Trends in health expectancies in Australia 1981-1993

Vol. 13, No. 1, 1996 Journal of the Australian Population Association T R E N D S IN H E A L T H E X P E C T A N C I E S IN A U S T R A L I A 1981-1993 Colin Mathers Australian Institute of Health and Welfare GPO Box 570 Canberra ACT 2601 Health expectancy indices combine the mortality and morbidity experience of a population into a single composite indicator. This paper summarizes and evaluates methods for the calculation of health expectancies and presents trends in the expectation of life with disability and handicap in Australia from 1981 to 1993. Unlike other countries for which recent health expectancy time series are available, Australian results indicate that the expectation of years with disability has increased for both males and females. Possible explanations for this are examined. The concept of a health indicator which combined information on mortality and morbidity was proposed by Sanders (1964) and the first example of such an indicator was published in a report of the United States Department of Health, Education, and Welfare (1969), which contained preliminary estimates of Disability-Free Life Expectancy (DFLE) calculated using a method devised by Sullivan (1971). This involved using the observed prevalence of disability at each age in the current population (at a given point of time) to divide the years of life lived by a period life table cohort at different ages into years with and without disability. During the first half of the 1980s, Sullivan's method was used to estimate disability-free life expectancy for the USA (Colvez and Blanchet 1983), Canada (Wilkins and Adams 1983), and France (Robine et al. 1986). Estimates of active life expectancy for the USA had also been made using double decrement life table methods by Katz et al. (1983). During the second half of the 1980s, there was a dramatic increase in the number of health expectancy calculations carded out, almost all using the Sullivan method (REVES 1993). An informal international research network, the Network on Health Expectancy (Rgseau Esp~rance de Vie en Santd or REVES) was established in 1989 with objectives including the harmonization of calculation methods and identification of the conditions necessary for comparison of health expectancy estimates, both across populations and over time (Bone 1992; Mathers and Robine 1993). Although the number of countries for which strictly comparable time series data are available has increased, there is still only a handful of countries for which there is information on changes in health expectancy over a reasonably long time period (Robine 1994). Three major hypotheses have been advanced for the evolution of population health in advanced societies where birth and death rates are low and death rates are continuing to fall, particularly at older ages, with consequent increasing life expectancies. The first hypothesizes declining health status, the second improving health status and the third a kind of status quo (Robine 1992). The 'pandemia' hypothesis (Gruenberg 1977; Kramer 1980) postulates that the decline in mortality is due to decreasing fatality rates for diseases and not to a reduction in their incidence or progression. Consequently the decline in mortality is accompanied by an increase in chronic illness and disability. Olshansky and colleagues (1991) have developed further arguments from evolutionary biology in favour of the expansion of morbidity hypothesis. The second hypothesis, compression of morbidity, was first proposed by Fries (1980, 1989) who suggested that adult life expectancy is approaching its biological limit so that, if the incidence of incapacitating disease can be postponed to later ages, then morbidity will be compressed into a shorter period of life. The third hypothesis was proposed by Manton (1982), who suggested that the decline in mortality may be partly due to decreased fatality rates, but at the same time the incidence and progression of chronic diseases may be decreasing, leading to a dynamic equilibrium. Health expectancy indices which combine mortality and morbidity into a single composite indicator are a very attractive tool for monitoring long term trends in the evolution of population health and for addressing the question of compression or expansion of morbidity. Methodology for Calculation of Health Expectancies Three major methods of calculating health expectancies have been used: Sullivan's method, the double decrement life table method and the multistate life table method. Sullivan's method uses the