MRC multicentre study of cognitive function and ageing (MRC-CFAS)
- The main aims of the multicentre study are:
- To estimate the prevalence and incidence of cognitive decline and dementia and the range of variation of those two measures throughout England and Wales. The study should be capable of identifying a two-fold or greater difference between locality in these rates.
- To determine the natural history of dementia, in particular the rate of progression of cognitive decline including the distribution of the interval between the identification of cognitive impairment and death.
- To evaluate the degree of disability associated with cognitive decline and the service needs this disability generates.
The subsidiary aims are:
- To form the basis for longer term studies of trends over time and by birth cohort of the prevalence and incidence of cognitive decline.
- To determine the contribution of different underlying pathologies to the rates of dementia and the geographic variation in these rates and to the burden of disability.
- To identify factors associated with differing rates of cognitive decline and with the risk of dementia.
- To act as a core resource and provide a framework to support specific sub-studies in each centre.
Subsequent development of aims:
The study was recognised as providing an unrepeatable opportunity to define the changes in the brain which occur in normal ageing and in those who have become mentally frail. Shortly after the study was funded additional funding was obtained for a neuropathologists' consortium to examine the neuropathological aspects of brain ageing.
Aims of Neuropathology study:
- To determine the prevalence and severity of pathological lesions in an unselected cohort of elderly individuals with and without cognitive impairment.
- To determine the frequency of specific pathological diagnoses in the cognitively impaired sample.
- To correlate severity of specific pathologies with patterns of cognition, function and behaviour in life independently of clinical and pathological diagnostic categories.
The breadth of the required measures allows the study to examine determinants of healthy ageing, including important components of quality of life.
Population studies can look at novel genetic findings to see if they are important. This was not part of the initial funding of the study, but the aim of setting up a population based DNA resource has been incorporated in a later phase of the study.
Aims of Genetics study:
To perform genetic association studies to discover genes which influence risks for four related clinical parameters:
- Alzheimer’s disease
- Vascular / mixed dementia
- Dementia of all types
- Cognitive decline over a 5 year period.
From 1985 to 2025 it is projected that there will be a substantial increase in the size of the population aged over 65 in the United Kingdom. The largest increase will occur in those aged over 80. The prevalence of chronic disease increases exponentially with age with a parallel rise in the cost of providing adequate health care. There is also an enormous hidden cost to the informal carers of infirm elderly in the community, who are often themselves of retirement age. The change in structure of the population therefore has major implications for the future provision of services.
The distribution of many chronic diseases shows considerable variation in the United Kingdom. It is well known, for example, that the standardised mortality ratios of several cancers, ischaemic heart disease, and cerebrovascular disease are high in the north and north west, and low in the south east.
In terms of their contribution to disability among the elderly, dementia and cognitive decline have been identified as of particular importance. Since there are large regional variations in other chronic disorders, variations in rates of dementia and cognitive decline might be expected. Routine statistics, however, are of little help in looking for regional differences in occurrence; death certification of these conditions is extremely unreliable and hospital statistics are of no value.
It is thought that senile dementia of the Alzheimer’s type (SDAT), is the most common form of dementia in the community, with vascular dementia (VD) next. While risk factors for SDAT, (apart from age itself, family history and Down’s syndrome), remain unclear, it is thought that hypertension is a risk factor for VD. There is documented variation in the levels of blood pressure in different areas in middle aged men, (Regional Heart Study), and there is also variation in other hypertension associated morbidity throughout the country. From this information alone a considerable variation in the occurrence of VD might be expected. Greater understanding of regional variation of this condition, its risk factors and natural history would have clear potential for prevention in the future. Any demonstrated true variation in the occurrence of Alzheimer’s disease by region could lead to a more targeted search for risk factors than has been possible to date.
Prevalence is not a function of incidence alone, but depends on survival. Recent studies of demented subjects report that survival is shorter than expected, but not as markedly as early studies suggested. Whether there are differences in survival according to differential diagnosis is not clear, with different studies producing conflicting results. Duration of survival has been found to be strongly associated with level of cognitive function in large community studies of the elderly using a crude measure of cognitive function. Survival in relation to cognitive decline has not yet been examined in this country. The Swedish longitudinal study has provided more evidence in favour of decline in many functions, including cognitive, in the period before death.
Migration is a further important factor, particularly in geographically delimited, rather than national, studies. Migration into retirement areas, or to be close to family, as well as into institutions can affect age specific prevalence figures dramatically. The prevalence of dementia in institutions may have increased over the last decades, and may account, at least in part, for reduced prevalence rates found in community surveys which have not sampled institutions. It is therefore of importance in any national study that institutionalised subjects are included, and that migration patterns of the elderly are studied in each area in order to estimate their effect on comparative prevalence rates.
3.2 Cognitive decline and dementia in the UK
Previously reported prevalence estimates for moderate to severe dementia among people over 65 years of age have varied from 2.5% to 14.0% with an indication of a decrease over time and of higher rates in the north of the country. If true, these variations could be of major importance both for health service planning and for the distribution of etiologic factors. Differences in methodology made any comparison of the published estimates suspect, and it was impossible to determine whether any of the observed differences are genuine or simply methodological artefacts. This emphasised the importance of performing a national study using uniform methods.
Time trends have not been examined in a consistent and comparable fashion in the UK. Cohort effects have been reported in Sweden indicating that successive groups of people at given ages perform better on cognitive tests. This effect is likely to be due to different educational experience, and altering patterns of chronic disease, but may also be due to other environmental factors such as improved nutrition. It is not known whether similar cohort effects exist in the UK and if present, whether they have any impact on the occurrence of dementia and cognitive decline. These observations emphasise the need for studies examining different sites and times with uniform methods
3.3 Studies being conducted in the UK at the time (1989) of planning for CFAS
In 1986, a review of the work being conducted by research teams examining large numbers of the elderly, both community and volunteer, was prepared on behalf of a Medical Research Council (MRC) working group. The following notes describe some of these.
At Liverpool, Copeland and his colleagues were funded by the MRC to conduct a longitudinal study of the incidence of mental disorder in the elderly using the Geriatric Mental State Examination, (GMS), (Copeland et al 1976). 5244 elderly residents were interviewed at baseline. This study has been incorporated into CFAS.
