Joumal of Intellectual Disability Research, 1992, 36, 337-347

Evaluation of a screening instrument fordementia in ageing mentally retarded personsH. M. EVENHUIS

Hooge Burch, Institute for Mentally Retarded People, Zwammerdam, The Netherlands

ABSTRACT. The value of a standardized hetero-anamnestic questionnaire, theDementia Questionnaire for Mentally Retarded Persons (DMR), as a screeninginstrument for diagnosis of early dementia, has been evaluated during a longitudinalfollow-up of 139 institutionalized ageing mentally retarded subjects. By means ofdecision analysis, provisional criteria have been derived for a diagnosis, based on absoluteDMR scores and on scorechanges through time. For different intellectual levels, separatecriteria are required for interpretation of absolute scores, but longitudinal scorechangesmight be independent of the intellectual level. The DMR was only evaluated for themildly and moderately retarded intelligence categories, because of the small number ofdementing persons among the severely and profoundly retarded subjects. Causes of felse-positivity and false-negativity are considered.

I N T R O D U C T I O N

Dementia is difiScult to diagnose in mentally retarded people. Because memoryimpairment and beginning disorientation are often neither mentioned by the patientsthemselves, nor recognized by their caregivers, diagnosis may remain uncertain untillate in the disease (Wisniewski & Rabe, 1986; Lai & Williams, 1989, Evenhuis, 1990a).Moreover, the multiple physical or psychiatric problems of old age may additionallyhamper diagnosis.

To promote observation and recording of essential data, a standardized informant-based questionnaire, to be completed by caregivers, has been developed: the DementiaQuestionnaire for Mentally Retarded Persons (DMR) (Evenhuis et al, 1990). Sincethe questions were intended to be generally applicable to mentally retarded personswith different intellectual levels and scoring is uniform, such an instrument is expectedto be less sensitive than individual interviews of the caregivers by an experiencedphysician or psychologist. The aim of the present study was to evaluate the DMRquestionnaire as a diagnostic screening instrument. Therefore, decision analysis wasapplied to evaluate criteria for a diagnosis concerning dementia, based on DMR scores.

SUBJECTS

All residents of the geriatric pavilions of Hooge Burch, an institute for mentallyretarded people, from 1983 to 1989 («=114) were included, as well as all residents of

Correspondence: Heken M. Evenhuis MD PhD, Hooge Burch, Institute for Mentally RetardedPeople, PO Box 2027, 2470 AA Zwammerdam, The Netherlands.

337

Mild (55-70)High-moderate (45-55)Low-moderate (35-45)Severe (25-35)Profound (<25)

434333146

338 H. M. Evenhuis

Table 1. Age per intelligence category at first participation in 139 mentally retardedsubjects

Retardation (IQ) n Mean age (years)

66-1 (51-93)65-4 (48-85)65-3 (52-93)58-2 (43-86)57-8 (49-76)

Total 139 64-5 (43-93)

a geriatric sheltered home, belonging to the same institute, from 1988 to 1989 (w=25).There were 79 female residents and 60 males. Patients with Down's syndrome wereexcluded because most of them had been referred to the geriatric pavilions in a stageof advanced dementia. Fifty-three people participated during the whole period of 6years, 31 died during the course ofthis period, and 55 were admitted after 1983,including the residents of the sheltered home. On the initial evaluation, 45 subjectswere younger than 60 years of age, 57 were aged between 60 and 69, 26 were 70-79,and 11 subjects were 80 years or older. The distribution of the severity of retardation,measured by several scales (Stutsman, 1931; Dutin, 1959; Leiter, 1969), and age perintelligence category is presented in Table 1.

The prevalence at final assessment of very limited or absent mobility (32%),insufficiently corrected severe hearing loss (loss bilaterally over 60 dB, 12%) andseverely debilitating physical disease (16%) was equally distributed over the intelligencegroups, as was the use of anti-epileptic drugs (16%). The use of psychotropic drugs(mean 24%) was relatively high in the mildly and high-moderately retarded, ascompared to the more severely retarded subgroups. The prevalence of insufficientlycorrected advanced visual loss (visual acuity bilaterally <0.1, mean 17%) was relativelyhigh among the low-moderately and severely retarded. Visual and hearing functioncould not be assessed in the profoundly retarded subjects. The prevalence of severepsychiatric sjmiptoms (13%) was evenly distributed over the mildly and moderatelyretarded subgroups, but no major behavioural problems were reported in the severelyand profoundly retarded.

