ASSESSMENT OF PATIENTS WITH MEMORYPROBLEMS USING A NURSE-ADMINISTEREDINSTRUMENT TO DETECT EARLY DEMENTIA

AND DEMENTIA SUBTYPESMICHAEL DENNIS1*, LESLEY FURNESS2, JAMES LINDESAY3 AND NEIL WRIGHT4

1Senior Lecturer and Honorary Consultant Psychiatrist, Division of Psychiatry for the Elderly,Leicester General Hospital, Leicester, UK

2Research Community Nurse, Division of Psychiatry for the Elderly, Leicester General Hospital, Leicester, UK3Professor of Psychiatry and Honorary Consultant Psychiatrist, Division of Psychiatry for the Elderly,

Leicester General Hospital, Leicester, UK4Lecturer, Department of Psychiatry, Leicester General Hospital, Leicester, UK

SUMMARY

Background. With the development of pharmacological treatments for Alzheimer's disease there will be an increasein the numbers of patients requiring assessment from specialist services. Could the role of the specialist clinician besupported by other health professionals screening those who might bene®t from treatment?

Method. Sixty-four consecutive referrals to the Leicester University Memory Clinic were assessed at home by acommunity psychiatric nurse using a semi-structured interview. The nurse then reported her ®ndings to a psychiatristand a diagnosis was agreed. This diagnosis was then compared to the Memory Clinic diagnosis and a standardized(ICD-10) diagnosis recorded by another psychiatrist examining the clinic records.

Results. The nurse assessment procedure performed well in detecting dementia, with a kappa statistic (k) of 0.75when compared to the standardized and Memory Clinic diagnoses. There was, however, only moderate concordancebetween the ICD-10 diagnosis and nurse (k� 0.46) and the Memory Clinic and nurse (k� 0.60) for Alzheimer'sdisease. The relatively low k value for Alzheimer's disease was principally a result of di�cult in di�erentiating vasculardementia.

Conclusions. A single supervised community psychiatric nurse, using a structured assessment instrument, canadequately detect early dementia in a sample of patients referred with memory problems. Subtypes of dementia arenot, however, accurately di�erentiated. # 1998 John Wiley & Sons, Ltd.

Int. J. Geriat. Psychiatry, 13: 405±409, 1998.

KEY WORDSÐDementia; Alzheimer's disease; assessment; community psychiatric nurse; services

The acetylcholinesterase inhibitor donepezil hasrecently been licensed for the treatment of Alz-heimer's disease, both in the United States and theUnited Kingdom. Other similar products are likelyto be launched in the near future. These drugs arespeci®c for Alzheimer's disease and are most likelyto be of bene®t only in the mild and moderatestages of the illness (Kelly et al., 1997). It is likelytherefore that there will be a substantial increase in

the numbers of patients referred to specialistservices for assessment, and a precise diagnosiswill need to be determined to identify thosewhowillbene®t most from treatment. As well as di�erenti-ating the `worried well' and depressed, subtypes ofdementia will need to be accurately diagnosed.

There is as yet no de®nitive test for Alzheimer'sdisease, but an 80±90% accurate clinical diagnosisof dementia subtype is possible by a specialistclinician, assisted by neuropsychological assess-ment and investigations such as neuroimaging(Martin et al., 1987; Joachim et al., 1988; Morriset al., 1988). However, existing specialist servicessuch as memory clinics and district psychogeriatric

CCC 0885±6230/98/060405±05$17.50 Received 29 October 1997# 1998 John Wiley & Sons, Ltd. Accepted 28 January 1998

INTERNATIONAL JOURNAL OF GERIATRIC PSYCHIATRY, VOL. 13: 405±409 (1998)

*Correspondence to: Michael Dennis, Division of Psychiatryfor the Elderly, University of Leicester, Leicester GeneralHospital, Leicester. Tel: 0116 2584597, Fax: 0116 2731115,e-mail: msd4@le.ac.uk

services are unlikely to be able to manage theincreased demand without unacceptably longwaiting lists. Could the role of the clinical specialistbe supported by another health professionalscreening those who are referred for assessment?Community psychiatric nurses (CPNs) have beenshown to di�erentiate dementia e�ectively fromother psychiatric disorders (Collighan et al., 1993),and Seymour et al. (1994) found a good level ofagreement between a psychogeriatrician's diag-nosis and the diagnosis of a trained communitynurse utilizing the CAMDEX structured interview.

