IJE Advance Access originally published online on January 25, 2007
International Journal of Epidemiology 2007 36(2):358-365; doi:10.1093/ije/dyl307
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Educational differences in the dynamics of disability incidence, recovery and mortality: Findings from the MRC Cognitive Function and Ageing Study (MRC CFAS)
1 Leicester Nuffield Research Unit, Department of Health Sciences, University of Leicester.
2 Epidemiology and Public Health Group, Peninsula Medical School, Exeter
3 MRC Biostatistics Unit, Cambridge.
4 Department of Public Health and Primary Care, University of Cambridge.
5 Medical Research Council Cognitive Function and Ageing Study http://www.cfas.ac.uk
* Corresponding author. Leicester Nuffield Research Unit, Department of Health Sciences, University of Leicester, 22–28 Princess Road West, Leicester LE1 6TP, UK. E-mail: cxj{at}le.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Background This study aims to establish the extent of educational differences in the disability transitions of incidence, recovery and mortality in people aged 65 years and over, whether these can be explained by differentials in disease burden and their relative contribution to educational differences in prevalence and disability-free life expectancy (DFLE).
Methods A stratified random sample of 13 004 participants in five areas in England and Wales were interviewed in 1991–94 and followed up at 2, 6 (one centre only) and 10 years. Two levels of disability were analysed: mobility difficulty and activities of daily living (ADL) disability. We fitted logistic regression models to model educational differences in disability prevalence, incidence, recovery and mortality transitions. DFLE was calculated to assess the combined effect of the dynamic transitions.
Results Those with 
9 years education had higher ADL and mobility disability prevalence and higher incidence and lower recovery of mobility disability. Differences in disability incidence remained after adjustment for comorbidity. Women with the lowest education had shorter life expectancies (1.7 years less at the age of 65 years) than the most educated and had even shorter DFLE (1.9 years free of ADL disability and 2.8 years free of mobility difficulty at the age of 65 years).
Conclusions Differentials in education continue to contribute to prevalence of disability at ages beyond 65 years in both men and women and independently of diseases. These appear to be driven predominantly by differentials in disability incidence that also compound to produce greater differentials in DFLE between education groups than in total years lived.
Keywords MRC CFAS, socioeconomic factors, disability, old age, self-report, activities of daily living
Accepted 18 December 2006
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The nature of the links between less privileged socio-economic status and health have been extensively studied in middle aged populations, but rather less so in older people, especially in the UK. For mortality, strong links have been demonstrated between socio-economic status and overall survival in older people.1 Higher prevalence rates of disability (having difficulty undertaking everyday activities) have also been linked to various markers of less privileged social position in many studies, especially in the USA2,3 but also in Europe.4,5 This relationship has been demonstrated in Britain too, in terms of higher prevalence,6 earlier onset7,8 as well as the impact on life expectancy with and without disability9–12.
However, disability is not a fixed state,13–15 but rather a dynamic one. The pool of prevalent disability is determined by incidence, recovery and mortality, both in those who were disabled, and those who were not. There is an established association between socio-economic status and disability incidence,16 but there is sparse evidence for a relationship with recovery from disability, and mortality from disabled states. In the established population for epidemiologic studies of the elderly (EPESE) populations in the USA, Melzer et al.17 demonstrated that the education excess of mobility disability was attributable to higher incidence rates in less privileged groups, and not differences in the other dynamic transitions. Similar findings were reported in Taiwanese18 and Dutch19 cohorts. If these findings apply more widely, using a variety of markers of social position, then there would be clear implications for the nature and timing of efforts to reduce health inequalities in old age. However, in a Chinese population, years of education were associated with both onset of and recovery from activities of daily living (ADL) difficulty.20
In Britain, there have been several reports of higher incidence rates of disability by socio-economic status. Grundy and Glaser13 reported evidence of higher incidence by social class, educational qualifications and housing tenure. Grundy and Holt21 showed that disability status was associated with socio-economic and geographic variables, such as proportion of adult life spent unemployed and residence outside the Southeast of England. Ebrahim et al.22 found that manual social class plus lifestyle factors were strongly and independently associated with increased odds of incident locomotor disability over 12–14 years, in a cohort of British men whilst Adamson et al.23 reported higher incidence rates in locomotor disability by socioeconomic status in the younger elderly in Scotland. Breeze et al.,24 using the Whitehall study male cohort, similarly reported raised rates of poor physical performance in old age by civil service grade, over a 29 year follow-up. Finally, a subjective report of adequacy of income was found to have the strongest relationship with incident ADL disability in Melton Mowbray.8 Rates of recovery or mortality with or without disability have attracted little attention, due mainly to the lack of good large-scale cohort data. The only UK study to report rates of recovery by various measures of social disadvantage is a single-centre longitudinal study that found that both mortality and remission rates were higher for some measures of social disadvantage though numbers of transitions were small.9
