Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

To facilitate the diagnosis of dementia in persons with intellectual disabilities (ID), based on observations of caregivers, since 1980 the Dementie Vragenlijst voor Zwakzinnigen (DVZ) has been developed by Heleen Evenhuis, ID physician, and Margeen Kengen and Harry Eurlings, behavioral therapists, all working in De Bruggen center for people with ID, Zwammerdam, the Netherlands [1]. The Dementia Questionnaire for People with Learning Disabilities (DLD) is an English translation of this instrument. Formally known as the Dementia Questionnaire for Mentally Retarded Persons (DMR) . After many years of distribution through De Bruggen, its publication has now been taken over by Harcourt Test Publishers [2]. In this chapter, we review the development of the DMR (DLD) along with its clinical applications.

Background

In people without a preexisting cognitive impairment, the diagnosis of dementia is primarily based upon an interview with the patient and his/her family. Collected information concerns memory, orientation, thought, mood, interest and activities, self-care, speech, and practical abilities. Completed with neuropsychological assessment, and physical and laboratory assessment to exclude physical causes of deterioration, a diagnosis of probable dementia can be made in an early stage in a vast majority of cases. Our practical experience at that moment, later confirmed by research, was that in principle, dementia has in people with ID the same course and similar symptoms as in other people [3, 4]. Therefore, the diagnostic procedure should be comparable. Because neuropsychological tests, at least those available in those years, were not applicable to persons with developmental ages lower than around 5 or 6 years, we considered a careful interview of observations by the family or other carers of even more importance for a diagnosis than in other people. To help us and others ask the right questions, we decided to develop a list of items, which should be normally asked in each proxy-based interview.

DMR (DLD) Designing Process

We started with the normal way in patient interviews, designing our item list accordingly: what is the situation now, and what was it before that? Before long we were confronted with the problem, that in this population, the preexisting cognitive level varies considerably between individuals. Therefore, the current functional level will always, more explicitly than in other people, have to be compared with the former level of functioning. This can only be realized in case of continuous and capable observations by persons who are familiar with the individual person and with symptoms of dementia. However, in practice, the average carer worked no longer than 2 years with the same clients, whereas, in the 1980s, nobody had any experience with dementia. Especially memory and orientation were seldom explicitly noted. As a result, observations were always incomplete and relevant data had been unsatisfactorily recorded. We concluded that looking back did not provide us with reliable, objective information, and that we had to work the other way round: structured recording of functioning before any deterioration was apparent, and again in case of deterioration. This required questions in a “here and now” format. Moreover, they had to be formulated in such a way, that they could be answered for persons with mild, moderate as well as severe ID.

These considerations resulted in a first draft with 77 items, to be completed by a family or staff member, who was familiar with the person. The questions were primarily based on first international guidelines for dementia diagnosis [5, 6] and were originally clustered in seven clinical subscales: short-term memory, long-term memory, spatial and temporal orientation, speech, practical skills, mood and inactivity, and behavioral disturbance. Further, the choice of items was based on our practical experience with interests and communicative capacities of people with mild to severe ID. Together with the methodologist Prof. L.J.Th. van der Kamp of the psychology department of Leiden University, and his graduate student Josien de Boer, the format was completed and first evaluation studies were performed. To prevent response tendencies, the items were placed in an arbitrary sequence. The questionnaire was provided with a simple linear score system, in which the items had three response categories: 0 points, no deficit; 1 point, moderate deficit; 2 points, severe deficit. Therefore, higher scores correspond to more severe deterioration. Appendix B shows the format of questions 1–5.

The subject’s behavior during the past 2 months had to be judged. If an item could not be defined, e.g., in case of a lack of expressive capacities of the subject, this could be scored as “not to be determined” in the early version.

