Scottish Government

Chapter Three Demographics

3.1 This Chapter examines the baseline data on demography, household structure and health in older age.

3.2 Demography, and more specifically the ageing of the population, will play an important role in the evolution of the costs of care. This has been acknowledged in much of the previous research on the costs of long-term care, including the costs of personal and nursing care. See for example, the Royal Commission into Long-Term Care (1999), the CDG Report (Care Development Group 2001) and the Range and Capacity Review (Scottish Executive 2004b).

3.3 Household structure is an aspect of demography which is also an important factor in determining formal care needs. Older people living alone are more likely to require formal support than those living in multi-adult households, where care and support can be provided informally.

3.4 The health of older people also affects the demand for care. The larger the proportion of the population who need assistance with Activities of Daily Living ( ADLs) or Instrumental Activities of Daily Living ( IADLs), the greater will be the demand for care services.

3.5 In this chapter we review each of these three issues. In particular, we examine the evidence that was available to the CDG at the time that it drew up the report and comment on how this information might be modified in the light of more recent information. We also describe, where possible and relevant, the evolution of this information since 2001. We begin by considering the demographic information contained in the CDG report.

Demography

3.6 Population estimates were an essential component of the CDG's estimate of the costs of personal and nursing care in 2001 and the projections of these costs to 2026. The CDG's estimates were based on the 1998 Government Actuary's Department ( GAD) demographic projections. These estimates were therefore extremely important in the decision to proceed with legislation to provide free personal and nursing care. GAD forecasts were also used in the Range and Capacity Review ( RCR) which produced forecasts of service users, workforce and costs for various scenarios projected forward to 2019.

3.7 Between the publication of the CDG report and the RCR, the results of the 2001 Census became known. These indicated that previous GAD projections had underestimated the size of the population of older age groups in Scotland. Thus the CDG report stated that in 2001 "there are around 790,000 people aged 65 or over in Scotland" ( CDG 2001, p41, para 5.11). The 1998 GAD projection, on which this statement was based, reflected the 1991 Census updated with information from the deaths registrations.

3.8 The 2001 Census indicated that there were in fact 805,000 Scottish residents in this age group. Table 3.1 shows the estimates of the 2001 Scottish population made at various times prior to 2001 and the results from the 2001 Census. This underestimate of the size of the older age groups in Scotland clearly biased the estimates of the costs of care downwards. The numbers aged 85 and over were also underestimated. This is particularly important for cost estimates because this group have relatively high levels of disability and it is therefore more costly to provide care for this age group. The CDG report suggested that there were 86,000 in this age group, whereas the Census estimate was 88,000.

Table 3.1 Forecasts of the Scottish population in 2001 by age group

Notes to Table
Source: GAD Population Projections Database

3.9 Although the differences between the population data used by the CDG and the 2001 Census results are only around 2 per cent, the 2001 Census had a significant effect on the forward projections of the population of older people in Scotland. These were reflected in the RCR projections which were able to take account of the 2001 Census information. It is difficult to reconcile the RCR with the CDG projections directly because they use different age groups, and different methodologies. The CDG demography concentrated on the 60+ age group at five year intervals from 2001 to 2026, whereas the RCR focussed on the 65+ age group at five year intervals from 2004 to 2019.

3.10 The need for social care is a risk which is faced by the whole population. One can deal with this risk either through 'pay as you go' or 'funded' insurance provision. With 'pay as you go' current payments into the scheme are used to fund its outgoings. A funded scheme is one where a lump sum is built up from individual contributions. This is subsequently drawn down by those individuals to whom the adverse event occurs. Scotland has relied on a 'pay as you go' mechanism with local authorities bearing much of the financial risk. The private market for long-term care insurance policies is very small and shows no sign of growth.

