Does socio-economic advantage lead to a longer, healthier old age?

ARTICLE IN PRESS

0277-9536/$ - se

doi:10.1016/j.so

�Correspondfax: +4411625

E-mail addr

[email protected] (C

(R.M. Hancock

Social Science & Medicine 62 (2006) 2489–2499

www.elsevier.com/locate/socscimed

Does socio-economic advantage lead to a longer,healthier old age?

Ruth J. Matthewsa,�, Carol Jaggera, Ruth M. Hancockb

aUniversity of Leicester, Leicester, UKbUniversity of Essex, UK

Available online 20 December 2005

Abstract

The effect of socio-economic disadvantage on mortality is well documented and differences exist even at older ages.

However, whether this translates into differences in the quality of life lived at older ages is less well studied, and in

particular in the proportion of remaining life spent without ill health (healthy life expectancy), a key UK Government

target. Although there have been studies exploring socio-economic differences in disability-free life expectancy (DFLE)

worldwide, these have tended to focus on a single measure of socio-economic advantage, for example, education, race,

social class or income, with the majority based on cross-sectional data from younger populations. In this prospective study

we examine differences in DFLE and total life expectancy (TLE) at older ages using a range of measures of socio-economic

advantage. We use a longitudinal study of 1480 participants aged 75 years or over in 1988 registered with a UK primary

care practice, who were followed up until 2003 with measurements at up to seven time points. Disability was defined as

difficulty with any one of five activities of daily living.

The largest differences in DFLE for both men and women were found for housing tenure. Women aged 75 years living in

owned or mortgaged property could expect to live 1 year extra without disability compared with those living in rented

accommodation, while for men the difference was almost 1.5 years. The effect of socio-economic advantage on disability-

free and total life expectancies appeared to be larger for men than women. In women, socio-economic advantage had more

effect on DFLE than total life expectancy for all indicators considered, thus the socio-economically advantaged

experienced a compression of disability.

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Healthy life expectancy; Longitudinal studies; Socio-economic status; Older people; UK

Introduction

Life expectancy is increasing steadily in theUnited Kingdom, and ensuring that this increase

e front matter r 2005 Elsevier Ltd. All rights reserved

cscimed.2005.11.019

ing author. Tel.: +441162525479;

25423.

esses: [email protected] (R.J. Matthews),

. Jagger), [email protected]

).

is not accompanied by an increase in poor health forolder people is a target for three Governmentstrategies, in particular Standard Eight of theNational Service Framework (NSF) for OlderPeople (Department of Health, 2001). Evidence todate suggests that the increase in total life ex-pectancy (TLE) has not been reflected in healthy lifeexpectancy for the UK, with the extra years livedbeing spent with ill health (Bissett, 2002; Kelly,Baker, & Gupta, 2000). For disability, a major

.

www.elsevier.com/locate/socscimed

ARTICLE IN PRESSR.J. Matthews et al. / Social Science & Medicine 62 (2006) 2489–24992490

health concern in later life, research in the UK hasshown that while the level of severity of disabilityhas diminished overall, the extra years of life gainedhave not been years free of disability (Grundy,Ahlburg, Ali, Breeze, & Sloggett, 1994), in contrastto evidence from Europe (Cambois, Robine, &Hayward, 2001) and the United States (Crimmins,2004) where a compression of morbidity is sug-gested. The UK Treasury’s long-term projections ofthe costs of an ageing population assume that theproportion of life spent in long-term care willremain constant but acknowledge that this is acautious assumption and do not rule out acompression of morbidity for the UK (Treasury,2004).

The association between increased mortality andlower socio-economic status is well documented andthere is growing evidence of a similar relationshipwith disability in older age groups (Arber & Ginn,1993; Grundy & Glaser, 2000; Grundy & Holt,2000). Whether these relationships translate intodifferences in the proportion of remaining life spentwith disability has been less conclusively studied inthe UK. Disability free life expectancy (DFLE)combines mortality and morbidity by dividing TLEinto various states to estimate the average years ofremaining life a person can expect to live withoutdisability. Health expectancies, of which DFLE isone, are independent of the size and age structure ofpopulations and so allow direct comparisons of thedifferent socio-economic groups that make up thepopulation.

