The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand: An Analysis of Counterfactual Projection Scenarios Based on German Health Insurance Routine Data The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand An Analysis of Counterfactual Projection Scenarios Based on German Health Insurance Routine Data* Alexander Barth Abstract: Although demand for long-term care (LTC) in Germany is expected to increase over the coming decades, the LTC sector will struggle to provide suffi cient capacity. Evaluating the impact of different risk factors on future LTC demand is necessary in order to make informed policy decisions. With regard to LTC need, dementia and lower extremity injuries (LEI) are common risk factors. Both are used to demonstrate their maximum attainable effi cacy in mitigating the future increase in overall LTC need, both at home and in nursing homes. We use a multi-state projection model for which the estimation of the underlying transition and mortality rates is based on longitudinal health claims data from AOK, Germany’s largest public health insurance provider, between 2004 and 2010. We project six different scenarios of LTC for ages 75+ in Germany for the period from 2014 to 2044, including counterfactual scenarios that remove the effects of LEI, de- mentia, or both. Our multi-state projections distinguish between home-based and institutional LTC. Removing the effect of LTC risk factors mitigates the increase in total LTC de- mand and postpones demand until a later age. Removing dementia markedly shifts future care demand from institutional LTC to LTC at home and even increases de- mand for LTC at home at older ages beyond the baseline projection due to the dual function of dementia as a risk factor for both LTC demand and mortality. Removing LEI has less of an effect on overall and sectoral LTC demand. Removing both risk factors at the same time results in the greatest impact, which is even more marked than that of both individual scenarios combined, thus indicating a synergistic rela- tionship between dementia and LEI on LTC risk. The type of LTC demand (home-based or institutional) shows considerable plas- ticity when specifi c risk factors are removed. We demonstrate the degree to which Comparative Population Studies Vol. 44 (2019): 235-268 (Date of release: 03.12.2019) Federal Institute for Population Research 2019 URL: www.comparativepopulationstudies.de DOI: 10.12765/CPoS-2019-19en URN: urn:nbn:de:bib-cpos-2019-19en5 * This article contains supplementary material in the form of a Data Appendix: DOI 10.12765/ CPoS-2019-20en; URL: http://www.comparativepopulationstudies.de/index.php/CPoS/article/ view/304/290. http://www.comparativepopulationstudies.de/index.php/CPoS/article/view/304/290 • Alexander Barth236 LTC demand can be affected in favour of LTC at home, using dementia and LEI as examples of potentially modifi able risk factors, and thus show how the effi cacy of potential intervention targets for policy-makers can be assessed. This study provides evidence on the degree of plasticity of future long-term care demand at home and in institutions that would hypothetically be attainable when completely removing specifi c cognitive or physical risk factors of care need (de- mentia or lower EI). It is based on large-scale health claims data, which contain longitudinal individual level data on morbidity and long-term care status. A close link exists between the cognitive risk factor of dementia and the type of LTC, as its absence shifts care demand to home-based care at older ages. The study also dem- onstrates the usefulness of counterfactual projections based on health claims data in assessing the hypothetical maximum effi cacy of different intervention strategies. Keywords: Long-term care · Counterfactual projections · Dementia · Extremity injuries 1 Introduction Increasing life expectancy and declining fertility are driving population ageing in developed countries. Across OECD states, the share of people aged 80 years and older is expected to triple from 2005 to 2050, causing a marked increase in LTC de- mand (Fujisawai/Colombo 2009; Comas-Herrera et al. 2006). Ageing nations need to manage this increase, e.g. by providing opportunities for informal care, giving professional assistance to informal caregivers or by increasing capacity in nursing homes. The growth rate in LTC expenditure was the highest among all healthcare sectors (OECD 2017) and total LTC expenditure is expected to increase from 1 per- cent to up to 4 percent of GDP between 2005 and 2050 (Fujisawai/Colombo 2009). Between 2030 and 2050 in particular, LTC demand is expected to increase markedly, as large baby boomer cohorts born between the mid-1950s and 1960s will reach ages at which LTC becomes prevalent (Doblhammer 2012; Rothgang et al. 2012b). As both personnel and funding in the LTC sector are in short supply, further strate- gies to mitigate increases in LTC demand are of vital interest. Since LTC dependency often arises due to illnesses (Koller et al. 2014; Barth et al. 2016), lowering either the prevalence of a disease or its impact on LTC demand could mitigate the increase in demand. However, most studies focus on total future LTC demand or labour supply (Doblhammer 2012; Fujisawai/Colombo 2009). Evi- dence on the effi cacy of specifi c conditions is sparse. This article therefore uses counterfactual projections that remove either dementia, lower extremity injuries (LEI), or both to assess their effi cacy in mitigating the increase in total and sectoral LTC demand. This study uses a large sample of German statutory health insurance routine data, which provide morbidity, mortality and LTC status. It allows all transition risks that are required for the counterfactual projections to be modelled based on the The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 237 same source. In terms of LTC, dementia is a greater risk factor than LEI (Barth et al. 2016), which leads us to hypothesise that its overall mitigating effect will be greater. Since dementia often requires a higher degree of care, its removal should result in a marked reduction in institutional LTC demand. Because health insurance routine data cover all diseases and are available in many countries, this approach could also be applied to different diseases or countries. 2 Background 2.1 Individual level LTC risk factors In order to examine trends in population health, health expectancy measures such as healthy, active or disability-free life expectancy are common, which divide the number of remaining life years into unimpaired and impaired years. While results largely depend on the exact measure in question, fi ndings generally show that a compression of morbidity occurs. Against the background of increasing life expec- tancy, this means that the life years gained are mostly spent in good health. These gains pertain in particular to serious conditions, while for minor impairments, a dy- namic equilibrium (additional life years distributed evenly) or even an expansion of morbidity can be found, which is characterised by multimorbidity and chronic impairments (Unger/Schulze 2013; Christensen et al. 2009). LTC dependency often arises due to chronic functional impairments when basic tasks cannot be performed alone. We focus on two common LTC causes: one related to mental health – demen- tia; and one related to physical mobility – lower extremity injuries (LEI). Dementia is the primary cognitive cause of LTC and mostly occurs at the age of 75 and older. It describes a syndrome that subsumes different diseases that all cause a severe decline in cognitive functioning or affect personality traits. Most pa- tients require assistance and supervision (Doblhammer et al. 2012). LEI are a major physical health risk factor as regards LTC. Mobility limitations are often caused by LEI such as a broken hip, often resulting from falls (Barth et al. 2016), which are com- mon in older age and especially frequent among women (Stevens/Sogolow 2005). As