DEMENTIA CARE NEEDS IN AN AREAPOPULATION: CASE REGISTER DATAAND MORBIDITY SURVEY ESTIMATES

BRIAN COOPER1* AND ROB FEARN2

1Hon. Research Fellow, Institute of Psychiatry, London, UK2Research Nurse, Institute of Psychiatry, London, UK

ABSTRACT

Objective. To compare case register data on the frequency and distribution of known dementia cases in ametropolitan area population with expected total numbers computed from a national disability survey.

Method. Known cases were enumerated by a cross-sectional census of the Camberwell Dementia Register. Expectedtotal numbers were calculated using the Cognitive Disability (CD) Planning Model, based on the OPCS nationalsurvey of disability, 1985±86.

Results. Cases ascertained by the Dementia Register census comprised one-®fth of expected total prevalence. Theproportion of such cases was higher for persons in long-stay care (1 in 3) than for those in private households (1 in 7).According to the CD Planning Model, cases known to specialist agencies were on average no more severely disabledand dependent than those who were unknown. In terms of absolute numbers, the district nursing and home helpservices appeared to be the most important untapped sources of case detection, but other research indicates thatgeneral practice contacts (not included in the planning model) may be at least equally important.

Conclusions. At any given time, a high proportion of dementia cases, whether in long-stay care or in the community,will be outside the purview of specialist services. Primary care agencies are a major potential source, and a systematichealth screening of persons aged over 75 years could be used to realize this potential.# 1998 JohnWiley & Sons, Ltd.

KEY WORDSÐdementia; Alzheimer's disease; epidemiology; case registers; disability surveys; health care needs

While population-based service statistics provide auseful resource for planning, in the mental health®eld, as elsewhere in medicine, realistic estimates ofneed cannot be based on service contacts alone.Large disparities may be found between existinglevels of provision, in terms of the numbers ofmentally ill persons in contact with specialistagencies, and total prevalence as revealed by areamorbidity surveys. In an ageing society, the pointapplies with particular force to late-life dementia,which because so many cases remain unidenti®edhas sometimes been referred to as a `silentepidemic' (Caird and Cargill 1987). The presentstudy examines this question in a British metro-politan area, by comparing case-register data with

estimates for the same population computed from anational disability survey.

OBJECTIVE AND METHOD

The research aims were, ®rst, to establish the ratiobetween numbers of dementia cases (severe ormoderately severe) known to specialist services andexpected total numbers among persons aged over65; secondly, by comparing known and expectednumbers in relation to type of residence andreliance on community support, to ascertainwhich service agencies and professional groupsare potential sources of case detection.

The study was based on Camberwell, an area ofSouth London which at the 1991 national censushad a population of 212 560, of whom 27 471(12.9%) were aged 65 or over. A dementia caseregister for this population (Holmes, 1996) was setup in 1993 to provide cumulative data on all

CCC 0885±6230/98/080550±06$17.50 Received 10 February 1998# 1998 John Wiley & Sons, Ltd. Accepted 8 May 1998

INTERNATIONAL JOURNAL OF GERIATRIC PSYCHIATRY

Int. J. Geriat. Psychiatry 13, 550±555 (1998)

*Correspondence to: Dr B. Cooper, Section of Old AgePsychiatry, Institute of Psychiatry, London SE5 8AF, UK.

Contract/grant sponsor: The Wellcome Trust; Contract/grantnumbers: Project Grant No. 036000, Grant No. 046759.

con®rmed cases known to the mental healthservices; their distribution, clinical features, courseand outcome, and their service contacts. Duringthe register's ®rst 2 years, probable cases noti®edby local hospitals, nursing homes, old people'shomes, psychogeriatric day centres and communitypsychiatric nurses (CPNs) were assessed by meansof a standardized examination, scrutiny of hospitalrecords, or both. Con®rmation as a `case' requiredeither a clinical consultant diagnosis or a score of24 or below on the Mini Mental State Examination(Folstein et al., 1975), followed by application ofstandard diagnostic criteria. The number anddistribution of cases given here were derivedfrom an enumeration of persons noti®ed to theDementia Register who were present in the at-riskpopulation on 31.10.94 (Holmes et al., 1995) andwere rated as severe or moderately severe casesaccording to the `CAMDEX' guidelines (Rothet al., 1988).