Worldwide there have been a number of studiesexploring socio-economic differences in DFLE(Crimmins & Cambois, 2003), although almost allthe evidence is based on cross-sectional data, andthere are few studies of the older population. Moststudies have focussed on a single measure of socio-economic status, e.g. education, race, social class orincome. A number of studies have explored theassociation between education and DFLE, althoughthese have mostly been based on younger adults(Crimmins & Saito, 2001; Doblhammer & Kytir,1998; Sihvonen, Kunst, Lahelma, Valkonen, &Mackenbach, 1998; Valkonen, Sihvonen, & Lahel-ma, 1997). Evidence on older populations arelimited to the US, where poorer educational statushas been linked to shorter DFLE and TLE(Guralnik, Land, Blazer, Fillenbaum, & Branch,1993). However, in the UK most of today’s olderpopulation left school at the minimum age, limitingthe extent to which differentiation by education as

measured by school leaving age is possible (Grundy& Holt, 2001). There are also disadvantages in usingoccupational measures such as social class in peopleof retirement age. Therefore, it has been recom-mended that studies focussing on socio-economicinequalities in older people should focus on a set ofmeasures rather than a single indicator of socio-economic status (Huisman, Kunst, & Mackenbach,2003).

In the UK the focus has been on social class withone study finding that differences in DFLE betweensocial classes were larger than differences in TLE(Bebbington, 1993). More recent estimates have alsofound that the more privileged populations definedby social class could expect to live longer lives with ashorter period of disability (Melzer, McWilliams,Brayne, Johnson, & Bond, 2000). However, both ofthese studies are based on cross-sectional data, andso only take into account the progression to orrecovery from poor health implicitly in the pre-valence rates. The use of longitudinal data allows usto explicitly model transitions to and from disabilitythrough to death.

In our earlier work, the association between onsetof disability and a range of individual, householdand area level measures of socio-economic statuswere explored, and a subjective measure of incomeadequacy was found to have the strongest associa-tion with onset (Matthews, Smith, Hancock, Jagger,& Spiers, 2005). Using the same population ofpeople aged 75 years and over followed for 15 years,this paper extends the previous work to incorporaterecovery from disability and mortality, and is thefirst to use UK longitudinal data for an olderpopulation and a range of socio-economic measuresto examine differences in the proportion of remain-ing life spent with disability at older ages by socio-economic status.

Methods

The setting for this study was a single largegeneral practice (list size 33,000), which is the soleprovider of primary care services for the town ofMelton Mowbray, Leicestershire and the surround-ing area. Data for this study come from twosources. A cross-sectional study undertaken in1988 of all patients aged 75 years and overregistered with the practice on 31 December 1987(Jagger, Clarke, & Clarke, 1991) provides baselinedata. During 1990, the practice began offering aroutine health assessment to all patients aged 75

ARTICLE IN PRESSR.J. Matthews et al. / Social Science & Medicine 62 (2006) 2489–2499 2491

years and over as part of a contractual requirementof general practitioners in the UK to offer annualhealth assessments to patients of this age group(Department of Health and the Welsh Office, 1989).Data from six waves of health assessments areavailable for analysis and provide longitudinal datafrom 1990 to 2003. Full details of the organisationof the health assessments have been reportedpreviously (Jagger, Clarke, O’Shea, & Gannon,1996) but are briefly described here. All participantsin the original 1988 survey were eligible for everyhealth assessment, unless they were known to havedied, or were no longer registered with the practicebecause they had moved out of the area. For eachround of health assessments, the practice popula-tion is divided into 12 geographic areas, and ataround monthly intervals, all patients aged 75 yearsor over from one of the areas are offered a healthassessment carried out by a practice nurse in thepatient’s own home. Although the population isapproached systematically, there are no time con-straints, as in longitudinal studies, to complete eachwave of assessments within a certain time. There-fore, the time intervals between assessments forindividuals vary, as well as within waves betweenindividuals.

Three states were used for the calculation ofDFLE: disabled, not disabled and dead. In the 1988survey, trained interviewers asked participantsabout difficulties with performing a number ofactivities of daily living (ADLs) including getting toand from the toilet, getting in and out of a chair,getting in and out of bed, dressing and feeding.These questions were also asked in the six waves offollow up health assessments. Participants weredefined as not disabled if they self-reported beingable to perform (without difficulty, aids or help) allof these five ADLs. All patients were flagged at theNational Health Service Central Registry, whoprovided information on the deaths of studymembers.

The 1988 survey included a number of measuresof socio-economic status like housing tenure,income, receipt of means tested state benefits, socialclass and whether participants reported difficultiesmanaging on their income. Social class was deter-mined from the last (or usual) occupation for men.For married women, social class was based on theirhusband’s job, and for single women their own job,or if never employed their father’s job. Participantswere asked if they were willing to answer questionsabout their income and financial circumstances.