First Studies, Leading to Publication of the Final Version

In 1983, single completions of the first version of the DMR (DLD) were performed by pairs of two independent carers for 98 institutionalized older persons with mild to profound ID, to test the interrater reliability, internal consistency of the subscales, and the relationship of intellectual levels and scores. The interrater reliability appeared satisfactory (see below). Items that correlated insufficiently with the other items within the same subscale, were omitted, as well as items that in a majority were scored as “not to be determined” and items which discriminated insufficiently (i.e., mostly scored as “0”), leading to a final list of 50 questions (Table 3.1).

Table 3.1 Dementia questionnaire for people with learning disabilities
Full size table

As expected, a negative correlation was found between intellectual levels and scores: the lower the intellectual level, the higher the scores. Based on internal consistency outcomes, the original subscale “Mood and Inactivity” was split up into the subscales “Mood” and “Activity and Interest.” [7] In a second study, again with single completions, in two institutionalized populations of, respectively, 271 and 263 older persons with mild to profound ID, the relationship of the expert diagnosis “dementia” with DMR (DLD) scores was studied. Results of a discriminant analysis showed that the subscales “Short-term memory,” “Orientation,” “Speech,” “Practical skills,” and “Mood” discriminated best between groups with and without a diagnosis “dementia.” If scores of all individual participants were classified according to the results of the discriminant analysis, in an average of 72% of subjects a correct diagnosis was made. A correct diagnosis based on DMR (DLD) scores appeared particularly difficult in case of a severe or profound ID, extreme apathy, or clouded consciousness [8].

Psychometric Properties

Reliability

The interrater reliability was studied by measuring the Pearson correlation coefficient for the different subscales. In this stage of the development of the DMR (DLD), the subscales “Mood” and “Activity and interest” were one subscale. The correlation coefficients for the different subscales varied between 0.44 and 0.94 (Table 3.2). Only for subscale “Behavioral disturbance,” the correlation between raters was relatively low (0.44). It appeared that this low correlation resulted from differences within one of the six pairs of raters. The results for the other subscales were satisfactory [7].

Table 3.2 Interrater reli ability [7]
Full size table

“Gold Standard”: Expert Diagnosis

Because no other diagnostic instruments for dementia were available, evaluated for people with ID, a specialist diagnosis by a physician and/or psychologist with expert knowledge in the field of dementia and ID was used against which to judge the sensitivity of DMR (DLD) scores. A specialist diagnosis “dementia” was made in case of a permanent and increasing deterioration of the cognitive and social functioning, according to DSM-III-R and later DSM-IV criteria [5, 9]. These criteria had to be slightly modified (Table 3.3), because of the variance of original cognitive functioning as part of the ID. Additionally, because no or hardly any neuropsychological test methods are available to reliably assess abstract thought, judgment, aphasia, apraxia or constructive insight in this population, we omitted the criterion “disturbances of abstract thought and judgment,” whereas aphasia and apraxia could only be observed in daily circumstances.

Table 3.3 Modified diagnostic criteria for dementia (modified DSM-III-R) [7, 8]
Full size table

Sensitivity and Specificity

In two prospective longitudinal studies, the sensitivity and specificity of different criteria for interpretation of DMR (DLD) scores have been studied in older groups with Down syndrome (DS) and with other causes of ID, both for multiple and for single completions [10, 11]. In these studies, persons with a clinical expert diagnosis of “dubious dementia ” were classified as demented. The diagnosis “dubious dementia” was made in all cases of progressive functional deterioration, in which a diagnosis “dementia” could not be made according to modified DSM-III-R/IV criteria. This usually involved persons with insufficient capacities to express themselves, e.g., by severe generalized motor impairment or severe chronic depression, or persons with a beginning dementia who did not meet DSM criteria during the study, but did afterwards.

Diagnostic Criteria

The following diagnostic criterion for a diagnosis “dementia,” based on scorechange as compared with original DMR (DLD) scores, led to the best sensitivities and specificities [11].

An increase of the Sum of Cognitive Scores (SCS) of 7 points or more and/or an increase of the Sum of Social Scores (SOS) of 5 points or more, independent on the original level of ID. Results of application of this criterion are presented in Table 3.4.