Clarke and colleagues at the Department of Community Health at Leicester University, in collaboration with the Department of Psychiatry, have established cohort studies in Melton Mowbray.
At Cambridge, the Cambridge City over 75 cohort (CC75C) was ongoing (Brayne and colleagues). This large sample of over 75’s were initially seen as part of the Hughes Hall Project for Later Life. The measure of change in this study is the difference in the scores on Mini-Mental State Examination (MMSE) between the first and the second interview, approximately two years apart. The diagnostic instrument being used in this study is Cambridge Examination for Mental Disorders of the Elderly (CAMDEX), (also used in the CFAS first wave interviews).
General practice is used as the sampling frame for all these community studies.
Since the CFA Study started numerous other longitudinal studies have either started, or been made longitudinal by the addition of further waves and appropriate measures.
3.4 International context in 1989
A study of the elderly in New York and London using GMS had reported higher rates of dementia in New York than in London using both clinical ratings and computerised diagnostic procedures. This difference could have been due to differing effects of etiologic factors, but a role for factors such as selective migration cannot be excluded from these results. No studies had compared the incidence of dementia and cognitive decline across different sites using the same carefully controlled methodology. It was therefore not possible to compare figures with any confidence. The European Action Group on the Epidemiology and Prevention of Dementia, (EURODEM), established a common protocol for an international study examining the incidence of risk factors for dementia in the community, of which CFAS formed a part.
3.5 Dementia, cognitive decline or cognitive impairment
One of the reasons why dementia and cognitive decline are difficult to study rigorously is the absence of a biological marker for the condition, and the heterogeneous nature of the underlying disorders. There are many causes of dementia and potentially more of cognitive decline. Studies have measured cognitive impairment as opposed to decline, and others have not measured cognition at all but relied on other diagnostic methods to arrive at a diagnosis of dementia. Few community studies have attempted differential diagnosis of dementia even though accurate differential diagnosis is often possible at post mortem. Several studies making a series of clinical diagnoses and using sophisticated and rigorous investigative procedures have diagnosed SDAT or VD with only 70-80% accuracy when compared with diagnosis from post mortem. The benefits of more accurate diagnosis through a parallel neuropathology study also extends to the larger group of diverse dementias and their related risk factors.
3.6 Implications for planning services
All the areas mentioned above are of importance in trying to plan services for the elderly, both in the short and long term. A further area which is relatively unresearched is the impact of dementia and cognitive decline on use of services. Any study which plans to be aid long term planning of services must take into account not only the factors mentioned here, but also variation in cultural expectations of the ageing process and of service provision.
There has been much discussion about cognitive decline and dementia between the Department of Health (DH), MRC and other relevant experts. The studies outlined above can address only a few of these areas. CFAS and RIS of CFAS attempts a more comprehensive response to the interests of the DH, as well as focusing on the scientific aspects of the epidemiology of cognitive decline and dementia in the UK.
4.1 Rationale for the approach
Studies of incidence and natural history require cognitive assessment at a minimum of two time points, sufficiently separated for change in cognitive function to be identifiable given the inherent variability in the measures used. For the estimation of the incidence of disabling cognitive decline, new cases need to be identified and the precision of the estimate will depend on the number of such cases. Identification of new cases requires a change in cognitive function of sufficient magnitude to be distinguishable from chance variation. Too short an interval between separate assessments would have the double disadvantage that changes in cognitive function may not be large enough to be separated from background noise, and that not enough genuine events will have occurred. Too long an interval, on the other hand, would result in many deaths occurring between the assessments, particularly of the older age groups in the study. Since survival of people with dementia is shorter that that of the general population, the risk of biasing the estimate of incidence downwards would be severe. An interval of two years between assessments, with a sustained effort to obtain information on intercurrent deaths, is felt to be optimal for the estimation of incidence. For the study of natural history, sequential observations are needed on all those identified with suspected or confirmed dementia, together with a comparison group selected from those with no observable cognitive impairment. Yearly assessment is considered necessary, both to estimate the variation in the rate of progression of the disease and to assess the advancing degree of disability and the consequent service requirements.
Estimation of geographic variation throughout the country can be approached in two ways. Either selection of a genuinely national sample or selection of a number of centres and sampling in their environs. Since the study aims to maximise the number of subjects from whom post mortem brain tissue is obtained, the first approach represents great logistic problems. The second approach is thus the one of choice, leaving two issues to be resolved - how many centres, and how to choose them. The first criterion is that each centre should have the necessary expertise in psychiatric epidemiology and field studies. The second is that the centres should cover most of the national variation in factors which might influence risk for the disease. Since these are largely unknown, surrogates have been used which relate to national variation in the rates of other chronic diseases, (e.g. cardiovascular disease, stroke, cancer of the lung and stomach), geographic location, socio-economic level and degree of urbanisation.
On the basis of these criteria the study is based in the following centres: two rural samples in Cambridgeshire, and Gwynedd (administered from Liverpool), and urban samples in Liverpool, Newcastle, Nottingham and Oxford. Inclusion of Cambridge and Liverpool enables the study to dovetail into existing MRC funded studies at these centres. The inclusion of Newcastle was motivated by the earlier study (Kay 1964) conducted there. The choice of the number of centres to include in the study is based on pragmatic grounds. A much larger number would give a fuller picture of whatever pattern of variation exists, but the additional information that extra centres might contribute would be difficult to justify on grounds of cost. A reduction below six in the number of centres would risk missing essential elements in the pattern of variation. The choice of six appears a reasonable compromise, sufficient to establish national levels and degree of variability in the rates of cognitive decline. Reduction of the number below six would seriously jeopardise national coverage.
It is unrealistic to attempt estimates of time trends in the rates for dementia within a five year time span. The experience from time trends in other chronic diseases such as lung cancer, cardiovascular disease and stroke suggests that a minimum of ten years is required to establish, with any degree of precision, the rate of change in disease occurrence. It is certainly of importance to investigate time trends, both for the value to epidemiological studies and for health service planning, but the most economic approach will depend largely on the degree of variation in incidence seen in the five year study. It is therefore anticipated that about four years after the start of this study, a proposal would be developed for monitoring time trends (this was submitted, but not funded).
4.2 Study Management
October 1990 to December 2000
Up to December 2000 CFAS was overseen by a steering committee with a Study Management Group to ensure smooth running of the study.