Those who were willing were asked about their totalincome (of self and spouse, if any). Income wascategorised into three weekly income groups foranalysis, reflecting levels of the basic state pensionin 1988. Following the questions on income andfinancial circumstances, participants were askedwhether they had any difficulties managing on theirincome: ‘‘Do you find this adequate or is it difficultto manage on that income?’’

The Townsend deprivation score (Townsend,Phillimore, & Beattie, 1988) was chosen as ameasure of area level deprivation. Postcodes ofsubjects at the time of the 1988 survey were assignedto enumeration districts (areas of approximately 200households) to calculate Townsend deprivationscores using 1991 National census data on unem-ployment, overcrowding, lack of car and non-owneroccupation. Tertiles of Townsend deprivation scorewere constructed based on the population ofEngland and Wales for analysis.

Statistical methods

To examine the impact of socio-economic ad-vantage on the individual transition rates, wefirst calculated the prevalence of, incidence to,and recovery from disability and mortality bythe measures of socio-economic status with inci-dence, remission (recovery from disability), andmortality rates calculated using the person-yearsmethod.

ImaCH software (Lievre & Brouard, 2003) wasused to calculate DFLE and TLE by each socio-economic indicator from the 1988 survey, andincorporating up to six waves of follow up fortransitions to and from disability through to death.Multinomial logistic regression models were usedto model the transition probabilities between thethree states (disability free, disabled and dead) byage and socio-economic category. The embeddedMarkov process was used to partition the timeintervals between successive health assessmentsinto shorter steps of 1 month. Transitions betweenstates were then represented as the product ofthese 1-month transitions to approximate theunderlying continuous time process. Under theMarkov assumption, the probability of each transi-tion, i.e. recovery from (or onset of) disability isindependent of the duration of current disability(or good health) as well as the occurrence ofprevious episodes of disability (or good health).The software is able to deal with the variable time


intervals between assessments of disability whichoccurred due to the rolling nature of the healthassessments, and is also able to deal with missingassessments of disability at any of the six follow uphealth assessments.

Participants with missing data for each socio-economic indicator were excluded from the calcula-tion of DFLE for that indicator. Generally, theproportion of participants excluded for this reasonwas small (o4%), but income (13.5% missing) andreceipt of means-tested benefits (9.8%) had a higherproportion of missing data. Comparison of thosewith and without income data found no differencesin age, gender, disability in 1988 or death over thestudy period. However, women, those aged 85 yearsand over, those with disability and those who diedover the study period were all more likely to havemissing benefits data. A number of participants(n ¼ 143, 9.6%) were resident in warden-assistedhousing or living in a friend or relative’s home in1988. These participants were excluded from thecalculation of DFLE as their tenure could not beestablished and in any case, would be likely to beassociated with their health rather than socio-economic status.

Results

Of those eligible for the baseline survey, 96(5.7%) did not participate, and 1579 were inter-viewed. Those living in residential care (79 par-ticipants) are excluded here, as they were not askedthe relevant questions on socio-economic status.One participant was excluded because they wereyounger than 75 years at the baseline interview,three participants had missing ADL data in 1988and 16 participants did not receive a follow uphealth assessment and were still alive, so did notcontribute to the calculations of DFLE. Therefore,a total of 1480 participants were included in theanalysis, of whom 34.9% (517) were male and65.1% (963) were female. The median age of menand women in 1988 was 79 and 80 years,respectively.

Table 1 shows the prevalence of disability in 1988,incidence, remission, and mortality rates (personyears) by gender and indicator of socio-economicadvantage. Mortality rates were higher in men (15.1per 100 person years) than women (12.2 per 100person years). However, there was a higher pre-valence of disability in women at baseline (30.3%vs. 23.0%), and more women experienced onset of

disability over the 15 years of follow up than men(5.1 per 100 person years vs. 4.0 per 100 personyears). Generally, mortality rates were higher forthe socio-economically disadvantaged, with theexception of women reporting difficulties managingon their income whose mortality rate was veryslightly less than women who reported their incomewas adequate. The prevalence and incidence ofdisability were also generally higher for the socio-economically disadvantaged, although women inthe 2nd tertile of Townsend deprivation scoreappeared to have a higher prevalence and incidenceof disability than the most deprived. Interestingly,remission rates were lower for the socio-economic-ally advantaged as measured by receipt of means-tested benefits, social class, self-perceived difficultiesmanaging on income and Townsend deprivationscore in women. However, the numbers reportingrecovery were small.