Socio-economic differences at an individual level are usually defined by education, occupation, income and material circumstances, or some combination of these markers.25 Income and wealth (including dynamic changes in these) have been linked to health in old age15,26,27 and Robert and House28 have provided evidence of an increasing relative impact of income over education on some measures of health with increasing age. In the UK, occupation has been used as a marker of social status, and household tenure or car ownership have served as markers of material circumstances.8,9,29 However, especially for older people, contemporaneous measures of social position can be misleading, failing to reflect changing status from middle age or earlier.30 Full time education, as a marker, has the advantage of generally being completed early in adulthood, and therefore less likely to suffer from reverse causation. It is also a good measure of long-term economic position, at least in the USA.27
The aim of the analysis presented here was to measure the strength of association between educational status and each of the dynamic transitions of disability, in an older UK population with follow-ups over 10 years. Since a higher burden of disease and ill-health are often associated with lower socio-economic position and disability, we sought to explore whether socio-economic differences in disability transitions were a result of higher disease burden. We further examined socio-economic differences in disability-free life expectancy (DFLE), itself a combination of the dynamic transitions.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The Medical Research Council Cognitive Function and Ageing Study (MRC CFAS) is a longitudinal, population-based, multi-centre study whose original aims were to examine the descriptive epidemiology of dementia in England and Wales. Random samples of people aged 65 years or over were selected from the Family Health Service Authority lists in each of three urban (Newcastle, Nottingham and Oxford) and two rural centres (Cambridgeshire and Gwynedd) with over-sampling of those aged
75 years. A further urban centre (Liverpool) was part of MRC CFAS, but this study had a different design and is excluded from this analysis. A full description of the CFAS study design can be found elsewhere.31 All subjects were screened in their own homes by trained interviewers using a structured interview during 1992–1994, and provided information on socio-demographics (including level of education) and activities of daily living (ADL). Participants were classified into three groups based on the number of years of full-time education undertaken (0–9 years, 10,11 years and 12+ years), reflecting basic/higher education as 9 years was the statutory time for this generation. Surviving individuals were re-interviewed at 2 and 10 years and for one complete centre 6 years.
We used two measures of disability, mobility disability and ADL disability, collected for all participants either at a further interview or more detailed assessment, and identically at follow-up interviews. Mobility disability was defined as having some difficulty or requiring help to get up and down stairs. Participants were classified as having ADL disability if they were unable to perform at least one of the following five ADL without human help: transfer to and from a chair; put on shoes and socks; prepare a hot meal; get around outside and have a bath or an all over wash. Full details of the ADL disability classification are available elsewhere.8
Comorbidity was examined using the number of health conditions reported [from stroke, angina, heart attack, intermittent claudication, chronic bronchitis, asthma (except in childhood), visual impairment, hearing impairment, treated diabetes, Parkinson's disease, high blood pressure confirmed by GP, depression and arthritis]. Cognitive impairment was assessed using the Mini-Mental State Examination (MMSE).32 Responses to MMSE items of ‘don't know’, ‘no answer’ and items that could not be answered due to sensory or dexterity problems were recoded to zero. Individuals were assigned to an MMSE category on the basis of completed items if this could be done unambiguously, otherwise the full score was recoded to missing.
We used MRC CFAS version 8.0 for this analysis, which has information on 13 004 people who took part in the initial screen. Only small amounts of data were missing at baseline: education (n = 337, 2.6%), ADL disability (n = 163, 1.3%), mobility disability (n = 397, 3.1%), MMSE (n = 257, 2.0%) and comorbidity (n = 715, 5.5%). The response rate (as a percentage of those surviving) at 2 years was 79% and at 10 years, 75%, and the median (maximum) time to last interview was 38 (143) months. Overall 12 060 individuals (92.7%) had complete data at baseline and of these, 7229 (60.0%) had died by 31 December 2004 and 1024 (8.5%) had missing information on ADL or mobility status at all follow-ups, but were known to be alive. MRC CFAS oversampled those aged 75 years and over by design, so responses were reweighted to the population age sex distribution in each centre, to correct for the oversampling.
