I have read the journal’s policy and the authors of this manuscript have the following competing interests: CB is an Academic Editor on PLOS Medicine’s editorial board. FM is a paid statistical consultant on PLOS Medicine’s statistical board.
‡ These authors are joint senior authors on this work.
¶ Members are listed in the Acknowledgments.
Previous research has examined the improvements in healthy years if different health conditions are eliminated, but often with cross-sectional data, or for a limited number of conditions. We used longitudinal data to estimate disability-free life expectancy (DFLE) trends for older people with a broad number of health conditions, identify the conditions that would result in the greatest improvement in DFLE, and describe the contribution of the underlying transitions.
The Cognitive Function and Ageing Studies (CFAS I and II) are both large population-based studies of those aged 65 years or over in England with identical sampling strategies (CFAS I response 81.7%,
In this study, we observed improvements to DFLE between 1991 and 2011 despite the presence of most health conditions we considered. Attention needs to be paid to support and care for people with cognitive impairment who had different outcomes to those with physical health conditions.
Holly Bennett and colleagues investigate trends in life expectancy and disability-free life expectancy between 1991 and 2011 for adults age 65 years or older with long term health conditions using data from two population-based studies in England.
There have been many advances in healthcare since the 1990s, which means many people with health conditions are living longer lives.
We wanted to find out whether the extension to life for those with health conditions involves an increase in healthy or unhealthy years.
We used information from 2 large studies of people aged 65 years and over, the first began in 1991 and the second began in 2011.
We estimated how many years people with different health conditions would live with or without disability in 2011 compared to 1991.
We found improvements for people with most health conditions, with the exception of cognitive impairment where the increase in years lived with disability were equal (men) or greater than (women) the improvement in years lived without disability.
In the UK, the percentage of people with cognitive impairment has decreased over time.
However, if people with cognitive impairment are now living longer with disability than before, this may not mean that the demand for services and care will decrease.
Life expectancy (LE) and disability-free life expectancy (DFLE) have been increasing over time, but this has occurred unequally across the population [
As medical advances and public health practices have contributed to substantial reductions in mortality from leading causes of death such as cardiovascular disease (CVD) [
An overview of existing literature indicates that most studies on single LTCs and DFLE examine the impact primarily of diabetes, ischaemic heart disease, respiratory diseases, and arthritis, and only at one time point. Temporal comparisons are, as far as we are aware, limited to 3 studies, reporting improvements in remaining healthy years for people with diabetes [
Global estimates of trends in disability-adjusted life years (DALYs) suggest that increasing diabetes prevalence has contributed to rising DALY rates and that ischaemic heart disease, stroke, and chronic obstructive pulmonary disease remain the leading causes of DALYs in those aged 50 and older [
We use longitudinal data from the Cognitive Function and Ageing Studies (CFAS I and II), 2 large population-based studies of people aged 65 years or older in England, to identify trends in LE and DFLE for a wide range of LTCs separately for men and women. Our aims included determining whether people with specific LTCs have experienced longer LE and more years free of disability between 1991 and 2011, and, secondly, which LTCs if eliminated will lead to compression or expansion of disability. In addition, the longitudinal data for each study enables us to explore how the underlying transitions to and from disability, and to death, contribute to the observed trends by LTC.
CFAS I and CFAS II are 2 large population-based studies of people aged 65 years or older living in England [
Demographics included age group (65 to 69 years, 70 to 74 years, 75 to 79 years, 80 to 84 years, 85 to 89 years, and ≥90 years), sex, years in education (<10 years, 10 to 11 years, and ≥12 years), social class based on occupation (skilled, semi-skilled, and unskilled), and place of residence (community, semi-dependent housing, and care settings). Area-level deprivation was measured through the Townsend deprivation index [
Disability was categorised into any disability or disability-free using the modified Townsend activities of daily living (ADL) scale [
In total, 9 LTCs were considered, including arthritis, coronary heart disease (CHD—angina or heart attack), cognitive impairment, diabetes, hearing difficulties, peripheral vascular disease (PVD), respiratory difficulties (asthma except childhood only or chronic bronchitis), stroke, and vision impairment, based on previous analysis [
Demographics were inverse probability weighted for nonresponse. The nonresponse weights included age, sex, deprivation, and whether the participant lived in care settings. Health condition prevalence from CFAS I and CFAS II was weighted for nonresponse and age and sex standardised to the CFAS I population (1991). We used logistic regression to compare the prevalence of each LTC at baseline between the 2 studies and the extent to which age group, sex, and time contributed to differences in prevalence. All logistic regression models were weighted for nonresponse.