mobility is a basic requirement for independence, LEI can cause LTC depend- ency. They can also act as a proxy for impairments that increase LTC risk. Falls are linked to age, because they are often caused by other age-related conditions like muscle weakness, problems with gait or balance, impaired vision or mild cognitive impairment. Restoring mobility after a LEI can be diffi cult, and future mobility may be reduced as a precaution after suffering a LEI (Neuman et al. 2014; Guralnik et al. 1995). A study comparing both risk factors found that dementia is more infl uential than LEI as far as individual LTC risk is concerned (Barth et al. 2016). Dementia and LEI can also be causally related (Barth et al. 2016; Lautenschlager et al. 2008). LEI can make performing daily tasks or maintaining an active lifestyle diffi cult, thus undermining the protective effect that such a lifestyle can have against dementia (Wang et al. 2002) and making LEI a risk factor for dementia (Barth et al. 2016). Mobility impairments like gait instability can be a predictor of cognitive de- • Alexander Barth238 cline. Delirium can occur after a LEI, which in turn can cause dementia (Krogseth et al. 2011). Dementia is often found in older fracture patients (Bruijn et al. 2013; Hamer/Chida 2009; Zhou et al. 2016) because it shares direct and intermediate risk factors with fractures. Dementia can negatively affect gait and balance, and its treatment can increase fracture risks (Friedman et al. 2010). If dementia and LEI are present concurrently, they can synergistically increase LTC risk (Barth et al. 2016; Inagawa et al. 2013). Looking at trends of years lived with disability (YLD) as a result of these condi- tions helps to contextualise their importance with regard to LTC demand. World- wide, YLD due to falls, which are the primary cause of LEI in older age, have in- creased by 46 percent from 1990 to 2010. Falls account for 2.5 percent of all YLD, making them the 10th most frequent cause of YLD globally and the 3rd most fre- quent cause in Western Europe. YLD due to dementia increased by 80 percent, thus being responsible for 0.9 percent of all YLD worldwide. Globally, dementia is the 24th most frequent cause of YLD, and the 10th most frequent cause in Western Europe. While the increases in total YLD due to both conditions are driven in part by population ageing, YLD rates per 100,000 persons have also increased by more than 12 percent in the case of falls, and by more than 38 percent for dementia (Vos et al. 2012). In terms of disability-adjusted life years (DALYs) lost, which weight and combine life years lived with health impairments with those lost due to premature mortal- ity, and looking at the population aged 60 years and older, both dementia and falls are among the fi fteen most frequent causes. Dementia accounts for 1.7 percent (9th most frequent cause), and falls for 2.2 percent (7th most frequent cause) of all DALYs lost in 2010 (Prince et al. 2015). Along with population ageing, dementia and falls are therefore increasing in importance as causes of YLD. 2.2 Long-term care in Germany LTC dependency is defi ned in accordance with German statutory public LTC insur- ance (“Pfl egeversicherung”), which was introduced in 1995. Benefi ts are granted based on one of three dependency levels, which are assigned by an objective as- sessment that focuses on limitations in activities of daily living (ADL). For care level one, a minimum of 90 minutes of assistance are required, at least half of which must relate to basic care tasks. Level three is assigned when fi ve hours are required, four hours of which must relate to basic care tasks. Thus, not everyone in need of assis- tance is covered. The higher the care level, the more time and skill are required by the caregivers. Payments can be used for professional care at home, in institutions or at the discretion of the recipient, e.g. for informal caregivers, utilities or renova- tions. The benefi ts do not cover all care-related costs. Institutional care in particular causes cost-benefi ts gaps (Rothgang et al. 2012b). In 2014, more than 2.6 million people received benefi ts, around two thirds of whom were female. Figure 1 shows the age structure by care type at age 75 and older. More than half of all LTC recipients are at least 80 years of age, and more than one third are aged 85 years and older. Over two thirds are in ambulatory care. For The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 239 two thirds of recipients, care was provided on an informal basis by relatives. Infor- mal care at home is thus the most frequent type of care, which is best suited for a lower level of care need. Professional care services provide either partial or full care for the remaining recipients living at home. Less than 30 percent of LTC recipients live in institutions, and about half are 85 years or older (Statistisches Bundesamt 2015b). 2.3 Infl uences on future long-term care supply Both LTC sectors face diffi culties in the coming decades. Informal care depends on partners, spouses, daughters or in-laws (Doblhammer/Scholz 2010). The share of older people living with a partner is expected to increase slightly up to 2030, due to a sharper increase in the life expectancy of males compared to females (Schulz 2010; Pötzsch 2011). As a result, no decrease in partner-based care potential is ex- pected to arise until then. Even if couples are living together longer, home care might be too demanding when the need exceeds the partner’s potential to give care. Between the 1937-1942 and 1963-1967 birth cohorts, the share of childless women has nearly doubled to 20 percent, leading to a reduction in the care provision poten- tial by own children or their spouses (Statistisches Bundesamt 2015a) from the late 2030s onwards when the latter cohort begins to reach age 75 and older. Informal care depends on the labour market participation of caregivers, because working full-time jobs restricts available time. Since most informal care is provided by females (Dukhovnov/Zagheni 2015; OECD 2017), female labour market participa- tion is a key factor. Female labour market participation between the ages of 50 and 65 is expected to increase from 44 percent to 61 percent (2003-2050). In absolute terms, however, the number of women will decline due to lower fertility and a high- er prevalence of childlessness, which both decrease the care potential of females Fig. 1: Population aged 75 and older by care need in 2014 75 77 79 81 83 85 87 89 91 93 95+ No care Home care Institutional care 0.000.200.400.60 M F 0.600.400.200.00 F in millions Source: Long-term care statistics (“Pfl egestatistik”), own design. • Alexander Barth240 (Schulz 2010). When informal caregivers are themselves parents, they also have to care for their young children (Dukhovnov/Zagheni 2015). Providing care requires the caregiver to live close to the recipient. Job-related mobility might increase distanc- es, especially in rural areas which younger people leave to move elsewhere, while their parents remain behind (Maretzke 2016). A decreasing population, fewer people of working age and fewer young people entering the labour market indicate that the recruitment of care workers will prove challenging over the next few decades (Geerts et al. 2012; OECD 2017; Rothgang et al. 2012a). Between 2010 and 2030, the working age population will decrease by nearly three million, and the economically active population will decrease by 1.4 million (Bundesministerium für Arbeit und Soziales 2013). The level of job vacancies in the care sector is already more than twice as high as the overall average (Bun- desministerium für Gesundheit 2015). All sectors will compete for fewer potential employees, and care is currently not one of most appealing career choices due to low wages and the high demands placed on carers. The shortage of LTC staff is expected to total between 140,000 and 200,000 full-time jobs by 2025 (Bundesmin- isterium für Gesundheit 2015) and exceed 500,000 by 2030 (Vereinigung der Bayer- ischen Wirtschaft 2012). 