No ®gures for dementia prevalence in thispopulation are available from ®eld survey research.Expected numbers for area populations in Britainand elsewhere have sometimes been derived fromthe EURODEM reanalysis of pooled prevalencedata (Hofman et al., 1991), but these rely onaverages from surveys in a number of countriesand, moreover, are in¯ated by an unknown pro-portion of mild and borderline cases. We decidedto rely instead on estimates for dementia-relateddisability taken from the 1985±86 O�ce of Popula-tion Censuses and Surveys (OPCS) national surveyof disability in Great Britain (Martin et al., 1988),in which cognitive functioning was rated using asimple 11-item battery, the SEVINT scale. A scoreof 4 or below on this scale (`cognitive disability'),which approximates to one of 17 or less on thebetter-known Mini Mental State Examination (Elyet al., 1996), provides comparability with theDementia Register categories of moderate and

severe dementia. This approach has the advantagesthat it draws on a representative national sampleand that a special programme based on OPCSsurvey ®ndings, the Cognitive Disability (CD)Planning Model (Ely et al., 1996; Melzer et al.,1997), can be used to compute expected casenumbers for any local British population.

In essence, the model is simply an algorithmwhich can be used to estimate `expected' casenumbers in any area population, on the assump-tion that local frequencies for each age±sex groupre¯ect national averages. To apply it, one needs®gures for the age±sex structure of a given areapopulation, taken from the 1991 census returns ormore recent updating. In this instance, the problemwas complicated by the fact that Camberwell nolonger exists as an administrative area, being nowdivided between the London Boroughs of South-wark and Lambeth. Fortunately, we were givenaccess to a special analysis of 1991 census data forCamberwell, prepared for the Medical ResearchCouncil's Social Psychiatry Unit. The planningmodel was applied to this distribution in order togenerate expected case numbers.

RESULTS

Table 1 reveals a large disparity between numbersof cases noti®ed to the Dementia Register andexpected totals as computed from the planningmodel. Overall, cases known to the mental healthservices constitute only one-®fth of the expectednumbers for each sex, the corresponding ratiosbeing 8.3 and 40.4 per 1000 for men, and 13.1 and67.6 per 1000 for women. In all, persons aged 75 orover accounted for 83.4% of total prevalence. Theexpected numbers given here are already under-estimates and will become increasingly so in theyears ahead. The planning model assumes an

Table 1. Dementia in Camberwell elderly population: known cases and expected total numbers, according to sex andage group (prevalence ratios per 1000 in brackets)

Age group (yr) Male Female

Dementia Register 1994 Expected total number* Dementia Register 1994 Expected total number*

65±74 24 (3.4) 174 (24.5) 27 (3.1) 189 (21.9)

75±84 48 (13.9) 203 (58.7) 91 (15.3) 474 (79.8)

85 � 20 (36.0) 72 (129.7) 97 (54.0) 443 (246.5)

All 65 � 92 (8.3) 449 (40.4) 215 (13.1) 1106 (67.6)

*Cognitive Disability Planning Model (Ely et al., 1996).

# 1998 John Wiley & Sons, Ltd. Int. J. Geriat. Psychiatry 13, 550±555 (1998)

DEMENTIA CARE NEEDS IN AN AREA POPULATION 551

average increase in prevalence of just over 8% overthe decade 1996±2006, corresponding to anadditional 125 cases in Camberwell.

The age trends shown in Table 1 conform fairlywell to those reported from pooled survey analyses(Jorm et al., 1987;. Hofman et al., 1991), bothknown and expected frequencies rising steadilywith advancing age above 65. Increase with ageappears steeper among Dementia Register patients,so that the ratio of known to total case frequency islowest in the `young-old' age group (65±74 years)and highest among over-85-year olds. The absolutenumber of unknown cases is, however, highest inthe age group 75±84 years.

In Table 2, distributions for known cases andexpected numbers are set out according to agegroup and type of residence, ie whether in any formof long-stay care or living in a private household.Here, the estimated totals are based on nationalaverages in the OPCS survey, there being noaccurate information on the numbers of elderlyCamberwell residents (or former residents) in long-stay care. Among old persons in private house-holds, known cases amount to fewer than one inseven of the estimated total. The correspondingproportion among those in long-stay care, thoughsubstantially higher, still amounts to less than one-third of the expected ®gure. The proportionateincrease in known cases with rising age, shown inTable 1, appears to be characteristic for commu-nity cases but not for those in residential care.