LEs were estimated from longitudinal multistate models analysing transitions from disability-free to disability, recovery from disability to disability-free, and from either disability state to death in Interpolated Markov Chain (IMaCh) software version 0.99r19 [
This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (
This study was conducted as secondary data analysis of the Cognitive Function and Ageing Studies. The current ethics for MRC CFAS (including CFAS I centres) is from Eastern MREC, reference number 05/MRE05/37 and for the mortality data Wales REC 7, reference number 14/WA/1154. The current REC reference number for CFAS II is 07/MRE05/48 from Cambridge REC 4. For further information on past ethical approvals, please visit the CFAS website (
In CFAS I, there were 7,635 participants at baseline, 60.8% were women, and average age was 75.6 years. Before the 2-year follow-up interview, 10.7% (
CFAS I | CFAS II | ||||
---|---|---|---|---|---|
n | % | n | % | ||
Age group (years) | 65–69 | 1,981 | 25.0 | 1,939 | 23.0 |
70–74 | 1,776 | 22.8 | 1,873 | 22.7 | |
75–79 | 1,725 | 22.5 | 1,624 | 20.5 | |
80–84 | 1,308 | 17.7 | 1,278 | 17.5 | |
85–89 | 615 | 8.5 | 737 | 10.5 | |
≥90 | 230 | 3.5 | 311 | 5.8 | |
Sex | Men | 3,045 | 39.2 | 3,534 | 43.9 |
Women | 4,590 | 60.8 | 4,228 | 56.1 | |
Education (years) | <10 | 5,529 | 74.1 | 2,047 | 29.4 |
10–11 | 1,238 | 16.6 | 3,923 | 50.2 | |
≥12 | 692 | 9.3 | 1,667 | 20.4 | |
Social class | Skilled | 1,921 | 26.2 | 1,958 | 25.4 |
Semi-skilled | 3,855 | 52.6 | 3,962 | 54.3 | |
Unskilled | 1,555 | 21.3 | 1,370 | 20.2 | |
Place of residence | Community | 6,599 | 86.0 | 7,083 | 89.5 |
Semi-dependent | 683 | 9.1 | 482 | 7.2 | |
Care settings | 346 | 4.8 | 197 | 3.3 | |
Deprivation tertiles | Least | 2,561 | 33.5 | 2,940 | 33.2 |
Mid | 2,525 | 33.2 | 2,659 | 33.3 | |
Most | 2,549 | 33.4 | 2,163 | 33.5 | |
Health condition count | 0–1 | 3,523 | 45.7 | 3,420 | 41.9 |
2+ | 4,102 | 54.3 | 4,311 | 58.1 |
CFAS, Cognitive Function and Ageing Study.
The prevalence of most LTCs in those aged 65 years and over increased between 1991 and 2011, with the prevalence of diabetes and PVD more than doubling (
Models adjusted for sex and age group.
Unadjusted | Adjusted for age and sex | |||||
---|---|---|---|---|---|---|
OR | 95% CI | OR | 95% CI | |||
Arthritis | 1.1 | (1.0, 1.2) | 0.002 | 1.1 | (1.1, 1.2) | <0.001 |
Cognitive impairment | 0.7 | (0.6, 0.7) | <0.001 | 0.6 | (0.5, 0.6) | <0.001 |
CHD | 1.4 | (1.2, 1.5) | <0.001 | 1.3 | (1.2, 1.4) | <0.001 |
Diabetes | 2.6 | (2.3, 2.9) | <0.001 | 2.5 | (2.3, 2.8) | <0.001 |
Hearing difficulties | 1.3 | (1.2, 1.4) | <0.001 | 1.2 | (1.1, 1.3) | <0.001 |
PVD | 2.3 | (2.1, 2.6) | <0.001 | 2.3 | (2.0, 2.6) | <0.001 |
Respiratory difficulties | 1.1 | (1.0, 1.2) | 0.04 | 1.1 | (1.0, 1.2) | 0.03 |
Stroke | 1.3 | (1.2, 1.4) | <0.001 | 1.2 | (1.1, 1.4) | <0.001 |
Vision impairment | 1.2 | (1.1, 1.3) | <0.001 | 1.2 | (1.1, 1.3) | 0.002 |
CFAS, Cognitive Function and Ageing Study; CHD, coronary heart disease; CI, confidence interval; OR, odds ratio; PVD, peripheral vascular disease.