2.4 Projection of long-term care need for Germany A number of scientifi c or administrative LTC projections are available (Doblhammer 2012; Doblhammer/Scholz 2010; Pfaff 2010; Geerts et al. 2012; Doblhammer/Ziegler 2010; Schulz 2010; Rothgang et al. 2012a). They are based on population scenarios that are combined with different variants or trends of age-specifi c prevalence of total or sectoral LTC demand, or which incorporate determinants of LTC use such as household composition. They vary in their defi nition of care need, with some using the objective LTC insurance defi nition, and others using self-reported criteria, some- times based on survey data like SHARE (Geerts et al. 2012). The focus lies mostly on the number of LTC recipients. Some projections differentiate between care settings at home and in institutions (Geerts et al. 2012; Rothgang et al. 2012a). Because care need depends on the size of the cohorts reaching older ages, the primary source of uncertainty is the trend in life expectancy, while assumptions on the cohorts’ health status are the second key factor. Despite their methodological differences, all projections agree on a marked increase in LTC demand ranging from 22 percent to 62 percent between 2005 and 2030, and from 45 percent to 123 percent between 2005 and 2050. Health improvement scenarios assume that the increase in demand for LTC will be reduced by between 400,000 and 600,000 individuals in 2030, and by around one million in 2050 (Doblhammer 2012). Age structure therefore dictates a marked increase in LTC demand that no realistic degree of compression of morbid- ity can compensate. LTC demand will increase markedly, but the increase can be mitigated by health improvements. The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 241 2.5 Aim of this study Increasing LTC demand as a result of population ageing does not allow macro level interventions besides making changes to legislation regarding entitlement. Individ- ual level risk factors such as diseases are potentially modifi able, which shifts the focus on their effi cacy in mitigating future LTC demand. Our fi rst goal is to assess the plasticity of future LTC demand that is achievable by removing one or both of the two risk factors, namely dementia and LEI. First, we assess the impact on the total number of LTC recipients and second, on whether care demand arises at home or in institutions. The third aim is to demonstrate the general use of counterfactual projections based on claims data for the assessment of interventions and to show their use for policy planning. Germany is a good example of a country where LTC demand is growing. With Japan and Italy, Germany is one of the three countries most advanced on the path of population ageing and already has an age structure that other countries will be reaching (United Nations, Department of Economic and Social Affairs, Population Division 2017). Germany’s mandatory public LTC insurance system has been estab- lished over 20 years ago, meaning suffi cient real life experience has been gained and is available for analysis in routine data. Germany has a high and expanding ca- pacity of institutional LTC beds per 1,000 people aged 65 and over, surpassed only by a few countries such as the Netherlands or Sweden (Statistisches Bundesamt 2017; OECD 2017). Informal and institutional LTC demand can therefore be studied using data for Germany. Recent LTC reforms included persons formerly below the threshold for the lowest care level, which shows that providing care to those who require it remains important for policy-makers (Straub 2018). Given its status as a country where population ageing is at an advanced stage and continuing, Germany is a good example for retrospective studies and was also used as example to com- paratively study the costs of establishing national entitlement to LTC at home in different countries (Pickard et al. 2007). 3 Methods 3.1 Defi nition of long-term care Everyone with at least care level one is defi ned as LTC dependent. This study does not distinguish between care levels, but between LTC settings. LTC recipients not living in an institution are defi ned as receiving home care whereas benefi ciaries liv- ing in an institution are defi ned as institutional LTC recipients. The remainder of the population is considered as non-LTC dependent. 3.2 Defi nition of dementia Dementia is defi ned as one or more of the following: Alzheimer’s disease (ICD-10 codes F00/G30), vascular dementia (F01), Lewy body dementia (G31.82), circum- • Alexander Barth242 scribed brain atrophy (G31.0), dementia as a side-effect, e.g. of Parkinson’s dis- ease (F02, F05.1, G23.1), or other/unspecifi ed dementia (F03). To ensure validity, only diagnoses fl agged as “verifi ed” for outpatients or “discharge” for inpatients are considered. Second, at least two diagnoses issued in the same quarter either by an inpatient and an outpatient physician or from different outpatient specialists are required: inpatient diagnosis and neurological specialist, inpatient and general physician, inpatient and other type of specialist, neurological specialist and gen- eral physician, neurological specialist and other type of specialist, or other type of specialist and general physician. Alternatively, at least two verifi ed diagnoses at dif- ferent quarters from inpatient care, an outpatient neurological specialist, a general physician or other specialist are required. In these cases, the diagnosis is deemed valid from the time of the fi rst diagnosis onwards. If the fi rst verifi ed diagnosis oc- curs in the last observed quarter, it is also deemed valid. Dementia is coded as be- ing present from the fi rst validated occurrence onwards. 3.3 Defi nition of lower extremity injury Lower extremities are defi ned from the hip downwards. Relevant injuries are frac- tures, wounds, luxations, contusions, burns, frostbites and amputations (ICD codes S70–S99, relevant parts of T). LEI is coded as being present from the fi rst occur- rence onwards. 3.4 Population projection Population projections often use the cohort-component method, which projects the population by age and sex. Multi-state projections like those used here extend this method and share many characteristics. Age is divided into groups of equal size, in this case single years. The projection steps are of the same length as the age groups. The highest age group is open-ended, in our case 95 years and older. The projection requires a starting population that is specifi ed by age and sex for the fi rst year, and parameters of population development, i.e. fertility, migration and mortal- ity. The subgroups defi ned by age and sex are projected into the future for each step forward in time – in this case, single calendar years – by applying mortality, fertility and migration rates. In this case, fertility is disregarded because the interest of this study lies in the projection of LTC demand of cohorts that have already been born. According to the German Federal Statistical Offi ce, between 2000 and 2018 migration of persons in the 75 and older age group contributed between 0.3 percent and 0.8 percent to total annual immigration, and between 0.8 percent and 1.4 percent to total emigration. This corresponded to annual net totals of between +1,200 and -6,500 persons, or between 0.01 percent and 0.09 percent of the entire population in this age group, which increased from nearly 6 to 9.5 million persons. All variants of the most recent 14th offi cial coordinated population projection also assume net migration of simi- larly small amounts in the highest age group going forward. The impact of migration on population change at older ages is considered to be of minor importance and The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 243 is thus disregarded here. Our focus lies on assessing the effects of health changes on LTC demand: including concurrent migration fl