Table 3.4 Sensitivity and specificity of the DMR (DLD) (95% confidence intervals between parentheses) [11]
Full size table

A sensitivity of 100% means that all cases with an expert diagnosis of dementia will be correctly identified by the DMR (DLD). A specificity of 75% indicates that 75% of persons without dementia are correctly classified as “no dementia” by the DMR (DLD). However, 25% is incorrectly classified as “dementia” (the so-called false-positives). In such cases, further diagnostic assessment usually identified a functional deterioration by other conditions. Although of course a specificity of 100% would be preferable, this is in practice realized in hardly any diagnostic instrument [12, 13]. Which specificity is acceptable, will vary per condition. For example, a false-positive diagnosis of cancer would have to be avoided as much as possible. However, in the case of dementia in persons with ID, a specificity of 75% is acceptable. Indeed, in a majority of cases with incorrect diagnoses of dementia, further diagnostic assessment resulted in relevant and often treatable other diagnoses (severe sensory impairments, severe motor impairments, severe physical disease, and psychiatric conditions). As a conclusion, with the DMR (DLD), functional deterioration as a result of cognitive as well as noncognitive aspects is identified. Longitudinal judgment of scorechanges is more reliable than single completion and is therefore preferable.

Results of the last evaluation suggested that the DMR (DLD) is less accurate in case of specific causes of dementia, other than dementia in Alzheimer disease (DAD) (e.g., vascular dementia). However in this stage, such a conclusion can only be speculation because of the small subgroups.

Judgment by Committee on Test Affairs Netherlands

The quality of the Dutch DMR (DLD) has been recently rated by the Committee on Test Affairs Netherlands (COTAN) of the Dutch Institute of Psychologists. The purpose of these ratings is twofold. Test users are informed about the quality of available instruments, which information can help them in choosing an instrument. Besides, the ratings supply feedback to test-developers about the quality of their products. An English translation of the rating procedure has been published in the International Journal of Testing, 2001, pp. 155–182. Outcomes for the DMR (DLD) (2B.13 DVZ) were as follows: theoretical basis and soundness of test development procedure, satisfactory; quality of testing materials, good; comprehensiveness of the manual, good; norms, satisfactory; reliability, satisfactory; construct validity, satisfactory; criterion validity, satisfactory.

Applications of the DMR (DLD)

Dementia

The DMR (DLD) has been designed in principle for the diagnosis of dementia in adults with ID. However in practice, because DAD is the most prevalent cause of dementia, we have primarily evaluated the sensitivity for DAD. Due to small subgroups, the sensitivity for rarer types of dementia has been evaluated insufficiently.

Early Detection

Our longitudinal evaluation shows, that in all cases, a diagnosis based on DMR (DLD) scores was made prior to or at the same time as an expert diagnosis according to international criteria could be made (DSM-III-R/DSM-IV).

Screening Instrument and Effect Instrument

We stress that the DMR (DLD) is not an instrument for a definite diagnosis of dementia, because severe progressive physical and other psychiatric conditions, or a combination of less severe conditions, may influence the scores as well. Therefore, the DMR (DLD) has to be used as a screening instrument, i.e., for selection of persons for further specialist diagnostic assessment. Recently, the instrument has been proven satisfactory to evaluate effects of interventions [14, 15].

Repeated or Single Completion

The basis for a diagnosis of dementia is always a deterioration from the former individual level of cognitive functioning. Indeed, the DMR (DLD) is most sensitive in case of multiple measures.