METHODS

Diagnosis of dementiaObservational data about each subject's mood and behaviour, changes of short- andlong-term memory, temporal and spatial orientation, recognition of familiar people,self-help and vocational skills, and activities and contacts in- and outdoors wererecorded from annual interviews of the nursing stafiFby the author. In addition, generalphysical and neurological check-ups were performed. A diagnosis of dementia wasmade by the author when a permanent and progressive decline of cognitive and social

Dementia questionnaire 339

Table 2. Diagnostic criteria for dementia (modified DSM-III-R)

A. Demonstrable evidence of decline of original level of short- and long-term memory(observed in daily circumstances)

B. At least one of the following(observed in daily circumstances):

1. disturbance of original level of spatial or temporal orientation

2. aphasia

3. apraxia

4. personality change

C. The disturbance in A and B significantly interferes with work or usual social activities orrelationships v*ath others

D. Not occurring exclusively during the course of delirium

functioning over several years of observation was observed, according to DSM-III-Rcriteria (American Psychiatric Association, 1987). These criteria had to be modified(Table 2) because of the range of the level of cognitive impairment as a part of theoriginal mental handicap and the lack of neuropsychological methods for mentallyretarded adults available in the Netherlands. The approximate time of onset of thedementia was defined retrospectively, and the probable cause was determinedaccording to NINCDS criteria (McKhann et al., 1984) and DSM-III-R criteria.

'Dubious dementia' was diagnosed when a person's social and practical functioningdeteriorated progressively, but a reliable impression of his cognitive functions couldnot be obtained (e.g. in case of severe sensory loss or psychiatric illness).

Dementia Questionnaire for Mentally Retarded Persons (DMR)

The questionnaire is a standardized interview to be completed by caregivers (Evenhuiset al, 1990) It consists of 50 items in eight subscales (Table 3). As an example, onepage of the DMR is shown in the figure. Most items have three response categories:0, no deficit; 1, moderate deficit; 2, severe deficit. Therefore, higher scores correspondto more severe deterioration. Data on former evaluations of the DMR appear in thebackground section.

The DMR was completed annually by the caregivers of each subject and wasindependent of the clinical diagnosis. Scores on single subscales, as well as the 'sumof cognitive scores' (SCS: short- and long-term memory and orientation, subscales1-3) and the 'sum of social scores' (SOS: speech, practical skills, mood, activity andinterest and behaviour, subscales 4-8) (Table 3) were recorded and stored.

Decision analysis of DMR scores in relation to diagnosis

Decision analysis offers strategies for the interpretation of diagnostic tests aridprocedures (Griner et al, 1981; Sackett et al, 1985). Generally, when a test is usedfor the purpose of screening, it must be sensitive. The test criterion that produces the

340 H. M. Evenhuis

CATEGORY

24. Is on familiar terms with one ormore persons, living in the home:0 = often1 = sometimes2 = never

25. Understands simple instructions;0 = normally yes1 = sometimes2 '= nomially no

26. Recognizes staff members whomhe/she has known for more thana year (calls them by name orrecognition is apparent frombehaviour):0 = normally yes1 = sometimes2 = normally no

27. Gets angry easily:0 = never1 = sometimes2 = often

28. Knows that today is a weekendor a weekday:0 = normally yes1 = sometimes2 = normally no

29. Can tell something about whathe/she has been doing today:0 = normally yes1 = sometimes2 = normally no

Total page 7

1 2 3 4 5 6 7

Figure 1. Example of a page of the Dementia Questionnaire for Mentally Retarded Persons(DMR).

highest sensitivity should be selected unless its specificity is unacceptably low (that is,its false-positive rate too high), resulting in a too large number of patients falselyidentified as suffering from the disease.