ows would act as a confounding factor. Further, data to estimate age, sex and LTC-type-specifi c migration rates for older people are not available. Multi-state projections divide the population by age, sex and several functional states. We distinguish different LTC types using the states healthy (no LTC), LTC at home and LTC in an institution. Transitions (mobility between states) are possible from healthy to LTC at home (1st transition path) and to institutional LTC (2nd), and additionally from LTC at home to institutional LTC (3rd). Each state can lead to death (4th-6th). While transitions from any care to no care or from institutional to home care are possible, they are very rare. Ehing et al. (2015) show that the yearly transition prob- abilities to either a lower care level or from institutional to home care are very low, and Rothgang et al. (2017) describe the absolute number of changes from institu- tional to home care per year as marginal compared to the overall institutional care demand. The possibility of decreasing care need is thus disregarded and all transi- tions are considered as irreversible. Multi-state projections require all parameters of the cohort-component method for each functional state. Assumptions on the age- and sex-specifi c transitions into every other accessible state are required, too. The additional data we require for a multi-state approach consist of age-, sex- and state-specifi c parameters of mor- tality, and age- and sex-specifi c parameters for transitions between different LTC states. For all transitions in all scenarios, age- and sex-specifi c transition and mor- tality rates were estimated separately using parametric survival regressions with age and age-squared as covariates with an exponentially distributed baseline haz- ard (Stata procedures streg and predict) and calendar time as process time, with h(t\age) = eß0+ß1*age+ß2*age 2 used to estimate all required rates. For counterfactual scenarios, only data spells before the onset of the respective conditions are used: e.g. the dementia-free scenario only uses spells before the onset of dementia. Each model was run separately for both sexes. The estimated transition rates are applied unchanged for the entire projection time span, while mortality rates are reduced by 1.5 percent each year to account for increasing life expectancy. For Germany, a reduction of this amount is congruent with the decrease in mortality observed for ages 75 and over in the past decade. The 2014 starting population is based on the general population structure at the end of 2014 (census) and LTC statistics. It consists of males and females by age in all states (no care, care at home or care in institutions). 3.5 Health insurance routine data All parameter estimates are based on longitudinal health insurance routine data. A random sample was drawn, of 250,000 individuals who were at least 50 years old in the fi rst quarter of 2004 and were insured with Germany’s largest public health insurance provider AOK (“Allgemeine Ortskrankenkasse”). Besides ICD-10 records of medical diagnoses, information on age, sex, LTC status and date of death are • Alexander Barth244 included. Claims data is process-generated and used for the reimbursement of out- patient practitioners and inpatient clinics. The AOK covers about one third of the population aged 50 years and older, with this proportion increasing to as much as 50 percent among people of older ages. AOK data are not completely representa- tive of the German population, as there are differences regarding socio-economic and health status compared to other public sickness funds, and especially to private health insurance schemes (which cover about 11 percent of the population). The literature indicates that on average, compared to other public or private health in- surance providers, AOK members tend to be slightly older, have a lower socio-eco- nomic position and exhibit a somewhat higher prevalence of common conditions like hypertension, diabetes or cardiovascular diseases (Neubauer et al. 2017; Hoff- mann/Icks 2012). The sample was drawn from all insured persons, including those in nursing homes, regardless of whether they were seeking treatment and includes data up to the end of 2010 with one data spell per quarter. After data cleaning, consistency checking, removal of interrupted observations (e.g. due to a change of insurance provider) and the implementation of a two-year validation period for incident dementia, there remained around 122,000 individuals aged 65 and older at the fi rst quarter of 2006. 3.6 Scenarios Table 1 gives an overview of the properties for all scenarios. Each scenario covers the period from 2014 to 2044. Three counterfactual scenarios (2A-4A) are used to assess plasticity of LTC demand when removing specifi c risk factors: without de- mentia (2A), without LEI (3A) and the absence of both (4A). The status quo scenario 1A is the baseline for comparison. It shares the mortality decline with all counterfac- tual scenarios without removing any LTC risks. Two additional status quo scenarios with alternative mortality assumptions are used to compare the impact of removing risk factors with the effect of different rates of mortality decline (1B, 1C). Results are shown for the population in the relevant age group for LTC (75 years and older). All scenarios were calculated using the PDE Population Module tool (International Institute for Applied Systems Analysis (IIASA) 1997). The PDE project fi les, which include all necessary transition rates for all scenarios and variants presented here, as well as the base year population, are available as an online data appendix. 3.7 Use of counterfactual scenarios In order to assess the impact on LTC plasticity of potentially modifi able risk fac- tors, we use counterfactual projections that eliminate the respective risk factors. All differences between a counterfactual scenario and the baseline are caused by the elimination of the specifi c risk factor. The counterfactual scenarios do not strive to be realistic but are instead tools used to assess different options. They show the maximum degree to which the increase in LTC demand can be mitigated. The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 245 3.8 Hypotheses 3.8.1 General considerations Removing a LTC risk factor should mitigate future LTC demand. The amount of plas- ticity is likely to depend on the strength of the risk factor in question. The effect on total LTC demand of conditions that markedly increase the risk of LTC should be greater than that of weaker factors. Different risk factors can affect LTC risk differ- ently depending on age, sex or care setting. As a result, LTC plasticity can be ob- served in terms of total or sex-specifi c demand, age structure of LTC recipients and setting. LTC risks are also mortality risks. Their removal also increases life expectan- cy without and with LTC need due to other causes. The removal of a specifi c LTC risk can thus counterintuitively increase LTC demand due to other causes. Removing an LTC risk can lead to greater relative increases in older-age LTC demand compared to the baseline scenario. As the mortality decrease of 1.5 percent per year is the same for all counterfactual scenarios, an increase in older-age LTC demand is expected as a global effect for all main scenarios (1A-4A). Tab. 1: Overview of assumptions of all projection scenarios and variants A B C 1. Status Quo Mortality rates Mortality rates Mortality rates -2 % -1.5 % each year -1 %each year each year Includes dementia Includes dementia Includes dementia & & LEI cases & LEI cases LEI cases 2. No Dementia Mortality rates -1.5 % each year Excludes dementia cases 3. No Lower Mortality rates Extremity -1.5 % each year Injuries Excludes LEI cases 4. No Dementia or Mortality rates Lower Extremity -1.5 % each year Injuries Excludes dementia & LEI cases Source: Own design • Alexander Barth246 3.8.2 Dementia (2A) (1) Total demand Dementia is a risk factor