There are as yet no safe and effective forms of treatment to improve cognition and memory in the rising number of patients suffering from Alzheimer's disease (Growdon, 1992). Cognitive training programs may be therapeutic, especially when combined with phar- macological, psychosocial, and psychotherapeutic strategies. There are only a few case reports that show a few, short-term benefits of cognitive training relating specifically to trained cognitive subfunctions in patients suffering from Alzheimer's disease (Cameron & Stevens, 1990). Since conventional cognitive training is based mainly on abstract tests, the lack of relevance to everyday life has been an obvious target of criticism. Accordingly, we developed and tested a computer-based cognitive training program specially tailored to individual patients, employing personal and biographical material that relates directly to the patient's local and social environment and that is of personal importance to the patient, e.g., pictures of the patient's surroundings, home or usual shopping route, or photographs of the patient at an earlier age (Hofmann et al., 1995; Hofmann et al., 1996). The training involves three major elements depending on the stage of impairment: (1) social competence in patients with initial deficiencies, simulating, for example, a shopping situation; (2)
Correspondence to: Dr M. Hofmann. Tel: +41 61 325 51 11; Fax: +41 61 325 52 58.
493
494 M. Hoffmann et al.
orientation in patients with moderate impairment; and (3) emotional aspects in severely demented patients using, for example, biographical material.
Subjects and Methods
Ten patients on a specialized psychogeriatric research ward, each with a diagnosis of probable Alzheimer's disease according to DSM-III-R (American Psychiatric Association, 1987) and NINCDS-ADRDA criteria (McKhann et al., 1984), participated in the training program on a voluntary basis once informed consent had been given by themselves and their carers. All the patients had undergone extensive diagnosis, including laboratory tests, lumbar puncture, and MRI. Apart from signs of global cortical atrophy in all patients and signs of bilateral atrophy of the temporal lobe in six patients, there were no significant pathological findings. Each patient's cognitive performance was characterized at baseline (week 0) with the CDR (Hughes et al., 1982) and psychometric texts comprising SIDAM (Zaudig et al., 1989), MMSE (Folstein et al., 1975), Digit-span (Wechsler, 1987), the Trail- making test (Raitan, 1956), and immediate and delayed recall and recognition of words (Wechsler, 1987) (German versions in NAI, Oswald & Fleischmann, 1995). The MADRS (Montgomery & Asberg, 1979) was used to measure depressive symptoms and the IDDD (Interview for the Deterioration in Daily Living Activities in Dementia) (Teunisse & Derix, 1991) to reflect ADL functions. The MMSE, MADRS and IDDD were repeated at post- tests I and II. The patient's characteristics, clinical symptoms, baseline test results, and the training programs are shown in Table 1.
Each task was specified in consultation with the patients, all of whom lived in their own homes and had a close relative or other carer to look after them. Four patients (two LO and two EO) were trained in programs individually tailored to them, employing photographs of their own local and social environment. Six patients (three LO and three EO) were trained in a complex computer program simulating a walk into the center of town and also involving social-competence tasks and tests of orientation and memory. Between 50 and 150 photographs were taken for each training program, illustrating successive stages in a series of scenes and tasks, scanned into the memory of an IBM-compatible 486-PC and integrated into a presentation-software program. The pictures were displayed on a special large color monitor and allowed the interactive operation of the progam via touch-screen functions. The patients were instructed to "move through" the simulated scenes step-by- step by touching the correct area on the screen (e.g., a picture) and to complete tasks such as shopping or to answer questions (by touching multiple-choice answers), with as little assistance as possible from the instructor. At the same time, patients were encouraged to be as fast as possible by avoiding misleading stimuli. The time required for the tasks, the number of mistakes, and the number of times the test instructor gave advice were all recorded. Baseline performance (pre-test) was established following one introductory train- ing session. Training frequency was three or four times a week. The training phase lasted three weeks and was immediately followed by a first post-test (I). A second, delayed, post- test (II) was carried out after three weeks, without training, to see whether the training had had any lasting effect. At each point, a set of psychometric tests and scales documented cognitive performance, mood, and competence in daily activities.