3.11 Changes in the GAD forecasts of the number of older people in Scotland had a significant effect on the estimates of the future cost of free personal care. Table 3.2 shows the GAD forecasts (from which both the CDG and the RCR were derived) for 2019. The forecasts were made between 1998 and 2004. They show that the estimate of the number of Scots aged 60+ in 2019 increased by 7 per cent, from 1. 3m to 1. 39m between the 1998 and 2004 forecasts. Over the same period the forecasts of the number aged 65+ in 2019 increased by 9 per cent and, crucially, the forecast of those aged 85+ increased by 18 per cent.

Table 3.2 Forecasts of the Scottish population in 2019 by age group

Notes to Table
Source: GAD Population Projections Database

3.12 In recent decades, demographers have observed increases in life expectancy among the 'oldest old' in many developed countries. These have been well in excess of previous demographic predictions (Kannisto 1994). Thus, it was not surprising that the size of the oldest age groups in Scotland at the 2001 census was larger than expected. Recent research such as that of Mesle, Vallin and Andreyev (2002) on improving demographic forecasts may enhance the robustness of estimates of the numbers of the oldest old.

3.13 Although neither the CDG nor the RCR reports looked as far as 2041, the increases in the forecasts of the population aged 85+ for this year have been even more dramatic. Table 3.3 shows population projections for 2041 made by GAD using different base populations. They show that moving from the 2001 base to the 2004 base increased the GAD estimates of the size of the population aged 85+ in Scotland in 2041 by 46 per cent, from 186 thousand to 271 thousand. Again, this outcome is the result of forward projection of an unexpectedly large number of the 'oldest old' in the 2001 Census. If the presumption that care costs increase with age is correct, then the increasing size of this age group would suggest substantially increasing costs in the future. However, this assumption is currently the subject of considerable controversy. Seshami and Gray (2004), using longitudinal data on hospital costs in Oxford, argue that costs rise with proximity to death rather than with age. Stooker et al. (2001) similarly argue that care costs increase substantially in the last year of life. This is an extremely important issue, both for care and health costs. However, it has not been seriously explored in Scotland and we would recommend that the Executive commission a literature review of this issue, including consideration of lessons from elsewhere for the Scottish context.

Table 3.3 Forecasts of the Scottish population in 2041 by age group

Notes to Table
Source: GAD Population Projections Database

3.14 The CDG report included a "funnel of doubt" in its estimates of the future costs of personal care in Scotland. However, this doubt was dependent on factors such as healthy life expectancy and the costs of care - not on demography. The substantial increase in the GAD forecasts of older people in Scotland since the publication of the CDG forecasts suggests that uncertainty over demography is also an important element of the overall uncertainty associated with the costs of personal care. The difficulties of forecasting the size of the older population were perhaps not appreciated at the time at which the CDG report was produced, but the uncertainty around demographic forecasts should certainly be borne in mind in future costing exercises.

3.15 Table 3.4 shows the breakdown of the Scottish population living in communal establishments or private households from the 2001 Census. Around 34,000 of those aged 75 and over live in communal establishments - these will largely be care homes or hospitals. More than a fifth of those aged 85 and over lived in communal establishments in 2001. Unfortunately the breakdown by age group does not enable us to make a precise comparison by age group.

Table 3.4 Population living in communal establishments by age group

Notes to Table
Source: 2001 Census

Household structure

3.16 Household structure has an important bearing on the costs of care. Living arrangements have an important impact on the use of formal services by older people (Chappell 1985). Those living with others are more likely to use informal sources (Carriere et al. 2001). Childless older people and those living apart from their children are more likely to use social services than those who live with their children (Choi 1994).

3.17 The General Register Office for Scotland ( GROS) is responsible for projecting the number of different types of households in Scotland. These projections are typically shorter than the population projections produced by GAD. Thus the 1998-based household projection covers the period from 1998 to 2012, whereas the most recent, 2002-based projection extends only as far as 2014.