for LTC. Its removal should decrease total demand compared to the baseline. As dementia is a greater LTC risk than LEI, it should have a greater effect on total demand. (2) Sectoral demand Dementia is strongly linked to institutional LTC. Its removal should decrease institutional demand compared to the baseline. (3) Older-age demand Institutional older-age LTC demand is expected to decrease due to the strong link between dementia and institutional LTC. Total older-age demand should increase compared to the baseline, because dementia’s removal should de- crease mortality and cause later entry and longer life in LTC. 3.8.3 LEI (3A) (1) Total demand LEI are a risk factor for LTC. Their removal should decrease total demand compared to the baseline, but not as much as dementia, since LEI are the risk factor which has less of an impact. Since LEI are more frequently found among women, their removal is likely to affect overall LTC demand among females in particular. (2) Sectoral demand LEI are less of a risk for institutional LTC than dementia; accordingly a minor decrease in institutional demand is expected. (3) Older-age demand A shift from older-age institutional demand to LTC demand at home is ex- pected, although to a lesser extent than without dementia. Total older-age demand should increase compared to the baseline as removing LEI ought to decrease mortality, delay entry into LTC and enable longer life in LTC. 3.8.4 Dementia & LEI (4A) (1) Total demand When present concurrently, dementia and LEI act synergistically and increase LTC risk to a disproportionate degree. In the absence of both factors, total demand should be markedly lower than in the baseline and also lower than in both individual scenarios combined. (2) Sectoral demand Due to the removal of dementia, a marked shift towards LTC at home is ex- pected. (3) Older-age demand A strong shift to older-age LTC at home is expected. Total demand should increase more than in any other scenario due to lower mortality, and is ex- The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 247 pected to be even higher than in both individual counterfactual scenarios combined. 4 Results 4.1 Total long-term care demand The plasticity of LTC demand that is attainable when removing dementia, LEI or both can be shown from different perspectives. Figure 2 shows the total demand over time, by scenario. Figure 3 subdivides the total demand by LTC type. Table 2 shows the endpoints of all scenarios in total, by care type as well for the younger (75-84) and older (85+) age group and by sex. Figure 4 shows the sex-specifi c popu- lation age structure by LTC type at the endpoint of each scenario. The benchmark for all counterfactual scenarios is the main status quo scenario (1A). In this scenario, LTC demand increases by 78 percent in total, from 1.82 to about 3.24 million people, between 2014 and 2044 (Fig. 2, Table 2) with LTC demand Fig. 2: Total number of LTC recipients in millions over time, all scenarios 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 2014 2019 2024 2029 2034 2039 2044 Year Status Quo (1A) No Dementia (2A) No LEI (3A) No Dementia or LEI (4A) Status Quo (1B) Status Quo (1C) in millions Source: Own calculations • Alexander Barth248 among males more than doubling and female demand growing by 64 percent. Like the status quo scenario (1A), all counterfactual scenarios also show increasing de- mand over time, indicating that no counterfactual scenario can mitigate the increase caused by population ageing. Beyond this demographically-driven increase, miti- gating plasticity is visible. Figures A1-A4 (appendix) show the respective base year transition probabilities for scenarios 1A-4A. Figure A1 indicates that without LTC need, the transition to LTC typically starts with home care, the probability of which doubles roughly every six years. The probability that recipients of LTC at home will transition into institutional LTC also increases with age, although not as rapidly as seen for the transition into LTC at home for persons without LTC dependency. Direct transition from no LTC de- pendency to institutional LTC dependency is far less common. The removal of LTC risk factors decreases the probability of transitioning into LTC, and also from LTC at home to institutional LTC. Removing individual LTC risk factors in the counterfactual scenarios (2A-3A) therefore slightly mitigates the increase in overall long-term care demand compared to the baseline scenario. The lower the probability of transition into LTC, the greater the mitigating effect on total LTC demand will be at each end- point. Changes in the transition probabilities to LTC are smallest in the LEI-free scenario (3A), where they are more apparent for females than males (Fig. A3). Mitigation of LTC demand in this counterfactual scenario, which results in a 72 percent increase in total LTC demand compared to 78 percent in the baseline scenario, is primarily attributable to female LTC demand (Table 2). The dementia-free scenario (2A) shows a more considerable decrease in LTC transition probabilities from no care to home care, and especially from home care to institutional LTC (Fig. A2) and results in a slightly more marked mitigation of future overall LTC demand, with an increase of 71 percent. The dementia and LEI-free scenario (4A) shows the biggest reduction in the transition probabilities into LTC (Fig. A4) and thus results in a marked mitigation of the increase in total LTC demand to 61 percent, compared to 78 percent in the baseline. These fi ndings support hypotheses 2A-1, 3A-1 and 4A-1 concerning the impact on total LTC demand. The more pronounced increase in male than female LTC demand is consistent in all scenarios. The alternative status quo scenarios, which assume faster (1C, 103 percent increase in total LTC demand) and slower (1B, 54 percent) mortality decreases without eliminating a specifi c LTC risk factor, indi- cate that the future mortality trend is a more important factor than individual LTC risks, because its effect on future LTC demand is greater. 4.2 Sectoral long-term care demand Figure 3 shows LTC demand by type over time. It illustrates that the major differenc- es between the counterfactual scenarios are found in their effects on LTC setting. The dementia-free (2A) and the dementia- and LEI-free counterfactual scenarios (4A) indicate that removing dementia affects LTC type, because these scenarios markedly decrease the transition probabilities to institutional LTC and thus show a shift towards LTC at home compared to the baseline scenario 1A (Fig. 3, A2, A4). The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 249 Compared to a 103 percent increase in institutional LTC within 30 years, which equates to about 1.3 million people, the dementia-free scenario (2A) reduces by half the increase in institutional demand, to 51 percent, and the dementia- and LEI-free variant (4A) lowers the increase to 32 percent (Table 2). At the status quo endpoint, 39 percent of LTC recipients are institutionalised, while the dementia- and LEI-free scenario (4A) and the dementia-free scenario (2A) show that 28 percent and 30 per- cent respectively of LTC is provided in institutions. In turn, home care is more fre- quent than in the baseline. Without dementia, LTC at home increases by 81 percent (2A), and without dementia and LEI (4A) by 76 percent, while the baseline indicates a 65 percent increase. Removing dementia markedly shifts demand from LTC at insti- tutions to LTC at home, coupled with a small decrease in total LTC demand growth. This supports hypotheses 2A-2 and 4A-2. The LEI-free scenario (3A) does not show Fig. 3: Total number of long-term care recipients (millions) by care type over time, all scenarios 20 14 20 18 20 22 20 26 20 30 20 34 20 38 20 42 Year Home Care Institution Status Quo (1A) 0.0 1.0 2.0 3.0 4.0 20 14 20 18 20 22 20 26 20 30 20 34 20 38 20 42 Year Status Quo (1B) 0.0 1.0 2.0 3.0 4.0 20 14 20 18 20 22 20 26 20 30 20 34 20 38 20 42 Year Status Quo (1C) 0.0 1.0 2.0 3.0 4.0 20 14 20 18 20 22 20 26 20 30 20 34 20 38 20 42 Year No Dementia (2A) 0.0 1.0 2.0 3.0 4.0 20 14 20 18 20 22 20 26 20 30 20 34 20 38 20 42 Year No LEI (3A) 0.0 1.0 2.0 3.0 4.0 20 14 20 18 20 22 20 26 20 30 20 34 20 38 20 42 Year No Dementia or LEI (4A) 0.0 1.0 2.0 3.0 4.0 in millions in millions in millions in millions in millions in millions Source: Own calculations • Alexander Barth250 such a marked shift to LTC at home, because the growth in institutional demand which it projects (88 percent) is only 15 points less than the 103 percent increase in the baseline, which tends to support hypothesis 3A-2. The LEI-free scenario (3A) is an exception, as it primarily mitigates the increase in female LTC demand in both care settings; the other effects on care setting are largely identical for both sexes. Of the two conditions considered, a marked mitigation of total and institutional LTC demand is only achievable by targeting dementia. 