Computer Training in AD Patients 495
[--
c~
{
o
,.~ ~} o ~ ~ ~ .o
~_~ ~ - ~ .~_~ o ~
..~.~ ~.6 - : .~ ~ '~ ~ -~o ;-.{~ o ~ ",: o o o ~ ~..~
• . ~ ~ ~ - - ~ ~ ~ = " ~
~.~. ~} ~ ~ ~ ~ ., ~ ~,~ ~.-
~ . - ~ . ~ : ~ ~ < ~ ' ~ _ ~ " ~ .~
.o ~ o o ~ , _ ~ " ~-'" ~'- ~ ~ I- ~ O ~ o ~ ~ o o . . o • ~ ~ ~ : ~ < ~ ~ a ~ ~ o . ~ a ~ - : ~
.~_.~ ~.~..o ~ -~.8. ~ ~ o ~ ~ ~ o • -= ~ ~ ~ ' ~ s o ~ ~ o- ~ . ~ e~.~. ~ : ~
Z }- Z ~ ~ ~ -~ ~:~ ~ ~ ~ " = '~ ° ~ = ~ ~ ~
eq o o e~l o o ~o o
:~ -~ .~.
" ~ 0 .~ . 0 ~
~ o . ~ & ~ "~ ~ .~ o ~'.~
~ 2 ~ ^ _ ~
~-~ ~:~ ~-~ ~,~ ~ -~- c~ ~ ~Ir ~
0 0 0 0 0 0 0 0 0 0
.~ B "" Ii
~ o ~
, .o~.
• < ~ o ~ ~ o
o
~ =
~ o . ~ .
• ~ ~ ._~ -
~ - - ~ 8
496 M. Hoffmann et al.
During the third week (post-test II), we gave the patients and carers a questionnaire that we had developed to evaluate subjective results and the degree to which the training had been accepted. (The questionnaire has not been published.) Patients and carers were asked to rate the following three questions (the version in brackets is the one addressed to relatives and carers):
1. Do you think this training has improved your (your relative's) memory? 2. Could you (your relative) transfer to real-life situations what you (he/she) have (has)
been training to do on the computer? 3. Did you like the training? (How much, do you think, did your relative like the training?)
On a scale of seven, each question could be rated from "Not at all" (= - 3) to "Indifferent" (= 0) and "Very much" (= + 3), with a possible combined minimum score o f - 9 points and a combined maximum of + 9 points.
Statistical analysis was carried out with two-way analysis of variance (ANOVA) for repeated measures based on Greenhouse-Geisser correction degrees of freedom (the original degrees of freedom are reported). Post-hoe comparisons were performed by Fisher's Pro- tected Least Significant Difference (PLSD) test (p at least < 0.05). Analyses of covariance were conducted on the pre- and post-test scores, using the pre-test results as covariate (ANCOVA). The Pearson's correlation coefficient was calculated between the patient's and the care-giver's questionnaire rating.
Results
Mean (standard deviation) age was 69 y (13.74) and mean duration of the disease 4.25 y (1.4). The severity of the disease ranged from mild (0.5-1) to moderate (2) according to the CDR-Staging (Hughes et al., 1982). Mean and standard deviations (in brackets) for the psychometric tests at the baseline assessment were: MMSE (n = 10): 19.4 (4.0); SIDAM (n = 10): 33 (9.4); Digits forward (n = 8): 5 (1.1); trail-making test (n = 8): 61.3 (22.2) sec; immediate recall (n = 8): 2.6 (1.4); delayed recall (n = 8): 1.6 (2.1); immediate recognition (n = 8): 3.4 (2.7) and delayed recognition (n = 8): 2.8 (2.2). The MADRS ratings ranged from 1 to 12 (mean: 7.7, std dev: 3.1) points, showing no major depressive disorders in this sample (see also Table 1).
