Health expectancies in the UK and its constituent countries, 2001

H eal th Stati sti cs Q u a r t e rly 2 9 Spring 2006 Health expectancies in the UK and its constituent countries, 2001 Claudia Breakwell and Madhavi Bajekal Office for National Statistics A previous article set out proposals for constructing a new series of health expectancies which aimed to widen the coverage to the UK and all four of its constituent countries and to improve the methods used in the calculations. This article investigates the impact of applying the new methods by comparing estimates of healthy life expectancy (HLE) and disabilityfree life expectancy (DFLE) based on the old and new methods for one year (2001). It then goes on to present and compare health expectancies based on the new methodology across England, Wales, Scotland and Northern Ireland and the entire UK in 2001. INTRODUCTION In recent years, health expectancy estimates (such as expected years in good health or without a disability) have become an increasingly important measure of population health at both national and international level. The interest in health expectancies, a concept first introduced in the 1960s and further developed in the 1970s has grown as a response to uncertainties surrounding the impact of an aging population on future demand for health and social care.1 At the time, two competing scenarios were offered: that new cohorts of older people would be healthier, and therefore make fewer demands on care (compression of morbidity) and second, that increased survival to older ages was simply a result of medical advances with people being kept alive longer, but in poorer health (expansion of morbidity), and would result in escalating demand for care services.2 As a result, health expectancies were developed as a population health indicator that combined both lifespan and healthrelated quality of life into a single summary index for which data are readily available from death registrations and population surveys and censuses. The information produced is useful for many policy areas, including monitoring health trends, examining equity between subgroups of populations and providing a basis for health care planning.3 Health expectancy, defined as expected years of life either in good or fairly good health (healthy life expectancy, HLE) or free from limiting long-standing illness (disability-free life expectancy, DFLE), has been calculated from 1980 onwards for Great Britain and England by ONS. However, recent developments in the primary data used in the health expectancy estimates made it possible to review and improve the methodology used to produce the time series. Proposals for methodological improvements and for incorporating new sources of data were outlined in a previous article.4 NNaat ti oi onnaal l SSt taat ti si st ti ci css 18 H e a l t h S t a t i s t i cs Q u a r t e r ly 2 9 The aims of this article are first, to quantify the overall change in old and new estimates and isolate the impact of each separate component of change; and second, to present for the first time health expectancy estimates for the entire UK and all four constituent countries and compare their relative ranking. Spring 2006 Box one QUESTIONS USED TO CALCULATE ILL-HEALTH RATES METHODS Limiting long-standing illness Life expectancy GHS and CHS: (1) Do you have any long-standing illness, disability or infirmity? By long-standing I mean anything that has troubled you over a period of time or that is likely to affect you over a period of time. Yes/No The Government Actuaryʼs Department (GAD) produces interim life tables annually for the UK and its constituent countries. The interim life tables are based on mid-year population estimates, revised following the 2001 Census, and death data aggregated over a period of three consecutive years. Interim life tables for Scotland and Northern Ireland use mortality data by date of registration, while England and Wales life tables are based on deaths by date of occurrence. More information about the calculation of these tables can be found on the GAD website.5 Ill-health rates in the general population The General Household Survey (GHS) and the Continuous Household Survey of Northern Ireland (CHS) provide estimates of the ill-health for residents in households. Responses to two questions asked every year in the GHS and CHS were used to provide age- and sex-specific ill-health rates separately for England, Wales, Scotland and Northern Ireland and for higher level aggregations of these, namely Great Britain and the UK. The age patterns for these health rates did not differ greatly between the four constituent countries. For DFLE, the ill-health rate was the proportion of people who said they had a long-standing illness, disability or infirmity (Box 1, question 1a) which limits