We describe a pilot study evaluating the perfor-mance of a community psychiatric nurse (CPN)utilizing a semi-structured assessment instrumentin diagnosing early dementia and di�erentiatingdementia subtypes in a memory clinic population.

METHOD

A series of consecutive referrals to the LeicesterUniversity Memory Clinic were initially assessedat home by a CPN using a semi-structured inter-view designed for this study. The Memory Clinicwas chosen as the sample frame for the study as itscaseload is likely to re¯ect the population referredfor assessment and treatment of Alzheimer'sdisease in the future, ie borderline cases, `worriedwell' and those with mild cognitive impairment ofearly dementia. In addition, patients at theMemoryClinic undergo a comprehensive assessment,comprising psychiatric examination (incorporatingthe CAMCOG), physical examination includinglaboratory investigations, neuropsychologicalevaluation assessment by a speech therapist andneuroimaging (CT scan or MRI).

The assessment instrument used by the nursecomprised a number of sections:

. Reasons for referral, and carer/relative-generated problem list

. Informant history (memory impairment, generalmental functioning, instrumental activities ofdaily living, past medical, psychiatric, family,treatment and social histories)

. 15-item Geriatric Depression Scale (GDS;Sheikh and Yesavage, 1986)

. Symptom checklist for anxiety, psychosis andmania

. Physical checklist

. Activities of daily living, and indicators of socialcare

. The MRC extended version of the Mini-MentalState Examination (Medical Research Council,1987)

. Clock drawing

The MRC extended version of the Mini-MentalState Examination (MMSE; Folstein et al., 1975)has 19 additional items involving memory, verbal¯uency and praxis from the CAMCOG (Roth et al.,1986).

The instrument is designed to be administeredwith an informant present, and takes 40±60minutesto complete. Following the assessment, the CPNreported her ®ndings to a consultant psychiatrist(not involved with the Memory Clinic) andfollowing case discussion a joint diagnosis wasagreed (this is referred to below as the `nurse diag-nosis'). The nurse involved in the study wasexperienced in assessing elderly patients in the com-munity, having worked for some years as a G-gradeCPN.

Another psychiatrist, blind to the research nurseassessment, retrospectively obtained a standardizeddiagnosis for each case from the Memory Cliniccase notes by applying ICD-10 research criteria(WHO, 1993) and the Newcastle criteria for Lewybody dementia (McKeith et al., 1992).

A total of 64 patients were seen and interviewedby the research nurse. Three patients were excludedas they had been referred by a consultant neurol-ogist with referral information containing theresults of extensive investigations. The nurse diag-noses, Memory Clinic diagnoses and standardizedICD-10 diagnoses were compared for the remain-ing 61 patients and levels of agreement calculatedusing the kappa statistic (Altman, 1991). Descrip-tive data for the sample were calculated using SPSSfor Windows (SPSS, 1993).

RESULTS

The majority of referrals to the Memory Clinicwere from general practitioners (50), but inaddition there were four patients referred fromgeneral psychiatrists, two from consultant physi-cians and ®ve from CPNs. The content of thereferral information was variable, no diagnosis wasindicated in 26 cases and the referrers wereuncertain in a further 12. A provisional diagnosisof unspeci®ed dementia appeared in nine referralletters, Alzheimer's disease in eight, vascular

# 1998 John Wiley & Sons, Ltd. INT. J. GERIAT. PSYCHIATRY, VOL. 13: 405±409 (1998)

406 M. DENNIS, L. FURNESS, J. LINDESAY AND N. WRIGHT

dementia in two, depression in three and normalpressure hydrocephalus in one.