observed age-specific prevalences of health states in a population at a given point in time ('crosssectional' prevalences) to calculate the years of life lived in the various health states at each age by a period life table cohort (Mathers 1991, Appendix B). Double decrement life table methods were used by Katz et al. (1983) and are based on the observation, during the study period, of either of two outcomes: disability or death. This method assumes that the disability state, as well as death, is irreversible. Thus the disability state used with this method must either be irreversible (for example, senile dementia) or one where the probabilities of recovery can be assumed to be negligible. Multistate life table methods were proposed by Rogers, Rogers and Belanger (1989, 1990) to take into account reversible transitions between good health and one or more disability (or other health) states. They used multistate life table methods to calculate active life expectancy for the USA in 1984 using data from the 1986 Longitudinal Survey of Ageing which reinterviewed 5,150 people who had previously been interviewed in 1984 and who were aged 70 or over at that time. The transition probabilities were thus calculated for a two year interval at each age. Their data showed that transition rates from dependence to independence can be surprisingly high, even for the older old. In addition, the multistate life table method allows one to calculate health expectancies for population subgroups in a specific health state at a given age, such as those not disabled at age 65, whereas the Sullivan method gives only the average health expectancy for the entire population at a given age. With the proliferation of health expectancy calculations in recent years, there has been vigorous debate about methods of calculating health expectancies, and in particular, the validity and limitations of the Sullivan method. Problems relating to the validity of the Sullivan method were first raised in 1989 by Bebbington (1992) and Brouard and Robine (1992). Bebbington compared the Sullivan method with the double decrement life table method using data where the disability incidence rate was rising over time. He demonstrated that the Sullivan method gives a lower estimate of disability than the double decrement method, because in effect, the disability prevalence rate used in the Sullivan method reflects the past experience of each cohort, and not the current incidence rates. Brouard and Robine (1992) similarly argued that the prevalence of disability is a stock dependent on past history, whereas the incidence of disability is a flow which can be used to compute a 'pure period proportion' of disabled people not dependent on past flows, which in turn could be used to compute a pure period indicator of disability-free life expectancy. They noted that the question of whether to measure stocks or flows is a common dilemma in demography and in other disciplines such as economics. Brouard (1980) had previously demonstrated the large bias resulting from the use of stocks rather than flows to calculate trends in working life expectancy for French women and has reminded those debating the merits of the Sullivan and multistate methods of the very similar discussions relating to the use of prevalence versus longitudinal data in 'working life tables' (Bonneuil, Brouard and Robine 1992; Brouard 1990). In fact, very similar problems had been encountered decades previously by demographers attempting to calculate working life expectancy (WLE), the average number of years a person could expect to spend in the workforce. When Wolfbein (1949) calculated WLE for US males in 1940, he stated that it was necessary to know 'the probability of accession and separation from the labour force at different ages', although he went on to derive these transition rates from the 'differences in successive worker rates' (that is, prevalence rates) using a number of assumptions. The inapplicability of these assumptions to women precluded the calculation of WLE for females. More recently, Hoem (1977) criticized these unrealistic assumptions and proposed a multistate method for calculating WLE, demonstrating that the results obtained differed greatly from the conventional (Wolfbein) results for those who are in the labour force. Schoen and Woodrow (1980) clearly distinguished labour force-based methods from population-based methods and argued that the multistate (increment-decrement) method was generally preferable to conventional working life tables based on age-specific proportions in the labour