3.18 For this project, the number of households where the head of household is aged over 65 is of particular interest. This information was not used to inform the CDG estimates of the costs of free personal care. Given the importance of household structure noted above, one might have expected more attention to be paid to estimates of the future number and composition of households in Scotland. The 1998-based forecasts suggest that the proportion of single adult households where the head of household is aged 65-74 will rise from 47.3 per cent of households in 2002 to 49 per cent in 2012, while those aged 75+ will increase from 69.9 per cent to 73.3 per cent over the same period. Such increases would tend to put upward pressure on the costs of care but were not taken account of by the CDG.

3.19 The 2002-based household projections are able to incorporate additional information from the 2001 Census. They show a sharp fall in the estimates of the proportion of older single households in Scotland. The estimate of the population of households with a single adult aged 65-74 in 2002 fell to 44.3 per cent of the total households with either a single adult or two adults. For those aged 75+, the proportion fell to 65.1 per cent. Importantly, while all other aspects of the demography of older people appear to be increasing, the 2002 household based projections suggest that in 2012 there will be 34,000 fewer single adult households where the householder is aged 65 or over than was forecast in the 1998 projections.

3.20 The methodology for estimating numbers of households did not change between 1998 and 2002. It is based on a two-point exponential model, which uses information from the two most recent censuses. No other information is used and therefore the projections of the number of households in Scotland do not incorporate any additional information on the likely impact of social trends. Thus the reason that the 2002-based projections show no significant trend in the proportion of single adult older households between 2002 and 2016 reflects information on trends in household formation between the 1991 and 2001 censuses that were not available for the 1998 projection.

3.21 Unlike the more recent population projections among older age groups, the recent household projections might have caused the CDG to reduce the size of the estimated increase in the costs of care, since formal care requirements are likely to be lower when a partner is available to provide informal care. It is difficult to forecast the net effect on the costs of care of the reduction in the numbers of older people living alone relative to the increase in the population of older people.

3.22 The relationship between household size and care costs is not well understood in Scotland. There is some information for other countries, but further research in this area as it relates to older people would be of value in better understanding how these costs may evolve; it could usefully be linked with the research on general household formation that is planned by GROS.

Healthy life expectancy

3.23 Information on the length of time that individuals can expect to remain disability free has an important bearing on the current and expected future costs of care. However, this is an area that is beset by data difficulties and where there is little consensus on future trends. To anticipate our conclusions, the classification systems for health and disability are numerous and relevant surveys are few, often do not focus on the relevant population, and tend to ask questions that do not necessarily help to usefully classify individuals.

3.24 The CDG had a limited amount of information on disability and trends in health expectancy among older people. Available datasets had limited utility for detailed analysis of health expectancy. The CDG relied on work done by Stearns and Butterworth (2001) which concluded that:

"Overall, disabled elderly people (in all living situations combined) decreased as a proportion of the population between 1985 and 1996/7. The best estimate of the reduction overall is of 0.2 or 0.3 percentage points per year in the UK, though evidence indicates that the rate of reduction may have been slightly greater in Scotland". (Stearns and Butterworth 2001 p 155)

3.25 This improvement in the overall health of older people was included in the estimates of the CDG and thus played an important role in determining the estimated costs of free personal and nursing care (see CDG 2001: 43). It was based on a comparison of the Family Resources Survey ( FRS) 1996/97 Disability Follow-Up Survey and the Office of Population Census and Survey ( OPCS) 1985 Survey of Disability among Adults in Private Households.

3.26 The size of the Scottish sample for the OPCS survey was 691, while that for the FRS Disability Survey was 345 and the sample frame excluded households living North of the Caledonian Canal. Stearns and Butterworth readily admit that there are difficulties in making strong conclusions from such small samples. Nevertheless, their estimates played a crucial role in determining the final costings arrived at by the CDG.