Of 140 Dementia Register patients in privatehouseholds, 61 (43.6%) were living alone and 79(56.4%) with one or more other persons. Expectedtotals computed from the planning model, andadjusted for the proportion of elderly personsliving alone in Camberwell (46% compared with anational average of 37%), were 344 persons(33.8%) living alone and 673 (66.2%) living withothers. The proportionate di�erence is consistent

with a lower level of informal support amongdementing old people living alone, who are morelikely to be dependent on professional caregiversand hence to be known to specialist services.

The CD Planning Model can generate frequencyestimates for some dementia-related disabilities orabnormalities of behaviour which tend to result independency on constant care and supervision. InTable 3, expected frequencies for such indicatorsare compared with corresponding ®gures derivedfrom the Dementia Register data. Comparisonscannot be exact, since information on the twosamples was obtained by use of di�erent schedulesand rating scales. Approximations are, however,probably close at the severe end of the spectrum.Here, `severe limitation of mobility' means bedfast,chairbound or able to move about indoors onlywith help; `frequent urinary incontinence' means atleast once daily; `aggressive behaviour' meansphysical violence and not just verbal aggression

Table 2. Dementia in Camberwell elderly population: known cases and expected total numbers in institutional careand in private households, according to age group

Age group (yr) In institutional care In private households

Dementia Register 1994 Expected total number* Dementia Register 1994 Expected total number*

65±74 16 71 35 292

75±84 72 214 67 463

85� 79 253 38 262

All 65� 167 538 140 1017

*From Dementia Planning Model (Ely et al., 1996).

Table 3. Frequency of severe disabilities and dependencyassociated with dementia: known cases and expectedtotal numbers in Camberwell elderly population (pro-portion of a�ected cases in brackets)

Nature of disability/

dependency

Dementia

Register census,

31.10.94

(N� 307)

Expected total

number*

(N� 1555)

Severe limitation

of mobility

46 (15.0%) 334 (21.5%)

Faecal incontinence or

frequent urinary

incontinence

112 (36.5%) 559 (35.9%)

Disturbed, aggressive

behaviour

30 (9.8%) 164 (10.5%)

High dependency for

basic self-care

139 (45.3%) 816 (52.5%)

*Cognitive Disability Planning Model (Ely et al., 1996).

# 1998 John Wiley & Sons, Ltd. Int. J. Geriat. Psychiatry 13, 550±555 (1998)

552 B. COOPER AND R. FEARN

or temper outbursts; `high dependency for basicself-care' means inability to feed, wash, dress orundress oneself unaided.

The table is impressive in two respects. First, itpoints to disparities between known and expectedtotal case numbers at least as great as those inTables 1 and 2, the ratio being in each instance 1 :5or lower. Secondly, it suggests that cases ofdementia unknown to the mental health servicesare as likely to be associated with severe disabilityin daily living as those which are known. Indeed,the proportion with severe limitation of mobility isactually lower for Dementia Register patients thanamong the expected total group (chi-sq. 6.65;d.f.� 1; p5 0.01). The table thus provides noevidence that persons noti®ed to the DementiaRegister are predominantly the most dependent oncare among all those a�ected. This unexpected®nding may be due to a selective under-recordingof dementing old people with severe physicalillness, because these are referred preferentially togeriatric or other medical departments and notnoti®ed to the register.

Table 4, restricted to persons living in privatehouseholds, places numbers and proportions ofDementia Register patients known to be receivingcertain forms of community care alongside totalscomputed from the CD Planning Model. (Notethat the six types of service here included wereselected from a total of 15 recorded in the OPCSnational survey, and were of course not mutuallyexclusive. The fact that the expected percent-ages in Table 4 add up to just over 100 is purecoincidence.)

Comparison is again inexact, for two reasons. Tobegin with, the two datasets relate to di�erent

points in time (1994 and 1985±86 respectively)during a period of changing provision. Secondly,the planning model is based on national averagesand cannot take account of regional or localvariations, either in the level of provision ofdi�erent community services or in the prioritiesthat these observe. Nonetheless, the ®gures give arough indication of the extent to which dementingold persons who require community support arelikely to be known to the specialist services. Thatthe number of known cases under CPN supervisionactually exceeded the expected total ®gure isunsurprising, since (a) Camberwell has a relativelyhigh level of this type of provision, and (b) anycontact with a CPN should result in noti®cation tothe Dementia Register. Day care attending is alsoto some extent associated with noti®cation, mostreferrals to the area psychogeriatric day centresbeing made by consultants in old age psychiatry.