4.2.1 Steering Committee
Remit: To ensure that the research addresses, in the most appropriate manner, the objectives set out in the proposals approved by the Neurosciences Board (NB) and the Health Service Research Committee (HSRC) in response to Department of Health (DH) interests. At the first meeting they committee approved the working protocol for the study in advance of start of fieldwork. Thereafter the committee has monitored the timetabling and procedure of the project and safeguarded its interests with regard to add-on studies.
Composition: Chair, Chair of the Study Management Group, the Scientific Coordinator of CFAS, representation from SMG, MRC representative , representative from the Neurosciences Board, DoH representative.
Meetings: The committee should meet at a minimum of yearly intervals.
4.2.2 Study Management Group
Remit: To oversee the running of the study. Initial task to agree a definitive protocol and arrangements for co-ordination of the study. Thereafter it should be responsible for ensuring that these were being followed satisfactorily and for resolving any difficulties in the management or co-ordination of the study. Proposals for using the multicentre resources of the study, i.e. study population and the associated documentation on each individual, sampling frames etc., for linked investigation will be referred to the SMG.
Composition: Chair (appointed by MRC), Principle Investigators from each centre, representatives from MRC Biostatistics Unit, Study Manager, MRC representative.
Meetings: The committee should meet at least 3 times a year.
4.2.3 Instruments sub-group
Remit: To design the instruments.
Meetings: At the beginning of the study in order to design the instruments. Subsequently the group was disbanded and reformed as required.
4.2.4 Advisory Committee
Remit: To advise SMG in ensuring appropriate storage of blood and brain tissue, documentation, and scrutiny of proposed collaborative projects.
Composition: Chair, the Chair of the Study Management Group, the Scientific Coordinator of CFAS, representation from SMG, Chair or nominated deputy from the Steering Group, MRC representative (later from Technology Transfer Group when this is deemed appropriate), representative from the Neurosciences Board, independent geneticist. DoH declined to nominate a representative, but wish to be kept informed.
Meetings: As required - communication by email.
December 2000 onwards
In January 2001 the MRC CFAS Co-operative Group was formed. The management structure of the Group is as follows:
4.2.5 CFAS Cooperative Group
Co-operative management Committee (CMC).
This Group reports to the Cross Board of the Medical Research Council.
Remit: To ensure that the research addresses the objectives set out in the proposals; to approve procedures and monitor progress; to consider applications for additional components and collaborative studies; with recommendations from the Advisory Committee for use of the blood and brain resource; to safeguard the original objectives of the study, including the Health Services research component; to ensure that use of CFAS sample / material at only one site will not compromise the study as a whole.
Composition: Chair (approved by MRC and DoH), one PI from each active component , associate PI’s - investigators from each centre involved in the core activity, one representative from the Neurosciences Board or Cross Board Group , one representative from the Advisory Committee, one MRC and one DoH representative. Co-opted to the CMC: Tony Johnson, Study Statistician, Judith Nickson, Neil Walker.
Meetings: The CMC should meet at least three times a year.
Remit: To advise CMC in ensuring appropriate management of blood and brain tissue, documentation, and scrutiny of proposed collaborative projects.
Composition: Chair, Chair of CMC, Representative from CMC, MRC representative (later from Technology Transfer Group when this is deemed appropriate), representative from Neurosciences or Cross Board Group, independent geneticist. DoH declined to nominate a representative, but wished to be kept informed.
Meetings: as required - communication mainly by email.
Analysis Management Group (AMG).
This group reports to the Co-operative management Committee.
Remit: To oversee and advise on all issues relating to analysis of the data, with delegation to the various local analysis groups (IPH Cambridge Analysis Group - ICAG, Neuropathology Analysis Group - NPathAG, Newcastle Analysis Group - N/CAG, Liverpool Analysis Group - LPAG, etc); to review all papers before submission; to control management of the dataset and release accredited versions to CFAS researchers and external collaborators.
Composition: Chair, members of CMC, Tony Johnson, Study Statistician, Data Manager, Assistant Data Manager /Statistician (to be appointed), Judith Nickson, Neil Walker, PI's from active components, a representative of each AG.
ICAG, N/CAG, NPathAG, LPAG, etc.
Remit: To co-ordinate all local management and analysis matters.
Composition: Active researchers.
Study Co-ordination and Administration.
Remit: To enable all aspects of the study to run smoothly and to time. To liaise at all levels within CFAS, maintaining communications between PI's at all centres via email and distribution of paperwork; to minute study group meetings; to organise, train, manage and monitor interviewers at all centres to ensure fieldwork progresses as planned, to report to CMC and AMG on progress, etc. To receive new proposals for components and collaboration and to enable confidential consideration by advisory committee of genetic projects.
Click here to see a diagram of the management structure of the CFAS Co-operative Group.
4.2.6 Study Components
The study programme consists of the main study of cognitive function and ageing plus five linked studies.
This was conducted by the Centre for Health Services Research Unit, at Newcastle in collaboration with the Centre for Health Economics, York, and aimed to investigate the resource implications to families and health and personal services of mental and physical frailty in elderly people. Individuals and their carers were interviewed three months after screening, with further interviews at 6, 12, 18, and 24 months. The study took place in Cambridge, Newcastle, Nottingham and Oxford. This project was funded by the MRC for five years.
2. Network Study
This was conducted within the Centre for Social Policy Research & Development at University Wales at Bangor. Its aims were to describe the nature and distribution of support networks of elderly people and to examine the impact senile dementia has on these networks. It postulated that some support networks will be able to cope better with the incidence of dementia and attempted to establish differing service needs on the basis of different types of network. Collaboration took place in Liverpool and Gwynedd. This project was funded by DH/Welsh Office from 1 June 1989 to 31 May 1994.
This study aimed to examine the relationship between sociological, psychological and biological variables in a representative sample of the normal elderly population. This was a collaborative project between Cambridge and Nottingham. It was funded by ESRC for a period of three years.
A joint study of neuropathology has been undertaken at Oxford. The main purpose of this study is to diagnose different types of dementia through examination of tissue from the CFAS brain donation phase, relating the quantity of plaques and tangles with the degree of dementia. The information gained from this study would underpin the validity of the epidemiological findings and provide a firmer foundation for predictions of prevalence of SDAT, VD and other common disorders producing cognitive decline in the elderly. Collaboration is taking place in all six centres, separately funded for the first five years by the MRC: subsequently this linked study was absorbed into the core CFAS activity, and brain collection and examination is ongoing.
