tests are assessed in more detail by other CAMCOG items. CAMCOG also contains the Abbreviated Mental Test (AMT) of Hodkinson (1972), derived from the original Dementia Scale of Blessed et al. (1968) and its predecessor (Roth & Hopkins, 1953). CAMCOG-R differs from CAMCOG in three ways: (1) It includes two additional items to assess executive function in more detail; a verbal measure (ideational fluency) and a non-verbal measure (visual reasoning). (2) In addition to the six original items assessing remote memory (for the period of the 1930s and 40s), it contains six alternative items (for the period of the 1950s and 60s) which are intended for more recently born cohorts. (3) The tactile perception item (recognising two coins placed in the subject’s hand) has been omitted. This item proved problematic as new coins were introduced and normal elderly people often had difficulty in recognising them. The omission of this item brings the CAMCOG total down from the original 107 to the new total of 105. Scoring q q q q q To assess the range of cognitive functions required for a diagnosis of dementia. To assist in differential diagnosis within the dementias. To incorporate items which are graded in difficulty within a cognitive domain in order to assess the full range of cognitive ability. To permit the measurement of cognitive decline from very high levels of premorbid ability by minimizing ceiling effects. To facilitate comparison with some other widely used brief cognitive tests, by including them within its framework. To examine profiles of cognitive performance by deriving scores on subscales which assess different cognitive abilities. Content The items contained in the CAMCOG were selected to sample the areas of cognitive functioning which are specified in operational diagnostic criteria, such as those elaborated in DSM-IV (American Psychiatric Association, 1994) and ICD-10 (World Health Organization, 1993). These cognitive functions include memory, language, attention, perception, praxis and thinking (now called executive functioning). The CAMCOG also samples important domains within an area of cognitive functioning; for example, memory items include assessment of remote and recent memory, semantic and episodic memory, intentional and incidental learning, and recall and recognition measures of retrieval. The CAMCOG incorporates items which are commonly used in neuropsychological assessment to examine dissociable funtions. Thus, for example, there are measures of language comprehension and language expression and these are CAMCOG provides subscale scores for hypothetically dissociable functions, as well as a total score with a maximum of 105 points. Each item contributes between 1 and 6 points to the relevant subscale and to the total score. For the animal fluency item, where elderly subjects may produce over 40 different animals, recoding is necessary, as specified in question 158. Despite the addition of further tests of executive function, the original CAMCOG scoring and total score has 82 Detailed description Table 1 CAMCOG items and subscales Subscale 1 Orientation 2 Language Maximum Score 10 30 Sections Time Place Comprehension: motor response verbal response reading Expression: naming fluency (animals) definitions repetition writing to dictation Remote Recent New learning: incidental New learning: intertional Serial sevens Counting backwards Calculation Copying Drawing Actions to command Similarities Tactile recognition* Visual recognition Unusual views Recognise person Maximum Score 5 5 4 3 2 6 6 6 1 2 6 4 12 5 5 2 2 3 3 6 8 2 2 6 1 3 Memory 27 4 Attention and calculation 9 5 Praxis 12 6 Abstract thinking 7 Perception 8 11 Total 107 * This item (recognising two coins) has been omitted from CAMCOG-R, changing the total from 107 to 105. not been changed, but a separate executive function score may be calculated (see p. 66). Extended assessment of executive function promised relatively early in the course of dementia. Impaired executive function together with a relative preservation of memory is indicative of dementia of the frontal type (e.g. Gregory & Hodges, 1993). Executive function is very difficult to assess briefly, both because of the many different processes subsumed under this rubric, and because most of the existing measures (many still at an early stage of development) tend to be time-consuming. The original CAMCOG contains two items which can be regarded as measures of executive function. These are the Similarities questions which assess abstraction, and the fluency item (animals) which assesses initiation and categorisation. In view of the increasingly prominent role of executive function measures in cognitive assessment and dementia diagnosis two further items have been added. The ideational