Date Published: May 24, 2018
Publisher: Oxford University Press
Author(s): Andrew Kingston, Louise Robinson, Heather Booth, Martin Knapp, Carol Jagger.
models projecting future disease burden have focussed on one or two diseases. Little is known on how risk factors of younger cohorts will play out in the future burden of multi-morbidity (two or more concurrent long-term conditions).
a dynamic microsimulation model, the Population Ageing and Care Simulation (PACSim) model, simulates the characteristics (sociodemographic factors, health behaviours, chronic diseases and geriatric conditions) of individuals over the period 2014–2040.
about 303,589 individuals aged 35 years and over (a 1% random sample of the 2014 England population) created from Understanding Society, the English Longitudinal Study of Ageing, and the Cognitive Function and Ageing Study II.
the prevalence of, numbers with, and years lived with, chronic diseases, geriatric conditions and multi-morbidity.
between 2015 and 2035, multi-morbidity prevalence is estimated to increase, the proportion with 4+ diseases almost doubling (2015:9.8%; 2035:17.0%) and two-thirds of those with 4+ diseases will have mental ill-health (dementia, depression, cognitive impairment no dementia). Multi-morbidity prevalence in incoming cohorts aged 65–74 years will rise (2015:45.7%; 2035:52.8%). Life expectancy gains (men 3.6 years, women: 2.9 years) will be spent mostly with 4+ diseases (men: 2.4 years, 65.9%; women: 2.5 years, 85.2%), resulting from increased prevalence of rather than longer survival with multi-morbidity.
our findings indicate that over the next 20 years there will be an expansion of morbidity, particularly complex multi-morbidity (4+ diseases). We advocate for a new focus on prevention of, and appropriate and efficient service provision for those with, complex multi-morbidity.
Healthcare delivery was built, and generally remains centred, on the treatment of single diseases. Over the last decade, the growing number of older people (aged 65 years and over) has become a considerable challenge to health and social care service provision and funding, as over 50% have at least two chronic conditions (multi-morbidity) [1, 2]. Moreover numbers of the very old, aged 85 years and over, are set to double over the next 20 years , with multi-morbidity the norm in this age group . Multi-morbidity increases the likelihood of hospital admission, length of stay and readmission, raises healthcare costs, reduces quality of life, and increases dependency, polypharmacy and mortality [2, 5]. In addition to multi-morbidity, many of the very old have sensory impairment and incontinence , making a single disease-focused model of healthcare unsuitable .
Between 2015 and 2035 increases of more than 50% are projected in the number of older people affected by most individual diseases and impairments, the largest increases being for numbers having cancer (179.4%) and diabetes (118.1%) (Table 1); exceptions are CHD (22.1%), depression (−15.1%) and CIND (25.6%). Arthritis and cancer will see the greatest rise in prevalence of 14.0 and 15.1 percentage points respectively. In the 85+ age group, all diseases apart from dementia and depression more than double in absolute numbers between 2015 and 2035 (see Appendix Table 1 in the Supplementary data, available in Age and Ageing online).
Table 1.Prevalence of (and numbers with) individual diseases and impairments in 2015, 2025 and 2035 and percentage change in numbers between 2015 and 2025, and 2015 and 2035, population aged 65 years and over2015% (n)2025% (n)2035% (n)% Change (2015–2025)% Change (2015–2035)Diseases Arthritis48.6 (4,721,300)60.3 (7,059,300)62.6 (9,046,300)49.591.6 Cancer12.6 (1,224,900)19.6 (2,297,700)23.7 (3,422,000)87.6179.4 CHD18.3 (1,778,700)16.6 (1,937,800)15.0 (2,172,500)8.922.1 Dementia6.8 (659,700)7.8 (918,800)8.5 (1,227,500)39.386.1 Depression2.3 (225,700)1.3 (155,500)1.3 (191,600)−31.1−15.1 Diabetes14.7 (1,428,400)19.8 (2,317,900)21.6 (3,115,400)62.3118.1 Hypertension49.0 (4,768,200)54.9 (6,423,400)55.9 (8,080,400)34.769.5 Respiratory18.0 (1,747,400)21.5 (2,520,000)24.4 (3,520,300)44.2101.5 Stroke7.5 (726,100)8.7 (1,021,700)9.3 (1,337,500)40.784.2Impairments CINDa2.7 (264,100)2.3 (273,500)2.3 (331,600)3.625.6 Hearing12.4 (1,201,800)11.6 (1,354,400)12.5 (1,812,400)12.750.8 Vision6.2 (600,000)5.2 (613,400)5.4 (777,700)2.229.6Total Population9,723,9001,170,58001,444,9900aCognitive impairment no dementia.