At the five year follow up interview of the assessed group, bloods were taken and stored for future analysis. Part of the sample has had DNA extracted and genetic analyses are being conducted.
4.2.7 Study organisation
The Liverpool centre hosted the national co-ordinator between 1990-97. The national co-ordinator was responsible for the overall organisation of the project, the training of the local co-ordinators in the use of the instruments and ensured reliability of the data collected and comparability amongst the centres. This role has now evolved into that of a Study Manager, based at Cambridge. In 2001 a new structure was formed consisting of a lead interviewer, an administrator and ancillary support.
The MRC Biostatistics Unit in Cambridge has acted throughout as the statistical centre for the study, responsible for the centralisation, consolidation and maintenance of the overall study data set. Definitive versions of the dataset are produced, enabling development of past analyses. The computerisation of the interviews and development of the fieldwork administrative database management system has been carried out within the Biostatistics Unit.
Each centre employed a local co-ordinator for the first five years who was responsible for selecting the sample and organising recruitment, training and quality control of interviewers; initiating and maintaining the administrative database; and organisation of the fieldwork at the local level.
4.2.8 Team Members
4.3.1 Sampling Frame
The purpose of the study is to estimate age specific rates of prevalence and incidence of cognitive decline and dementia among those aged 65 and over. The population is thus all those aged 65 years and over on the index date, living within a specified geographic location. Background information on the demographics of the populations sampled was collected from the Office of Population Census and Surveys (OPCS), 1990-91 census now Office of National Statistics, (ONS), to relate to regional and national data.
It is essential that the population lists used for sampling were as accurate as possible. It has already been stated in Section 3.1, that accurate prevalence rates can only be achieved if those who are confined to institutions are included in the sampling frame. Family Health Service Authority (FHSA) lists were used as the sampling frame. The frame would be incomplete if eligible members of the population were not registered with a GP. However individuals in long stay hospitals remain registered with their GP two years after institutionalisation so sampling from FHSA lists should ensure their inclusion. Each centre looked into the practices of long stay hospitals in their area to confirm this. Since the catchment area of general practices have ill defined and overlapping boundaries, it is essential to use the FHSA list of individuals for geographic sampling. Each centre defined a precise geographic area, and the study population was drawn from all those who were resident within it. Problems of inaccuracy, patients who died or moved away but were still on the FHSA list, were resolved by asking GP surgeries to check the lists. In the event of a GP being unable to update the sampling list, the original list was used.
4.3.2 Sample size
The size of the sample for each centre was calculated to satisfy the following aims:
- To estimate differences in prevalence rates with sufficient precision that if the estimated difference is two fold the 95% confidence interval has 1.5 as a lower bound.
- To establish two fold difference in incidence rates with a high degree of power (80% for a 5% significance level).
We might expect a range of prevalence rates from 2.5% to 5% of moderate and severe dementia among all those over 65 years of age. If the average survival of these dementia is 2.5 years, then the corresponding incidence rates would be approximately 1% and 2% per year.
On this basis, a sample of sufficient size to yield 2500 interviews of individuals aged 65 years and over, stratified by age (equal numbers aged 65-74 and 75 plus) was chosen from the FHSA lists for each selected area. With observed prevalence rates of 2.5% and 5%, this would give a 95% confidence interval for the relative risk of 1.5 to 2.67.
The exception to this is Liverpool where a sample size of 6000 was used. This sample was also stratified, by age and sex, to ensure equal numbers within each five year age band.
The index date for the study was 1.10.1990. The English FHSA’s provided computer accessible records for all patients with a date of birth prior to 1.10.1926. In Gwynedd the paper records stored in cabinets in GP surgeries were accessed for the information on the sampled individuals. The records provided name, title, address, sex, date of birth, NHS number and GP code. In order to reduce inaccuracies this information was requested in two batches, at an interval of one year. In the first year one half of the sample was drawn from the first frame (with an index date of 1.10.1990). In the second year the remainder of the sample was drawn from the second frame (with an index date of 1.10.1991).
The sample included only those individuals living within a specific geographic location. In Cambridgeshire this included the rural area of Ely city and surrounding villages. Liverpool included an urban sample from the city of Liverpool based on postal districts L1-L19, L24, L25-27, and drew a rural sample from specific rural areas (Ynys Mon and Dwyfor) in Gwynedd north Wales. Newcastle sampled all postcodes north of the river and within the city boundary. Nottingham was sampled from the entire city excluding four wards that were used in a previous survey. In Oxford the sample was drawn from the city on postcodes 1-4.
Each annual sample selected 1850 individuals at random from the frame, stratified into two age groups, (under 75 and 75 and above), 925 per group. This allowed for loss before interview of 600 people. These losses were anticipated through GP refusal, non eligibility of individuals, (death, incorrect date of birth, incorrect address), as identified by GP records before interview, and loss to interview, (death, incorrect address, uncontactable, refusal), after sampling. These figures were calculated using information taken from the first ten weeks of interviewing in Liverpool where the final GP consent rate was 89%, 95% of individuals were identified as eligible by their GP, and 80% of individuals were actually available for interview.
1850 x 0.89 x 0.95 x 0.80 = 1251.34
Therefore drawing an initial sample of 1850 yielded a working sample of approximately 1250. This oversampling fraction was provided for use in the absence of accurate information for each centre. Adjustments to the sample size were made towards the end of the year to allow for inadequate or excessive numbers.
GP practices were randomly assigned to two equal sized groups, and the sample drawn from practice group 1 in year one and from practice group 2 in year two. This avoided the necessity of contacting a GP twice in two years and eased administration when attempting to assign interviews in the same area. The actual numbers of GP’s in each group on the index date were different, because of resignations and new appointments. New GP’s were assigned to groups according to their practices. This resulted in minimal movement of patients between year groups, since they usually stay with the same practice. New practices were assigned randomly.