In contrast, only a fraction of the expectednumber of dementing elderly in contact with eachgeneralist service (district nursing; local authoritysocial work; home help support; meals on wheels)could be ascertained from the Dementia Register.Table 4 suggests that the largest group of unidenti-®ed cases is to be found among patients of thedistrict nursing service and the second largestamong old people visited by home helps. Oneshould note, however, that the OPCS survey datado not allow estimation of the number of casesknown to general practice teams, so that this is notincluded in the planning model.

DISCUSSION

In using the CD Planning Model to computeexpected numbers for a local area population, twomain problems of method arise. First, the OPCSsurvey estimates were based on a cognitive screen-ing whose sensitivity and speci®city were notclearly established. The scale has good face validityas a dementia screen and its application yielded forthe population aged over 65 a prevalence estimateof 5.6% which approximates to the median forprevalence surveys of dementia in the past twodecades (Livingston, 1994; Cooper, 1995; Kay,1995). Projecting the OPCS survey data onto the1996 population structure gives a revised ®gure of6.2% slightly below the 6.6% (CI 5.9±7.3) newlyreported from a large-scale British project (MedicalResearch Council Cognitive Function and AgeingStudy 1997). The expected totals computed for

Table 4. Persons with dementia living in private house-holds in Camberwell, according to community servicesreceived. Comparison of Dementia Register cases andexpected total numbers

Type of service received Dementia

Register 1994

(N� 140)

Expected total

number*

(N� 1017)

CPN supervision 55 (39.3%) 36 (3.5%)

Day centre attendance 84 (60.0%) 140 (13.8%)

District nurse visiting 41 (29.3%) 362 (35.6%)

Social worker visiting 24 (17.1%) 132 (13.0%)

Home help 51 (36.4%) 266 (26.2%)

Meals on wheels 34 (24.3%) 96 (9.4%)

*Cognitive Disability Planning Model (Ely et al., 1996).

# 1998 John Wiley & Sons, Ltd. Int. J. Geriat. Psychiatry 13, 550±555 (1998)

DEMENTIA CARE NEEDS IN AN AREA POPULATION 553

Camberwell are thus unlikely to be serious over-estimates. Nevertheless, reliance on a brief assess-ment of this kind must be assumed to result insome degree of misclassi®cation, with both `falsepositives' and `false negatives'.

Secondly, application of the model to small-areapopulations, while allowing for local variations inage and sex distribution, cannot take into accountdi�erences in occupational class, education orother social factors which might also in¯uencethe risk of dementing disorders in the elderlypopulation. Hence the research technique, whileeconomic of resources, should not be considered asatisfactory substitute for ®rst-hand data gatheringby means of area ®eld surveys.

Does it provide a realistic picture of the currentBritish scene? The available information, althoughscanty, seems compatible with our data. Accordingto the OPCS disability survey, of the 60% ofpersons with `advanced cognitive impairment' atthat time living in private households, only 12%had been seen by psychiatrists and 2% by CPNswithin the preceding 12 months. It is unclear,however, what proportion of the remaining 40% inlong-stay care were also known to psychiatricservices. A more relevant cross-bearing is providedby a recent census survey of the Forth Valley, anarea of Central Scotland whose population includessome 40 000 persons aged 65 or over (Gordon et al.,1997). Of a total of 2060 ascertained cases ofprobable dementia, 530 (25.7%) were noti®ed byspecialist medical departments (geriatric and psy-chogeriatric), a number which represents just underone-®fth of an expected total of 2708 cases, com-puted from EURODEM prevalence ratios. Thissimilarity in proportion suggests that the Camber-well ®ndings are not grossly atypical.

The ®gures for case frequency set out in Table 1are compatible with a selective process of help-seeking, diagnosis and referral according to themodel of `levels' and `®lters' developed for generalpsychiatry by Goldberg and Huxley (1994), theestimated totals corresponding to level 1 (thecommunity) and the known cases to level 4 (casesreferred to mental illness services) in that model.Clearly there is no possibility, given existing scalesof provision, that specialist hospital departmentscould exercise direct responsibility for the medicaland nursing care of all a�ected persons, even if thatwere thought desirable. Of greater practical importis the need for e�ective illness recognition andassessment of need by those primary care agenciesin closest contact with the elderly.

Table 4 points to the importance in this contextof district nursing and home help services. Aserious omission from the table is, however, anyestimation of cases seen in general practice. In theForth Valley survey, 235 (29.5%) out of 797 casesin private households were noti®ed by GPs, whilethe primary care teams as a whole, includingattached district nurses and health visitors, noti®ed437, or 54.8% of community cases (D. S. Gordon,personal communication). The primary care teamevidently has a major part to play in case detectionand management, and its role could in future beenhanced by a more systematic assessment ofcognitive function as part of the contractualscreening of patients aged over 75 years (Ili�eet al., 1994).