4.3 Older-old and younger-old long-term care demand Table 2 shows all scenario endpoints separately for younger (75-84 years) and older (85 years and older) LTC recipients. The relative increase in older-age demand com- pared to the baseline scenario (1A) is generally greater than for the younger group. The reasons are the same in all scenarios: large cohorts reach this age group and mortality risk progressively decreases in all scenarios. For instance, the baseline scenario (1A) shows an increase in total LTC demand of 69 percent for the younger and 86 percent for the older group. The plasticity when removing risk factors primar- ily lies in shifting the majority of the increase in demand to the older age group. For instance, removing dementia (2A) shows an increase in younger total LTC demand of 27 percent, and of 111 percent for the older age group. Larger relative increases of this kind in counterfactual scenarios are accompanied by a greater LTC demand in absolute terms for the older age group. For instance, the absolute difference in total demand between the endpoints of the baseline (1A) and the dementia-free sce- nario (2A) for the 85 and older age group is close to 250,000 individuals. Thus, for both genders and overall, more people reach the highest ages in the counterfactual scenarios due to the removal of conditions that also act as mortality risks. The share of institutional LTC recipients among those larger older cohorts is smaller than in the baseline due to the removal of LTC risk factors. The differences between the development of younger and older-age LTC de- mand are also apparent for the different LTC settings. Compared to the status quo (1A), where older-age institutional demand increases by 119 percent and demand for LTC at home by 64 percent, the older-age demand in the counterfactual scenarios resembles the setting-specifi c effects seen for overall demand: the removal of LTC risk factors mitigates the increase in institutional demand compared to the baseline scenario (1A) at the cost of more recipients of home care, due to both mitigation and postponement of the increase in LTC transition probability (Fig. A1-A4). Removing dementia (2A) mitigates the increase in institutional demand to 94 percent, while demand for LTC at home increases to 122 percent, double the increase seen in the status quo scenario. The dementia and LEI-free scenario (4A) shows even stronger plasticity: it extenuates the increase in institutional demand even more (70 percent), but shows the largest increase in home care demand of any particular scenario (129 percent). The LEI-free scenario (2A) shows the smallest degree of plasticity, which supports hypotheses 2A-4, 3A-4 and 4A-4. These outcomes occur in largely similar fashion among both genders, except for the LEI-free scenario (3A), where no The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 251 mitigation, but instead an even greater increase in older-age male institutional LTC demand occurs. Consequently, the most substantial effect is that in the case of the younger group, both LTC types see a mitigated increase in demand in all counterfactual scenarios compared to the baseline, whereas for the older group, decreases in institutional demand growth come at the cost of increased demand at home. This indicates that removing LTC risk factors delays entry into any type of LTC and also delays transi- tioning from home to institutional LTC. The results for older-age LTC demand show that the total demand cannot be mitigated, because decreases in institutional de- mand always come at the cost of more recipients of LTC at home. However, the sector where higher increases in LTC demand occur can be infl uenced by targeting specifi c risk factors, as the dementia-free (2A) and, in particular, the dementia- and LEI free (4A) scenarios demonstrate. Male Home Care % Institutional % All Care % All Results (75+) 2014 379,061 135,684 514,745 Status Quo (1A) 764,203 102 332,710 145 1,096,913 113 No Dementia (2A) 809,016 113 239,982 77 1,048,998 104 No LEI (3A) 754,897 99 335,201 147 1,090,098 112 No Dem. or LEI (4A) 775,677 105 237,137 75 1,012,814 97 Status Quo (1B) 668,933 76 256,566 89 925,499 80 Status Quo (1C) 862,926 128 423,509 212 1,286,435 150 75-84 2014 236,473 73,116 309,589 Status Quo (1A) 434,471 84 146,819 101 581,290 88 No Dementia (2A) 364,164 54 64,868 -11 429,032 39 No LEI (3A) 410,410 74 131,733 80 542,143 75 No Dem. or LEI (4A) 328,809 39 56,366 -23 385,175 24 Status Quo (1B) 387,602 64 118,075 61 505,677 63 Status Quo (1C) 481,827 104 179,195 145 661,022 114 85+ 2014 142,588 62,568 205,156 Status Quo (1A) 329,732 131 185,891 197 515,623 151 No Dementia (2A) 444,852 212 175,114 180 619,966 202 No LEI (3A) 344,487 142 203,468 225 547,955 167 No Dem. or LEI (4A) 446,868 213 180,771 189 627,639 206 Status Quo (1B) 281,331 97 138,491 121 419,822 105 Status Quo (1C) 381,099 167 244,314 290 625,413 205 Tab. 2: Overview of scenario results by sex, care type, age group and in total • Alexander Barth252 5 Discussion 5.1 Summary The ageing of large baby boomer cohorts will increase the demand for LTC services in the coming decades (Nowossadeck et al. 2016; Doblhammer 2012). To ensure that the LTC system will be able to respond to this challenge (Bundesministerium für Gesundheit 2015; Fujisawai/Colombo 2009), policy-makers require tools to assess the effi cacy of interventions aimed at mitigating the increase in LTC demand. Coun- terfactual projections that eliminate specifi c health-related risk factors can be used to assess the maximum degree of plasticity attainable with specifi c conditions. This study investigates the effi cacy of dementia and LEI in mitigating the increase in LTC demand, thus using one physical and one cognitive health-related determinant of Tab. 2: Continuation Female Home Care % Institutional % All Care % All Results (75+) 2014 818,263 489,398 1,307,661 Status Quo (1A) 1,208,259 48 934,980 91 2,143,239 64 No Dementia (2A) 1,359,754 66 705,818 44 2,065,572 58 No LEI (3A) 1,206,675 47 842,629 72 2,049,304 57 No Dem. or LEI (4A) 1,336,821 63 587,928 20 1,924,749 47 Status Quo (1B) 1,114,848 36 771,747 58 1,886,595 44 Status Quo (1C) 1,299,872 59 1,117,975 128 2,417,847 85 75-84 2014 386,858 167,286 554,144 Status Quo (1A) 597,527 54 279,258 67 876,785 58 No Dementia (2A) 530,333 37 134,828 -19 665,161 20 No LEI (3A) 544,103 41 230,525 38 774,628 40 No Dem. or LEI (4A) 472,033 22 113,215 -32 585,248 6 Status Quo (1B) 561,031 45 242,396 45 803,427 45 Status Quo (1C) 632,364 63 317,628 90 949,992 71 85+ 2014 431,405 322,112 753,517 Status Quo (1A) 610,732 42 655,722 104 1,266,454 68 No Dementia (2A) 829,421 92 570,990 77 1,400,411 86 No LEI (3A) 662,572 54 612,104 90 1,274,676 69 No Dem. or LEI (4A) 864,788 100 474,713 47 1,339,501 78 Status Quo (1B) 553,817 28 529,351 64 1,083,168 44 Status Quo (1C) 667,508 55 800,347 148 1,467,855 95 The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 253 LTC as two examples which are responsible for signifi cant shares of life years