The basic study plan was to examine twice the cognitive function of a population of individuals aged 65 years and over, with an interval of two years between the two examinations. One month after the initial interview a 20% subsample was assessed and followed up on an annual or biannual basis. At the two year follow up of the 80% not assessed at first interview, a further 20% subsample was assessed one month later and again followed up on an annual or biannual basis.
The first examination was conducted via a screening interview of all available and consenting individuals, half of whom were screened in year one and the remainder in year two. The purpose of this first interview was to (i) screen for possible cases of dementia, (ii) provide a baseline for the study of cognitive decline (iii) provide an assessment of risk factors for decline in the population for future nested case control studies of incident cases and (iv) to provide a sample for the Resource Implication Study. Two years, or as near as practicable, after their first interview all individuals who were still alive and had not had the assessment interview were screened again - half the sample in year three and the remainder in year four. The purpose of this interview was to (i) screen for possible incident cases of dementia and (ii) provide information on change in cognitive function. Where an initial interview is refused on another’s behalf a proxy screening interview was sought. All in-scope subjects (i.e. alive and living in the area on sample date) were flagged on the NHS Central Register by ONS and quarterly returns of notifications of death and causes coded to ICD9, cancellations of cipher, emigration, long-term institutionalisation, etc. are received. From January 2001 cause of death is coded to ICD10.
Between two and four weeks after screening, 20% of individuals had a diagnostic interview. These individuals were selected on the basis of cognitive function. All those who had AGECAT scores of 03 and above; plus all those who had an incomplete MMSE and MMSE scores of less than or equal to 21 were chosen for assessment. Individuals who appeared severely cognitively impaired were asked a limited range of questions. Prevalence of dementia was estimated on the basis of this 20% sample plus extrapolation to the screening interview.
A smaller sample of those assessed was selected to take part in an interim interview at one year, and another at three years, to provide information on those who drop out at different stages of the study.
Use of common instruments and sampling procedures, with joint training of the interviewers ensured rigorous comparability among centres.
4.5 Assessment instruments
The two main instruments for assessment in psychiatric epidemiology among the elderly of organic disorders and cognitive change are the Geriatric Mental State (GMS) with the associated History and Aetiology Schedule (HAS) and the diagnostic procedures in AGECAT and the Cambridge Mental Disorders Examination (CAMDEX).
The usual approach when using a two-stage assessment is to choose a sensitive but relatively non-specific screening test, which is low cost and easy to apply, in order to select efficiently those to whom the more accurate instrument will be applied. The final classification was based on the second stage assessment for those to whom it was applied; those selected at the first stage for no further testing were classified according to the result of the first stage test, by extrapolation from the second. It was essential for the informativeness of the study that the initial screening test be almost as sensitive as the second-stage test. Lack of specificity of the initial test does not seriously affect the validity of informativeness of the study, but it does affect the cost. These considerations meant the first-stage test was aimed specifically at cognitive function, such as the extended MMSE (Mini Mental State Examination). The second-stage was a more complete psychiatric assessment including the same cognitive test.
The first stage examination was conducted via the screening interview. This consists of items from the organicity scale of the GMS; the MMSE and additional cognitive items; essential Activities of Daily Living (ADL); questions regarding sociodemographic variables such as social class, socioeconomic group and educational level; and questions regarding risk factors that are in line with EURODEM. The study in Liverpool differs in that the entire screening and assessment phase are contained in the first interview. For this reason not all the variables above are included although the core essentials remain the same.
The second-stage assessment was conducted using GMS (B3), CAMCOG and HAS. Some modifications were made to the HAS so that the Blessed Dementia Rating Scale and Hachinski could be calculated for all respondents. Some minor additions to the GMS (B3) were made so that the CAMDEX format could be extracted for CAMDEX diagnoses. The information collected using these instruments allowed application of the DSM III and ICD 10 diagnostic algorithms. The GMS was chosen in order to increase comparability with European collaborative studies. The annual follow-up interview comprised GMS (B3), CAMCOG and extra cognitive items.
4.6.1 The sample
The FHSA provided computerised records of all patients born before 1.10.1926. The sample was drawn and the relevant GP’s consent requested. The size of the sample was reduced by invalid GP codes, GP’s subsequently assigned to another group and addresses outside the sampling location. There was further post hoc elimination of subjects who were found to have died or to have moved outside the area before the index date. These losses were allowed for by oversampling, (Section 4.3.3). Actual response rates were calculated on the basis of this reduced, in-scope sample.
Each local co-ordinator organised the local fieldwork. The administrative database management system, which was developed in Cambridge for use in all centres, enabled allocation of batches of interviews to individual interviewers. Records were kept for each subject; when the letter was sent, date and time of interview, which interviewer was allocated the interview, whether the interview was successfully completed and whether the individual was asked to participate in the Resource Implications study or the ESRC Study of Healthy Ageing, or for follow-up in the main study. Local co-ordinators were responsible for maintaining a detailed audit trail of the progression of the study, e.g. method of sampling from FHSA, sampling fraction used, outcome for each approach, etc. These audits trails were collated by the national co-ordinator.
4.6.3 Approach for interview
Each potential participant was sent an introductory letter signed by the GP (organised by the study centre). At some centres a letter from the centre was also sent, explaining the study, giving the date and timespan for a potential interview and a contact telephone number for queries or refusals. In other centres the GP letter was followed by a personal visit from the interviewer to explain the study and to arrange a date for interview. This letter was viewed as an option to improve compliance and not a necessary condition of the study.
4.6.4 The interviews
The screening interview was conducted in the respondent’s place of residence, using portable computers with software customised centrally by the MRC Biostatistics Unit. The software delivered the questionnaire to the screen and stored the answers onto a 3.5" floppy disk. Also recorded were details of the interviewer, time of start and finish and outcome of interview. This allowed the local co-ordinator to examine variability between interviewers in relation to length of interview and outcome. At the end, the interviewer was notified on screen whether the subject was eligible for the RIS or ESRC Study, and consent was requested from both the subject and their carer (HSRU provided Biostatistics with the eligibility criteria for the Resource Implications Study). Interviews lasted for between 30 and 45 minutes, but would be much shorter if the respondent was confused or demented or too frail to answer the complete set. If responses to the first set of questions indicated that the respondent was unfocussed in time and place, the program automatically invoked ‘priority mode’ and a very short subset of questions were asked. If the interviewer felt that the respondent was frail and tiring, or becoming agitated, the short ‘priority mode’ set of questions could be invoked manually. Interviews were conducted by lay interviewers, recruited for the purpose and trained by both the local and national co-ordinator. Reliability checks were made by both the local and the national co-ordinator. Proxy screening interviews were conducted where an interview was not possible with the named participant, due to e.g. extreme confusion or frailty. If after four attempts to contact, an interview was not arranged, the approach was abandoned.