TLE and DFLE by each measure of socio-economic advantage at age 75 are shown inTable 2. DFLE was not calculated separately formen and women by income, as there were conver-gence problems due to small numbers. The differ-ences in TLE and DFLE for the socio-economicallydisadvantaged compared to those advantaged ran-ged from 0.2 years to 0.7 years for TLE in womenand 0.2 years to 1.4 years in men, whilst differencesin DFLE ranged from 0.4 years to 1.1 years inwomen and 0.2 to 1.5 years in men (Fig. 1).Although these differences appear small, they areof similar magnitude to the increases in lifeexpectancy at age 75 years of 1.0 year for menand 0.8 years for women observed over the 10-yearperiod 1991–2001.

Differentials in DFLE by housing tenure werelargest for both men and women. Women aged 75years living in owned or mortgaged property couldexpect to live 1 year extra without disabilitycompared with those who lived in rented accom-modation, while in men the difference was justunder 112 years. The effect of social class on DFLEwas greater for men than women; with a loss inDFLE of nearly a year in men of manual social classcompared with those from non-manual social class,in contrast with a difference of just over half a yearin women. The lower remission rate for the moresocio-economically advantaged women by Town-send score along with a lower mortality rate(Table 1) carried through to produce lower DFLEin this group compared to the socio-economicallydisadvantaged (Fig. 1).

ARTICLE IN PRESS

Table 1

Prevalence of disability in 1988, incidence and remission of disability and mortality over 15 years of follow up by gender and socio-

economic status for 1480 participants

Prevalence % (n) of

disability

Incidence ratea (number

of incident cases)

Remission ratea (number

of remissions)

Mortality ratea (number

of deaths)

Women Men Women Men Women Men Women Men

Overall 30.3 (292) 23.0 (119) 5.1 (230) 4.0 (95) 4.2 (52) 5.2 (20) 12.2 (831) 15.1 (472)

Townsend deprivation score

Best off tertile 25.4 (107) 21.9 (55) 4.6 (101) 3.6 (44) 3.5 (17) 5.5 (10) 11.6 (362) 13.6 (216)

2nd tertile 38.3 (95) 21.1 (28) 5.8 (56) 4.7 (28) 4.7 (18) 6.7 (6) 12.9 (213) 16.4 (129)

Worst off tertile 30.7 (89) 27.1 (36) 5.5 (73) 4.2 (23) 4.7 (17) 3.4 (4) 12.5 (252) 17.0 (126)

Housing tenure

Owns/mortgaged 24.7 (107) 19.1 (57) 4.5 (108) 3.6 (55) 6.1 (26) 5.1 (10) 10.6 (354) 13.6 (261)

Private/council rented 31.3 (95) 24.7 (36) 5.7 (76) 4.4 (26) 3.4 (16) 4.4 (5) 12.2 (263) 17.3 (139)

Living with others/

warden assisted

40.2 (90) 36.1 (26) 5.9 (46) 5.9 (14) 3.0 (10) 6.3 (5) 15.9 (211) 17.9 (70)

Income

Married 4£70/Not

married 4£50

28.8 (68) 24.2 (64) 4.9 (56) 4.1 (51) 3.7 (11) 6.5 (15) 11.8 (198) 13.7 (236)

Married £61-£70/Not

married £41-£50

31.2 (146) 20.3 (33) 5.2 (115) 4.4 (32) 5.3 (33) 4.5 (4) 11.7 (399) 16.6 (91.3)

Married o£61/Not

married o£41

30.5 (39) 30.4 (7) 5.0 (29) 5.6 (4) 2.1 (3) 0 13.7 (115) 22.0 (21)

Receiving means tested benefits

No 23.8 (76) 20.7 (56) 5.0 (84) 3.5 (46) 4.0 (12) 4.6 (8) 11.2 (263) 14.7 (243)

Yes 31.9 (172) 24.6 (51) 5.1 (126) 4.6 (42) 4.9 (38) 5.7 (10) 12.1 (467) 15.1 (191)

Social class

Non-manual 27.6 (92) 17.7 (32) 4.4 (74) 4.3 (41) 3.8 (16) 3.6 (4) 11.7 (288) 13.5 (160)

Manual 32.0 (190) 26.3 (86) 5.5 (149) 3.7 (52) 4.4 (34) 6.0 (16) 12.5 (511) 16.0 (303)

Self perceived difficulties managing on income

Income adequate 29.0 (219) 21.2 (90) 4.8 (179) 3.6 (74) 3.7 (34) 5.1 (16) 12.1 (654) 14.5 (384)

Income not adequate 33.3 (63) 30.3 (27) 6.1 (48) 6.7 (21) 6.4 (18) 5.7 (4) 11.7 (157) 18.0 (84)

aPer 100 person years.