Originally, we have also tried to develop criteria for a single completion of the DMR (DLD), which would simplify large-scale screenings, e.g., in connection with research projects. This is only possible under the condition that reliable and interindividually comparable data from former intelligence tests, performed prior to any deterioration, is available. In our own evaluation studies, the participants’ level of ID had been ascertained with several tests: Stutsman Mental Measurement of Preschool Children [16], Peabody Picture Vocabulary Test [17], and Leiter International Performance Scale [18]. The results may not be completely comparable to other scales, used nowadays and in other countries to test functional levels. Therefore, a diagnosis based on a single application of the DMR (DLD) is now considered insufficiently valid, and is strongly discouraged by us.

Criteria for Persons to Be Tested

The DMR (DLD) is applicable to persons with mild, moderate, or severe ID (developmental ages around 2–10 years). It is not applicable to persons with profound ID (developmental age lower than 2 years) and to persons with severe ID (developmental age 2–3 years) combined with severe other disabilities, such as motor impairment or hearing loss. In such cases, DMR (DLD) scores may approach extreme levels before any functional deterioration (“ceiling effect ”).

Who Answers the Questions?

The questionnaire has to be completed by a family or staff member who is familiar with the person. Carefulness and objectivity are very important. This may be advanced by DMR (DLD) completion not by a single person, but by a family member together with a staff member, or by several carers together, and preferably guided by the investigator.

Who Interprets the Answers?

Interpretation of the results is only useful in combination with other diagnostic data, as applies for each diagnostic instrument. Therefore, this should be done by the diagnosing physician, psychologist, or behavioral therapist.

Directions for Diagnostic Use

Because longitudinal judgment of DMR (DLD) scores provides the most reliable diagnosis, it is advised to routinely perform a first scoring of the DMR (DLD) before any functional deterioration is observed. This might be done when somebody moves to a home for several persons with ID, or joins a day activity center. Any observed deterioration should prompt repeated completion of the DMR (DLD). If no scorechange is found, consistent with a diagnosis of dementia, further diagnostic assessments are to be aimed primarily at other causes of deterioration, such as a depression or sensory impairment. Dependent on the development of symptoms, a next DMR (DLD) scoring and judgment is advised after 6–12 months.

In case of a DMR (DLD) diagnosis “dementia,” referral for specialized psychiatric and general physical examination is advised, according to national or international guidelines [19,20,21,22]. In any case, visual and hearing functions are to be actively tested, because of increased risks of age-related sensory impairments in this population, which are missed in many persons with ID [23, 24].

Rating

The questionnaire is provided with a simple linear score system, in which the items have three response categories: 0 points, no deficit; 1 point, moderate deficit; 2 points, severe deficit. The subject’s behavior during the past 2 months has to be judged. If an item cannot be defined, e.g., in case of a lack of expressive capacities of the subject, the score has to be “2”.

The items are clustered in eight subscales (Table 3.1) and placed in an arbitrary sequence, to prevent response tendencies. Combined scores on the first three subscales (short-term memory, long-term memory, and orientation) are indicated as the SCS . Combined scores on subscales four through eight (speech, practical skills, mood, activity and interest, and behavioral disturbance) as the SOS . The questionnaire is provided with a short instruction for completion. Completion takes 15–20 min.

Other Studies of the DMR (DLD)

Since the availability of an English translation of the DMR (DLD), it is clinically used in many countries around the world. Several researchers have evaluated the DMR (DLD) for their country, or used it in epidemiological or intervention studies.

The DMR (DLD) in Diagnostic Test Batteries

Since the 1990s, other diagnostic instruments, both informant-based and to be administered directly to persons with ID, have been applied or developed to assess for dementia. Most of these tests are aimed at specific symptoms, such as maladaptive behavior, memory decline, or verbal fluency, or are specifically designed for persons with DS. Combinations of such tests in diagnostic batteries have been recommended by several groups [25,26,27,28]. The DMR (DLD) in all cases was presented as the most promising informant-based screening tool in most adults with ID, including those with DS. It is the only informant-based scale available for assessing orientation [27].