The sample was relatively young, with a meanage of 67.7 years (SD+ 10.8, range 36±88), with27 males and 34 females. As expected, the meanCAMCOG score (73, SD+ 24.0, range 7±106) andthe mean MMSE score (21, SD+ 7.2, range 3±30)recorded in the Memory Clinic were relatively high.There was no signi®cant di�erence between theMMSE scores recorded by the research CPN andthe Memory Clinic (two-tailed paired sample t-test,p� 0.37). The mean delay from seeing the researchCPN to ®rst Memory Clinic appointment was6.4 weeks (SD+ 4.5, range 1±27).

Table 1 shows the diagnoses recorded by theresearch CPN following case discussion with aconsultant psychiatrist, the Memory Clinic and thestandardized ICD-10 diagnosis. According to thestandardized diagnosis, 40 (66%) of the patientswere su�ering from dementia and 24 (39.5%)of these from Alzheimer's disease; the four casesgiven a diagnosis of `other' were hypothyroidism,

post-epileptic confusion, chronic brain injury andParkinson's disease. In ®ve cases the MemoryClinic was unable to reach a diagnosis, opting for arepeat cognitive assessment at 6-month follow-up.Although a nurse diagnosis was made in 60 casessome doubt was expressed in 20, and the consultantand nurse felt that further investigation such asroutine blood tests, neuropsychological assessmentor neuroimaging would be required to con®rm thediagnosis. Table 2 shows the concordance betweenthe three diagnoses. There was good agreementbetween the nurse and standardized diagnoses forthe diagnosis of dementia, but only moderate con-cordance for Alzheimer's and poor concordancefor vascular dementia. There were four cases ofAlzheimer's disease diagnosed as vascular dementiaby the nurse assessment and four cases of vasculardementia diagnosed as Alzheimer's disease. Whencomparing Memory Clinic diagnosis with nursediagnosis (Table 3), there was better concordancefor Alzheimer's disease, but it was still poor forvascular dementia.

Table 1. Subjects' diagnoses, according to ICD-10, the Memory Clinic and nurse assessment procedure

Diagnosis ICD-10 Memory Clinic Nurse

N Frequency (%) N Frequency (%) N Frequency (%)

Early onset Alzheimer's 7 11.5 8 13 8 13

Late onset Alzheimer's 17 28 18 29.5 18 29.5

Dementia with Lewy bodies 2 3.5 0 0 2 3.5

Vascular dementia 12 19.5 7 11.5 8 13

Mixed dementia 2 3.5 1 1.5 1 1.5

Depression 4 6.5 4 6.5 7 11.5

Anxiety 1 1.5 0 0 3 5

Well 11 18 10 16.5 9 15

Other 4 6.5 8 13 4 6.5

No diagnosis 1 1.5 5 8.5 1 1.5

Total 61 100 61 100 61 100

Table 2. Concordance between ICD-10 diagnosis and the nurse diagnosis and Memory Clinic diagnosis

Diagnosis No. of

ICD-10

cases

Nurse diagnosis Memory Clinic diagnosis

Agreement Speci®city Sensitivity k statistic Agreement Speci®city Sensitivity k statistic

(%) (%) (%) (%) (%) (%)

Dementia

(all)