R.J. Matthews et al. / Social Science & Medicine 62 (2006) 2489–2499 2493

Discussion

Earlier work exploring the association between anumber of socio-economic factors and disabilityusing the same longitudinal survey of people aged75 years and over in 1988 found a subjectivemeasure of difficulty managing on income wasthe most important predictor of onset of dis-ability (Matthews et al., 2005). In this paper, weextended the analysis to examine the effect ofthese indicators on the number of years an olderperson can expect to live without disability, andfound that housing tenure showed the greatestdifferential in DFLE in both men and womenalthough the subjective measure of financial diffi-

culty also reflected important differences in DFLE.Differentials between the socio-economically ad-vantaged and disadvantaged in disability-free andTLE at age 75 were generally less than a year, andmay at first appear insignificant. However, increasesin life expectancy in this age group will not besubstantial and indeed these differences are ofsimilar magnitude to the gains in life expectancyat age 75 over the 10-year period 1991–2001. Thesefindings provide a baseline to track whether thedifferentials in DFLE and TLE remain constantover time or whether the socio-economically ad-vantaged experience a compression of morbiditywhilst the disadvantaged experience an expan-sion. If the latter case holds this has important

ARTICLE IN PRESS

Table

2

Disabilityfree

life

expectancy

(DFLE),totallife

expectancy

(TLE)andproportionofremaininglife

spentactivebysocio-economic

statusatage75years

Women

Men

All

TLE(SE)

DFLE(SE)

DFLE/TLE

(%)

TLE(SE)

DFLE(SE)

DFLE/TLE

(%)

TLE(SE)

DFLE

(SE)

DFLE/TLE

(%)

Overall

11.33(0.30)

8.67(0.30)

77

8.84(0.35)

7.25(0.37)

82

10.32(0.24)

8.11(0.23)

79

To

wnse

nd

dep

riva

tion

sco

re

Bestofftertile

11.03(0.40)

8.18(0.34)

74

9.23(0.45)

7.64(0.45)

83

10.54(0.30)

8.37(0.29)

79

2ndtertile

11.37(0.52)

8.65(0.48)

76

8.53(0.60)

7.02(0.55)

83

9.93(0.40)

7.77(0.35)

78

Worstofftertile

11.55(0.46)

8.66(0.45)

75

8.34(0.58)

6.74(0.61)

81

10.23(0.37)

7.85(0.36)

77

Ho

usi

ng

ten

ure

Owns/mortgaged

11.93(0.38)

9.40(0.37)

79

9.44(0.42)

7.97(0.43)

84

10.77(0.29)

8.70(0.29)

81

Private

orcouncilrented

11.27(0.48)

8.33(0.45)

74

8.01(0.54)

6.52(0.54)

81

10.00(0.38)

7.61(0.35)

76

Inco

me

Married4£70/N

otmarried

4£50

——

——

——

10.14(0.34)

8.03(0.33)

79

Married£61-£70/N

otmarried

£41-£50

——

——

——

10.50(0.33)

8.11(0.31)

77

Marriedo£61/N

otmarried

o£41

——

——

——

10.28(0.61)

7.88(0.62)

77

Rec

eivi

ng

mea

ns

test

edben

efits

No

11.52(0.43)

9.00(0.41)

78

8.74(0.44)

7.34(0.46)

84

10.15(0.32)

8.21(0.30)

81

Yes

11.36(0.38)

8.64(0.35)

76

8.93(0.51)

7.13(0.51)

80

10.53(0.31)

8.12(0.29)

77

So

cial

cla

ss

Non-m

anual

11.54(0.45)

9.08(0.41)

79

9.68(0.53)

7.89(0.56)

82

10.77(0.35)

8.59(0.33)

80

Manual

11.26(0.35)

8.49(0.34)

75

8.47(0.40)

6.98(0.41)

81

10.09(0.27)

7.87(0.26)

78

Sel

fper

ceiv

eddif

ficu

ltie

sm

anagin

gon

inco

me

Incomeadequate

11.49(0.33)

8.88(0.33)

77

9.00(0.37)

7.42(0.39)

82

10.42(0.26)

8.25(0.25)

79

Incomenotadequate

10.90(0.56)

8.08(0.46)

74

8.03(0.66)

6.13(0.73)

76

9.89(0.44)

7.45(0.38)

75

—notestimable

dueto

smallnumbers.