fluency item (‘How many different uses can you think of CAMCOG takes approximately 20 minutes to administer. Strict adherence by the interviewer to the printed instructions for administration and coding is required to ensure reliable scores. Administration Executive function is the term used to describe a variety of high level cognitive processes including planning, organisation, abstraction, categorization, initiation, reasoning, mental flexibility, sequencing and the allocation of attentional resources. Neuropsychological evidence suggests that the frontal lobes play a key role in performance on tasks of executive function (e.g. Cummings, 1993) and that performance on such tasks may be com- CA M C O G : Detailed description 83 for a bottle?’) is taken from a test battery developed by Crawford et al. (1995). It assesses initiation and mental flexibility in the verbal domain. To assess executive function in subjects who may have language problems, a non-verbal test of visual reasoning has been added. It uses a format somewhat similar to Raven’s Progressive Matrices (Raven et al., 1976) and assesses sequencing, categorisation and abstraction. The ideational fluency test has been validated in patients with head injury, where it was found to be more sensitive to impairment than other fluency tests (Crawford, 1995). Validation studies are currently underway in normal elderly and demented patients. The tests are included in advance of published results, on the basis of their validity in other contexts and their brevity. They are inserted after the Similarities items (questions 197–200) and numbered 200(a) and (b). Because of their provisional status, scores on these tests do not contribute to the total CAMCOG score, which remains unchanged. A separate executive function score may be derived by adding the scores on these two items to the scores on similarities and animal fluency (see p. 66). For this purpose, both ideational fluency and animal fluency are recoded, bringing the maximum executive function score to 28. CAMCOG Applications Cooper et al., 1992). It is currently being used in the UK in the Medical Research Council Multi-Centre Study of Cognitive Function and Ageing (MRC CFA Study) and data will soon be available on over 3000 people aged 65 years and older from a nationally representative sample, half of whom are being re-assessed annually with CAMCOG. This study will provide norms on an unselected elderly population as well as data on large numbers of individuals with dementia. To date, British norms are available only on a very elderly cohort in Cambridge City aged over 75 years (Huppert et al., 1995, 1996). Data from these studies are presented in Tables 2 and 3. CAMCOG Profile For clinical work, it is often useful to have a visual profile of an individual’s cognitive strengths and weaknesses. Jean Hooper and Romola Bucks, two clinical psychologists working with older adults in Gloucestershire, have developed the ‘Cognitive Profile’ (Hooper & Bucks, 1993) based on CAMCOG subscales, which is scored manually. A CAMCOG profile can also be obtained using computer scoring and can be printed out from the CAMDEXR disk. We are currently developing a computerised method for examining an individual’s obtained versus expected scores on the CAMCOG total and subscales, based on his/her sociodemographic characteristics. Health variables may also be included to examine the extent to which an individual’s cognitive impairment can CAMCOG has been used in many published investigations both clinical (e.g. Hunter et al., 1989; Jobst et al., 1992a, b) and population-based (e.g. O’Connor et al., 1989; Brayne & Calloway, 1990; Clarke et al., 1991; Table 2 Performance of an elderly population sample on CAMCOG subscales Percent obtaining maximum 51 0 0 31 27 80 20 16 Subscale Orientation Language Memory Attention Praxis Calculation Abstract thinking Perception Total Maximum value 10 30 27 7 12 2 8 10* 106* Mean (SD) 9.2 (1.0) 21.9 (2.6) 20.7 (3.6) 5.1 (1.9) 10.1 (2.0) 1.8 (0.4) 5.2 (2.3) 7.8 (1.8) Median score 10 25 21 5 10 2 5 8 Range 3–10 9–29 4–26 0–7 0–12 0–2 0–8 1–10 * For this community sample one item (recognising two people in the room) had to be omitted, reducing the number of items from 4 to 3, and the maximum score from 11 to 10. The total CAMCOG score was accordingly reduced from 107 to 106. Source: Huppert et al. (1995) 84 Detailed description Table 3 Mean scores of an elderly population sample on CAMCOG subscales Age Group CAMCOG Subscales Orientation Language Memory Attention Praxis Calculation Abstract thinking Perception Source: Huppert et al., (1995) 77–79 (n = 135) 9.5 25.4 21.7 5.5 10.5 1.8 5.6 8.3 80–84 (n = 191) 9.3 25.2 21.0 5.1 10.3 1.8 5.2 8.1 85–89 (n = 64) 8.8 23.5 18.9 4.7 9.3 1.7 4.4 6.8 90+ (n = 28) 8.9 23.8 18.6 4.5 8.8 1.6 4.6 6.0 All Ages 9.2 24.9 20.7 5.1 10.1 1.8 5.1 7.8 be explained by physical disorder or depression as opposed to dementia (see Jorm, 1994). Camcog Scores and Sociodemographic variables CAMCOG scores, like scores on any cognitive test with an adequate range, are markedly influenced by age, sex, education and social class. This can be seen in Table 4. For total CAMCOG score, all four variables exert a significant effect even when the other three variables have been controlled for. For CAMCOG subscales, the relationship is more complex, with age exerting a significant effect on almost all subscales while the other variables show selective effects. These findings make it clear that sociodemographic variables must be taken into account when judging whether an individual is impaired or not impaired compared to the population average. CAMCOG versus MMSE CAMCOG examines a wider range of cognitive functions than the MMSE and includes items graded in difficulty. CAMCOG total score is more normally distributed than MMSE scores (Fig. 1) and avoids ceiling effects. Figure 2 shows that elderly people who obtain maximum scores (29 or 30) on the MMSE are widely distributed in terms of CAMCOG score. Reliability Total score on the CAMCOG was found to have excellent internal reliability (Cochran’s alpha 0.82, 0.89 in different samples) and test–retest reliability (Pearson correlation 0.86). The reliability of the individual sub- Figure 1 (a) Distribution of CAMCOG scores (b) Distribution of MMSE scores CA M C O G : Detailed description 85 Table 4 Sociodemographic variables and CAMCOG performance CAMCOG score No. Sex Male Female Age (years) 77–79 80–84 85–89 90+ Education* (age at leaving school) 12/13 14 15 16 17+ Social class* Professional/managerial Skilled non-manual Skilled manual Semi-skilled/unskilled manual 137 281 135 191 64 28 52 242 52 44 26 82 74 148 103 Female (%) Mean (SD) 87.4 (9.5) 83.7 (11.5) 88.3 (9.1) 86.1 (9.8) 78.1 (13.3) 76.6 (10.7) 80.0 (12.8) 84.3 (10.2) 88.0 (10.9) 87.5 (12.3) 90.4 (7.1) 86.1 (10.9) 88.2 (10.3) 85.7 (10.1) 81.4 (12.1) Median 90 86 89 88 81 80 83 86 90 92.5 91 88 90 87.5 85 Range 52–102 31–102 48–102 55–102 31–100 51–89 31–98 48–102 51–102 52–100 67–100 51–102 48–102 51–100 31–102 67 64 68 67 82 71 64 73 66 88 68 73 62 69 * It was not possible to establish education for 2 respondents, and social class for 1 respondent. Source: Huppert et al. (1995) scales, which corresponded to different cognitive abilities, and which may sample qualitatively different processes, was also acceptable. (Pearson test-retest reliability 0.46–0.80). Reliability data are presented in more detail in Huppert et al. (1996). CAMCOG and Dementia respectively (Table 6). CAMCOG score also predicts dementia diagnosis. For each 1 point decrease in score, there is a 20% increase in the probability of dementia diagnosis. These findings are described in more detail in Huppert et al. (1996). CAMCOG scores are very effective in differentiating between demented and nondemented individuals. Huppert et al. (1996) report that in an elderly population sample the CAMCOG total score, as well as each subscale score, differed significantly between nondemented individuals and those with the diagnosis of mild dementia or minimal dementia (Table 5). As well as differentiating between groups, CAMCOG also differentiates successfully between individuals. CAMCOG total scores showed high levels of sensitivity and specificity in differentiating between non-demented individuals and those with a diagnosis of mild dementia. The cut-point which produced the highest levels of both sensitivity and specificity was 80/81, with values of 93% and 87%, Figure 2 CAMCOG score for high MMSE scorers (29 or 30) 86 Detailed description Table 5 Means (and SDs) on Cambridge Cognitive Examination (CAMCOG) and its Subscales as a Function of Dementia Severity Non-demented Subscale Orientation** Language** Comprehension** Expression* Memory** Remote Memory** Recent Memory** Learning** Attention/Calculation* Praxis** Abstract Thinking** Perception* CAMCOG** Mean 9.73 26.13 8.66 17.44 22.18 4.75 3.82 13.56 6.91 1.57 5.78 7.87 89.69 (SD) (0.57) (2.27) (0.63) (2.00) (2.67) (1.21) (0.46) (1.88) (2.14) (1.64) (2.13) (1.80) (8.49) Minimal dementia Mean 9.03 24.01 8.33 15.68 18.41 3.97 3.37 11.09 5.16 9.40 4.64 6.89 77.95 (SD) (0.96) (2.20) (0.82) (1.83) (4.20) (1.44) (0.85) (3.04) (2.38) (1.90) (2.41) (1.88) (9.72) Mild dementia Mean 7.19 22.16 7.67 14.58 14.74 2.94 2.23 9.30 4.32 8.10 2.58 6.20 65.46 (SD) (1.83) (3.75) (1.34) (2.83) (4.55) (1.47) (1.23) (3.55) (2.37) (2.11) (2.15) (1.53) (1.69) Note: Because of missing data, n = 291–322 for non-demented, 64–71 for minimal dementia and 41–53 for mild dementia. ** All group means differ, P