We have previously reported that in the period between CFAS I and CFAS II, men’s LE at age 65 years increased by 4.6 years (95% confidence interval (CI): 3.7 to 5.5 years,
DFLE and DLE at age 65 years for men (A) and women (B) with a health condition in the CFAS I and CFAS II. Models stratified by sex and study and adjusted for age and the health condition. Results also shown in Tables C and E in
Between CFAS I and II, women experienced an increase in LE at age 65 years of 2.1 years (95% CI: 1.1 to 3.0 years,
The large improvements in DFLE seen in men with respiratory difficulties (4.5 years) appear to be a result of a decrease in the probability of death from a disability-free state (relative risk ratio (RRR): 0.2, 95% CI: 0.1 to 0.7,
Estimates from models stratified by sex and health condition with age and study covariates. Results also shown in
Estimates from models stratified by sex and health condition with age and study covariates.
Men | Women |
||||||
---|---|---|---|---|---|---|---|
RRR | 95% CI | RRR | 95% CI | ||||
No disability -> Disability | Arthritis | 0.9 | (0.7, 1.1) | 0.36 | 0.7 | (0.6, 0.8) | <0.001 |
Cognitive impairment | 1.0 | (0.7, 1.5) | 0.91 | 0.9 | (0.7, 1.1) | 0.36 | |
CHD | 0.7 | (0.5, 0.9) | 0.02 | 0.6 | (0.4, 0.8) | 0.004 | |
Diabetes | 0.7 | (0.4, 1.4) | 0.26 | 0.7 | (0.4, 1.1) | 0.17 | |
Hearing difficulties | 0.8 | (0.5, 1.2) | 0.32 | 0.7 | (0.5, 0.9) | 0.02 | |
PVD | 0.7 | (0.4, 1.3) | 0.24 | 0.6 | (0.3, 1.2) | 0.15 | |
Respiratory difficulties | 0.7 | (0.5, 1.1) | 0.08 | 0.7 | (0.5, 0.9) | 0.02 | |
Stroke | 0.5 | (0.3, 1.0) | 0.02 | 0.6 | (0.3, 1.2) | 0.15 | |
Vision impairment | 0.7 | (0.4, 1.1) | 0.17 | 0.6 | (0.4, 0.8) | 0.004 | |
Disability -> No Disability | Arthritis | 1.1 | (0.7, 1.8) | 0.69 | 1.0 | (0.7, 1.3) | 0.82 |
Cognitive impairment | 1.4 | (0.7, 2.9) | 0.35 | 0.9 | (0.6, 1.5) | 0.65 | |
CHD | 0.7 | (0.4, 1.2) | 0.20 | 1.0 | (0.6, 1.7) | 0.99 | |
Diabetes | 1.5 | (0.5, 4.4) | 0.46 | 1.9 | (0.6, 6.4) | 0.29 | |
Hearing difficulties | 1.1 | (0.6, 2.0) | 0.76 | 1.0 | (0.6, 1.8) | 0.94 | |
PVD | 1.6 | (0.4, 6.3) | 0.50 | 0.5 | (0.2, 1.5) | 0.18 | |
Respiratory difficulties | 1.7 | (0.9, 3.2) | 0.10 | 0.9 | (0.5, 1.4) | 0.69 | |
Stroke | 1.2 | (0.5, 3.1) | 0.70 | 2.8 | (1.1, 7.2) | 0.03 | |
Vision impairment | 1.0 | (0.4, 2.6) | 0.93 | 0.8 | (0.4, 1.5) | 0.51 | |
No disability -> Death | Arthritis | 0.7 | (0.4, 1.1) | 0.17 | 0.5 | (0.2, 1.2) | 0.13 |
Cognitive impairment | 0.5 | (0.2, 1.2) | 0.13 | 0.6 | (0.2, 1.8) | 0.36 | |
CHD | 0.5 | (0.3, 0.8) | 0.006 | 0.7 | (0.2, 2.5) | 0.58 | |
Diabetes | 0.5 | (0.3, 0.8) | 0.006 | 0.5 | (0.1, 2.8) | 0.41 | |
Hearing difficulties | 0.4 | (0.2, 1.0) | 0.03 | 0.7 | (0.2, 2.3) | 0.57 | |
PVD | 0.8 | (0.3, 2.0) | 0.64 | 0.2 | (0.0, 2.0) | 0.17 | |
Respiratory difficulties | 0.2 | (0.1, 0.7) | 0.001 | 0.9 | (0.2, 3.4) | 0.88 | |
Stroke | 0.4 | (0.1, 2.1) | 0.24 | 0.5 | (0.1, 2.2) | 0.38 | |