The assessment and annual interviews were also conducted using portable computers customised by the MRC Biostatistics Unit. These interviews were conducted by interviewers other than those used for the screening instrument also recruited and trained for the purpose. Interviewers did not know the outcome of the first interview. The interviews lasted from 45 to 90 minutes, again with a ‘priority mode’ route. The annual interview consisted of a combined screen and assessment, where information on changes since last interview were recorded. At the assessment and biannual follow up interview permission was sought to approach and interview a relative or carer to ask for an objective account of the respondent’s health and abilities.
Six years after the initial screening interview, all respondents in the assessed groups were reinterviewed using the combined screen and assessment interview and at the end of that interview signed permission was requested to take a sample of blood or saliva. Permission was also sought to access GP and hospital notes.
As the power of the study lies in finding incident cases, many attempts were made to achieve follow-up interviews. Strategies included asking at the GP surgery, checking local institutions, hospitals, neighbours, phone book and the FHSA. If the individual was known to have moved to an area close to one of the other centres in the study, reciprocal arrangements were made for an interview.
All respondents who were selected to be assessed were eligible for approach for a declaration of intent to donate brain tissue at post mortem (DOI). The paper records were first examined to ensure that there were no reasons why approach should not be made. The GP surgery was informed that the research nurse was in the area, and GP managers were very helpful in locating respondents. First contact was by phone, possibly via the relative or carer who had previously been interviewed at the assessment interview stage, to arrange an informal visit.
The visit begins as an informal chat: no mention of DOI is made if there has been a recent emotional upset, or the respondent is depressed, or if there is lack of insight due to confusion, dementia, etc., or there is another person present who indicates disapproval in any way.
If appropriate, at subsequent meeting, the subject of ‘donating brain tissue after death’ is raised. The family are mentioned in the course of the discussion. Sharing the topic with the family is encouraged and leaflets are left for respondents and family members to consider. Further contact is encouraged to answer further questions. Those who make an independent decision are encouraged to speak to their family about it.
If any member of the family does not agree, DOI is not signed, wishes of all involved are respected.
If a DOI is signed, more than one relative’s name should be obtained as next of kin. A thank you letter is sent and a donor card is given, with a suggestion that it be placed with legal papers which would be accessed in the event of death. Confidentiality is stressed - only GPs and hospitals will be informed.
In nursing/residential homes, a sticker is put on the notes and a copy of the DOI plus instructions is attached. GP notes have a sticker attached. Regular contact is kept with GPs and occurrences such as change of address are noted. Outside the area practice staff are contacted - there is greater potential for missing brains in these instances.
4.6.6 Retrospective Informant Interview (RInI)
When a respondent who has signed a DOI dies, and the brain is donated and stored at the centre’s brain bank, the next of kin may request information from the initial examination of the brain. Some time later, not earlier than 3 months from the date of death and avoiding birthdays and anniversary of death, an approach is made to the relative to ask if they would be willing to give some information about the deceased in the period prior to last illness. The aim of this interview is to gather details of any incident conditions or deterioration which may be significant in the final diagnosis from the brain examination.
4.7 Data processing and analysis
Data from each interview was fed back via the computerised system. The administrative file was set up to record the date of interview and next anticipated step. The confidential data from the interview was separated into anonymised files and transferred to the MRC Biostatistics Unit at regular intervals. Coding of textual information, i.e. occupation, social class, SEG, medication was organised by the administrative centre and Biostatistics Unit. Clearly defined versions of the dataset are released as necessary.
See list of Publications
4.9 Amendments required by the Anglian and Oxford MREC July 1999
This is a continuation of the Study which began in 1990. No new participants will be recruited, and the interview schedules have been used in the last wave of the study. Ethical approval has been sought and given at each of the six centres in the past.
1. Support of interviewees who might become distressed during the course of interviews. This topic is addressed in interviewer training, and the advice is to offer to halt the interview, and to address the distress as appropriate, possibly involving a relative or carer. If necessary the interview will be abandoned, or rescheduled for a later date, possibly in the presence of a relative or carer. Alternatively an interview with a relative or carer may be carried out instead. The interviewees will have been interviewed at least 4 times since 1991, using an interview schedule very similar to the current one.
2. Criteria for recruitment of the 500 subjects involved in the study. The baseline population for the study (13006) was recruited in 1991-3 via an approach to a randomly sampled group from FHSA lists, (see study diagram). At the end of the first interview 20% were selected on the basis of cognitive function for a assessment, or diagnostic, interview (see section 4.4). This group have been followed up intensively since then, and if it was considered appropriate were approached for a declaration of intent to donate brain tissue at post mortem (DOI) (see section 4.6.5). The 80% not selected for assessment were reinterviewed after 2 years (incidence screen) and 20% of them were selected for assessment. The cases in this group were approached for a DOI. Procedures for approach for a DOI are described in section 4.6.5, of the protocol. So the people whom we are interviewing in this phase of the study were recruited in 1990-1, first interviewed in 1991-3, and subsequently at yearly or two yearly intervals. They have already signed a declaration of intent. They were last seen over 2 years ago, and our aim is to maintain contact to continue to map individual progress and to minimise time between last interview and death.
3. Blood taking. There will be no blood taking at this phase. Bloods were taken (with signed permission) two to three years ago. Bloods were taken in Oxford at the second wave of interviews and in other centres at the third wave.(see study diagram). We plan to measure homocysteine, folate, vitamin B-12 and methylmalonic acid (MMA) levels (if possible) on all 1670 Oxford blood samples. We also plan to measure homocysteine, folate and vitamin B-12 on the remaining 1000 samples from the other CFAS centres. We will be analysing candidate genes for dementia, cognitive decline and Alzheimer disease. Samples will be analysed either by case-control association studies (Alzheimer disease and dementia) or using a quantitative trait approach (cognitive decline).