lived in disability in Western Europe and worldwide, and whose importance will only increase further as population ageing continues (Vos et al. 2012). We analyse their effects on both total and sectoral LTC demand at home and in nursing homes. The results indicate the varying effi cacy of dementia and LEI in mitigating the increase in total, sector-specifi c and age group-specifi c LTC demand, and show that their outcomes are broadly similar for both genders. Only in the LEI-free scenario (3A) do some results differ by sex, with the increase in female LTC demand mitigated primarily. This indicates that, in line with the literature, LEI affect older women much more frequently than they affect older men (Stevens/Sogolow 2005; Ismail et al. 2002). Tab. 2: Continuation Total Home Care % Institutional % All Care % All Results (75+) 2014 1,197,324 625,082 1,822,406 Status Quo (1A) 1,972,462 65 1,267,690 103 3,240,152 78 No Dementia (2A) 2,168,770 81 945,800 51 3,114,570 71 No LEI (3A) 1,961,572 64 1,177,830 88 3,139,402 72 No Dem. or LEI (4A) 2,112,498 76 825,065 32 2,937,563 61 Status Quo (1B) 1,783,781 49 1,028,313 65 2,812,094 54 Status Quo (1C) 2,162,798 81 1,541,484 147 3,704,282 103 75-84 2014 623,331 240,402 863,733 Status Quo (1A) 1,031,998 66 426,077 77 1,458,075 69 No Dementia (2A) 894,497 44 199,696 -17 1,094,193 27 No LEI (3A) 954,513 53 362,258 51 1,316,771 52 No Dem. or LEI (4A) 800,842 28 169,581 -29 970,423 12 Status Quo (1B) 948,633 52 360,471 50 1,309,104 52 Status Quo (1C) 1,114,191 79 496,823 107 1,611,014 87 85+ 2014 573,993 384,680 958,673 Status Quo (1A) 940,464 64 841,613 119 1,782,077 86 No Dementia (2A) 1,274,273 122 746,104 94 2,020,377 111 No LEI (3A) 1,007,059 75 815,572 112 1,822,631 90 No Dem. or LEI (4A) 1,311,656 129 655,484 70 1,967,140 105 Status Quo (1B) 835,148 45 667,842 74 1,502,990 57 Status Quo (1C) 1,048,607 83 1,044,661 172 2,093,268 118 Source: Own calculations • Alexander Barth254 Fig. 4: Population aged 75 and older by care type at end points, by scenario -600,000 -200,000 0 75 77 79 81 83 85 87 89 91 93 95+ No careHome careInstitutional care 0.00 0.200.400.60 0.600.400.200.00 M F Status Quo (1A) 0.00 -600,000 -200,000 0 75 77 79 81 83 85 87 89 91 93 95+ 0.00 0.200.400.60 0.600.400.200.00 M F No Dementia (2A) 0.00 -600,000 -200,000 0 75 77 79 81 83 85 87 89 91 93 95+ 0.00 0.200.400.60 0.600.400.200.00 M F Status Quo (1B) 0.00 -600,000 -200,000 0 75 77 79 81 83 85 87 89 91 93 95+ 0.00 0.200.400.60 0.600.400.200.00 M F No LEI (3A) 0.00 -600,000 -200,000 0 75 77 79 81 83 85 87 89 91 93 95+ 0.00 0.200.400.60 0.600.400.200.00 M F Status Quo (1C) 0.00 -600,000 -200,000 0 75 77 79 81 83 85 87 89 91 93 95+ 0.00 0.200.400.60 0.600.400.200.00 M F No Dementia or LEI (4A) 0.00 in millionsin millions in millionsin millions in millionsin millions Source: Own calculations The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 255 Dementia is a more infl uential risk factor than LEI, both in terms of its impact on total as well as on sectoral and age group-specifi c LTC demand. In the status quo scenario that is used as the baseline, the relative increase in institutional LTC de- mand is nearly double the increase in LTC demand at home. The removal of demen- tia shifts the majority of the increase in LTC from institutions to LTC at home, and also results in a modest mitigation of the increase in total LTC demand. Additionally, the removal of dementia causes a later onset of and a longer life with LTC depend- ency. Both circumstances contribute to most of the increase in LTC demand shifting from the 75-84 to the 85 and older age group. Less institutional LTC dependency in the younger age group comes at the cost of higher demand for LTC at home for the 85 and older group. Mitigating the increase in total LTC demand by removing a specifi c risk factor does not therefore necessar- ily translate into a mitigation for all age groups or for all sectors, but can be a com- posite effect of trends differing by both age group and LTC type. The effi cacy of LEI in mitigating the increase in total LTC demand is smaller than the effect seen when removing dementia, and its impact on sectoral LTC demand is different. The removal of LEI mitigates the increase in institutional LTC demand in comparison with the baseline, yet the fi gure is higher than if dementia is removed. The increase in demand for LTC at home is not signifi cantly affected. As a result, the effi cacy of LEI lies almost completely in mitigating the increase in institutional LTC demand, albeit to a lesser degree than dementia. Eliminating LEI also results in a larger increase in LTC demand for the 85 and older age group while the mitigation of the increase in LTC demand for the 75-84-yearold age group is modest. This is the same general pattern as seen for dementia, although to a lesser degree. The largest impact on the plasticity of LTC demand is seen as a consequence of the concurrent removal of both dementia and LEI. This scenario not only retains the core characteristics of both individual counterfactual scenarios, but also shows that both conditions are linked synergistically (Barth et al. 2016). When both are removed at the same time, the mitigation of the increase in LTC is even more pro- nounced than could be expected by combining both individual scenarios, both in terms of plasticity of total LTC demand, and regarding the shift in LTC setting from institutions to the home. 5.2 Strengths and weaknesses The main strength of this study is that all input parameters are estimated using the same reliable and large-scale longitudinal empirical data that include medical diag- noses, sex, age and type of LTC. It is unaffected by typical survey biases like unit- nonresponse or recall uncertainty, which would have been particularly problematic for dementia patients. The data covers the whole population, including those in nursing homes. It should be reiterated that AOK data cannot be entirely representa- tive of the German population, as those insured with AOK tend to be in slightly worse health and occupy lower socio-economic positions than members of other public or private health insurance schemes (Neubauer et al. 2017; Hoffmann/Icks 2012). Basing our estimates on a group that is in a somewhat less favourable posi- • Alexander Barth256 tion might cause our fi ndings to be slightly biased in two ways. First, in terms of overestimating the risk (or mitigating effect) that dementia and LEI (or their absence) exert on people’s LTC need, e.g. because they might develop dementia at a younger age or to a more severe degree or be more likely to fall and sustain a severe injury than persons insured elsewhere, who might be healthier or have a more protec- tive lifestyle in terms of compensatory cognitive reserves or dementia or fall risk factors. This could affect both the absolute size as well as the sectoral distribution of future LTC demand in the counterfactual scenarios. Second, persons in higher socio-economic positions might have greater fi nancial reserves, which could mean that they are able to avoid or delay having to apply for LTC benefi ts, or institutionali- zation in cases where the care need is severe, and remain without formal LTC status or in home care longer. This would result in a slight reduction in the transition from no care to home care, or the marked sectoral shift from institutional to home care in dementia-free scenarios. One weakness that affects our projections of LTC is its defi nition, which has changed since the data we used was collected. In the timeframe covered by our data, dementia played a special part in the assessment of offi cial LTC dependence. Until 2008, persons with dementia who did not meet any of the ADL dependency requirements were not formally eligible for LTC insurance benefi ts. In 2008, LTC de- pendence level 0 was introduced to include dementia as a criterion for LTC depend- ency, but this is not available in our data. Thus, not everybody with dementia as the sole impairment