40 89 90 88 0.75 85 91 85 0.73

Alzheimer's

disease

24 74 76 71 0.46 90 89 92 0.80

Vascular

dementia

12 80 92 33 0.29 92 100 58 0.69

# 1998 John Wiley & Sons, Ltd. INT. J. GERIAT. PSYCHIATRY, VOL. 13: 405±409 (1998)

NURSE-ADMINISTERED ASSESSMENT INSTRUMENT FOR DEMENTIA 407

DISCUSSION

The results of this study support the view thatcommunity assessment by a trained psychiatricnurse and case discussion with a psychiatrist canaccurately detect dementia and di�erentiate thisfrom other psychiatric disorders (Collighan et al.,1993; Seymour et al., 1994). In fact, this sample isa more stringent test of this approach to assess-ment, since these patients were less impaired thanthose who are usually referred to a communitymental health team. They were predominantly ayoung-elderly population, with relatively highmean MMSE and CAMCOG scores re¯ecting aspectrum of patients with early dementia, border-line impairment, functional psychiatric disordersand the `worried well', in addition to other caseswith more signi®cant cognitive impairment.

With the advent of treatments speci®c forAlzheimer's disease, it is important to di�erentiatethis disorder from other causes of dementia, and todo so in the early stages. Here there were cleardi�culties with the standardized CPN assessment.There were too few patients with dementia withLewy bodies in this sample to evaluate the accuracyof nurse assessment in detecting this condition, butthere were conspicuous problems in identifyingvascular dementia, as the low kappa values demon-strate. Similarly, the relatively low kappa valueassociated with Alzheimer's disease when compar-ing the nurse and standardized diagnoses was aresult of this particular di�culty. This is similar tothe experience of Seymour et al. (1994), who alsohad some disagreements in di�erentiating vascular/mixed dementia from Alzheimer's disease. Thefour cases of vascular dementia diagnosed asAlzheimer's by the nurse assessment procedure allhad CT scan evidence of cerebrovascular disease,and two had physical signs suggestive of stroke.Two further cases of vascular dementia `missed' bythe nurse assessment also had cerebral infarctionon CT scanning. These cases emphasized the

importance of both the physical examinationand neuroimaging in distinguishing betweenAlzheimer's disease and vascular dementia. Thesensitivity and speci®city of the nurse assessmentwould have been signi®cantly improved if it hadbeen performed in conjunction with neuroimaging.There was a higher level of concordance betweenthe Memory Clinic diagnosis of Alzheimer'sdisease and that of the research nurse, however,this may have been as a result of the Memory Clinicdeferring diagnosis in borderline cases.

It should also be emphasized that in more than30% of cases the psychiatrist supervising the nursethought further investigations were necessary toincrease con®dence in the diagnosis. In one of thesecases, although a diagnosis of possible Alzheimer'swas recorded, the nurse detected hearing di�cultiesand psychomotor slowing. This case was sub-sequently diagnosed as su�ering from hypo-thyroidism. Diagnosis is not an error-free process;there were some disagreements between the `goldstandard' ICD-10 diagnosis and the consensus-derived Memory Clinic diagnosis, and in at leasttwo cases vascular risk factors detected by the CPNwere not recorded in the Memory Clinic notes. Theusefulness of home assessment was also illustrated;paranoid symptoms and alcohol abuse respectivelywere detected in two cases by the CPN but sub-sequently missed in the Memory Clinic.

The di�culties highlighted in this study indi�erentiating dementia subtypes emphasizes theneed for patients with mild cognitive impairmentto receive a comprehensive specialist assessmentincluding neuroimaging. Clearly, without such acareful assessment there would be concern that newtreatments might be prescribed inappropriately.Nurse screening utilizing standardized assessmentinstruments may be an important and cost-e�ective®rst stage in this process, but it does not provideall the information necessary for the accuratediagnosis of dementia subtypes. The communitynurse, and other members of the community team,

Table 3. Concordance between Memory Clinic diagnosis and nurse diagnosis

Diagnosis No. of

Memory

Clinic cases

Agreement

(%)

Nurse diagnosis k statistic

Speci®city Sensitivity

(%) (%)

Dementia (all) 34 89 81 94 0.76

Alzheimer's disease 26 80 83 77 0.60

Vascular dementia 7 85 90 43 0.43

# 1998 John Wiley & Sons, Ltd. INT. J. GERIAT. PSYCHIATRY, VOL. 13: 405±409 (1998)

408 M. DENNIS, L. FURNESS, J. LINDESAY AND N. WRIGHT

will also have key roles in monitoring the e�cacy ofany prescribed antidementia treatments.