their activities (Box 1, question 1b). For HLE an ill-health rate was the proportion of respondents who said that their health was “Not good” to the general health question (Box 1, question 3). The main difference in the calculation of ill-health rates in the household population between the old and new series was the availability of agespecific ill-health rates for children under the age of 16 in both surveys. Previously, the ill-health rate of those aged 16–19 was used as a proxy for all those aged 0–15. Combining the GHS and CHS data Since 2000 the GHS has used a two-step approach for weighting, accounting for both survey non-response and ‘grossingʼ up the sample to national population estimates. In order to combine these two sets of data and calculate UK rates in the correct proportion, the CHS needs to be weighted to the population resident in households in Northern Ireland (i.e. excluding residents of institutions). Estimates of the population resident in households, derived from mid-year population estimates and the Labour Force Survey, are used to weight the GHS and have also been used to weight the CHS by age (in five-year age bands) and sex. The standard mid-year population provided by ONS can not be used in this situation as they include residents of communal establishments while both the GHS and CHS only cover the non-institutional population. It should be noted that the ill-health rates for Great Britain, derived from the GHS, have been weighted for survey non-response, but not those for Northern Ireland as the CHS survey does not include non-response weighting.4 If ‘Yes’: a) What is the matter with you? b) Does this illness or disability (do any of these illnesses or disabilities) limit your activities in any way? Yes/No 2001 CENSUS: (2) Do you have any long-term illness, health problem or disability which limits your daily activities or the work you can do? Yes/No General Health GHS, CHS and 2001 CENSUS: (3) Over the last 12 months would you say your health has on the whole been good, fairly good, or not good? Ill-health rates in the institutional population Ill-health rates The GHS and CHS only include residents in the household population. Residents in communal establishments providing medical, nursing and social care (categorised as medical and care in the 2001 Census) represent a significant proportion of the elderly – about 4 per cent of those aged 65 or over. By definition these establishments include those with higher rates of ill-health than in the household population. The method used to calculate the ill-health rates for residents of medical and care establishments for the new data point was as follows. In the 2001 Census, for the first time, residents in communal establishments were asked to rate their general health in addition to being asked whether they had a long-term illness that limited their activities (Box 1 questions 2 and 3). Responses to these questions were used to derive the ill-health rates applied to residents of communal establishments for the 2001 estimates. The important difference in the communal population ill-health rates between the old and new series was the use of of actual rates of perceived ‘not goodʼ health for the HLE calculation, rather than using limiting long-standing illness rates as a proxy as done previously. Institutional population size and age distribution For calculating the health expectancy estimates, the size and age distribution of the communal establishment population was derived from the 2001 Census. The method used in the previous series was substantially different: an estimate of the institutional population was derived first through extrapolation from the previous 1991 Census counts, and this figure was then deducted from the mid-year population estimate to derive the general population estimate. This approach has been shown to over-estimate the size of the institutional population in the inter-censal period.4 Thus, along with the change to the calculation of ill-health rates, the communal population adjustment in the new series is substantively different from that used in the old series. 