In the present study, the diagnosis was used against which to judge the sensitivityof DMR scores. Because the DMR is intended as a screening instrument with the

Min—maxscores

0-140-160-14

0-44

0-80-160-120-120-12

0-60

0-104

Inter-raterreliability

0-840-870-86

0-680-940-740-740-44

Dementia questionnaire 341

Table 3. Dementia Questionnaire for Mentally Retarded Persons (DMR): subscales andinter-rater reliability

Subscales

1. Short-term memory (seven items)2. Long-term memory (eight items)

3. Spatial and temporal onentation (seven items)

Sum of cognitive scores (SCS)

4. Speech (four items)5. Practical skills (eight items)6. Mood (six items)7. Activity and interest (six items)

8. Behaviour disturbance (six items)

Sum of social scores (SOS)

Total scorepurpose of selecting persons with possible dementia, cases with a diagnosis of 'dubiousdementia' in the analysis were classified as 'dementia'.Decision analysis of longitudinal scorechanges

The most important concept for diagnosing dementia in mentally retarded persons isa decline in functioning over time. Therefore, in this longitudinal study, an analysisof DMR scorechanges during follow-up has been performed. Scorechanges in theindividuals who, according to the diagnosis, developed dementia or dubious dementiaduring the present study, were compared with variations of these scores in the non-demented individuals. Persons with dementia, in whom no DMR scorings before theonset of dementia were available, were excluded. The initial scores of each subject wereused as base value and were compared to scores during the first year of dementia inthe dementing subjects and final assessment scores in non-demented subjects. Thesensitivity and specificity of the DMR were calculated for several cutoff values ofscorechanges of SCS and SOS; criteria with the highest sensitivity, accompanied bya high specificity, were determined.

Decision analysis of absolute scores

To study whether a single completion of the DMR has any diagnostic value, absolutescores were also analysed. One measure firom each subject was selected for thisanalysis. Since interest was focused on the sensitivity of the DMR in early dementia,measures during the first year of dementia were used in those with an onset ofdementia (or dubious dementia) during the study, and initial scores in those who weredementing before participating in the study. For calculations of specificity, finalassessment scores of the non-demented subjects were used. Since DMR scores aredependent on the intellectual level, as has been demonstrated in earlier studies

342 H. M. Evenhuis

Table 4. Sum of cognitive scores (SCS) in 23 subjects with dementia (dem) or dubiousdementia (dub) (the imderlined scores have been used for sensitivity calculations ofabsolute scores)

Years of (dubious) dementiaDegree of ',retardation 0 0 1 2 3 4 5 6 7 >7

Mild

18

2136 37

33

18 _44

123456789

1011

demdubdemdemdubdubdubdemdemdemdem

High-moderate12131415161718

dubdemdubdubdubdemdem

Low-moderate19202122

demdemdubdem

Severe23 dem

804

122208

120

187

15

100

2424

2

182141

012

33

123727

32

3434

3242

35

1708

4116

3021

3521

34

392730

4441241937

21

242839

3720

3740

40

_ 41 40

20 _08 21 27 37 40

30 21 2038

38 38 37 39

(background section), this analysis has separately been performed for each intelligencecategory. For diflFerent cutoff values of single subscores, SCS and SOS, the sensitivityand specificity were calculated, and criteria with the highest sensitivity, combined witha favourable specificity, were determined.

RESULTS

Diagnosis of dementia

Nine individuals met the criteria for a diagnosis of dementia at the onset of theinvestigation, and five more demented during the course of the investigation. Theprobable type of dementia was 'Alzheimer's' in eight individuals. Vascular' in four, andunclear in two cases. Only one of the severely retarded and none of the profoundlyretarded subjects demented. A diagnosis of 'dubious dementia' was made in nine

Dementia questionnaire 343

Table 5. Sensitivity and specificity of longitudinal DMR scorechanges in 111 mildly toprofoundly retarded subjects with multiple measures

Criterion

Increase of SCS >seven pointsIncrease of SOS >four pointsIncrease of SOS >seven points