3.27 Table 3.5 shows Stearns and Butterworth's estimates of the numbers of disabled people in Scotland broken down by household type. The estimates of the numbers in private households draw on the information in the OPCS and FRS Disability surveys. It suggests that there was a rapid increase in special needs housing between 1985 and 1996/97 and also a significant reduction in the number of disabled individuals living in private households. This reduction is largely responsible for an estimated overall reduction in the number of disabled people in the Scottish population from 386,000 to 315,000 over this period.

Table 3.5 Disabled people aged 65 and over in Scotland

Notes to Table
Source: Stearns and Butterworth (2001)

3.28 If this exercise was repeated today, the 1996/97 FRS Disability Follow-Up would still be a main source of data since it remains the only survey which asked specific questions related to personal care. However, it could possibly be supplemented by data from the Scottish Health Survey. In the 1998 Scottish Health Survey, the sample aged 65 and over was 1463. Unfortunately its sample frame was the population of Scotland aged between 2 and 74 living in private households. Residents living in institutions were excluded from the survey. When the 2001 Scottish Health Survey becomes available, it will provide a more authoritative analysis of the prevalence of disability among the Scottish population at the time of the introduction of free personal care than was available to the CDG.

3.29 Both estimates of the disabled population are well in excess of measures of activity in caring for, or providing resources to, older disabled people in Scotland. For example, there are 134,000 individuals aged 65+ that currently claim Attendance Allowance in Scotland, which is a non-means tested DWP benefit payable to those aged over 65 who have an illness or disability and need help with personal care - they do not have to receive personal care. The Scottish Executive (2004a) estimated that there were 40,000 recipients of free personal care living in private households in June 2004. That these numbers are much smaller than the overall estimates of disability is not a surprise, since being disabled does not necessarily imply a need for care. However, if the ultimate objective is to forecast care needs, then a further filter is required in order to calculate the share of the disabled population that does require care.

3.30 We have reanalysed the 1996/97 FRS Disability Survey. Table 3.6 shows the percentage breakdown of the population with disabilities in Scotland by age group and whether the disability is mild, moderate or severe. The allocation is based on functional disability using measures relating to Activities of Daily Living ( ADLs). It clearly shows that disability increases with age in the sample. This is our attempt to reproduce the estimates constructed by Stearns and Butterworth using the same dataset. While our estimates are broadly in agreement with theirs, they do not coincide exactly. This is probably because of slight differences in the allocation of a small number of individuals to categories of disability. Table 3.6 also shows the sample size for Scotland in each age group and category. The numbers are so small that misallocating relatively small numbers of individuals could have a substantial effect on the estimates of the proportions of the population in each age/disability category. These proportions play a vital role in estimating the future incidence of disability in the Scottish population and the consequent demand for care. The sample upon which they are based is so small that it is inevitable that these proportions are unstable. However the 1996/97 FRS Disability survey was, and remains, the most detailed survey of disability among the population of older people in Scotland.

Table 3.6 FRS Disability Survey - Results for Scotland

Notes to Table
Source: Family Resources Survey

3.31 There are other possible approaches to calibrating disability. These include estimates of healthy life expectancy, which have recently been produced by the Scottish Executive (Clark et al. 2004). Estimates of life expectancy after aged 65, healthy life expectancy after age 65, and the difference between these measures for both Scotland and England are shown in Table 3.7. It shows that among the over 65s, average spells of long term limiting illness (as measured by the difference between life expectancy and healthy life expectancy) are generally longer in England than in Scotland. This is likely to result from lower overall life expectancy in Scotland and from the higher incidence of acute illness in Scotland. This does imply greater demand for long-term care in England than in Scotland. It is difficult to make inferences about the demand for formal care services from such differences in demand, since healthy life expectancy measures typically rely on self-assessed measures of the presence or absence of long-term limiting illness, or self-assessed health.

Table 3.7 Life expectancy ( LE) and healthy life expectancy ( HLE) in Scotland and England based on absence of long-term limiting illness ( LLI)

Notes to Table
Source: Clark et al. (2004) and OPCS (2004).
Data for Scotland are for 2000 and for England 2001.
Estimates of Healthy Life Expectancy ( HLE) have been calculated by the Office of National Statistics for England and for Scotland by Clark et al. (2004).