19 National Statistics H eal th Stati sti cs Q u a r t e rly 2 9 Spring 2006 Health expectancies Health expectancies in the UK and its constituent countries have been calculated using Sullivanʼs method (see Box 2).6 In this method ill-health rates are used to partition life expectancy into expected years of good and not good health. Confidence intervals at the 95 per cent level were also calculated. Two sets of estimates of health expectancy were calculated for Great Britain and England for 2001: health expectancy based on the old and new methodology. These two sets of 2001 data points provide a statistical ‘bridgingʼ year to assess the affect that changes in methods have had on the estimates. Box two SULLIVAN’S METHOD FOR CALCULATING HEALTH EXPECTANCY 1. For each age/sex group obtain the life table schedules lx and the expectation of life ex for the year of interest. Calculate: L is the conventional life table measure of the average n x number of person years lived in the age interval x to x+n. Obtain the ill-health rate ndx in each age group observed in a survey or census. If they are excluded, add the numbers in communal establishments catering for the sick and disabled. Calculate the average number of persons aged x to x+n living without ill-health in each age/sex group as n LWDx = nLx (1- ndx) Calculate life expectancy without ill-health as Similar patterns of overall (or net) change between the old and new estimates for 2001 were observed for Great Britain and England (see Table 1). We have therefore focused on the interpretation of results for Great Britain. For females, both HLE and DFLE estimates at birth and at age 65 were higher for the new estimates, but the increase was much larger for HLE (1.1 years at birth and 0.9 years at age 65) than for DFLE (0.2 years for both). For males, on the other hand, both HLE and DFLE at birth were lower in the new compared to the old estimates (by 0.2 years and 0.1 years, respectively), but were higher at age 65 for HLE (by 0.3 years) and remained unchanged for DFLE. The total net changes appear to be relatively modest and, as the 95 per cent confidence intervals of the two sets of estimates overlap substantially, this suggests that the new methodology did not result in a statistically significant change.7 This was not true of the HLE estimates HLEx = (∑nLWDx)/ lx where the summation is from age x upwards. Table 1 To understand the separate contribution of each component of change in methodology to the net overall change between the old and new estimates, changes were introduced a step at a time. Taking the old estimates as the starting point, a single methodological change was introduced at each successive step and the health expectancies recalculated. In this way, the contribution of each successive change was isolated and its contribution to the net overall change quantified. The changes to Census data were implemented first, followed by changes in the survey data. The order in which these changes are made may alter their separate contribution to the net overall change, because of the interaction between each of the components. However comparisons can be made between the effect each component of change has on any of the health expectancy figures in this article as they are implemented in the same order in each. Impact of revised methodology on health expectancy estimates for Great Britain L = ex lx - ex+n lx+n 3. Quantifying the components of change between the old and new estimates RESULTS n x 2. Estimates using the new methodology could not go further back than 2001 for two reasons. First, before 2000, the general health question was not asked of those under 16. Second, the GHS was suspended in 1999 and as health expectancies are calculated by combining three years of data (centred on the middle year) a 2000 point could not be estimated. Life expectancy (LE), ‘old’ and ‘new’ healthy life expectancy (HLE) and disability-free life expectancy (DFLE), at birth and age 65: by sex, 2001 Great Britain, England Life Expectancy Healthy life expectancy Old Series At birth Males Females At age 65 Males Females Disability-free Life Expectancy New Series Old Series New Series Years Years 95% Confidence Interval Years 95% Confidence Interval Years 95% Confidence Interval Years 95% Confidence Interval Great Britain England Great Britain England 75.7 76.0 80.4 80.6 67.0 67.3 68.8 69.0 (66.8–67.3) (67.0–67.6) (68.5–69.1) (68.7–69.4) 66.8 67.1 69.9 70.1 (66.5–67.1) (66.8–67.4) (69.6–70.2) (69.8–70.4) 60.5 60.8 62.7 62.9 (60.2–60.8) (60.5–61.2) (62.4–63.1) (62.5–63.2) 60.4 60.7 62.9 63.0 (60.0–60.7) (60.3–61.1) (62.5–63.2) (62.6–63.4) Great Britain England Great Britain England 15.9 16.1 19.0 19.2 11.6 11.7 13.2 13.3 (11.4–11.8) (11.5–11.9) (12.9–13.4) (13.1–13.6) 11.9 12.0 14.0 14.2 (11.7–12.1) (11.7–12.2) (13.8–14.3) (14.0–14.4) 8.8 8.9 10.1 10.2 (8.6–9.0) (8.7–9.2) (9.8–10.3) (9.9–10.5) 8.8 8.9 10.3 10.4 (8.6–9.1) (8.7–9.2) (10.0–10.5) (10.2–10.7) Source: Office for National Statistics; Government Actuary’s Department National Statistics 20 H e a l t h S t a t i s t i cs Q u a r t e r ly 2 9 The proxy rate has been shown to be higher than the actual ill-health rate for girls and lower than the actual ill-health rate for boys.4 for women as the confidence intervals did not overlap and the number of years in good health in the new series was over 1 year more than the old series. To place the scale of these differences in context, on average health expectancies have been increasing by about 0.2 years (or two and