Patient eight did not undergo post-test II for compliance reasons (missing data), meaning that nine patients (4 LO, 5 EO) were left for the variance analyses of training variables ("mistakes", "time", "advice"). Comparing the two patient groups [two-way-ANOVA for repeated measures with one grouping factor (EO vs LO) and one repeated factor (weeks)], EO-AD patients only showed a tendency to a more prominent improvement in mistakes [EO vs LO: F (i. 7) = 0.48, n.s.; weeks: F (2,J4) = 5.09, p = 0.06; EO/LO x weeks: 5.09, p = 0.06] in comparison with the LO-AD patients. Therefore, the two groups (EO/LO) were combined and analysed by one-way ANOVA. There was a tendency for a reduced number of mistakes after weeks three (post-test I) and six (post-test II) ["Mistakes": F (1, 2) = 4.2, p -- 0.07]. The reduction in time and advice required was significant after weeks three and six (PLSD-test) ["Time": F ~z 16) = 12.84, p --- 0.006; "Advice": F (2,16) -- 15.01, p = 0.002].
ANCOVA with pre-test scores as covariates revealed significant interaction terms of the
Computer Training in AD Patients 497
Mistakes
mln
Time needed Advice needed
week 0 week 3 week 6 week 0 week 3 week 6 week 3 week 6 pre-test post- test I post- test II pre-test post-test I post-test II post-test ! post-test II
n 6O
So
40
30
20
10
0 week 0
pre-test
Figure 1.
covariate with: "mistakes" ["Mistakes x cov": F ~2, 14) = 38.17, p = 0.0003], "t ime" ["Time x coy": F ~2, 14~ = 106.29, p = 0.0001] and "advice" ["Advice x cov": F ~2, ~4~ = 14.88, p = 0.0049]. Median division of the pre-test results indicated that patients with higher scores at week 0 (pre-test) showed a more prominent reduction in these variables at post-test I and II, compared to those with lower initial scores (see also Figure 1).
Table 2 depicts "mistakes", "t ime" and "advice" and MMSE, MADRS and IDDD- scores. There were no significant changes in these scores when calculated with one-way ANOVA (Table 2) or ANCOVA, with pre-test results as covariates (data not shown). As shown in Figure 1, nearly all patients achieved a marked reduction of the training variables during the training period. Only two patients made more mistakes at post-test I (five compared to one and four compared to two) but got faster (9 min compared to 25 min and
Table 2. Training Variables and Psychometric Scales
Variable Pre-test Post-test I Post-test II P-Value (G-G)
n = 9 n = 9 n = 9 Mean (Std, Dev.) Mean (Std. Dev.) Mean (Std. Dev.)
MMSE: Mini-mental State Examination (Folstein et al., 1975); MADRS: Montgomery and Asberg Depression Rating Scale (Montgomery & Asberg, 1979); IDDD: Interview for Deterioration in Daily Life in Dementia (Teunisse & Derix, 1991).
498 M. Hoffmann et al.
10
8
6
4
2
0
-2
~Patient
I-~Care-giver
Figure 2.
6 min compared to 15 min) and needed less advice (6 compared to 22 and 2 compared to 15).
The ratings by the patients and care-givers for the three acceptability questions are shown in Figure 2. Each column of the histogram represents the combined scores of the three questions for each single patient and his or her care-giver. There was no correlation between the patient's and the care-giver's ratings (r = 0.26, N = 10, n.s.).
Discussion
This pilot study addresses the question of computer-based cognitive training in patients suffering from Alzheimer's disease. It focuses on a new kind of training: interactive computer programs either individually tailored to the patient's circumstances and social environment, or a more general and complex simulation of everyday relevance, including social-com- petence tasks and tasks of memory and orientation. In accordance with B~ickmann's criteria for positive features of memory training in AD, this approach integrates the patient's personal (social and local) environment and can be adjusted to take account of the degree of impairment (B~ickmann, 1992).
The data collected in this study needs to be interpreted carefully, since it involves only a small and heterogeneous sample: the group consisted of early- and late-onset patients in different stages of the disease who were trained using five different programs.
Despite these limitations, a quite robust pattern of results emerged as far as the patient's performance was concerned: eight out of ten patients showed fewer errors after training, and all performed the tasks faster and required less assistance. These results were more pronounced in patients with higher scores at the beginning of the exercise, and there was no difference between LO- and EO-AD patients. Only two patients made more mistakes during the actual training than they had done at the beginning, but they became faster and worked more autonomously - - which could be interpreted as a change of work strategy or as a new way of coping with the task.