force. Table 1 of their paper shows large differences in labour force-based estimates of WLE by the two methods, but very small differences for population-based estimates. In fact, most of the criticisms of the prevalence-based method have come from authors who were primarily interested in the labour force transitions rather than in the estimates of WLE. It is now well understood that Sullivan's method, unlike the standard life table method for calculating period life expectancy, does not produce a pure cross-sectional indicator derived from the current health conditions of the population (Mathers 1991; Crimmins, Saito and Hayward 1993). This is because the prevalence rates are partly dependent on earlier health conditions of each age cohort, that is, incidence, recovery and state-specific mortality rates applying at earlier times (or ages). For these reasons, Bonneuil et al. (1992) and Brouard and Robine (1992) have correctly argued that the use of Sullivan's method does not permit true international comparisons or comparisons over time. To construct a purely cross-sectional indicator, one would have to use the equilibrium prevalences observed in a fictitious cohort which had always been exposed to the observed cross-sectional transition rates between health states. Mathers (1991) has shown that in an equilibrium or stationary population, where all transition rates are constant over time, Sullivan's method gives the same health expectancies as the multistate methods. In fact, the Sullivan method is potentially more accurate than the multistate life table method in this case, as the use of disability prevalence includes the effects of all periods of disability experienced, unlike the use of transition probabilities between the end-points of age intervals. The problems with Sullivan's method arise not because it uses prevalence and mortality data averaged over 4 all health states, but because the data it uses are dependent on past conditions in the population. Theoretically, the multistate life table method is to be preferred for calculating health expectancies, but its use requires longitudinal data which are expensive and time-consuming to collect and are rarely available. Robine and I have developed a simulation model using French data, which allowed us to compare the Sullivan estimate with the pure period estimate from the multistate life table for a population which has experienced realistic changes in transition rates over time (Robine and Mathers 1993). An example of such a simulation is shown in Figure 1. In this scenario, disability incidence rates are assumed to have jumped substantially during the two world wars, but otherwise decline at a constant annual percentage except during the 1960s when the epidemic of cardiovascular disease was peaking. In order to clearly illustrate the differences between the methods, we ignored any effects of the two world wars on mortality. The resulting estimates of DFLE obtained by the Sullivan method and the multistate method are shown in Figure 1. We concluded that the difference between the estimates produced by the two methods is small when transition rates are changing relatively slowly over time, as postulated by the principal scenarios for evolution of population health, and that Sullivan's method is acceptable for monitoring long term trends in health expectancies for populations. Disability and Handicap Prevalence Data The Australian Bureau of Statistics (ABS) conducted a third survey on disability and ageing in Australia in 1993 using similar methodology and definitions to its two earlier surveys in 1981 and 1988 (Australian Bureau of Statistics 1982, 1988, 1993). The 1993 ABS Survey of Disability, Ageing and Carers was a population sample survey comprising two components: 9a household sample of 17,800 private dwellings and 1,600 special dwelling units (giving a total sample of approximately 42,000 persons or 0.25 per cent of the total Australian population living in households). 9a sample of 700 health establishments (hospitals, nursing homes, hostels, retirement villages, etc.) giving a total sample of approximately 4,800 persons or 2.9 per cent of the total population of health establishments). The 1981 and 1988 ABS surveys on disability defined a person with a disability as having one or more of the following conditions which had lasted or were likely to last for six months or more: 9loss of sight (even when wearing glasses or contact lenses) 9loss of hearing 9speech difficulties in native language 9blackouts, fits or loss of consciousness 5 Figure 1 Comparison of trends in Disability-Free Life Expectancy (DFLE) for French males 1905-2010 calculated in a simulation model using Sullivan's method and the multistate life table method (Period DFLE) Numb~ ofyears 75 Scenario: Disability incidence rate decreases except during the two world wars a n d the 1960s 70 65 60 55 50 45 40 I ~162 Sullivan DFLE t.,'-X.. /o.,'" \ S.s"1 Period DFLE 35 Total LE at age 0 (French males) 30 25 1900 I I I I I 1920 t 940 1960 1980 2000 Calendar year Source: Robine and Mathers (1993). 