4. Mention of donation of brain tissue in letter of invitation. The 4 letters of approach for interview do not mention donation of brain tissue. Two centres will continue to approach for DoI during this phase of interviewing. The nurse telephones to arrange a visit and would not speak of brain tissue donation unless she judged it appropriate. In past waves of the study, local ethical committees have required that certain criteria concerning this be followed; e.g. at Cambridge, all procedures which were anticipated had to be stated at the beginning - thus the information sheet (given out in 1991-2) included a sentence about tissue donation.
MRC CFAS Activity from January 2001-2005
The MRC Cognitive Function and Ageing Study is a multicentre multidisciplinary research programme of health and frailty in ageing populations. In the initial Co-operative grant application the following aims were identified: to support essential fieldwork at ten years to allow measurement of cognitive, functional and health changes at ten years from baseline; to estimate incidence; to examine risk and natural history; to continue the brain donation programme across the centres and to support the neuropathological analysis; to collate and prepare data collected in previous waves of the study for archiving; to provide the core infrastructure to support component grants. The Cross-Board reviewed the co-operative proposals and a dedicated working group recommended, in addition, that CFAS should produce an outline of the strategy for further development of research and the Co-operative as a whole. In addition the study was required to support a Biological Resource Advisory Group with independent membership and to ensure lay membership of its management group. Once the Co-operative was established it became clear that dedicated resources were needed to manage and document the biological resource.
2.0 Objectives of the Co-operative
The full objectives were to:
- - produce a scientific strategy for the study
- - revise the terms of reference of the Biological Resource Advisory Group
- - seek lay members for the Management Committee of the Co-operative
- - reinterview all survivors in the five identical sites plus those from Liverpool who have declaration of intention to donate tissue (DOI) in place with incorporation of the data into the archive
- - document and make accessible to collaborators a core dataset and develop mechanisms to extend this to new data acquisition
- - document and support the biological resource of the study, in particular the brain banks and to support ethical and governance structures
- - develop the administrative and analytical database necessary to allow flexible and diverse component grant collaboration
3.1 The scientific strategy
This document was developed within the Co-operative during 2001 and a final version submitted to MRC January 2002 (http://www-cfas.medschl.cam.ac.uk). This laid out the strategy for science which concentrated on the strengths of CFAS – the collection of the major dimensions that determine quality of life in older people: cognition, function, affect and health, its population base, diverse sites, longitudinal nature and a unique biological resource. It outlined the openness to collaboration of the group and defined five major themes to encourage further work, each with a lead member. These were:
- - dementia and cognition (Matthews)
- - depression (Dewey)
- - disability and healthy active life expectancy (Jagger)
- - health and health policy (Bond)
- - cohort and new developments (not described in this document)
In addition two supporting themes were identified
- - ethics and governance (Parry)
- - data and analysis methods (Johnson)
Existing components and collaborations were linked to each of these, and new components and collaborations sought in each with regular updating from each theme group to the Co-operative. Each theme developed a set of aims and a summary of intentions and research activities.
3.2 Biological Resource Advisory Group.
This group (renamed the Biological Resource Advisory Committee, “BRAC”) has met each year and has reviewed the original Terms of Reference with several revisions. The current version is available on the CFAS website (http://www-cfas.medschl.cam.ac.uk/advisory_committee.htm). The membership has been extended to include four honest brokers, (two with genetics expertise and two with neurobiological expertise) and a lead for the lay members. BRAC meets annually and the honest brokers review applications for use of biological resources via email, before MREC submission. The Chair of the BRAC (Professor Ingrid Allen) is kept fully informed of the many issues relevant to research and ethical governance arising from the biological, particularly the brain, resource.
3.3 Lay membership of the study
Three lay representatives have become full members of the Co-operative Management Committee. Mrs Brenda Barber via the Alzheimer’s Society, The Venerable Lorys Davies via the Hospital Chaplaincy and Mrs Diana Lloyd via Age Concern. We acknowledge our lay members’ enthusiasm for the study and their willingness to extend their contribution beyond attendance at the Management meeting, in particular to address the complex ethical issues within the study that have arisen from external authorities.
3.4 Fieldwork and brain donation
During 2002/3 all survivors in the Cambridgeshire, Gwynedd, Newcastle, Nottingham and Oxford centres have been re-interviewed (Annex 5). The small numbers with declarations of intention for post-mortem donation in Liverpool were re-interviewed. Response rates were around 73%. Interviews were fed back to the centre shortly after completion and checked so that queries or problems were identified rapidly. This has been achieved by incorporation within the core team of a fieldwork lead ensuring quality, and an information and data manager within the MRC BSU to receive data, identify and resolve problems. A newsletter was provided for all our participants and stakeholders during the interviewing stage (commended by the MREC and our lay members).
An accurate audit trail for our participants throughout the study was created, which is provided on the website (http://www.cfas.ac.uk/pages/dataandanalysis/index.html). Respondents with declarations of intention to donate tissue in place have had this honoured wherever possible. The numbers of successful donations is now approaching 470. The success rate for donation after death from individuals with declarations of intention to donate tissue has been maintained at 74%.
3.5 Data documentation and release and web development
There has been extensive work on the datasets, with full audits of responses. Adequate support of core infrastructure for the study has enabled release of new data versions so that the aim of providing a core dataset of the early waves of the study with attendant documentation has been achieved. In addition data from the current fieldwork are already available ensuring timely analysis. Data generated returns to the archive at MRC Biostatistics Unit as a condition of collaboration, and publications must be quality checked and properly cite use of CFAS data. The data management team is coding and checking all scales originally embedded in the interview using original publications. These are released into a dataset alongside the versioned interview data, together with notes and references on the website. Data are sent to collaborators tailored to their requested analysis format than enables then preservation of labels and missing values.
The website (http://www.cfas.ac.uk) undergoes continuous development and now contains a full description of the study, its history, component parts, administrative and ethical structures. Support for collaboration is now well developed. The website shows the route to request collaboration, from first idea through to request for data or tissue. Research governance in relation to the data and study output has been improved through the use of formal data and tissue request forms, available on the web, which are discussed and decided upon by the management group (often by email).