Further evaluation of the assessment instrumentinvolving other community nurses and communityteam workers is clearly indicated following thispilot study. Ideally this would be with a largersample of patients referred to community mentalhealth teams and with access to the results ofinvestigations as required.

CONCLUSIONS

This study shows that a single supervised CPN,using a structured assessment instrument, canadequately detect dementia in a sample of patientsreferred with memory problems. Many of thesepatients were su�ering from early dementia. Sub-types of dementia were not, however, accuratelydi�erentiated. With the advent of Alzheimer-speci®c treatments it is important to di�erentiateAlzheimer's disease from other causes of dementia;this process relies upon comprehensive multi-disciplinary assessment and includes physicalexamination and neuroimaging.

REFERENCES

Altman, D. (1991) Practical Statistics for MedicalResearch. Chapman & Hall, London, pp. 403±405.

Collighan, G., Macdonald, A., Herzberg, J et al. (1993)An evaluation of the multidisciplinary approach topsychiatric diagnosis in elderly people. Brit. Med. J.306, 821±824.

Folstein, M., Folstein, S. and McHugh, P. (1975) `Mini-Mental State'. A practical method for grading the

cognitive state of patients for the clinician. J. Psy-chiatr. Res. 12, 189±98.

Joachim, C., Morris, J. and Selkoe, D. (1988) Clinicallydiagnosed Alzheimer's disease: Autopsy results in150 cases. Ann. Neurol. 24, 50±56.

Kelly, C., Harvey, R. and Cayton, H. (1997) Drug treat-ments for Alzheimer's Disease. Brit. Med. J. 314,693±694.

Martin, E., Wilson, R., Penn, R. et al. (1987) Corticalbiopsy results in Alzheimer's disease: Correlation withcognitive de®cits. Neurology 37, 1201±1204.

McKeith, I., Fairburn, A., Perry, R. et al. (1992)Neuroleptic sensitivity in patients with senile dementiaof Lewy body type. Brit. Med. J. 305, 673±678.

Medical Research Council (1987) Report from the MRCAlzheimer's Disease Workshop. Medical ResearchCouncil, London.

Morris, J., McKeel, D., Fulling, K. et al. (1988)Validation of clinical criteria for Alzheimer's disease.Ann. Neurol. 24, 17±22.

Roth, M., Tym, E., Mountjoy, C. et al. (1986)CAMDEX: A standardised instrument for the diag-nosis of mental disorder in the elderly, with specialreference to the early detection of dementia. Brit. J.Psychiat. 149, 698±709.

Seymour, J., Saunders, P., Wattis, J. and Daly, L. (1994)Evaluation of early dementia by a trained nurse. Int. J.Geriatr. Psychiat. 9, 37±42.

Sheikh, J. and Yesavage, J. (1986) Geriatric depressionscale (GDS): Recent evidence and development of ashorter version. In Clinical Gerontology: A Guide toAssessment and Intervention (T. Brink, Ed), HawthornPress, New York. pp. 165±173.

SPSS (1993) SPSS for Windows. Statistical Packagefor Social Sciences. SPSS, Chicago.

World Health Organization (1993) The ICD-10 Classi®-cation of Mental and Behavioural Disorders. DiagnosticCriteria for Research. World Health Organization,Geneva.

# 1998 John Wiley & Sons, Ltd. INT. J. GERIAT. PSYCHIATRY, VOL. 13: 405±409 (1998)

NURSE-ADMINISTERED ASSESSMENT INSTRUMENT FOR DEMENTIA 409