a half months) per annum over the last two decades of the 20th century. Components of change in healthy life expectancy estimates The inclusion of new ill-health rates for residents in communal establishments, rather than using the 1991 Census limiting long-term illness rates as a proxy, has had the largest positive effect on the new HLE estimates (see Table 2). At birth and at age 65, while HLE for males increased by 0.3 years, the effect for females was greater. The HLE for women increased by 0.8 years at birth and at age 65. The larger rise in HLE than DFLE for both sexes can be accounted for by the use of the actual rates of ‘not goodʼ health which are much lower than the limiting long-standing illness rates that were used as a proxy in the old series. The rise in HLE is greater for women than for men as women are more likely to be resident in a communal establishment especially at older ages where the largest increase was seen. The updated communal establishment population has also produced a rise in HLE of about 0.1 years in all cases apart from males at age 65 where the HLE shows no change. The estimates used in the old HLE series overestimated the population resident in communal establishments. Therefore applying the updated figures has resulted in a rise in HLE as the true proportion of people with higher ill-health rates experienced by the institutional population was lower than estimated. The introduction of new ill-health rates for children aged 0–15 has resulted in the HLE at birth falling by 0.3 years for males and, conversely, rising by this amount for females. The difference seen between the sexes in this case is because the ill-health rate of 16- to 19year-olds was used as the proxy for 0- to 15-year-olds in the old series. Table 2 Spring 2006 Lastly, the use of weighted data from the GHS has resulted in a fall in HLE in all cases, with the exception of females at age 65 where the HLE showed no change. The weighting has resulted in an increase in the not good health rate for all men apart from those aged 15–19; these increases have contributed to the fall in HLE for males at both birth and age 65. For women the pattern was slightly different, fewer age bands had an increased not good health rate after weighting and the increases were of a smaller magnitude than seen in males.8 Components of change in disability-free life expectancy estimates DFLE was not as sensitive to the changes in methodology as HLE. The estimate for males at birth in 2001 dropped slightly from 60.5 years to 60.4 years between the old and new series, while at 65 it has remained at 8.8 years (see Table 3). The estimates for women also show little change between the old and new series: at birth the number of expected years free from disability increased from 62.7 to 62.9 years and at age 65 from 10.1 to 10.3 years. The impact of individual component changes in methodology on DFLE estimates shows broadly similar patterns to that for HLE estimates, with two exceptions. The limiting long-standing illness rate for residents in institutions has risen between 1991 and 2001, especially at younger ages. This has resulted in a reduction in DFLE at birth for both men and women while DFLE at age 65 has remained the same for both sexes. The improved estimates of the size and age profile of the institutional population have had more of an impact on the estimates for DFLE than for HLE. DFLE at birth rose by approximately 0.2 years, double the rise seen in HLE. The larger effect for DFLE was because among residents in institutions the rate of limiting long-standing illness was more than twice the rate of not good health in the majority of age groups. Contribution of each component of change in method between the old and new series of healthy life expectancy (HLE) at birth and at age 65: by sex, 2001 Great Britain Males Healthy life expectancy description At birth Old series – plus new ill-health rates for residents in CEs – plus new population estimates of residents in CEs – plus new ill-health rates for children 0–15 – plus weighted GHS data (new) Total net change (new–old) At age 65 Old series – plus new ill-health rates for residents in CEs – plus new population estimates of residents in CEs – plus new ill-health rates for children 0–15 – plus weighted GHS data (new) Total net change (new–old) Females Value 95% Confidence Interval Change from previous value Value 95% Confidence Interval Change from previous value 67.0 67.3 (66.7–67.3) (67.1–67.6) 0.3 68.8 69.6 (68.5–69.1) (69.3–69.9) 0.8 67.4 67.1 66.8 (67.1–67.7) (66.8–67.3) (66.5–67.1) 0.1 -0.3 -0.2 69.7 70.0 69.9 (69.4–70.0) (69.7–70.3) (69.6–70.2) 0.1 0.3 -0.1 -0.2 1.1 -0.2 1.1 11.6 11.9 (11.4–11.8) (11.7–12.1) 0.3 13.2 14.0 (13.0–13.4) (13.8–14.2) 0.8 11.9 12.0 11.9 (11.7–12.2) (11.7–12.2) (11.7–12.1) 0.0 0.0 -0.1 14.1 