ACKNOWLEDGEMENTS

The Wellcome Trust provided a Project Grant(No. 036000) for development of the CamberwellDementia Register as a research tool, andadditional support to Professor A. H. Mann andDr B. Cooper for a register-based epidemiologicalstudy (Grant No. 046759). The Dementia Registeris currently supported by the Bethlem andMaudsley Hospital Trust and the Gatsby Founda-tion. We are grateful to Dr Clive Holmes (Instituteof Psychiatry) for help with the analysis ofDementia Register data, and to Dr David Melzer(Cambridge University Institute of Public Health)for making the CD Planning Model available forthis study.

REFERENCES

Caird, F. I. and Cargill, D. (1987) Chronic brain failure:the silent epidemic. Austral. NZ Med. J. 100, 13±16.

Cooper, B. (1995) Principles of service provision in oldage psychiatry. In Psychiatry in the Elderly, 2nd edn(R. Jacoby and C. Oppenheimer Eds). OxfordUniversity Press, Oxford, pp. 357±375.

Ely, M., Melzer, D., Brayne, C. and Opit, L. (1996)Estimating the numbers and characteristics of elderlypeople with cognitive disability in local populations.Res. Policy Planning 14, 13±18.

Folstein, M., Folstein, S. and McHugh, P. (1975) `Minimental state'. A practical method for grading thecognitive state of patients for the clinician. J. Psychiat.Res. 12, 189±198.

Goldberg, D. and Huxley, P. (1994) Common MentalDisordersÐa Biosocial Model. Routledge, London.

# 1998 John Wiley & Sons, Ltd. Int. J. Geriat. Psychiatry 13, 550±555 (1998)

554 B. COOPER AND R. FEARN

Gordon, D. S. , Carter, H. and Scott, S. (1997) Pro®lingthe care needs of the population with dementia: Asurvey in Central Scotland. Int. J. Geriatr. Psychiat.12, 753±759.

Hofman, A., Rocca, W., Brayne, C. et al. (1991)The prevalence of dementia in Europe. A collaborativestudy of 1980±1990 ®ndings. Int. J. Epidemiol. 20,736±748.

Holmes, C. (1996) The Camberwell Dementia CaseRegister. Int. J. Geriatr. Psychiat. 11, 369±375.

Holmes, C., Cooper, B. and Levy, R. (1995) Dementiaknown to mental health services: First ®ndings of acase register for a de®ned elderly population. Int. J.Geriatr. Psychiat. 10, 875±881.

Ili�e, S., Mitchley, S., Gould, M. and Haines, A. (1994)Evaluation of the use of brief screening instruments fordementia, depression and problem drinking amongelderly people in general practice. Brit. J. Gen. Pract.44, 503±507.

Jorm, A. F., Korten, A. E. and Henderson, A. S. (1987)The prevalence of dementia: A quantitative integrationof the literature. Acta Psychiat. Scand. 76, 465±479.

Kay, D. W. K. (1995) The epidemiology of age-relatedneurological disease and dementia. Rev. Clin.Gerontol. 5, 39±56.

Livingston, G. (1994) The scale of the problem.In Dementia (A. Burns and R. Levy Eds). Chapmanand Hall, London, pp. 21±35.

Martin, J., Meltzer, H. and Elliot, D. (1988) The Preval-ence of Disability among Adults. Report 1. HMSO,London.

Medical Research Council Cognitive Function andAgeing Study (MRC±CFAS) (1998) Cognitive func-tion and dementia in 6 areas of England and Wales:distribution of MMSE and AGECAT organicity levelsin the MRC±CFAS study. Psychol. Med. 28, 319±335.

Melzer, D., Ely, M. and Brayne, C. (1997) Localpopulation di�erences and the needs of people withcognitive impairment. Int. J. Geriatr. Psychiat. 12,883±887.

Roth, M., Huppert, F. A., Tym, E. and Mountjoy, C. Q.(1988) CAMDEX: The Cambridge Examination forMental Disorders for the Elderly. Cambridge Univer-sity Press, Cambridge.

# 1998 John Wiley & Sons, Ltd. Int. J. Geriat. Psychiatry 13, 550±555 (1998)

DEMENTIA CARE NEEDS IN AN AREA POPULATION 555