Table 6. Sensitivity and specificity ofand low-moderately retarded subjects

Sensitivity

9/10 (90%)6/10 (60%)4/10 (40%)

absolute DMR scores

Sensitivity

Specificity

89/101 (89%)75/101 (75%)84/101 (84%)

in mildly, high-moderately

Specificity

Mildly retardedSCS>7 10/11 (91%) 24/32(75%)SOS>10 8/11 (73%) . 17/32(53%)High-moderately retardedSCS>15 7/7 (100%) 35/36(97%)SOS>15 7/7 (100%) 27/36(75%)Low-moderately retardedSCS >20 4/4 (100%) 19/29 (65-5%)SOS >15 4/4 (100%) 17/29 (59%)

individuals: two of these suffered from advanced Parkinson's disease and one fromprogressive neurological disease of unknown origin, one individual had a diagnosis ofsevere bipolar affective psychosis, and four showed multiple physical pathology withdepression or clouded consciousness. In one subject with a decline of short-termmemory and possibly of long-term memory, a definite diagnosis of dementia could notbe made.

Decision analysis of longitudinal scorechanges

In 124 subjects, two or more measures were available, and of these, 23 persons hada diagnosis of dementia or dubious dementia. SCS scores of the 23 individuals arelisted in Table 4. Thirteen persons with dementia were excluded from this part of theanalysis as no DMR scores were obtained before the onset of dementia. Of the 10remaining subjects with dementia, seven were mildly, one high-moderately, and twolow-moderately retarded. Scorechanges that were accompanied by the highestsensitivity were an increase of seven points or over on the SCS, and of four points orover on the SOS. The specificity resulting from application of these cutoff values ispresented in Table 5. When an increase of SCS of seven points or over is consideredas a criterion for a DMR diagnosis of dementia, one person would have been 'false-negative' (subject 7 in Table 4): a mildly retarded woman with Parkinson's disease anda diagnosis of dubious dementia, who showed only limited cognitive scorechanges forseveral years. In this subject, an increase of SCS of 12 points occurred later. However,in the present study, only scorechanges during the first year of (dubious) dementia

344 H. M. Evenhuis

were considered. Conditions that accompanied 'false-positivity', usually more than onein each subject, were severe psychiatric disease (major depression and psychosis), non-ambulancy due to advanced Parkinson's disease or arthrosis, severe visual or hearingloss and advanced malignancy or other debilitating physical conditions.

Since the 'sum of social scores' did not change in four subjects during the first yearof dementia, a DMR diagnosis based on increase of the SOS had a maximal sensitivityof 60% (Table 5). Results of the analysis of single subscores are not presented ordiscussed.

Decision analysis of absolute scores

The underlined scores of the 23 dementing subjects in Table 4 have been used forsensitivity calculations. Because of the lack of subjects with a diagnosis of dementiain the severely and profoundly retarded groups, the analysis could not be applied tothese intelligence categories. Table 6 presents the cutofiF values of SCS and SOS thatproduced the highest sensitivity, combined with a favourable specificity, for the otherintelligence categories. If in the mildly retarded subgroup an SCS of seven points orover is selected as a criterion for a DMR diagnosis 'dementia', again, subject 7 (Table4) would have been missed. An SOS of 10 points or over identified eight out of 11mildly retarded dementing subjects. The other three people (subjects 3, 4 and 7)presented with an SOS of 10 points or over at a more advanced stage of (dubious)dementia.

DISCUSSION

The purpose of the DMR is to promote structured observation of the elderly mentallyretarded, and thus, to facilitate early recognition of possible dementia. This studyshows, that the DMR meets this purpose as a screening instrument for early dementiain mentally retarded persons, even at a single completion, provided that scores arerelated to the degree of mental retardation.The results suggest that scorechangesthrough time are independent of the intellectual level. However, a standardized, limitedhetero-anamnesis can never be as sensitive as a skillful psychiatric examination, andthe DMR was not designed to replace careful medical examination.