14.1 14.0 (13.9–14.3) (13.9–14.3) (13.8–14.3) 0.1 0.0 -0.0 0.3 0.9 0.3 Source: Office for National Statistics; Government Actuary’s Department 21 National Statistics 0.9 H eal th Stati sti cs Q u a r t e rly 2 9 Table 3 Spring 2006 Contribution of each component of change in method between the old and new series of disability-free life expectancy (DFLE) at birth and at age 65: by sex, 2001 Great Britain Males Disability-free life expectancy description At birth Females Value 95% Confidence Interval Change from previous value Value 95% Confidence Interval Change from previous value Old series – plus new ill-health rates for residents in CEs – plus new population estimates of residents in CEs – plus weighted GHS data (new) 60.5 60.3 (60.2–60.8) (60.0–60.7) -0.2 62.7 62.7 (62.4–63.1) (62.3–63.0) -0.1 60.6 60.4 (60.2–60.9) (60.0–60.7) 0.2 -0.2 62.9 62.9 (62.6–63.2) (62.5–63.2) 0.2 -0.1 Total net change (new–old) -0.1 -0.1 0.1 At age 65 Old series – plus new ill-health rates for residents in CEs – plus new population estimates of residents in CEs – plus weighted GHS data (new) Total net change (new–old) 0.1 8.8 8.8 (8.6–9.0) (8.6–9.0) -0.0 10.1 10.1 (9.8–10.3) (9.8–10.3) -0.0 8.9 8.8 (8.7–9.1) (8.6–9.1) 0.1 -0.1 10.3 10.3 (10.0–10.5) (10.0–10.5) 0.2 -0.0 0.0 0.2 0.0 0.2 Source: Office for National Statistics; Government Actuary’s Department Table 4 Life Expectancy (LE), healthy life expectancy (HLE) and disability-free life expectancy (DFLE), at birth and age 65: by sex and country, 2001 Country At birth Males Females At age 65 Males Females Life expectancy Healthy life expectancy Disability–free life expectancy Years Years 95% Confidence Interval Years 95% Confidence Interval United Kingdom Great Britain England Wales Scotland Northern Ireland 75.7 75.7 76.0 75.4 73.3 75.2 66.8 66.8 67.1 65.5 65.3* 65.1* (66.5–67.0) (66.5–67.1) (66.8–67.4) (64.2–66.8) (64.4–66.2) (64.5–65.6) 60.3 60.4 60.7 57.7* 58.8* 57.8* (60.0–60.6) (60.0–60.7) (60.3–61.1) (56.2–59.2) (57.7–59.9) (57.1–58.6) United Kingdom Great Britain England Wales Scotland Northern Ireland 80.4 80.4 80.6 80.1 78.8 80.1 69.9 69.9 70.1 69.4 68.6* 67.2* (69.6–70.1) (69.6–70.2) (69.8–70.4) (68.1–70.7) (67.6–69.5) (66.6–67.8) 62.8 62.9 63.0 61.1 62.6 59.6* (62.5–63.1) (62.5–63.2) (62.6–63.4) (59.6–62.6) (61.5–63.8) (58.9–60.3) United Kingdom Great Britain England Wales Scotland Northern Ireland 15.9 15.9 16.1 15.7 14.9 15.7 11.9 11.9 12.0 11.1 11.5 11.1* (11.7–12.1) (11.7–12.1) (11.7–12.2) (10.2–12.1) (10.9–12.2) (10.7–11.5) 8.8 8.8 8.9 7.3* 8.7 7.6* (8.6–9.0) (8.6–9.1) (8.7–9.2) (6.4–8.3) (7.9–9.5) (7.0–8.1) United Kingdom Great Britain England Wales Scotland Northern Ireland 19.0 19.0 19.2 18.7 18.0 18.7 14.0 14.0 14.2 12.8* 13.5 12.5* (13.8–14.2) (13.8–14.3) (14.0–14.4) (11.8–13.8) (12.8–14.2) (12.1–12.9) 10.2 10.3 10.4 8.5* 10.0 8.4* (10.0–10.5) (10.0–10.5) (10.2–10.7) (7.4–9.5) (9.2–10.8) (7.9–8.9) Source: Office for National Statistics; Government Actuary’s Department * Significantly different from England at the 95% level. Healthy Life Expectancy in the UK and constituent countries, 2001 Figure 1 shows the life expectancy (LE) at birth for males and females in the UK and its constituent countries subdivided into years in ‘good or fairly goodʼ health (‘healthyʼ) and in ‘not goodʼ health (‘unhealthyʼ). For both sexes, England, Wales and Northern Ireland were found to have similar estimates of LE at birth, but HLE in England was higher than Wales and was significantly higher compared with Northern Ireland (see Table 4). The position in Scotland was slightly different: HLE at birth National Statistics 22 in Scotland was significantly lower than in England and similar to the estimates for Wales and Northern Ireland. But as Scotland had the lowest LE, it also had the least number of years spent in poor health within the UK. Of the four countries, Northern Ireland had the highest number of years spent in ill-health, significantly higher than both England and Scotland. At age 65 (see Figure 2 and Table 4) England had a significantly higher HLE than Northern Ireland for both sexes, for women it was also significantly higher than Wales. HLE in Scotland was higher (but was not H e a l t h S t a t i s t i cs Q u a r t e r ly 2 9 significant at the 95 per cent level) than in Northern Ireland and Wales for both sexes. In other words, despite its lowest average expectation of life, Scotlandʼs healthier lifespan (in terms of both the number of years and the proportion of total life in good health) resulted in its HLE at 65 being second only to England in the UK. Similar patterns of variation in LE and HLE within the UK were found for males and females. Women were consistently found to have both a higher HLE and LE, and a higher number of expected years lived in poor Figure 1 Spring 2006 health. This difference