R.J. Matthews et al. / Social Science & Medicine 62 (2006) 2489–24992494


implications for future projections of health andsocial care spending.

Patterns of differences in TLE were similar toDFLE although differences in DFLE between theadvantaged and disadvantaged were larger thandifferences in TLE for women, while for men thedifferences were inconsistent. For women in parti-cular, the more socio-economically advantagedlived more years in total than the disadvantagedand also more years disability-free. Moreover, agreater proportion of their remaining life was free ofdisability, and thus the socio-economically advan-taged experienced a compression of disabilitycompared to their disadvantaged counterparts.

Evidence of the association between socio-eco-nomic status and DFLE has tended to focus onyounger populations and a single measure of socio-economic status. The few studies of older popula-tions focus on education, limiting the extent towhich our results can be compared with otherstudies. We found a difference of just over half ayear in the proportion of remaining life a womanaged 75 could expect to live without disabilitybetween those of manual and non-manual socialclass, and a difference of 1 year for men. The mostrecent cross-sectional evidence of people aged 65years and over from the UK (Melzer et al., 2000)showed slightly higher differences at age 65, with adifference of 2.1 years in DFLE between socialclasses I and II (professional/managerial) comparedwith III and IV (manual/partly and unskilledoccupations) in men, and a difference of 1.8 yearsin women at aged 65. We found very little differencein DFLE and TLE between income groups. Thisagrees with our earlier work, which found noassociation between income and onset of disability(Matthews et al., 2005), although this may be partlydue to the available income measure lackingvariation in this population. Evidence on incomeis restricted to two cross-sectional studies fromCanada (Nault, Roberge, & Berthelot, 1996; Wilk-ins & Adams, 1983). One study carried out in the1970s (Wilkins & Adams, 1983), found thatparticipants in the upper quintile of income couldexpect to have longer DFLE and TLE than those inthe bottom quintile.

The effect of social class on both DFLE and TLEwas greater for men than women. This confirms themost recent cross-sectional evidence from the UK(Melzer et al., 2000), which showed that thedifference between socio-economic groups basedon social class was larger in men than women. It has

been argued that measures such as social class maybe less relevant for older people of retirement age asit may not reflect their current status, particularly inwomen (Arber & Ginn, 1993). The difficulty inclassifying women who have never worked may alsocontribute to this finding. However, we also founddifferences to be greater in men using housingtenure and self-perceived difficulty managing onincome, although these gender differences weresmaller. While housing tenure may be associatedwith previous occupation or social class, the self-perceived measure of difficulty managing on incomeis more likely to reflect current status. However,more research is needed to explore the factors thatmay influence this subjective measure of socio-economic status. Nault et al. (1996) also reportedthe difference in healthy and TLE between incomegroups to be larger for men than women. Unfortu-nately, we were not able to confirm this associationbecause of problems with convergence for ourincome variable, due to small numbers.

Evidence from a number of studies has suggestedthat socio-economic status has a greater effect onthe proportion of remaining life without disabilitythan TLE (Bebbington, 1993; Cambois et al., 2001;Sihvonen et al., 1998; Valkonen et al., 1997). Wefound that differences in DFLE were larger thanTLE in women for all socio-economic indicators,with the exception of those in the 2nd tertile ofTownsend deprivation score. However, in men onlydifficulty managing on income showed substantiallylarger differences in DFLE than TLE, whilst thoseof manual social class had greater differences inTLE than DFLE.

There are a number of limitations when inter-preting our results. Firstly, the study is restricted tothe population of a small town and the surroundingarea with a predominately white population, andtherefore may not be representative of the currentolder population. However, earlier work usingNational Census data showed that Melton has asimilar age, sex and social class distribution to thatof England and Wales (Jagger et al., 1991).Secondly, we were limited by small numbers whenmen and women were analysed separately, withconvergence problems for some socio-economicindicators. Thirdly, a relatively high proportion ofparticipants were left out of calculations for incomeand means-tested benefits because they did notanswer these questions. While we found no differ-ence between those who did and did not answer theincome questions, we found that women, those 85

ARTICLE IN PRESS

Females-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

Loss in DFLE (years)

Loss in TLE (years)Yes vs. No

SOCIAL CLASS

TOWNSEND DEPRIVATION SCORE

2nd tertile vs. best off tertile

3rd tertile vs. best off tertile

HOUSING TENURE

Private/council rented vs. Owned/mortgaged

RECEIPT OF MEANS TESTED BENEFITS

Manual vs. Non manual

SELF PERCEIVED FINANCIAL DIFFICULTIES

Income not adequate vs. Adequate income

(a)

Males

-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6




HOUSING TENURE



Yes vs. No

SOCIAL CLASS

Manual vs. non manual


Income not adequate vs. Income adequate


Loss in TLE (years)

(b)

Fig. 1. Loss in years of disability-free and total life expectancy at Age 75 by Socio-economic status. (a) Females; (b) Males; (c) Total

population.