Vision impairment | 0.3 | (0.1, 1.0) | 0.04 | 0.4 | (0.1, 1.7) | 0.20 | |
Disability -> Death | Arthritis | 0.8 | (0.6, 0.9) | 0.03 | 1.0 | (0.9, 1.1) | 0.86 |
Cognitive impairment | 0.9 | (0.7, 1.0) | 0.25 | 0.9 | (0.8, 1.1) | 0.19 | |
CHD | 0.7 | (0.6, 0.9) | <0.001 | 0.9 | (0.7, 1.1) | 0.36 | |
Diabetes | 0.8 | (0.5, 1.3) | 0.36 | 0.9 | (0.6, 1.2) | 0.55 | |
Hearing difficulties | 0.9 | (0.7, 1.1) | 0.36 | 1.1 | (0.9, 1.3) | 0.31 | |
PVD | 0.7 | (0.4, 1.0) | 0.13 | 0.5 | (0.3, 0.8) | 0.006 | |
Respiratory difficulties | 0.9 | (0.7, 1.1) | 0.36 | 0.9 | (0.8, 1.1) | 0.19 | |
Stroke | 0.7 | (0.5, 0.9) | 0.02 | 0.8 | (0.6, 1.1) | 0.15 | |
Vision impairment | 0.9 | (0.7, 1.1) | 0.36 | 1.0 | (0.9, 1.3) | 0.64 |
1Models converged at 1-month steps apart from the women’s PVD and stroke models, which converged at 12-month steps.
CFAS, Cognitive Function and Ageing Studies; CHD, coronary heart disease; CI, confidence interval; LTC, long-term condition; PVD, peripheral vascular disease; RRR, relative risk ratio.
Men with arthritis (RRR: 0.8, 95% CI: 0.6 to 0.9,
Women with CHD experienced a decline in DLE (−0.8 years), possibly because of a decline in the likelihood of transitioning to disability (RRR: 0.6, 95% CI: 0.4 to 0.8,
The largest increase in DFLE occurred in women with stroke (3.5 years), and this may be partly explained by the observed substantial increase in the probability of recovery (transition from disability to no disability), though CIs are wide (RRR: 2.8, 95% CI: 1.1 to 7.2,
Comparing LE and DFLE for men and women with and without each LTC provides understanding of the impact on population health of eliminating each condition. More specifically, if the gains in DFLE from elimination of a condition (calculated by subtracting DFLE of those with the condition from the DFLE of those without the condition) exceeds the gains in LE, then elimination will lead to a compression of disability; if gains in LE exceed those in DFLE, then expansion of disability will result. We focus on the most recent period (CFAS II) to investigate the population impact of elimination of each LTC, and examine whether elimination of the LTC would result in the years with disability (DLE) significantly decreasing (compression) or increasing (expansion) if the LTC was eliminated.
From CFAS II, elimination of arthritis (−1.6 years, 95% CI: −2.4 to −0.8,
Difference in years with disability (DLE) between men (A) and women (B) with and without the health condition (DLE without health condition–DLE with health condition) in CFAS II with 95% CIs. Models stratified by sex and study and adjusted for age and health condition. Results also shown in Tables C and E in
To better understand why elimination of an LTC would contribute to compression of disability, we examined the RRRs for those with each condition (compared to those without the condition), separately for men and women and by study.