between the sexes was largest in Northern Ireland where at birth women expected to live an extra 2.8 years in poor health compared with men. By age 65 the proportion of life expected to be lived in poor health increased markedly but the pattern across the countries of the UK remained the same as at birth. The difference in the number of years lived in poor health between the sexes was reduced in all countries; Northern Ireland still had the largest gender difference with women expected to live an extra 1.6 years in poor health compared with men. Life expectancy and healthy life expectancy (HLE) at birth: by sex and country, 2001 Healthy 90 Not Healthy 80 70 Years of life 60 50 40 30 20 10 0 United Kingdom England Wales Scotland Northern Ireland United Kingdom England Wales Scotland Northern Ireland Scotland Northern Ireland Females Males Source: Office for National Statistics, Government Actuary’s Department Figure 2 Life expectancy and healthy life expectancy (HLE) at age 65: by sex and country, 2001 Healthy 25 Not Healthy Years of life 20 15 10 5 0 United Kingdom England Wales Scotland Northern Ireland United Kingdom England Males Wales Females Source: Office for National Statistics, Government Actuary’s Department 23 National Statistics H eal th Stati sti cs Q u a r t e rly 2 9 Spring 2006 DISABILITY-FREE LIFE EXPECTANCY IN THE UK AND CONSTITUENT COUNTRIES, 2001 The variation between countries of the UK in DFLE, at birth and at age 65, (see Figures 3 and 4) was similar to the variation in HLE. In all cases England had significantly higher DFLE than either Wales or Northern Ireland, with the exception of females at birth where Wales was not significant (see Table 4). Scotland had the least number of years lived with a disability of all four countries, resulting in the DFLE for Scotland being the second highest in the UK at birth and at age 65 for both sexes. Figure 3 DFLE at 65 was a year lower for men and two years lower for women in Wales and Northern Ireland compared with England. In fact both men and women at age 65 in Wales and Northern Ireland can expect to live over half of their remaining life with some form of disability. As found for HLE, Northern Ireland had the largest difference in DFLE between the sexes, with women on average living three years longer with a disability than men. While the differences between the sexes were reduced at 65, women in Northern Ireland were expected to live over two years longer with a disability compared to men. Life expectancy and disability free life expectancy (DFLE) at birth: by sex and country, 2001 90 Without Disability With Disability 80 70 Years of life 60 50 40 30 20 10 0 United Kingdom England Wales Scotland Northern Ireland United Kingdom England Wales Scotland Northern Ireland Females Males Source: Office for National Statistics, Government Actuary’s Department Life expectancy and disability-free life expectancy (DFLE) at age 65: by sex and country, 2001 Figure 4 25 Without Disability With Disability Years of life 20 15 10 5 0 UK England Wales Scotland Northern Ireland Males Source: Office for National Statistics, Government Actuary’s Department National Statistics 24 UK England Wales Females Scotland Northern Ireland H e a l t h S t a t i s t i cs Q u a r t e r ly 2 9 DISCUSSION The majority of the health expectancy estimates increased once the new methodology had been implemented with the exception of HLE and DFLE for males at birth, which fell by 0.2 and 0.1 years respectively. The fall in HLE for males at birth can be attributed to the combined effect of using the actual rates of ‘not goodʼ health for children aged 0–15, that were underestimated by a proxy in the old series, and the use of weighted data from the GHS. GHS will be integrated into the Continuous Population Survey.14 Both these developments are likely to have further implications for the health expectancy series, and will be investigated when these data become available. Key findings ● There is little discontinuity between the old and the new series with the exception of HLE for women. The number of years a woman could expect to live free from poor health increased at both birth and age 65 by over a year. This large rise in HLE is the effect of using the actual ‘not goodʼ health rates for residents in institutions, where women outnumber men especially at older ages. England, Wales and Northern Ireland had similar life expectancies while in Scotland it was lower. England also had the highest estimates of both HLE and DFLE while estimates for Scotland, Wales and Northern Ireland were lower. Because Scotland had the lowest life expectancy, this resulted in Scotland having the least number of expected years spent in ill-health of all four constituent countries in the UK. The mismatch between areas with high levels of mortality and those