9 slowness at learning or understanding 9 incomplete use of arms or fingers 9 incomplete use of feet or legs 9 long term treatment for nerves or an emotional condition 9 restriction in physical activities or in doing physical work 9 disfigurement or deformity 9 need for help or supervision because of a mental disability 9 long term treatment or medication (although still restricted in some way by the condition being treated) (Australian Bureau of Statistics 1982, 1990). These conditions include impairments, disabilities, and a handicap, as defined in the World Health Organization's International Classification of Impairments, Disabilities, and Handicaps framework, and even some health conditions, and should perhaps be viewed as defining a wider population likely to contain those persons with a disability. In the 1993 ABS Survey of Disability, Ageing and Carers, the list of screening questions for disability was expanded to include: 9difficulty gripping and holding small objects 9long term effects of head injury, stroke, or any other brain damage 9any other long term condition resulting in a restriction. The ABS surveys defined a handicapped person as 'a disabled person aged five years or over who was further identified as being limited to some degree in his/her ability to perform tasks in relation to one or more of the following five areas: self care, mobility, verbal communication, schooling, and/or employment'. Severity of handicap for persons aged five years or over was assessed, for self care, mobility, and verbal communication, as follows: (a) severe handicap - - personal help or supervision required or the person is unable to perform one or more of the tasks; in the 1993 survey this category was further divided into severe and profound handicap. In this section, the term 'severe handicap' refers to severe and profound handicap combined. (b) moderate handicap - - no personal help or supervision required, but the person has difficulty in performing one or more of the tasks. (c) mild handicap - - no personal help or supervision required and no difficulty in performing the tasks, but the person uses an aid, or has difficulty walking 200 metres or up and down stairs. All disabled children under the age of five years were regarded as being handicapped; the severity of their handicap was not assessed. Changes in Prevalence of Disability and Handicap from 1981 to 1993 Prevalences of disability and handicap reported in ABS publications from the 1993 survey are not directly comparable with those from the earlier surveys because of the addition of a number of items to the disability screening question as described above. For the analyses presented in this paper, estimates have been derived from the 1993 survey data, using definitions consistent with the 1981 and 1988 survey screening questions. The age-standardized prevalence of disability increased substantially between 1981 and 1988, from 14.9 to 16.8 per cent for males, and from 12.8 to 14.4 per cent for females (Figure 2). The increase in the reported prevalence of handicap was much greater, from 9.4 to 13.7 per cent for males Figure 2 Trends in age-standardized prevalence of disability, handicap and severe handicap, by sex, 1981 to 1993 Per cent Per cent 20 20 Females Males lity 15 Disability 15 " Handicap 10 5. Severe handicap Severe handicap I 1981 I 1988 I 1993 Year 0 . I 1981 I 1988 I 1993 Year Note: Age-standardized to total Australian population, 1988. Note that 1993 prevalence rates are derived on the basis of the 1988 form of the disability screening question. Input data sources: Tabulations supplied by the Australian Bureau of Statistics from the 1981 Survey of Handicapped Persons, the 1988 Survey of Disabled and Aged Persons and the 1993 ABS Survey of Disability, Ageing and Carers. and from 8.7 to 12.2 per cent for females. In contrast, the prevalence of severe handicap did not increase for men between 1981 and 1988, and increased by only a small amount for women. For both sexes, the age-standardized prevalence of disability increased slightly between 1988 and 