Logistic regression was used to model prevalence of disability (ADL and mobility), incidence to and recovery from disability and state-specific mortality. SAS version 9.1 was used for analysis. In all cases, models for education with adjustment for age were fitted, and then comorbidity and MMSE at baseline were added to examine whether education effects on disability were a result of greater disease burden in those with lower education. Additionally time between last report of no disability/disability and the event of interest (disability/no disability/death) was included as a covariate for analysis of incidence, recovery and mortality, respectively. All analyses were performed for men and women separately.
Disability-free life expectancies (DFLE) were calculated from the baseline, 2-, 6- and 10-year follow-up data using Interpolated Markov Chain (IMaCH) software version 0.98h.33 This technique partitions the time intervals between successive interviews into shorter steps and then models the resulting transition probabilities by multinomial logistic regression on age. Estimated transition probabilities then act as inputs to a multistate life table. We used education as a covariate, but analysed men and women separately.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Of the study population for analysis (n = 12060) 59.5% were female and 9.7% were aged 85 years and over. Years of education were negatively associated with age, 61.4% of the youngest age group having 0-9 years education compared with 66.9% of those aged 85+ years.
Prevalence of disability
Women reported more disability than men at baseline (mobility disability: men 26.3%, women 38.1%; ADL disability: men 11.5%, women 16.0%) (Table 1). In both men and women, the prevalence of ADL and mobility disability were inversely related to years of education, and were not wholly explained by age or greater comorbidity in the lower educated (Table 2). Much of the inequality in the prevalence of ADL disability by years of education was accounted for by cognitive function although inequalities in mobility difficulty remained even after adjustment for cognitive function.
|
|
Mortality
Low education conferred a higher risk of mortality from a disability-free state compared with the highest educated after adjustment for age (ADL disability: men OR = 1.2, 95% CI 1.1–1.4; women: OR = 1.3, 95% CI = 1.2–1.6) although the increased risk disappeared after further adjustment for comorbidity and MMSE. Similar effects were seen for mortality from no mobility (men: OR = 1.2, 95% CI 1.0–1.4; women: OR = 1.5, 95% CI = 1.3–1.8). Lower education did not appear to infer any increased risk of mortality from a disabled state, either ADL or mobility defined.
Incidence of disability
A total of 1480 transitions to ADL disability and 2182 to mobility disability were observed over the 10 year follow-up (Table 1). Having 
9 years of full-time education was associated with greater incidence of mobility disability in both sexes and ADL disability in women (Table 3). ORs were attenuated slightly after adjustment by MMSE and comorbidity but differences remained significant.
|
Recovery from disability
There were 714 transitions from mobility disability to no disability, but the number of transitions from ADL disability to no disability numbered only 211, and therefore modelling was not attempted for recovery from ADL disability (Table 1). Low-educated men and women were significantly less likely to recover from mobility disability although confidence intervals were wide, and the effect was attenuated in women after adjustment for comorbidity and MMSE (Table 3).
Disability-free life expectancy (DFLE)
As we had found evidence that the socially disadvantaged, as measured by years of education, experienced higher mortality and disability incidence, we examined the socio-economic differentials in DFLE. Compared with those with the highest education level (12+ years) women with the lowest level of education had shorter life expectancies (1.7 years less at age 65) and even fewer years free of disability (1.9 years free of ADL disability at age 65 and 2.8 years free of mobility disability at age 65) (Figure 1). In men, socio-economic differentials in life expectancy and mobility DFLE were of similar magnitude to those in women, the lowest educated men experiencing reductions of life expectancy at age 65 years of 1.1 years and 2.4 years fewer free of mobility disability, compared with those with the highest education (Figure 1). Differentials in ADL DFLE in men were smaller to those in women, but still exceeded the differences in total years of life lived.