with high reported ill-health may arise due to several factors. Variations in prevailing norms of health, which vary between population groups, are known to influence self ratings of health.9 Alternatively, mortality patterns may reflect the morbidity of previous decades while self rated health reflects current morbidity.10 Lastly, self rated health is sensitive to conditions that are poorly reflected by mortality. For example, the higher proportion of untreatable cancers with low survival in Scotland may contribute to lower LE, while on the other hand the higher proportion of treatable cancers in England and Wales may result in higher morbidity reflected in poorer self rated health, and higher LE.11,12 CONCLUSION The use of updated more accurate sources in the calculation of health expectancies by ONS has improved the quality and accuracy of the estimates. This change in methodology means that the new points are not strictly comparable to the previous series. However, given the small net change in estimates between the old and new series, the new series will continue to be used to monitor long-term trends in healthy life expectation. Combining CHS data for Northern Ireland along with GHS data for Great Britain has meant that for the first time health expectancies could be calculated for the entire UK. ONS will continue to produce new health expectancy estimates based on the updated sources for the UK, Great Britain, England, Scotland and Northern Ireland. The GHS survey design and questionnaire items for self-assessed health are in the process of being changed in response to European Union requirements for a harmonised Healthy Life Years Structural Indicator. These proposals have been incorporated into the 2005 GHS and include a new question on general health using a five point scale. This is in addition to the existing three point question used for HLE and an additional question to assess the severity of long-standing disability. In addition to these changes due to be implemented in 2006, the GHS will become longitudinal with one quarter of the annual sample being refreshed each year. Therefore after every four years the sample members will be different from the starting sample.13 In 2008 the longitudinal Spring 2006 ● ● Health expectancy in the UK in 2001 was highest in England. For example, disability-free life expectancy at birth was 60.7 years for males in England, compared to 60.3 years for the UK. Life expectancy in Scotland was lower than elsewhere in the UK in 2001. Because there was not as great a difference in the expected years spent healthy, people in Scotland on average spent fewer years in ill-health. For example, in Scotland males at birth could expect to live 14.5 years with a disability compared to 15.4 years for the UK as a whole. Improvements to the method for calculating health expectancies had the greatest effect on estimates of healthy life expectancy for women. The new method resulted in the estimated years women could expect live in good health, both at birth and at age 65, increasing by a year. References 1. Satio Y, Crimmins E M and Haywood M D (1999) Health Expectancy: An overview. NUPRI Research Paper Series No. 67. 2. Mathers C D (2002) Heath Expectancies: An Overview and Critical Appraisal, in Murray C J L, Salomon J A, Mathers C D and Lopez A D (eds.) Summary Measures of Population Health, World Health Organisation: Geneva, pp 177–204. 3. Crimmins E M (2003) The relevance of health expectancies, in Robine J M, Jagger C, Mather C D, Crimmins E M and Suzman R M (eds.) Determining Health Expectancies, Wiley, pp 102–109. 4. Breakwell C and Bajekal M (2005) Review of sources and methods to monitor Healthy Life Expectancy. Health Statistics Quarterly 26, 17–22. 5. www.gad.gov.uk/ 6. Jagger C (1999) Health Expectancy Calculation by the Sullivan Method: A Practical Guide. NUPRI Research Paper Series No 68. 7. No formal tests of statistical significance could be carried out on the difference between estimates as both use the same survey data and are not therefore derived from independent samples. 8. www.statistics.gov.uk/lib2000/index.html 9. Johansson S R (1991) The health transition: the cultural inflation of morbidity during the decline of mortality. Health Transition review 1(1), 39–68. 10. OʼReilly D, Rosato M and Patterson C (2005) Self reported health and mortality: ecological analysis based on electoral wards across the United Kingdom. BMJ 331, 938–939. 11. Quinn M, Wood H, Cooper N and Rowan S (eds.) (2005) Cancer Atlas of the UK and Ireland 1991–2000, SMPS No 68, Palgrave Macmillan: Basingstoke. 12. www.isdscotland.org/isd/info3.jsp?pContentID=402&p_ applic=CCC&p_service=Content.show& 13. www.statistics.gov.uk/events/gss2005/agenda.asp 14. www.statistics.gov.uk/ssd/downloads/Proposals_CPS.pdf 25 National Statistics