3.6 Brain donation and ethics
Since the year 2000 organ donation and retention after death in the UK has received intense public attention. Despite CFAS being one of the few studies where appropriate consent has been sought before death from respondents from the outset in a sensitive and timely manner it is not surprising that all aspects of the programme have been scrutinised in detail. The brain donation programme was originally funded through CFAS researchers in different centres who sought approval from local Ethic Committees. There was no requirement for central coordination of either the ethical processes or the archiving of actual tissues. Awareness of the need to address potential differences in these processes and account for the resource more completely led to the successful request for supplementation, described in 3.6.1. During 2003 an information sheet and consent form for those next of kin providing a Retrospective Informant Interview were introduced. In 2004 following recent advances in medical research, MREC required us to seek additional and more specific approval for use of donated brains those respondents who had a DOI in place and had not refused further contact with the study where approached and asked to sign a new DOI form. Copies of the new Rini and DOI forms can be found on the website. (http://www.cfas.ac.uk/pages/consentforms/index.html)
3.6.1 Support for neuropathological studies
Supplementation of the Core CFAS grant has funded a tissue resource co-ordinator, based in Sheffield, with the tasks of preparation of a tissue database, attendant documentation and preparation of material for individual CFAS neuropathology projects. The funding of this post has enabled us to collate dispersed documentation, from earliest stages of the neuropathology study to the present, enabling review by BRAC and MREC, as well as audit the biological resource which is distributed across several sites and document any local research use not previously known to CFAS. Cambridge and Newcastle retain their material at present as they both have active brain banks; the Oxford, Nottingham, Liverpool and Gwynedd materials are being moved to Sheffield.
3.7 Support for components
One of the major areas of core activity that has grown over the three years is the need to support component applicants and others in seeking ethical approval. The majority of recent applications to conduct research in CFAS have required new ethical submissions, including straightforward analytical work within the core (such as new genetic or actuarial analyses based on existing data). These processes, required for research governance and accountability, include the need for approval by CMC, BRAC and then MREC and LREC.
3.7.2 Data support
This has been critical to the development, and follow-through, of many of the projects mentioned below and is fully described under the report from the data and analysis theme (section 4.6). The Co-operative has functioned to provide multidisciplinary support as well as to provide the detailed knowledge of the study that makes meaningful analysis possible. This has included the addition of new centres and members to the Co-operative: Leicester (Jagger), Sheffield (Ince), London (Parry) and Edinburgh (Macdonald).
This section is presented by theme group as outlined above, summarising aims, and membership.
4.1 Dementia and cognition
This group incorporates the key aims of CFAS. The group members are multidisciplinary (biostatistics, epidemiology, public health, old age medicine, old age psychiatry, general practice, health services research, neuropsychology, neuropathology, genetics, molecular biology).
Investigation of depression within CFAS depends on data from the assessment stages of the study. The depression theme group includes: biostatistics, health service research, neuropathology, public health and epidemiology, old age psychiatry.
4.3 Healthy Active Life Expectancy, functional disability and survival
The aim of the group is to extend work on the areas of disability and measures of health expectancy, both cross-sectional and longitudinal, and including Healthy Active Life Expectancy (HALE), a key indicator in government policies for older people. The group expertise includes biostatistics, old age psychiatry, public health, epidemiology and health services research.
4.4 Health and health policy
There exists a large amount of data within CFAS and particularly the Resource Implications Study (RIS) data set concerning broader health and social policy issues including detailed data on direct and indirect costs, social support and social networks, use of health and personal social services and medication. Secondary analysis of these data and longitudinal analyses of summary data from different waves of the study provide substantial policy analysis on healthy ageing, informal care, and long-term care and resource consequences of ill health in later life. The focus of activity has been on the use of the sound CFAS epidemiological and RIS health economic and health services data (from the prevalence phase) in actuarial and economic modelling. Other analyses of the main CFAS dataset have also been included within this theme where they did not fit coherently with the remit of other theme groups. Members of theme group are drawn from a wide range of disciplines including: actuarial mathematics, biostatistics, health economics, health service research, health psychology, neuropsychology, old age psychiatry, public health, social gerontology, social policy and sociology.
4.5 Ethics and research governance
The theme is dealing with the legal, ethical and economic challenges for human biological sample collections, other activity supports the research of other theme groups and is described elsewhere. Dr Parry has written three papers and given several presentations arising from this work. It has had an impact within the work of the Genetics Knowledge Park in Cambridge, and is contributing to current national and international debates on legal, ethical and economic issues related to tissue banking.
4.6 Data and analysis theme
The support for the data and analytical aspects of the study provided by this theme has been key in the Co-operative. It has drawn on the long term involvement of core members of the collaborative group, particularly those in Cambridge in the MRC Biostatistics Unit, and the University Department of Public Health and Primary Care (and similarly for the Resource Implication Study data in Newcastle). This combined with a consistent approach to data cleaning and database development with release of versioned datasets has been central to study integrity.
The core support of the Co-operative for data management, liaison with collaborators (including support of students) and checking of analyses and papers at output stage has been critical to their progression. There has been methodological development of incidence analysis in studies with longitudinal subsampling, cognitive decline and Markov Chain methods for life expectancy, and multiple imputation methods have been applied to estimate the prevalence of depression.
Objectives and proposed activity for 2005 onward
Continuing support for active components, conduct of components.
Re-interviewing of those respondents with DOI in place at all centres,
Bringing together a hard copy archive for the main study and biological resource
Analysis of the ten year data
Further analysis of risk including blood based analyses within the ten year follow up.
Dissemination of research results
Other than oral presentations and publications, reports and book chapters we will communicate the results for the study to the local communities through appropriate bodies locally, regionally, nationally and internationally. Public lectures and discussions will be held in some centres. We have previously sent out newsletters and would propose to do so for those who are invited to interview during the current phase. This newsletter is distributed to health and social service bodies, voluntary agencies in the relevant areas as well as to national organisations.
Version 4 - 4/5/05
- Research Information
- Brief Introduction
- Research Themes
- Bolt-on Studies
- Glossary of Terms
- Design and Interview
- Study Protocol
- Scientific Strategy
- Study Protocol
- Study Information
- Consent Forms
- Neuropathology Forms
- CFAS I Questionnaires
- CFAS II Questionnaires
- Data & Analysis
- Liverpool Data
- Advisory Committee
- Biological Resources
- Ethics & Legal Aspects
- Archive Documents
- Management Structure