|
Although differences in ADL and mobility DFLE between those with the highest and lowest education narrowed with age, they were still evident across the whole age range in men and women and exceeded differences in life expectancy (Figures 1 and 2).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Using a population-based cohort of those aged
65 years, we have demonstrated that early-life socio-economic differentials in transitions to and, to a lesser extent, from disability continue to contribute to prevalence differences in both sexes at ages beyond 65 years. Differences in the incidence of disability remained between those with the highest and lowest levels of education even after adjustment for a range of co-existing conditions, suggesting these were not simply a reflection of higher disease burden in the less educated. Moreover, the differentials in incidence, recovery and mortality compound and result in differences in mobility DFLE at age 65 years are 2.8 years for women and 2.4 years for men, exceeding differences in life expectancy. Our findings for mobility DFLE are similar in magnitude to differences in DFLE based on limiting longstanding illness between the most and least deprived areas of England (3.0 years for women and 2.7 for men).12 Our findings add to the evidence from other studies that differences in disability prevalence across educational groups persist at older ages. Although we found that relative differentials narrowed as rates increased with age, especially in women, absolute differences in disability prevalence between groups by years of education were maintained. Low education was associated with higher mortality in those previously disability-free but not in those who had already reported disability. Our findings of continued differences in disability incidence at older ages are consistent with others,13,17–20 but we have shown that this is not simply a reflection of the greater prevalence of disease in the lower educated. Interestingly, Ebrahim et al.34 found that socio-economic differences, measured by social class and income, in the incidence of severe but not mild disability was independent of diagnosed disease. More recent research has shown little impact of socio-demographic factors, including education, on declines in mobility though the time interval for onset was short (1 year) and the study size small.35
MRC-CFAS is one of the largest surveys to date to report recovery rates for the elderly population, but confidence intervals are wide, due to relatively small numbers in the disabled groups. Whilst lower rates of recovery for those with less education have been reported in a Chinese population,20 there was no such association in the EPESE study,17 the Taiwan Survey of the Elderly18 or the Longitudinal Aging Study Amsterdam.19 Some transitions to and from disability will be missed here, because of the 2 and 10-year intervals between follow-ups. In particular, recovery from ADL disability is often short-lived,36 but CFAS provides some evidence that more years of education are associated with higher rates of sustained recovery in men and women.
The strengths of MRC-CFAS are that it is a total population sample, including those in institutions, is nationally representative and has 10 years of follow-up. As with any longitudinal study, our findings are limited by losses to follow-up though by 10 years dropouts increasingly contribute to mortality events. In models for short and longer term dropout cognitive status accounts for most, but not all of selective dropout by education and social class.9,37 Adjustment for cognitive status did not materially change our findings, but the disabled and educationally disadvantaged are likely to be under-represented making our differences conservative.
Socio-economic differences in DFLE, measured by education, income and occupation have been demonstrated in the US and Europe.38 Though many of the studies reviewed by Crimmins and Cambois were based on cross-sectional data, there was a consensus that differences in DFLE or healthy life expectancy between education groups are greater than in total life expectancy. As in our study, differences have been found to persist into the oldest ages.39 Though we could not formally assess whether differences in DFLE by level of education were explained by comorbidity, the analyses of incidence, recovery and mortality suggest that educational differences in DFLE would still remain after adjustment for comorbidity.
In the US, differences in healthy life expectancy by education appear to have been widening over time.40 Levels of education have changed substantially over the last two decades and compared with other demographic and socio-economic factors, education has been the most important contributor to the improvements in functioning in the US older population41 although gains appear to have been confined to those with the highest level of education.3 Although similar improvements in disability and functioning are yet to be seen in the UK, our findings suggest that differences in incidence rates between educational groups are the driving force for DFLE differences, as in the US.17
Blane42 proposes five possible causal processes to explain the relationship between education and health: (i) the long-term effect of childhood circumstances on adult health; (ii) education is mediated through its influence on later occupation and income which themselves affect adult health; (iii) education impacts on the ability to take in and act upon health education messages; (iv) a further background variable affects both the capacity to complete education and maintain health and (v) ill-health during childhood limits education and predisposes to later ill-health. The absence of childhood health measures limits our ability to confirm or dispute some of these causal mechanisms but previous findings of estimates of differences in DFLE by social class for men but not women in MRC CFAS10 disputes the second explanation. Similarly the sustained differentials in incidence of disability between education groups with a baseline of no disability for all subjects and even after adjustment for comorbidity would give less credence to the first and last explanations, although the adjustment may well be incomplete. This leaves the possibility that education manifests itself through, in our case, the ability to adapt to increasing disability either through modifying tasks or employing technical aids. That the differences in mobility DFLE were greater than the more severe ADL DFLE supports this since the ADL measure was based on requiring help from another person. Better self-report measures or objective performance measures would confirm this.