Evaluations of the DMR (DLD)

Evaluations by other authors concern mostly single completions of the DMR (DLD), referencing to Intelligence Quotient (IQ) levels. It appeared that such results were less satisfactory than in our own evaluations, probably due to application of varying tests for IQ or functional levels, or other criteria for levels of ID. For this reason, Prasher proposed for persons with DS in the United Kingdom modified higher cut-off scores for single DMR (DLD) scores [29].

Burt and colleagues [30] in the United States, specifically evaluating assessment of orientation in 138 adults aged 29–82 years, found fair to good agreement between DMR (DLD) scores on the subscale “Orientation” (single ratings) and direct assessment. The level of agreement was negatively influenced by lower functioning, DS, and higher age.

Deb and Braganza [31] in the United Kingdom compared ratings on several informant-based scales with the clinician’s diagnosis among 62 adults with DS. The diagnosis according to DMR (DLD) criteria (single ratings) showed sensitivity and specificity at the 0.92 level for both categories. In this study, the observer-rated scales appeared more useful for the diagnosis of dementia than the used direct neuropsychological test.

Silverman and colleagues [32] performed a study of dementia in 273 adults with ID, applying multiple tests 18 months apart. As opposed to our own findings, single ratings of the DMR (DLD), referencing to IQ measurements with Wechsler Adult Intelligence Scale and Stanford–Binet scales earlier in adulthood, distinguished more effectively between individuals with and without dementia than scorechanges during the study period. Sensitivity of scoreschanges over the 14–18 month period was less impressive than reported in the DMR (DLD) manual. However, we suspect that in this study, the dementia process in a number of cases might have started before the first rating. As a result, no predementia baseline data were available, as is recommended in the manual. The authors recognize this: “It might be worthwhile examining change in DMR (DLD) scores for incident cases for whom a predementia baseline is available, and to rely more on single assessment scoring otherwise.” In this study, effects of different IQ tests were also studied. Indeed, it appeared that the IQ testing procedure had a significant effect on classifications of nondemented participants (p < 0.05) and a nonsignificant effect in other dementia status groups, but the power was low.

Shultz and colleagues [33] in the United States and Canada evaluated several screening tools for dementia in a case-control study, with 38 matched participants with mild to profound ID in each group. Again, single ratings were used for the DMR (DLD), referencing to IQ measurements that were at least 5 years old, obtained with a variety of methods. Paired t-tests for both SCS and SOS ratings were highly significant, without correlating to gender, age, IQ level, or DS. In a logistic regression analysis of all tests used, the DMR (DLD) SOS was the variable that best predicted group membership.

Recently, Walker and colleagues [34] carried out the DMR (DLD) interview independently with 2 carers caring (at least 6 months) for 26 people with Down syndrome. Fifteen males, 11 females, mean age 50.5 years (range 40–69 years). Only 15% of the pairs of informants had good agreement. Better agreement for less abled participants. The authors recommended were not to rely only on carer interviews when assessing for dementia in persons with ID.

The DMR in Intervention Studies

Prasher and colleagues [14, 15] used DMR (DLD) scores as the primary outcome measures in a 24-week randomized controlled trial (RCT) of the cholinesterase inhibitor donepezil. The study group consisted of 27 persons with DS and mild or moderate DAD. There was a tendency that donepezil halted the rate of decline, but the sample size was too small for statistical significance. The trial was continued as an open-label study until a total of 104 weeks. Long-term use of donepezil significantly reduced the rate of decline (p < 0.001). A comparable 24-week effect study of rivastigmine has also been published by Prasher and colleagues [35]. Prasher concludes that the DMR (DLD) is sufficiently sensitive to measure scorechanges as a result of intervention (personal communication 2004).

An uncontrolled evaluation of treatment with different cholinesterase inhibitors in a network of specialist memory clinics for people with ID in Southwest England was recently reported [36]. Here too, the DMR (DLD) was used to monitor intervention effects, showing a significant deterioration of total scores in the last two assessments before treatment (p < 0.01), during a mean interval of 10.8 months. Treatment seemed to stabilize scores during a mean period of 7.4 months, whereas the SOS showed a significant improvement (p < 0.05).