PACSim provides, for the first time, projections of a range of fatal and non-fatal chronic diseases and geriatric conditions conditional on the sociodemographic characteristics, health behaviours and existing morbidities of a real population aged 35 years and over as they age. We estimate that, over the next 20 years, there will be greater numbers of older people aged 65 years and over, both with individual diseases and with multi-morbidity, particularly with four or more diseases for which numbers will double, as will numbers with cancer, respiratory disease and diabetes. In addition, around a third of those with four or more morbidities will have mental ill-health (depression, dementia or cognitive impairment with no dementia). There will be an expansion of morbidity, with the gain in life expectancy at age 65 between 2015 and 2035 (3.6 years for men, 2.9 years for women) being less than the gain in years spent with multi-morbidity (5.5 years for men, 5.0 years for women), and two-thirds or more of the gain in life expectancy will be spent with four or more diseases. These findings suggest a new focus on those with four or more long-term conditions which we will term ‘complex multi-morbidity’.
To our knowledge PACSim includes more major chronic diseases than any other microsimulation model, thereby enabling a realistic prediction of the future burden of multi-morbidity. Limitations are focussed around the morbidities included, the time period for calculation of transitions, the assumptions underlying the transitions and the lack of confidence intervals. Firstly, most of the morbidities are self-reported, though all three surveys ascertained doctor-diagnosed disease. The exception to self-reported morbidities are cognitive impairment and dementia, both of which were only available in CFAS and therefore were imputed for participants in the two other surveys. Although PACSim covers the main sociodemographic and lifestyle risk factors for disease, inclusion of other factors was limited by incomparability of items across the three surveys. Secondly, the transition rates for all characteristics were based upon observations from two consecutive waves of each survey, a time period of around 2 years. A longer time period might provide more precise estimates of coefficients in the transition models but we were constrained by CFAS having only two waves and yet being crucial for providing information on dementia and cognitive impairment. Thirdly, the model assumes that transitions between states of all characteristics are independent of time, though incorporating time-varying transitions is a future possibility. Finally, providing confidence intervals around all outcomes that account for the error in the transition rates is non-trivial in a dynamic microsimulation model with as many outcomes and characteristics as PACSim. However running the simulation multiple times has provided evidence that the range of the prevalence of multi-morbidity is small (less than one percentage point) albeit when the error in the transition rates is ignored. Strengths include that: PACSim is based on three large, nationally representative surveys; baseline disease prevalence is broadly comparable with the Health Survey for England 2014 ; basing the simulation on monthly transitions which provides more realistic evolutions of characteristics that are co-dependent; and the ability to add scenarios which will allow us to see the effect of future interventions.
Comparisons of the prevalence of multi-morbidity between studies is difficult due to the number and definition of diseases and the age groups included [1, 5]. Nevertheless, our estimate of the prevalence of multi-morbidity of 54% in those aged 65 years and over in 2015 in England is in keeping with others . Risk factors and the prevalence of individual diseases differ between countries, but multi-morbidity is an increasing challenge for all countries, not least low- and middle-income countries (LMIC) in which populations are ageing much more rapidly than in high-income countries . Indeed our finding of a greater likelihood of poor mental health with multi-morbidity, particularly four or more diseases, confirms findings for LMICs .
Geriatricians have long recognised the challenges of multi-morbidity in balancing treatment and intervention options with quality of life and function, particularly in very old and frail people. However, healthcare policy in England has transferred chronic disease management from specialist services to primary care, which through initiatives such as the Quality Outcomes Framework  has reinforced a long-standing single-disease paradigm, an approach which does not adequately address the needs of older people. For example, the application of single-disease guidelines from the National Institute for Health and Care Excellence (NICE) for an older person with five conditions (Type 2 diabetes, previous myocardial infarction, osteoarthritis, COPD and depression) may result in a minimum of 11 medications (with up to 10 other drugs routinely recommended), 8–10 routine primary care appointments and 4–6 GP appointments, as well as multiple self-care/lifestyle modifications . Moreover, these findings are not restricted to England, nor to older people: similar levels of polypharmacy and healthcare visits are reported in the US for those in mid-life (aged 45–64 years) with three chronic conditions . The recent NICE guidelines for management of multi-morbidity are, therefore, welcome, especially as they aim to involve patients’ goals and preferences in clinical decision-making , though implementation will require training, longer consultations and more funding as primary care, not only in England, is already over-stretched .