1993 (Figure 2). In contrast, the age-standardized prevalence of handicap declined slightly, from 13.7 to 13.3 per cent for males and from 12.2 to 11.7 per cent for females. The prevalence of severe handicap also declined slightly, to become very close to its levels in 1981 (Figure 2). For both sexes, the prevalence of severe handicap increased between 1988 and 1993 for people aged less than 40 years, but decreased slightly for people of 40 years and over. Health Expectancies in 1993 The 1993 survey data and ABS life tables for 1993 have been used to estimate health expectancies for Australia for 1993. Total life expectancy at birth was 75.0 years for Australian males and 80.9 years for Australian females in 1993. Disability-free life expectancy at birth was 58.4 years for males and 64.2 years for females (Table 1). The difference between these two sets of figures is the expectation at birth of years of disability: 16.6 years for men and 16.7 years for women. In other words, for both men and women, just under 80 per cent of life will be lived without disability on average, if death rates and disability prevalence rates at all ages remain constant at their 1993 levels respectively. Of the years of disability, 12.6 are years of handicap and 3.4 are years of severe or profound handicap for males. Females experience more years of handicap from birth (14.0) and 5.7 of these are years of severe handicap, almost double that for males. Men have a lower life expectancy at birth than w o m e n and also a lower expectation of years of disability, handicap and severe handicap (Table 1). Table 1 Health expectancies at birth, by sex, Australia 1993 Expectation of life at birth with severe handicap with handicap, not severe with disability, but not handicapped free of disability Total life expectancy at birth (LE) HE (years) Males Females HE/LE (%) Males Females 3.41 9.23 3.95 58.39 74.98 4.6 12.3 5.3 77.9 5.72 8.23 2.72 64.18 80.85 7.1 10.2 3.4 79.4 Input data sources: 1993 Australian life tables (ABS 1994) and tabulations supplied by the Australian Bureau of Statistics from the 1993 ABS Survey of Disability, Ageing and Carers. Similar patterns are evident for health expectancies at age 65, although the proportion of remaining life spent free of disability is much lower at 41 per cent for men and 47 per cent for w o m e n (Table 2). Although total life expectancies of females significantly exceed those of males at all ages, the sex differentials for health expectancies are much lower and decrease more rapidly with age. Indeed, life expectancy free of severe handicap is only 1.5 years greater for women at age 65. Trends in H e a l t h Expectancies f r o m 1981 to 1993 Between 1988 and 1993, life expectancy at birth increased from 73.1 to 75.0 years for males and 79.5 to 80.9 years for females. Over the same period, disability-free life expectancy remained unchanged for males, whereas handicap-free life expectancy increased by 1.4 years and severe handicap-free life expectancy increased by 1.7 years (Table 3 and Figure 3). In contrast, disability-free life expectancy increased for females (63.4 to 64.0 9 Table 2 Health expectancies at age 65, by sex, Australia 1993 Expectation of life at 65 years with severe handicap with handicap, not severe with disability, but not handicapped free of disability Total life expectancy at age 65 (LE) HE (years) Males Females HE/LE (%) Males Females 2.36 4.96 1.90 6.51 15.73 34.5 10.3 12.1 41.4 4.65 4.55 I. 19 9.09 19.48 23.9 23.4 6.1 46.7 Input data sources: 1993 Australian life tables (ABS 1994) and tabulations supplied by the Australian Bureau of Statistics from the 1993 ABS Survey of Disability, Ageing and Carers. / , Table 3 Trends in health expe~tancles at birth, by sex, Australia, 1981, 1988 and 1993 / Sex Health expectancy(years) Males Life expectancy Severe handicap expectancy Handicap expectancy Disability expectancy Disability-freeexpectancy Females Life expectancy Severe handicap expectancy Handicap expectancy Disability expectancy Disability-freeexpectancy Change 1988-1993 1981 1988 1993 71.4 2.9 7.8 12.2 59.2 73.1 3.2 12.1 14.7 58.4 75.0 3.4 12.6 16.6 58.4 + 1.9 +0.2 +0.5 +1.9 0.0 78.4 5.2 9.8 13.4 65.0 79.5 6.0 14.0 16.0 63.4 80.9 5.7 14.0 16.7 64.0 + 1.4 -0.3 0.0 +0.7 +0.6 Input data sources: Australian life tables (ABS Deaths Australia Bulletins) and tabulations supplied by the Australian Bureau of Statistics from the 1981 Survey of Handicapped Persons, the 1988 Survey of Disabled and Aged Persons and the 1993 ABS Survey of Disability, Ageing and Carers. years), as did handicap-free life