In CFAS II, cognitive impairment was the largest contributor to loss of years in men’s LE and DFLE (Table C in
For women in CFAS II, DLE was increased for those with arthritis or PVD in comparison to those without the LTC. Women with arthritis were more likely to become disabled (RRR: 1.6, 95% CI: 1.3 to 2.0,
Although LTCs have been reported as major drivers of disability, to the best of our knowledge, our study is the first to quantify the contribution of chronic conditions to trends in DFLE using longitudinal data. We used 2 large, population-representative studies to examine temporal trends in LE and DFLE for those with LTCs to discover whether LTCs are becoming more or less disabling or fatal. Cognitive impairment was the only LTC where the prevalence decreased between 1991 and 2011, but also the only LTC where the percentage of remaining years with disability increased for men and women. Other LTCs showed improvements for both men and women with the health condition, with the majority of LE gains being years free of disability. This was especially true for women with CHD where improvements to LE and DFLE occurred alongside decreases in DLE, providing evidence that compression of disability while increasing life span could be possible in the presence of health conditions. Positive trends in DFLE for men resulted from reductions in the probability of death with or without disability, whereas for women, the trends mainly resulted from reductions in the probability of incident disability.
Although it may not be achievable to fully eliminate a health condition, by comparing DLE of men and women with and without each LTC in 2011, we were able to theoretically explore whether elimination of the LTC would result in a reduction of years with disability (compression of disability). We found that arthritis in men and women, stroke and cognitive impairment in men, and PVD in women, if eliminated, could all result in a significant reduction in years with disability and therefore a compression in disability.
The increasing prevalence of LTCs, particularly stroke and diabetes, in the last decades has already been documented, at least in the UK and the US [
Of the LTCs we considered, the only one for which prevalence has decreased is cognitive impairment. Despite this, the negative association between cognitive impairment and DFLE appeared greater in CFAS II than in CFAS I. This could be due to the greater prevalence of other LTCs being present in those with cognitive impairment in CFAS II compared to CFAS I, although this amounted to only 5 or 6 percentage points on an already large proportion (over 80%) of those with cognitive impairment having MLTCs. Similar results have been reported for changes in comorbidity with dementia in CFAS I and CFAS II [
We found in both CFAS I and II that elimination of arthritis could increase DFLE. This is consistent with other studies reporting lower DFLE, higher DLE [
While the majority of the literature on the impact of health conditions on life and health expectancy focuses on DALYs, and since DALYs combine years with disability and life years, this does not allow for estimation of the different probabilities of incident disability or recovery from disability at different ages, one of the advantages of estimating DFLE from longitudinal data. For comparability, therefore, we focus on literature reporting trends in DFLE. Similar to our findings, other studies have found improvements in stroke [
CFAS I and II have identical sampling frames so are well placed to provide temporal comparisons, giving accurate estimates of changes over 2 decades without compromising the validity of results. Both CFAS I and CFAS II are large population-based studies, which meant that estimates of LE and DFLE could be stratified by sex and a broad range of health conditions could be considered even when prevalence was relatively low. Both studies included residents of care homes and assisted living facilities, important given the difference in prevalence of cognitive impairment in these places of residence. Item nonresponse from the participant interview could be substituted with information from an informant interview with a friend or family member. There were some limitations to this analysis. The presence of health conditions depends on self-report by the participant, which therefore relies on their memory and accuracy of reporting but also on definition and diagnostic practice for the condition. With regard to participant memory, missing information from the participant was substituted by information given by informants to limit the loss of data from this in both CFAS I and CFAS II, resulting in less than 4% missing data in every health condition for both studies. Changes in definition and diagnostic practice have occurred for both stroke and diabetes [
Our study observed improvements to DFLE in the presence of most of the health conditions we included. Improvements in DFLE for people with stroke could be from decreases in stroke severity, potentially from increased use of preventive medicines or earlier treatment [
Our study is the first, to our knowledge, to estimate temporal trends in LE and DFLE with health conditions from longitudinal data and separately for men and women. We found that the underlying transitions influencing trends in DFLE for those with health conditions differed between men and women. Improvements for women with health conditions may be related to reduced disability incidence and improvements for men from reductions in the probability of death. For women, reductions in incident disability were great enough that DFLE increased and DLE decreased in the presence of CHD. While these findings are positive, we also found a decline in the percentage of remaining years spent disability-free for men and women with cognitive impairment. Given that cognitive impairment was also the only LTC where prevalence decreased, this is a cause for concern and requires further investigation.