Other than the PoweU-Procter and Miller results, which showed training to have only
Computer Training in AD Patients 499
very short-term effects (Powell-Procter & Miller, 1982), our better post-test II findings suggest that most patients could maintain some of the trained skills over a three-week period without further training. One possible hypothesis to explain this is that the touch- screen program operation draws, to a certain degree, on motoric-automatized operations involving procedural memory processes. It has been suggested that motoric and procedural (implicit) memory processes may be relatively unaffected in AD due to a different anatomical localization (Eslinger & Damasio, 1986), and it is assumed that procedural learning depends to a much lesser degree than explicit learning on intact structures in the temporal lobe and hippocampal area. Karlsson proposed a positive interaction of the two learning processes, showing that a motor action during learning improved cued recall of sentences in AD patients (Karlsson et al., 1989). Thus, in addition to motivational factors, the motor gesture of touching the monitor screen might also have helped the patients to learn explicit items.
With regard to more general effects such as the influence of the training on "orientation", "social competence", "mood" and "quality of life" methodological aspects have to be discussed. First, since it is almost impossible to evaluate such effects directly, psychometric scales were used, as were formal and informal information in the form, respectively, of our questionnaire and anecdotal remarks and comments provided by the patients and their carers. Second, these scales were rated by the same people who conducted the training sessions, meaning that bias on the part of the person making the rating could not always be ruled out.
We chose the MMSE to quantify effects on "orientation". With one third (10 out of 30) of its questions addressing orientation in time and space, the MMSE seemed, at first sight, a useful tool, but, as it turned out, it was not a suitable method for reflecting the more complex competence of situation-associated spacial orientation in which we were trying to train our patients. Since the MMSE is not sensitive to small cognitive changes and is not validated for repeated measurement, we lack the data to address the issue whether or not more general cognitive changes may be induced by such training. In the first trial using this method (Hofmann et al., 1996), four patients who received individual training programs were repeatedly assessed with a more sophisticated psychometric battery including SIDAM, MARDS, SKT (Syndrome Short Test), digit-span, immediate and delayed recall and recognition, and the trail-making test - - without significant changes in the course of the training. We then assumed either (1) that the training was too limited to have any influence on general cognition; or (2) that even sophisticated psychometric testing cannot reveal small effects against the background of a disease whose rate of progress is unknown in each individual case. Similar considerations probably also apply to the present research.
We could not measure improvements in "quality of life" or competence in "daily living activities" using the IDDD scale. This preliminary data set does not yet allow us to conclude whether (1) there are any effects; (2) the effects are too small and unspecific; or (3) the scale is inappropriate. For the time being, therefore, we can conclude only that the patients were able to handle successfully simulations on a PC screen that came close to real-life situations, but we lack evidence to show that the effects of this training can be transferred to real-life situations and that they are of relevance to the patient's competence in daily life.
None the less, carers provided anecdotal reports about the transferral of the trained task to real-life situations: in a "reality test", for instance, Patient two found his way to his
500 M. Hoffmann et al.
carer ' s flat wi thout any p rob lem, a task in which he had been previous ly t ra ined using the touch-screen. But on the way back - - a task in which he had not been t ra ined - - he lost his way. To learn more a b o u t the t ransfer ra l to real s i tuat ions, " rea l i ty tes ts" will be used in our next s tudy to assess the effects o f t ra in ing more directly.
The pat ients ra ted occupa t ion with fami l ia r issues as mot iva t ing and emot iona l ly reward- ing. In par t icu la r , the four pa t ien ts t ra ined with indiv idual ized p r o g r a m s were highly mot iva ted , since they were conf ron ted with personal , eve ryday tasks. The six pa t ien ts t ra ined by means o f the more general and complex s imula t ion also easily succeeded in es tabl ishing a feeling o f persona l relevance. Even i f an uncri t ical ra t ing on the pa t i en t ' s side canno t be ignored, it is the carer ' s ra t ings tha t are p r o b a b l y more realist ic and tha t cer ta inly reflect a posi t ive evalua t ion . This, in our view, permi ts us to conc lude tha t the pa t ien ts l iked to use the c o m p u t e r with this pa r t i cu la r in teract ive software. By recognizing their own improved pe r fo rmance as a resul t o f their t ra ining, some par t i c ipan t s seemed to prof i t f rom a sense o f achievement which was, as one pa t ien t commented , "qu i te different f rom the feeling o f get t ing worse in every o ther aspect o f life."