expectancy (an increase of 1.4 years) and severe handicap-free life expectancy (an increase of 1.7 years). A similar pattern was found for health expectancies at age 65. The proportion of remaining life at age 65 which is free of disability remained almost constant for w o m e n at around 46 per cent, whereas that for men declined from 45 per cent in 1988 to 41 per cent in 1993. Unlike other countries for which health expectancy time series are available, Australian health expectancies do not yet provide any evidence for the occurrence of 10 Figure 3 Trends in health expectancies at birth, by sex, Australia 1981, 1988 and 1993 75 85 LE -- Females SHFLE 8o 70 _~..4 m....,~ ~. -~ J a,.,p~ . HFLE .x I~ 7o '~ 65 I&._..~.,,.... 60 9.... Z 55 1980 ~'""t "'t'"--41I ......... "=4 I 19s5 19~ DFLE I 60 1995 ~--"---41-- ''-'''~-'~1 I 19s0 19s5 SHFLE HFLE I I 1990 1995 Year Year Input data sources: See Table 3. compression of morbidity, when that is defined in terms of a fairly wide definition of disability. Discussion Possible factors involved in the substantial increase in reported disability and handicap prevalence levels in Australia during the 1980s have been discussed in detail by Mathers (1991). Self-assessment of limitations or need for assistance in relation to specified activities may have changed in line with changing community perceptions of disability and handicap. Changing attitudes may have resulted in people being more aware of disabling conditions, or more willing to report such conditions and may also have affected how people interpreted 'need' and 'difficulty', concepts used to determine presence and severity of handicap. Another factor may have been changes in the availability of aids for disabled people, as use of an aid is a determinant of mild handicap. Government programs for provision of aids expanded during the 1980s and may have contributed to the increase in selfreported prevalence of handicap. Data from the Australian disability surveys presented here suggest there has been little change in age-specific prevalences of disability and handicap over the last five years (1988 to 1993), although the numbers of disabled and handicapped people are increasing as the population ages. The three theories about the evolution of the health status of populations may be expressed in terms of the relationship between health expectancy and 11 life expectancy. Using disability as an example, the 'pandemic' theory corresponds to a decline in the ratio of disability-free life expectancy to life expectancy, 'compression of morbidity' to an increase in the ratio of disability-free life expectancy to life expectancy (compression of morbidity may be unrelated to the rectangularization of the survival curve). Taking into account levels of severity, the theory of 'dynamic equilibrium' implies a decline in the ratio of total disability-free life expectancy to life expectancy and a levelling off or increase in the ratio of severe disability-free life expectancy to life expectancy. In 1991, in a first attempt to compare international trends in the evolution of population health status, Robine divided United States, English, and Australian disability-free life expectancy time series into four levels according to severity of disability used: level I is very severe disability, level II is severe to very severe, level III is moderate to very severe, and level IV is mild to very severe disability. His results showed a large increase in life expectancy over a period of 25 years and, by contrast, a stagnation in disability-free life expectancy: thus the years of life expectancy gained were equivalent to extra years of disability. However, life expectancies without severe disability have been on a parallel course to total life expectancy. These results indicated a pandemic of light and moderate but not of severe disabilities and have tended to support the theory of 'dynamic equilibrium' (Robine 1991, 1994; Robine, Mathers and Brouard 1993). The international evidence suggests that increases in disability prevalence began in the late 1960s and 1970s at the time when mortality rates at older ages began to decline significantly, but that these increases were confined to the less severe end of the disability spectrum. During this period, secondary prevention, the early detection of disease and subsequent intervention to slow its progress, was emphasized for many major fatal and non-fatal diseases. In addition, greater awareness of chronic conditions due to improved diagnostic techniques, more frequent contacts with the health care system, and