We have shown that socio-economic differences in DFLE persist into old age and appear to be driven by differential rates of disability incidence that are not wholly explained by differential disease burden. If transition rates remain the same over time within educational groups then the UK may see an overall compression of disability as educational differentials in DFLE exceeded those in total years lived. Rising levels of obesity alongside reduced physical activity may counteract this by raising incidence and reducing recovery rates but this can only be confirmed by longitudinal data on more recent cohorts.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
MRC CFAS is supported by major awards from the Medical Research Council and the Department of Health.
Conflict of interest: None declared.
KEY MESSAGES
|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
1 Breeze E, Sloggett A, Fletcher A. Socioeconomic and demographic predictors of mortality and institutional residence among middle aged and older people: results from the Longitudinal Study. J Epidemiol Commun H (1999) 53:765–74.[Abstract]
2 Minkler M, Fuller-Thomson E, Guralnik JM. Gradient of disability across the socioeconomic spectrum in the United States. New Engl J Med (2006) 355:695–703.
3 Schoeni RF, Freedman VA, Wallace RB. Persistent, consistent, widespread, and robust? Another look at recent trends in old-age disability. J Gerontol B-Psychol (2001) 56:S206–18.
4 Picavet HSJ, Hoeymans N. Physical disability in The Netherlands: prevalence, risk groups and time trends. Public Health (2002) 116:231–37.[ISI][Medline]
5 Avlund K, Damsgaard MT, Osler M. Social position and functional decline among non-disabled old men and women. European Journal of Public Health (2004) 14:212–16.
6 Victor CR, Vetter NJ. Poverty, Disability and Use of Services by the Elderly—Analysis of the 1980 General Household Survey. Soc Sci Med (1986) 22:1087–91.[CrossRef][ISI][Medline]
7 Victor CR. Inequalities in Health in Later Life. Age Ageing (1989) 18:387–91.
8 Matthews RJ, Smith LK, Hancock RM, Jagger C, Spiers NA. Socioeconomic factors associated with the onset of disability in older age: a longitudinal study of people aged 75 years and over. Soc Sci Med (2005) 61:1567–75.[CrossRef][ISI][Medline]
9 Matthews RJ, Jagger C, Hancock RM. Does socio-economic advantage lead to a longer, healthier old age? Soc Sci Med (2006) 62:2489–99.[CrossRef][ISI][Medline]
10 Melzer D, McWilliams B, Brayne C, Johnson T, Bond J. Socioeconomic status and the expectation of disability in old age: estimates for England. J Epidemiol Commun H (2000) 54:286–92.
11 Bone M, Bebbington R, Jagger C, Morgan K, Nicolaas G. Health expectancy and its uses. HMSO (1995).
12 Bajekal M. Healthy life expectancy by area deprivation: magnitude and trends in England, 1994–1999. Health Statistics Quarterly (2005) 25:18–27.[Medline]
13 Grundy E, Glaser K. Socio-demographic differences in the onset and progression of disability in early old age: a longitudinal study. Age Ageing (2000) 29:149–57.
14 Anderson RT, James MK, Miller ME, Worley AS, Longino CE. The timing of change: Patterns in transitions in functional status among elderly persons. J Gerontol B-Psychol (1998) 53:S17–27.
15 Maddox GL, Clark DO, Steinhauser K. Dynamics of Functional Impairment in Late Adulthood. Soc Sci Med (1994) 38:925–36.[CrossRef][ISI][Medline]
16 Fried LP, Guralnik JM. Disability in older adults: evidence regarding significance, etiology, and risk. J Am Geriatr Soc (1997) 45:92–100.[ISI][Medline]
17 Melzer D, Izmirlian G, Leveille SG, Guralnik JM. Educational differences in the prevalence of mobility disability in old age: The dynamics of incidence, mortality, and recovery. J Gerontol B-Psychol (2001) 56:S294–301.
18 Zimmer Z, Liu X, Hermalin A, Chuang YL. Educational attainment and transitions in functional status among older Taiwanese. Demography (1998) 35:361–75.[ISI][Medline]
19 Huisman M, Kunst A, Deeg D, Grigoletto F, Nusselder W, Mackenbach J. Educational inequalities in the prevalence and incidence of disability in Italy and the Netherlands were observed. J Clini Epidemiol (2005) 58:1058–65.