R.J. Matthews et al. / Social Science & Medicine 62 (2006) 2489–24992496

ARTICLE IN PRESS

Total population

-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6


Loss in TLE (years)




HOUSING TENURE


INCOME

Married £61-70/Not married £41-50*

Married <£61/Not married <£41*


Yes vs. No

SOCIAL CLASS

Manual vs. Non manual


Income not adequate vs. Adequate income

*vs. Married >£70/Not married >£50 (c)

Fig. 1. (Continued)

R.J. Matthews et al. / Social Science & Medicine 62 (2006) 2489–2499 2497

years and over, those with disability and those whodied over the study period, were all less likely tohave answered the means-tested benefits question,suggesting that our differences are conservative.Although this study does not include participantsliving in institutions at the time of the 1988 study, itdoes include those who moved into institutionswithin the study period, thus avoiding the loss tofollow up of such people, a common problem inlongitudinal studies of the elderly.

Despite these limitations, this is the first studythat has used longitudinal data to estimate DFLE inolder people using a range of socio-economicindicators. Though this is a survivor population,and the most disadvantaged may well have diedbefore age 75, we have shown that inequalities in theremaining years spent with disability still remainbetween the socio-economically advantaged anddisadvantaged, even after age 75. The finding that

the greatest differentials in DFLE were betweenthose who owned and those who rented their homeis open to different interpretations. Owning one’shome may be an indicator of one or more ofmaterial advantage, social status and better hous-ing, each of which could have positive effects onhealth. There is also a question of whether theseresults will hold in future generations as owner ratesincrease and tenure becomes less of an indicator ofsocio-economic status.

This study has found that socio-economic effectson DFLE and TLE were greater for men thanwomen, but for women, socio-economic advantageon all the indicators considered, had more effect onDFLE than TLE. Our findings confirm the need tomonitor inequalities by taking account of thequality as well as the quantity of life lived atolder ages. Where possible, including a numberof measures of socio-economic status would be


recommended when monitoring inequalities in olderpeople, because of the problems associated withsome measures in this age group. Our findings implythat policies to address inequalities in health atolder ages need to be concerned not only with socio-economic differences in mortality but also withdifferences in health status, particularly for women.To the extent that these differences result fromcircumstances and events earlier in life, healthimprovement policies need to be targeted on bothyounger and older people in disadvantaged posi-tions. It is also important that older people in socio-economically disadvantaged positions should beable to access services and other help which mayalleviate the double disadvantage of low socio-economic position and poor health.

Acknowledgements

This work was undertaken by the authors at theUniversity of Leicester who received funding fromthe Department of Health. The views expressed inthe publication are those of the authors and notthose of the Department of Health. We thank LucySmith, Nicky Spiers, Geraldine Barker and GrahamMartin for their advice as part of the project team.The authors thank Alison Hipkin, Partners andstaff at Latham House Medical Practice for their co-operation, Jan Simms and the practice nurses whocarried out the health assessments, Tony Arthur formanagement of the health assessments and adviceon the data, and Joy Slater for data preparation.

References

Arber, S., & Ginn, J. (1993). Gender and inequalities in health in

later life. Social Science and Medicine, 36(1), 33–46.

Bebbington, A. (1993). Regional and social variations in

disability free life expectancy in Great Britain. In J.-M.

Robine, C. Mathers, M. Bone, & I. Romieu (Eds.),

Calculation of health expectancy: Harmonization, consensus

achieved and future perspectives. Montrouge: John Libbey

Eurotext.

Bissett, B. (2002). Healthy life expectancy in England at

subnational level. Health Statistics Quarterly, 14, 21–29.

Cambois, E., Robine, J. M., & Hayward, M. D. (2001). Social

inequalities in disability-free life expectancy in the French

male population, 1980–1991. Demography, 38(4), 513–524.

Crimmins, E. M. (2004). Trends in the health of the elderly

[Review]. Annual Review of Public Health, 25, 79–98.