Acknowledgements--We would like to thank Mr Markus Fischer for his help in programming and Mr Kurt Kr~iuchi for statistical advice.
References
American Psychiatric Association (1987). Diagnostic and statistical manual of mental disorders, 3rd revised edition. Washington, DC: American Psychiatric Association.
B~ickmann, L. (1992). Memory training and memory improvement in Alzheimer's disease: rules and exceptions. Acta Neurolooica Scandinavia, 139(Suppl.), 84-89.
Cameron, J. C., & Stevens, A. B. (1990). Spaced-retrieval: a memory intervention for dementia of the Alzheimer's type. Clinical Gerontologist, 10(1), 58-61.
Eslinger, P. J., & Damasio, A. R. (1986). Preserved motor learning in Alzheimer's disease: implications for anatomy and behavior. Journal of Neuroscienee, 6, 3006--3009.
Erzigkeit, H. (1986). Der SKT zur beurteilung therapeutischer effekte nootroper substanzen. Proceedings of the 3rd Symposium on Neurotropics, 3-5 March 1986 Dresden.
Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12, 189-198.
Growdon, J. H. (1992). Treatment for Alzheimer's disease?. The New England Journal of Medicine, 327(18), 1306- 1308.
Hofmann, M., Hock, C., K~ihler, A., & Mflller-Spahn, F. (1995). Computergest~tztes individualisiertes Ged- ~ichtnistraining bei Alzheimerpatienten. Nervenarzt, 66, 703-707.
Hofmann, M., Hock, C., & M~iller-Spahn, F. (1996). Computer-based cognitive training in Alzheimer's disease patients. In J. H. Growdon, R. Nitsch, S. Corkin, J. Wurtman (Eds), The neurobiology of Alzheimer's disease, annals of the New York Academy of Science, 777, 249-255.
Hughes, C. P., Berg, L., Danzinger, W. L., Coben, L. A., & Martin, R. L. (1982). A new clinical scale for the staging of dementia. British Journal of Psychiatry, 140, 566-572.
Karlsson, T., B~ickmann, L., Herlitz, A., & Nilsson, L. G. (1989). Memory improvement at different stages of Alzheimer's disease. Neuropsychologia, 27, 737-742.
McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D. and Stadlan, E. M. (1984) Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA work group under auspices of the Department of I-l'ealth and Human Service Task Force in Alzheimer's disease. Neurology, 34(7), 939-944.
Montgomery, S. A., & Asberg, M. (1979). A new depression scale designed to be sensitive to change. British Journal of Psychiatry,, 134, 382-389.
Powell-Procter, L., & Miller, E. (1982). Reality orientation. A critical appraisal. British Journal of Psychiatry, 140, 457-463.
Oswald, W. D., & Fleischmann, U. M. (1995). Ntlrnberger-Alters-Inventar (NAI) 3., ~iberarbeitete und erg~inzte Auflage, NAl-testmaterial und textband. Hogrefe Verlag fflr Psychologie. GOttingen, Bern, Toronto, Seattle.
Computer Training in AD Patients 501
Raitan, R. M. (1956). Trail-making test. Manual for administration, scoring and interpretation. Indianapolis: Indiana University Press.
Teunisse, S., & Derix, M. M. (1991). Measurement of activities of daily living in patients with dementia living at home: development of a questionnaire. Tijdschr-Gerontol-Geriatr, 22(2), 53-59.
Wechsler, D. (1987). WMS-R: Wechsler Memory scale - - revised (manual). San Antonio: The Psychological Corporation.
Zaudig, M., Mitterhammer, J., & Hiller, W. (1989). Structured interview for the diagnosis of dementia of the Alzheimer type, multi-infarct dementia and dementias of other etiology according to ICD-10 and DSM-R- III - - SIDAM. Munich: Logomed.