20 Liang J, Liu X, Gu SZ. Transitions in functional status among older people in Wuhan, China: socioeconomic differentials. J Clini Epidemiol (2001) 54:1126–38.[CrossRef]
21 Grundy E, Holt G. Adult life experiences and health in early old age in Great Britain. Soc Sci Med (2000) 51:1061–74.[CrossRef][ISI][Medline]
22 Ebrahim S, Wannamethee SG, Whincup P, Walker M, Shaper AG. Locomotor disability in a cohort of British men: the impact of lifestyle and disease. Int J Epidemiol (2000) 29:478–86.
23 Adamson J, Hunt K, Ebrahim S. Socioeconomic position, occupational exposures, and gender: the relation with locomotor disability in early old age. J Epidemiol Commun Health (2003) 57:453–5.
24 Breeze E, Fletcher AE, Leon DA, Marmot MG, Clarke RJ, Shipley MJ. Do socioeconomic disadvantages persist into old age? Self-reported morbidity in a 29-year follow-up of the Whitehall Study. Am J Public Health (2001) 91:277–83.[Abstract]
25 Lynch J, G K. Socio-economic position. In: Social Epidemiology—Berkman L, IK, eds. (1999) New York: Oxford University Press. 13–35.
26 Smith JP. Wealth inequality among older Americans. J Gerontol B-Psychol (1997) 52:74–81.
27 Smith JP, Kington R. Demographic and economic correlates of health in old age. Demography (1997) 34:159–70.[ISI][Medline]
28 Robert S, House JS. SES differentials in health by age and alternative indicators of SES. J Aging Health (1996) 8:359–88.
29 Grundy E, Holt G. The socioeconomic status of older adults: how should we measure it in studies of health inequalities? J Epidemiol Commun Health (2001) 55:895–904.
30 House JS, Lepkowski JM, Kinney AM, Mero RP, Kessler RC, Herzog AR. The Social-Stratification of Aging and Health. J Health Soc Behav (1994) 35:213–34.[CrossRef][ISI][Medline]
31 The Medical Research Council Cognitive Function and Ageing Study. Cognitive function and dementia in six areas of England and Wales: the distribution of MMSE and prevalence of GMS organicity level in the MRC CFA Study. Psychol Med (1998) 28:319–35.[CrossRef][ISI][Medline]
32 Folstein MF, Folstein SE, McHugh PR. Mini-Mental State — Practical Method for Grading Cognitive State of Patients for Clinician. J Psychiat Res (1975) 12:189–98.[CrossRef][ISI][Medline]
33 Lievre A, Brouard N. The estimation of health expectancies from cross-longitudinal surveys. Math Pop Studies (2003) 10:211–48.
34 Ebrahim S, Papacosta O, Wannamethee G, Adamson J. Social inequalities and disability in older men: prospective findings from the British regional heart study. Soc Sci Med (2004) 59:2109–20.[CrossRef][ISI][Medline]
35 Ayis S, Gooberman-Hill R, Bowling A, Ebrahim S. Predicting catastrophic decline in mobility among older people. Age Ageing (2006) 35:382–87.
36 Hardy SE, Gill TM. Factors associated with recovery of independence among newly disabled older persons. Arch Intern Med (2005) 165:106–12.
37 Matthews FE, Chatfield M, Freeman C, McCracken C, Brayne C. Attrition and bias in the MRC cognitive function and ageing study: an epidemiological investigation. BMC Public Health (2004) 4:12.[CrossRef][Medline]
38 Crimmins E, Cambois E. Social inequalities in health expectancy. In: Determining health expectancies—Robine J-M, Jagger C, Mathers C, Crimmins E, Suzman R, eds. (2003) Chichester: John Wiley.
39 Crimmins EM, Hayward MD, Saito Y. Differentials in active life expectancy in the older population of the United States. J Gerontol B-Psychol (1996) 51:S111–20.
40 Crimmins EM, Saito Y. Trends in healthy life expectancy in the United States, 1970-1990: gender, racial, and educational differences. Soc Sci Med (2001) 52:1629–41.[CrossRef][ISI][Medline]
41 Freedman VA, Martin LG. The role of education in explaining and forecasting trends in functional limitations among older Americans. Demography (1999) 36:461–73.[ISI][Medline]
42 Blane D. Commentary: explanations of the difference in mortality risk between different educational groups. Inte J of Epidemiol (2003) 32:355–56.[CrossRef]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||