(DOCX)
(DOCX)
(DOCX)
(DOCX)
Supplementary results, additional tables reporting results from figures with confidence intervals and Table A: Weighted prevalence (%) of long-term conditions by age in the Cognitive Function and Ageing Studies (CFAS I and CFAS II). Weighted prevalence of health conditions from CFAS I and CFAS II age and sex standardised to CFAS I population (1991) with 95% confidence intervals (95% CI). Table B: Prevalence of at least one other long-term condition in people with each specific long-term condition, by sex and study—Cognitive Function and Ageing Studies (CFAS I and CFAS II). Table C: Life expectancy (LE), disability-free life expectancy (DFLE), and life expectancy with disability (DLE) with 95% confidence intervals (95% CI) at age 65 for men with and without long-term conditions in the first and second Cognitive Function and Ageing Studies (CFAS I and CFAS II). Table D: Percentage of remaining years at age 65 spent disability-free (DFLE %) or with disability (DLE %) for men with and without long-term conditions in the Cognitive Function and Ageing Studies (CFAS I and CFAS II). Table E: Life expectancy (LE), disability-free life expectancy (DFLE), and life expectancy with disability (DLE) with 95% confidence intervals (95% CI) at age 65 for women with and without long-term conditions in the Cognitive Function and Ageing Studies (CFAS I and CFAS II). Table F: Percentage of remaining years at age 65 spent disability-free (DFLE %) or with disability (DLE %) for women with and without long-term conditions in the Cognitive Function and Ageing Studies (CFAS I and CFAS II). Table G: Relative Risk Ratios (RRR) for transition with each long-term condition (relative to without condition) from unadjusted models for men in the Cognitive Function and Ageing Studies (CFAS I and CFAS II), with 95% confidence intervals (95% CI). Table H: Relative Risk Ratios (RRR) for transition with each long-term condition (relative to without condition) from unadjusted models for women in the Cognitive Function and Ageing Studies (CFAS I and CFAS II), with 95% confidence intervals (95% CI).
(DOCX)
We thank the participants, their families, the family doctors and their staff, and the primary care trusts for their cooperation and support. We thank the CFAS data manager, project managers, and fieldwork interviewers at Cambridge, Nottingham, and Newcastle for their valuable contribution. The CFAS Management Committee includes Professor Carol Brayne, Professor Fiona Matthews, Professor Louise Robinson, Professor Tom Dening, Professor Carol Jagger, Professor Ian McKeith, Professor Antony Arthur, Professor Steve Wharton, Professor Bob Woods, Professor Linda Clare, Professor Sarah Pendlebury, Professor Blossom Stephan, Professor Bronwyn Parry, Dr Simon Harrison, Raphael Wittenberg, Adelina Comas-Herrera, and Linda Barnes. This research was undertaken within the UK National Institute for Health Research (NIHR) collaboration for leadership in applied health research and care for Cambridgeshire and Peterborough and the Cambridge Biomedical Research Centre infrastructures, Nottingham city and Nottinghamshire county NHS primary care trusts. CFAS received support from the UK NIHR comprehensive clinical research networks in West Anglia and Trent, and the Dementias and Neurodegenerative Disease Research Network in Newcastle.
The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.
activities of daily living
Cognitive Function and Ageing Study
coronary heart disease
confidence interval
Coronavirus Disease 2019
cardiovascular disease
disability-adjusted life year
disability-free life expectancy
life expectancy with disability
Interpolated Markov Chain
life expectancy
long-term condition
multiple long-term condition
Mini Mental State Examination
Office for National Statistics
odds ratio
peripheral vascular disease
relative risk ratio
years lived with disability
years of life lost