Evaluation of longitudinal scorechange

In non-mentally retarded dementing subjects, loss of cognitive function usuallyprecedes the decline of practical skills and speech (Kaplan & Sadock, 1989). If weassume that the natural history of dementia in mentally retarded people is comparable,it is to be expected that cognitive DMR scores (SCS, subscores 1-3) will have a highersensitivity to early dementia than scores on speech, practical skills, and social behaviour(SOS, subscores 4-8) (Table 3). Moreover, since functions of the SOS are more likelyto be influenced by disturbances other than dementia, a lower specificity of the SOSis to be expected.

In the present study, scorechanges of SCS do indeed give a more favourablesensitivity-specificity relationship than scorechanges of SOS (Table 5): an SCS increase

Dementia questionnaire 345

of seven points or over includes nine out of 10 subjects during the first year ofdementia. The false positive rate was not related to the intellectual level.

Therefore, relevant scorechanges seem to be independent of the intellectual level,in contrast with absolute scores. In a former longitudinal study among persons withDown's syndrome, scorechanges accompanying dementia were the same in twoseverely retarded subjects as in four moderately retarded subjects (Evenhuis, 1990b).However, this finding has to be verified with larger numbers of dementing subjects withmoderate and severe mental retardation.

Evaluation of absolute DMR scores

In this study, absolute SCS values, also provide the most favourable combination ofsensitivity and specificity (Table 6). In the low-moderately retarded, the specificity ofthe SCS is lower than in the other intelligence categories. This is caused by the factthat about one third of the non-demendng, low-moderately retarded subjects presentedwith high scores on the 'long-term memory' and 'orientation' subscales. This illustratesthat judgment of long-term memory and orientation is diflScult in subjects with moresevere mental retardation. Although people with severe and profound mentalretardation could not be involved in this part of the analysis in the present study, dueto a lack of dementing subjects in these groups, it is to be expected that absolute DAlRscores of cogtiitive functioning in this population will have a low specificity. As a matterof fact, SCS base values of the studied profoundly retarded subjects appproximatedmaximum values.

In the high-moderately retarded group studied, the specificity of both SCS and SOSwas higher than in the mildly and the low-moderately retarded groups, reflecting alower false-positive rate. This difference can not be explained by a younger age (Table1) or lower prevalences of physical or psychiatrical handicaps in this group. The findingmight be related to the selected cut-off values: if the number of subjects with eariydementia in this group had been higher than was the case in the present study, lowercut-off scores might have had to be selected to provide a high sensitivity, andspecificities would consequently have been lower. This aspect has to be consideredfurther during the continuation of this study.

Differential diagnosis

Differential diagnoses that caused false-positivity were severe psychiatric conditions,advanced impairment of mobility or loss of visual and hearing function, and severephysical disease. In a number of subjects, not included in the presented false-positivityrate, such conditions were accompanied by temporary scorechanges. These changeshad been reversible during the studied period after, for example, admission of apsychotropic drug or improvement of mobility. Therefore, any increase of SCS ofseven points or over suggests the need for a psychiatric and general physicalexamination.

Since the prevalence of additional handicaps was high in the institutionalizedgeriatric group studied, the specificity of the DMR resulting from this study is probablylower than in the general ageing mentally retarded population. Therefore, the

346 H. M. Evenhuis

proportion of false-positives to be referred for specialist consultation will probably belower in practice than in this study.

In conclusion, the DMR is useful for early recognition of dementia and forideritification of changes in physical and psychiatric status. As such, the DMR maycontribute not only to improvement of individual patient care, but also to providinginsight in the prevalence of dementia in the mentally retarded population.

Background studies of the DMR

In two former cross-sectional studies of institutionalized ageing subjects (Evenhuiset al, 1984; Kengen et al, 1987), the inter-rater reliability (Table 3) and internalconsistency of items were tested. The low inter-rater reliability on the subscale'behaviour disturbance' (Table 3) was caused by one out of six pairs of raters. Therewas a significant correlation between the intellectual level and scores on subscales 1-5. Subjects vAxh. a diagnosis of dementia (w=8), as a group, scored significantly higheron subscales 1-5 and seven than subjects with a diagnosis of dubious or no dementia.In a prospective 3-year longitudinal study of 17 middle-aged institutionalized subjectswith Down's syndrome, the relation of scorechanges during follow-up, and a diagnosisof dementia (modified DSM-III-R criteria) (w=6), has been studied by means ofdecision analysis (Evenhuis 1990b). Because scores on single subscales insuflBcientiydistinguished 'demented' firom 'not-demented' subjects, the 'sum of cognitive scores'(SCS:short- and long-term memory and orientation) and the total score were studied.An increase during follow-up of nine points or over on the SCS as a criterion for aDMR diagnosis of dementia was coupled with a sensitivity of 6/6 and a specificity of10/11. As for the total score, an increase of 13 points or over as cut-off point produceda sensitivity of 6/6 and a specificity of 8/11.