perhaps, better communication by doctors to patients has probably led to increased reporting of chronic disease conditions in surveys (Colvez and Blanchet 1981) and to behaviour modification. For example, over the last two decades there has been substantially increased screening for and treatment of high blood pressure which may have led more people to restrict their activity without any change in the underlying incidence or prevalence of high blood pressure. Reviews of the recent international data (Robine, Mathers and Brouard 1993; Robine 1994) have suggested that there is no evidence of expansion of morbidity based on more severe measures o f disability prevalence. The Australian health expectancy estimates presented above show that, although 12 disability-free life expectancy has declined slightly for males, it has increased slightly for females, as has handicap-free life expectancy for both sexes. Trends in severe handicap-free life expectancy continue to parallel those for total life expectancy, although there is some suggestion of a compression of severe handicap expectancy for older women, but not men. Unlike other countries for which health expectancy time series are available, Australian health expectancies do not yet provide persuasive evidence for the occurrence of compression of morbidity, when that is defined in terms of a fairly wide definition of disability. References Australian Bureau of Statistics. 1982. Handicapped Persons Australia 1981. Catalogue No. 4343.0. Canberra. Australian Bureau of Statistics. 1988. Disabled and Aged Persons Australia 1988. Preliminary Results. Catalogue No. 4118.0. Canberra. Australian Bureau of Statistics. 1990. Disability and Handicap Australia 1988. Preliminary Results. Catalogue No. 4120.0. Canberra. Australian Bureau of Statistics. 1993. Disability, Ageing and Carers: Summary of Findings Australia 1993. Catalogue No. 4430.0. Canberra. Australian Bureau of Statistics. 1994. Deaths Australia 1993. Catalogue No. 3101.0. Canberra. Australian Institute of Health and Welfare. 1993. Australia's Welfare 1993: Services and Assistance. Canberra: Australian Government Publishing Service. Australian Institute of Health and Welfare. 1994. Australia's Health 1994: The Fourth Biennial Report of the Australian Institute of Health and Welfare. Canberra: Australian Government Publishing Service. Bebbington, A.C. 1992. Expectation of life without disability measured from the OPCS disability surveys. Pp.23-34 in J.M. Robine, M. Blanchet and E.J. Dowd (eds), Health Expectancy. London: Office of Population Censuses and Surveys, Her Majesty's Stationery Office. Bone, M.R. 1992. International efforts to measure health expectancy. Journal of Epidemiology and Community Health 46:555-558. Bonneuil N., N. Brouard and J.M. Robine. 1992. Active Life Expectancy: A Method of Calculation. REVES Paper No. 92, presented to Fifth International Meeting of the Network on Health Expectancy (REVES), Ottawa. Brouard, N. 1980. Esp6rance de vie active, reprise d'activit6 f6minine: un module. Revue Economique 31:1260-1287. Brouard, N. 1990. Evaluation of Existing Analytic Approaches. REVES Paper No. 45, presented to Third International Meeting of the Network on Health Expectancy (REVES), Durham. Brouard, N. and J.M. Robine. 1992. A method for calculation of health expectancy applied to longitudinal surveys of the elderly in France. Pp.87-97 in J.M. Robine, M. Blanchet and E.J. Dowd (eds), Health Expectancy. London: Office of Population Censuses and Surveys, Her Majesty's Stationery Office. Colvez, A. and M. Blanchet. 1981. Disability trends in the United States population 1966-1976: analysis of reported causes. American Journal of Public Health 71:464-470. Colvez, A. and M. Blanchet. 1983. Potential gains in life expectancy free of disability: a tool for health planning Int. J. Epidemiol 12:224-229. Crimmins, E.M., Y. Saito and M.D. Hayward. 1993. Sullivan and multi-state methods of estimating active life expectancy: two methods, two questions, two answers? Pp.155160 in J.M. Robine, C.D. Mathers, M.R. Bone and I. Romieu (eds), Calculation of Health Expectancies, Harmonization, Consensus Achieved and Future Perspectives. Colloque INSERM (Institut National de la Sant6 et de la Recherche M&licale) Vol. 226. France: John Libbey Eurotext and Les Editions INSERM. 