Figure 3.1 Estimates of Unhealthy Life Expectancy in Scotland after Age 65 for Males and Females Based on Long-Term Limiting Illness ( LLI) 1980-2000

3.32 Figure 3.1 shows estimates of years of unhealthy life expectancy (the difference between life expectancy and healthy life expectancy) for those aged 65 and over in Scotland from 1980 to 2000 based on the estimates of Clark et al. (2004). It shows estimates for males and females based on survey responses to questions about the presence or absence of long-term limiting illness. It indicates that females are likely to experience 2.3 years of long-term limiting illness more than men. It also implies a slight increase in unhealthy life expectancy between 1980 and 2000 in Scotland. These estimates are based on the General Household Survey ( GHS), rather than the FRS, which again has a relatively small Scottish sample and a much less comprehensive analysis of disability than the FRS Disability extension. Nevertheless, these estimates do not suggest any improvement in the trend of broadly defined disability in Scotland, casting doubt on any assumption of improving health among the older Scottish population. In recent years, issues such as increased rates of Type-2 diabetes and obesity among older people have emerged as serious public health concerns.

3.33 The contrast between the GHS and FRS/ OPCS estimates of the trend in disability in Scotland illustrate the uncertainty of survey-based estimates of disability among the population of all older people in Scotland. Not only are sample sizes small, but also the surveys that attempt to calibrate disability do so in a confusing number of ways. Table 3.8 lists the questions asked by relevant social surveys that relate to disability. Different questions are asked in different surveys. They rely exclusively on self-assessment rather than professional assessment. Only the Scottish Health Survey uses medical staff to assess individuals' health and as previously mentioned the most recent version of this survey did not include individuals aged above 75 in its sample.

Table 3.8 Disability questions asked in major surveys

Notes to Table
Source: Own Evaluation

3.34 Not only are the national surveys which ask questions about disability confusing in terms of the questions that they ask, they are also deficient in their representation of the older Scottish population. Table 3.9 shows estimates of the older Scottish population with and without long-term limiting illness by five-year age group for the 2001 Census, the FRS, the BHPS and the GHS. The 2001 edition of each of these surveys is used and grossed up using the weights supplied. These weighted results suggest that all of the surveys are subject to significant error in their representation of the older Scottish population living in private households. In particular, the surveys show wide disparities from the Census in estimates of the size of age /gender cohorts. The FRS best represents the age structure of the older Scottish population: the BHPS and GHS are subject to very large errors for both genders and all 5-year age bands.

Table 3.9 Census, FRS, BHPS and GHS estimates of Scottish population in private households 2001 with and without limiting long-term illness

3.35 Further the surveys tend to differ from the Census in their estimates of the population with long-term illness, even though the questions asked across the surveys, including the Census, are closely aligned. The FRS tends to consistently underestimate long-term limiting illness whereas the BHPS and GHS estimates are very erratic. It is concerning that these surveys do not accurately represent the age structure of the Scottish population and also that each provides widely different estimates of those with long-term limiting illness from the figures obtained by the 2001 Census.

3.36 Another way to think about the usefulness of existing surveys of the older population as a means of calibrating disability is to consider how the questions compare with those that have been developed specifically to assess need. In Scotland, the Single Shared Assessment - Indicator of Relative Need ( SSA- IoRN) is a tool which can identify severity of disablement and potential resource costs (Scottish Executive 2005a). It is intended to promote fairer access to resources for older people and to improve equity of resource allocation across Scotland. It involves a detailed two-step classification model based firstly on a low, medium or high classification in respect of Activities of Daily Living ( ADLs) and then further sets of questions relating to personal care, mental well-being or bowel-management that are contingent on the first stage outcomes. But this form of assessment, because it is expensive, is only likely to be applied to those that are likely to require care. Annual data are available on the proportion of individuals in different care settings with different SSA- IoRN outcomes ( NHS National Services Scotland 2004). However, at the time of writing, the dataset remains limited.