Summary

Recently, the DMR (DLD) has been rated satisfactory to good by the COTAN. From secondary studies by other authors, we conclude that the DMR (DLD) is high-ranking in recommendations for diagnostic batteries [25, 26, 28]. Apart from cognitive items, it also scores noncognitive items. It is the only informant-based scale for assessment of orientation [27]. Authors use preferably single DMR (DLD) ratings, requiring reliable IQ levels for referencing [29, 31,32,33]. In that case, the choice of IQ tests or tests for functional levels may negatively influence sensitivity and specificity, because different tests lead to different dementia classifications based on the DMR (DLD) [32] Nevertheless, results in these studies are promising. Corroborated by the findings of Silverman and colleagues [32], we stress again that a sensitive DMR (DLD) diagnosis based on score-changes requires baseline ratings prior to onset of dementia and not during dementia. The DMR (DLD) is a sensitive instrument to monitor changes as a result of intervention [14, 15, 35].

Our own evaluation studies have shown that the DMR (DLD) is not sensitive in persons with profound ID, because of a “ceiling effect.” To our clinical experience, there is a “bottom effect,” too: in persons with very mild or borderline ID and beginning dementia, it may take years before DMR (DLD) scores reach the level of a dementia diagnosis. Apparently, the DMR (DLD) is not sensitive to more subtle functional deterioration, and the questions have been designed with capacities of people with moderate and severe ID (developmental ages 2–6 years) in mind.

During our evaluation studies, the DSM-III-R was replaced by the DSM-IV [9]. Did this influence the validity of the DMR (DLD)? In the DSM-IV, some of the former clinical criteria for a diagnosis of dementia were omitted, namely “disturbances of abstract thought in judgment” and “personality change.” According to the DSM-IV, deterioration from the original level of functioning has to be more explicitly taken into account. The only change in our modified criteria would therefore be the absence of the criterion “personality change” (Table 3.3). Because this aspect in practice has hardly played a decisive role in our specialist diagnoses, it is not to be expected that outcomes of our validity studies would have shown relevant changes by applying DSM-IV instead of DSM-III-R criteria.

In 1995, we participated in an international consensus group for diagnosis of dementia in people with ID, which advocated application of ICD-10 rather than DSM-IV criteria in this population [25, 26, 37]. The reason was, that, as compared to the DSM-IV, in the ICD-10 more emphasis is placed on noncognitive aspects of dementia (e.g., emotional lability, irritability, and apathy). In practice, these noncognitive aspects are often the first signs, reported in individuals with ID, rather than cognitive aspects. The consensus group concluded that in this way a “two-step” diagnostic procedure is introduced, in which a possible diagnosis of dementia will be reconsidered, if observed behavioral changes are not accompanied by evidence of cognitive decline. It was seen as an advantage that in this way, consideration of all possible causes of decline is required, including of those that are treatable. These recommendations are in line with the more recent recognition of the role of psychiatric and behavioral disorders in dementia syndromes in clinical research in the general population [38]. Aylward and colleagues [26] observed that ICD-10 and DSM-IV overlap completely on the part of cognitive decline. The DMR (DLD) was cited as a reliable method to detect a decline in memory and other cognitive abilities, a decline in emotional control or motivation, or a change in social behavior. Indeed, with the second part of the DMR (DLD), a range of noncognitive aspects can be assessed, among which the aspects, mentioned in the ICD-10.

We conclude that the distinction of “dubious dementia” and “dementia” in the expert diagnosis in our DMR (DLD) studies is in fact comparable to this “two-step” procedure. Our choice to classify “dubious dementia” as “dementia” for the assessment of sensitivity and specificity is in line with the considerations of the international consensus group. Therefore, it may be assumed that evaluation of the DMR (DLD) against a clinical diagnosis according to ICD-10 criteria would have resulted in comparable outcomes.