A C K N O W L E D G E M E N T S

Mrs M.M.F. Kengen and H.A.L. Eurlings participated in the development of theDMR. The author is grateful to N.W. van der Hammen, Ph.J. Hoevenaar and MrsJ. Gaasbeek, and their nursing staff, for the completion of the questionnaires, to ananonymous referee for his or her valuable and detailed suggestions to improve boththe analysis and the presentation, and to Professor L.J.Th.v.d.Kamp and J.W.StefifelaarM.D. for their stylistic criticism. This study was financed by Hooge Burch.

REFERENCESAmerican Psychiatric Association (1987) Diagnostic and Statistical Manual of Mental disorders.

Third Edition Revised. American Psychiatric Association, Washington D.C.Dunn L.M. (1959) Peabody Picture Vocabulary Test. American Guidance Service Inc.,

Washington, DC.Evenhuis H.M., Eurlings H.A.L & Kengen M.M.F. (1984) Diagnostiek van dementie bijbejaarde

zwakzinnigen. Ruit40, 14-24.Evenhuis H.M. (1990a) The natural history of dementia in Down's S5Tidrome. Archives of

Neurology 47, 263-7.

Dementia questionnaire 347

Evenhuis H.M. (1990b) Clinical studies of Alzheimer's dementia and hearing loss in Down'ssyndrome, Ph.D. Thesis, University of Amsterdam, Amsterdam.

Evenhuis H.M., Kengen M.M.F. & Eurlings H.A.L. (1990) Dementia Questiormaire forMentally Retarded Persons (including Manual). Hooge Burch, P.O. Box 2027, 2470 AAZwammerdam, The Netherlands [Translation of: Evenhuis H.M., Kengen M.M.F. &Eurlings H.A.L. (1991) Dementie Vragenlijst voor. Zwakzinnigen. Swets en Zeitlinger, Iisse,The Netherlands.]

Griner M.F., Mayewski R.J. Mushlin A.I. & Greenland J. (1981) Selection and interpretationof diagnostic tests and procedures. Annals of Internal Medicine 94, 553-600.

Kaplan H.I. & Sadock B.J. (eds). (1989) Comprehensive textbook of psychiatry, 5th ed.Williams and Wilkins, Baltimore, MD.

Kengen M.M.F., Eurlings H.A.L., Evenhuis H.M. & BoerJ.de (1987) De Dementie Vragenlijstvoor Zwakzinnigen (DVZ):een onderzoeksinstrument voor de diagnostiek van senieledementie bij zwakzirmigen. Ruit 43, 24-30.

Lai F. & Williams R.S. (1989) A prospective study of Alzheimer disease in Down syndrome.Archives of Neurology 46, 849-53.

Leiter R.G. (1969) Leiter International Performance Scale. Stoelting Company, Chicago, IL.McKhann G., Drachman D., Folstein M., Katzman R., Price D. & Stadlan E.M. (1984)

Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group.Neurology 34, 939-44.

Sackett D.L., Haynes R.B. & Tugwell P. (1985) Clinical epidemiology. Little, Brown andCompany, Boston, MA.

Stutsman R. (1931) Mental Measurement of Preschool Children. World Book Company,Chicago, IL.

Wisniewski H.M. & Rabe A. (1986) Discrepancy between Alzheimer type neuropathology anddementia in persons with Down's syndrome. Annals of the New York Academy of Sciences All,247-60.

Received 25 March 1991; revised 21 December 1991