13 Fries, J.F. 1980. Aging, natural death, and the compression of morbidity. New England Journal of Medicine 303:130-135. Fries, J.F. 1989. The compression of morbidity: near or far? Milbank Memorial Foundation Quarterly~Health and Society 67:208-232. Gruenberg, E.M. 1977. The failures of success. Milbank Memorial Foundation Quarterly/Health and Society 55:3-24. Hoem, J.M. 1977. A Markov chain model of working life tables. Scandinavian Actuarial Journal 1-20. Katz, S., L.G. Branch, M.H. Branson, et al. 1983. Active life expectancy. New England Journal of Medicine 309(20): 1218-1224. Kramer, M. 1980. The rising pandemic of mental disorders and associated chronic diseases and disabilities. Acta Psychiatrica Scandinavica 62 (Supplement 285):282-297. Manton, K.G. 1982. Changing concepts of morbidity and mortality in the elderly population. Milbank Memorial Foundation Quarterly~Health and Society 60:183-244. Manton, K.G. 1987. Response to 'an introduction to the compression of morbidity'. Gerontologica Perspecta 1:23-30. Mathers, C.D. 1991. Disability-Free and Handicap-Free Life Expectancy in Australia 1981 and 1988. Australian Institute of Health, Health Differentials Series No. 1. Canberra: Australian Government Publishing Service. Mathers, C. and J.M. Robine. 1993. Health expectancy indicators: a review of the work of REVES to date. Pp.l-21 in J.M. Robine, C.D. Mathers, M.R. Bone and I. Romieu (eds), Calculation of Health Expectancies, Harmonization, Consensus Achieved and Future Perspectives. Colloque INSERM (Institut National de la Sant6 et de la Recherche M6dicale) Vol. 226. France: John Libbey Eurotext and Les Editions INSERM. Olshansky, S.J., M.A. Rudberg, B.A. Carnes, C.K. Cassel and J.A. Brody. 1991. Trading off longer life for worsening health. Journal of Aging and Health 3:194-216. R6seau Esp6rance de Vie en Sant6 (REVES). 1993. Statistical World Yearbook 'Health Expectancy'. Montpellier: REVES. Robine, J.M. 1991. Changes in Health Conditions Over Time. Paper presented to Fourth International Meeting of the Network on Health Expectancy (REVES), Leyden. Robine, J.M. 1992. Disability-free life expectancy. Pp.81-82 in J.M. Robine, M. Blanchet and E.J. Dowd (eds), Health Expectancy. London: Office of Population Censuses and Surveys, Her Majesty's Stationery Office. Robine, J.M. 1994. Disability-free life expectancy trends in France 1981-1991, international comparison. Pp.43-52 in C.D. Mathers, J. McCallum and J.M. Robine (eds), Advances in Health Expectancies: Proceedings of the Seventh Meeting of the International Network on Health Expectancy (REVES), Australian Institute of Health and Welfare, Canberra, February 1994. Robine, J.M., A. Colvez, D. Bucquet, F. Hatton, B. Morel and S. Lelaidier. 1986. L'Esp6rance de Vie Sans Incapacit6 en France en 1982. Population 41:1025-1042. Robine, J.M. and C.D. Mathers. 1993. Measuring the compression or expansion of morbidity. through changes in health expectancy. Pp.269-286 in J.M. Robine, C.D. Mathers, M.R. Bone and I. Romieu (eds), Calculation of Health Expectancies, Harmonization, Consensus Achieved and Future Perspectives. Colloque INSERM (Institut National de la Sant6 et de la Recherche M6dicale) Vol. 226. France: John Libbey Eurotext and Les Editions INSERM. Robine, J.M., C. Mathers and N. Brouard. 1993. Trends and differentials in disability-free life expectancy. In International Union for the Scientific Study of Populations (IUSSP) Seminar, Health and mortality trends among elderly populations: determinants and implications, Sendai, Japan, 21-25 June 1993. Rogers, R.G., A. Rogers and A. Belanger. 1989. Active life among the elderly in the United States: multistate life-table estimates and population projections. Milbank Memorial Foundation Quarterly/Health and Society 67:370-411. Rogers, R.G,, A. Rogers and A. Belanger. 1990. Longer life but worse health? Measurement and dynamics. Gerontologist 30:640-649. Sanders, B.S. 1964. Measuring community health levels. American Journal of Public Health 54:1063-1070. 14 Schoen, R. and K. Woodrow. 1980. Labor force status life tables for the United States, 1972. Demography 17:297-322. Sullivan, D.F. 1971. A single index of mortality and morbidity. Health Services and Mental Health Administration (HSMHA) Health Reports 86:347-354. United States Department of Health, Education, and Welfare. 1969. Toward a Social Report. Washington DC: US Department of Health, Education, and Welfare. Wilkins, R. and O.B. Adams. 1983. Healthfulness of Life: A Unified View of Mortality, Institutionalization and Non-lnstitutionalized Disability in Canada, 1978. Montr6al: L'Institut de recherches politiques, 162. Wolfbein, S.L. 1949. The length of working life. Population Studies 3:286-294. World Health Organization (WHO). 1980. Classification of Impairments, Disabilities and Handicaps. Geneva: WHO. 15