3.37 However, the contrast between the questions used in sample surveys and those in professional assessments highlights the difficulties that are inherent in using the existing set of sample surveys as a basis for assessing levels and trends in disability among older people in Scotland. Absence of good data on prevalence of disability hinders effective forecasting of the future costs of care. This reduced the accuracy of CDG cost estimates and from the previous discussion it is clear that there has been little improvement since the publication of the CDG report.

3.38 Thus in terms of two key attributes, existing surveys are not particularly helpful in providing information on the prevalence of disability or ill-health that might be translated into a demand for care.

• They are not representative of older age groups in the Scottish population if we take the 2001 Census as the benchmark against which they should be measured. This is probably because relatively small samples do not provide sufficiently robust estimates when grossed up.
• The approach to the collection of information on disability, and the information actually collected, are not very useful in identifying care needs.

3.39 The problem of measuring disability among older people has been taken forward with the English Longitudinal Survey of Ageing in England (Institute of Fiscal Studies 2002). Scotland does not have an equivalent survey. In France, 417,000 individual were asked additional questions at the 1999 Census on functional limitations, type of assistance, and restrictions on activities (Ravaud, Letourmy, Ville and Andreyev 2002). The design of the survey was based on the notion of disability as a social construct that reflects historical, cultural and social factors. How one views disability inevitably influences how one collects information about it. The most useful information that a survey might collect is likely to have both social and medical aspects.

Conclusion

3.40 Although it fairly represented the state of knowledge in 2001, much of the information available to the CDG was subsequently found to be inaccurate. In particular, forecasts made in 1998 of the number of Scots aged 65 and over now seem too low. Forecasts of those aged 85 and over based on 2004 data are substantially higher than those based on 1998 data. Because the prevalence of disability is higher among this age group, the inference is that the CDG significantly underestimated the number of people eligible for FPC in the future.

3.41 The CDG also assumed a 0.25 per cent improvement in healthy life expectancy based on the analysis of the Disability Extension to the Family Resources Survey - a rather small and dated survey. This also had a significant impact on the projections of those likely to require FPC. Worryingly, there has been no substantial improvement on the data used for this analysis. There is a strong argument for improving our understanding of the circumstances of older people, including their medical conditions, both in domiciliary and institutional settings. Only when such data are regularly available will it be possible to draw firmer conclusions about the path of healthy life expectancy in Scotland.

3.42 There were issues which the CDG did not consider. Principal among these were trends in household formation among older people. More recent projections suggest that although the numbers of older people in Scotland will significantly increase during the next decade, most of this increase will take place in two person households. Providing that past patterns of caring by spouses within households continue, the demands on formal care provision are likely to be significantly lower than if the growth in population had been concentrated in single-person households. The forecasting of household formation patterns is not an exact science. Again this is a relatively under-researched area. However, the Registrar General plans to initiate some new work on household formation in Scotland in the near future.

3.43 Another important assumption inherent in the CDG model was that care costs would increase with age. There is now a body of evidence which suggests that it is proximity to death rather than age which exerts the strongest influence on health and care costs. This could have a profound impact on the evolution of health costs in Scotland. This is a research issue which must be monitored by the Scottish Executive, even if there is no primary research within Scotland at present due to lack of suitable datasets.

3.44 These arguments do not necessarily affect views of the sustainability of the policy: Rather they suggest that the 'funnel of doubt' is in reality considerably larger than that described in the CDG report. There are some reasons why costs may be higher than those forecast by the CDG. There are some reasons why they may be lower, and there is considerable doubt about the relative importance of the factors that have changed or were not included in the CDG analysis of costs. This argument strongly supports the need for continued monitoring not only of the costs themselves, but also of relevant research which might influence views about the evolution of costs.