Crimmins, E. M., & Cambois, E. (2003). Social inequalities in

health expectancy. In J.-M. Robine, C. Jagger, C. D. Mathers,

E. M. Crimmins, & R. M. Suzman (Eds.), Determining health

expectancies. Chichester: John Wiley and Sons.

Crimmins, E., & Saito, Y. (2001). Trends in healthy life

expectancy in the United States, 1970–1990: Gender, racial,

and educational differences. Social Science and Medicine,

52(11), 1629–1641.

Department of Health. (2001). National service framework for

older people. London: DoH.

Department of Health and the Welsh Office. (1989). General

practice in the National Health Service: A new contract.

London: HMSO.

Doblhammer, G., & Kytir, J. (1998). Social inequalities in

disability-free and healthy life expectancy in Austria. Wiener

Klinische Wochenschrift, 110(11), 393–396.

Grundy, E., Ahlburg, D., Ali, M., Breeze, E., & Sloggett, A.

(1994). Disability in Great Britain. London: DSS.

Grundy, E., & Glaser, K. (2000). Socio-demographic dif-

ferences in the onset and progression of disability in

early old age: A longitudinal study. Age and Ageing, 29(2),

149–157.

Grundy, E., & Holt, G. (2000). Adult life experiences and health

in early old age in Great Britain. Social Science and Medicine,

51(7), 1061–1074.

Grundy, E., & Holt, G. (2001). The socioeconomic status of older

adults: How should we measure it in studies of health

inequalities? Journal of Epidemiology and Community Health,

55(12), 895–904.

Guralnik, J., Land, K., Blazer, D., Fillenbaum, G., & Branch, L.

(1993). Educational status and active life expectancy among

older blacks and whites. New England Journal of Medicine,

329(2), 110–116.

H M Treasury. (2004). Long-term public finance report: An

analysis of fiscal sustainability. London: H M Treasury.

Huisman, M., Kunst, A., & Mackenbach, J. (2003). Socio-

economic inequalities in morbidity among the elderly: A

European overview. Social Science and Medicine, 57(5),

861–873.

Jagger, C., Clarke, M., & Clarke, S. J. (1991). Getting

older-feeling younger: The changing health profile of

the elderly. International Journal of Epidemiology, 20(1),

234–238.

Jagger, C., Clarke, M., O’Shea, C., & Gannon, M. (1996).

Annual visits to patients over the age of 75–who is missed?

Family Practice, 13(1), 22–27.

Kelly, S., Baker, A., & Gupta, S. (2000). Healthy life expectancy

in Great Britain, 1980–96, and its use as an indicator in

United Kingdom strategies. Health Statistics Quarterly, 7,

32–37.

Lievre, A., & Brouard, N. (2003). The estimation of health

expectancies from cross-longitudinal surveys. Mathematical

Population Studies, 10, 211–248.

Matthews, R. J., Smith, L. K., Hancock, R. M., Jagger, C., &

Spiers, N. A. (2005). Socioeconomic factors associated with

the onset of disability in older age: A longitudinal study of

people aged 75 years and over. Social Science and Medicine,

61(7), 1567–1575.

Melzer, D., McWilliams, B., Brayne, C., Johnson, T., & Bond, J.

(2000). Socioeconomic status and the expectation of disability

in old age: Estimates for England. Journal of Epidemiology

and Community Health, 54(4), 286–292.

Nault, F., Roberge, R., & Berthelot, J.-M. (1996). Life

expectancy and health expectancy by sex, marital status,

and socioeconomic status in Canada [French]. Cahiers

Quebecois de Demographie, 25(2), 241–259.


Sihvonen, A. P., Kunst, A. E., Lahelma, E., Valkonen, T., &

Mackenbach, J. P. (1998). Socioeconomic inequalities in

health expectancy in Finland and Norway in the late 1980s.

Social Science and Medicine, 47(3), 303–315.

Townsend, P., Phillimore, P., & Beattie, A. (1988). Health and

deprivation: Inequality and the north. London: Croom Helm.

Valkonen, T., Sihvonen, A. P., & Lahelma, E. (1997). Health

expectancy by level of education in Finland. Social Science

and Medicine, 44(6), 801–808.

Wilkins, R., & Adams, O. (1983). Health expectancy in Canada,

late 1970s: Demographic, regional and social dimension.

American Journal of Public Health, 73, 1073–1080.

Documents

Does socio-economic advantage lead to a longer, healthier old age?