may have been registered as being LTC dependent. However, the likelihood of receiving LTC benefi ts as a dementia patient is high, especially for older patients (Barth et al. 2016; Doblhammer et al. 2012; Fink/Doblhammer 2015; Roth- gang et al. 2012b). Since 2015, after the observation period of the data used here, dementia became a criterion for LTC dependency, which means that more people with dementia as their sole limitation are eligible for LTC benefi ts (Straub 2018). Under these terms, the effi cacy of dementia in mitigating the increase in future LTC demand might therefore be even higher than our results indicate. The assumptions of future developments of mortality are central to the results. This is the same for any projection, and we showed different scenarios to account for this. 5.3 Future research questions This paper explores the impact of two conditions on the plasticity of total, sectoral and age-specifi c LTC demand and fi nds distinct effects and differences. Dementia and LEI are important cognitive and physical health risk factors, but are only exam- ples. Further research should investigate the effects of more diseases that are com- mon in older age. The distinction of different LTC settings and the differentiation of younger and older-age LTC demand should be upheld, because, as the results for dementia indicated, signifi cant differences might occur. In this case, they were caused by a decreased risk of transitioning from home to institutional LTC. Further research could thus not only consider the risk of requiring any kind of LTC, but also focus on factors for the increase in care need, i.e. the transition from LTC at home to institutional LTC or to a higher care level. Going beyond health-related risk factors The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 257 for LTC, other potentially modifi able determinants of LTC demand, e.g. individual living conditions, individual lifestyle factors or social interactions, could be incor- porated. Household composition is an important aspect of living conditions that should be taken into account. It is not only important for the possibility of receiving care from the partner. Living alone in older age is also a risk factor for LTC need. Com- pared to households with more than one resident, people living alone in older age are often in poor health, have problems performing ADL tasks, have lower levels of physical activity, are at greater risk of being isolated and suffer falls more often – all of which are risk factors for care need and for determinants of care need, like dementia (Kharicha et al. 2007). It is plausible that for a comparatively milder LTC risk factor, a dependent person living alone has a higher institutionalization risk than one living with a partner. Consequently, the inclusion of household composition in models used to estimate LTC transition risks would be desirable. However, we were unable to consider it in this paper because information on household structure is not included in health insurance routine data. Each country’s care regime and their defi nitions of care need are unique. How diseases or impairments translate into LTC entitlement, how access to LTC is regu- lated, whether LTC support is means-tested, which level of government organises LTC provision, whether care is provided on a public or private basis or a mixture of the two, how LTC demand is related to specifi c care settings and what level of qual- ity is aimed for all depend on the LTC regime (Riedel/Kraus 2011). Future research could also assess whether our fi ndings are unique to the German LTC regime be- tween 2004 and 2010, or if and to what degree they can be generalised. 5.4 Conclusion The total demand for LTC services will continue to increase across developed coun- tries over the coming decades (Fujisawai/Colombo 2009; Comas-Herrera et al. 2006). We show that the amount of the increase in total and sector-specifi c LTC de- mand can be mitigated by targeting potentially modifi able risk factors. In particular, the removal of dementia and, to a lesser degree, the removal of LEI, signifi cantly mitigate the increase in institutional LTC dependency at the cost of additional in- creases in LTC at home. Because nursing homes are both cost-intensive and labour- intensive, identifying effective intervention strategies is important for policy-mak- ers striving to decrease LTC costs and provide suffi cient care services when faced with a potential shortage in professional LTC staff (Fujisawai/Colombo 2009; Geerts et al. 2012). According to the German Federal Ministry of Health, up to 214,000 jobs for care professionals will be vacant in 2025 due to insuffi cient labour supply (Bun- desministerium für Gesundheit 2015), while another projection indicates half a mil- lion vacant full-time jobs by 2030 (Vereinigung der Bayerischen Wirtschaft 2012). Due to smaller cohort sizes in recent decades, the total number of people of working age is decreasing and will continue to do so well into the 2030s (Bundesministe- rium für Arbeit und Soziales 2013). Policy-makers have already taken action to make the care sector more competitive, for instance by campaigning for higher wages or • Alexander Barth258 underlining high job security (Bundesministerium für Gesundheit 2015; Fujisawai/ Colombo 2009). Even if the increase in institutional LTC demand were able to be mitigated by tar- geting risk factors like dementia, this would come at the cost of a marked increase in demand for LTC at home. Although this does not necessarily require care profes- sionals, it depends on the availability of informal caregivers, primarily partners and children (Dukhovnov/Zagheni 2015). Relatively speaking, more older people will live with a partner in the future (Pötzsch 2011), so the availability of partner-based infor- mal LTC potential is not the most serious problem for care needs that are classed as being non-severe. Children make up the other major group of informal caregivers. As far as the availability of informal care potential is concerned, increasing childless- ness in particular might be a problem once cohorts that have seen a decline in fertil- ity reach ages where LTC becomes prevalent by the end of the 2030s (Statistisches Bundesamt 2015a). Projections of informal care potential by adult children indicate a decrease of more than 30 percent between 2009 and 2040, and up to 40 percent by 2060 (Dudel 2015). This is not only a result of fewer children, but also of increasing labour-market participation, especially by middle-aged women (Schulz 2010), which reduces their availability as caregivers (Bundesministerium für Arbeit und Soziales 2013). This decline in informal care potential is estimated to be equal to 125,000 full-time care professionals in 2030 (Vereinigung der Bayerischen Wirtschaft 2012). Additionally, greater distances between where a dependent person and supporting child live might make the provision of informal care more diffi cult (Nowossadeck et al. 2016). As a result, professional LTC will be required above all by persons with LTC dependency who are single and childless, or whose children do not live close by. These factors should be taken into account when planning LTC services from a regional perspective. More fl exibility in mixed work-care scenarios might help in enabling more children of working age to provide care to their parents. This also requires fi nancial losses incurred as a result of fewer working hours to be partially compensated by transfer payments. Since most informal care is still provided by females (Dukhovnov/Zagheni 2015), efforts to promote more male provision of in- formal care could also help to recruit more informal carers. Aside from trying to tap into all available informal care resources and increasing the attractiveness of the care sector, other areas could be explored. For instance, innovations in housing and technology could help LTC dependent persons to live at home longer (Barth/Doblhammer 2016; Fujisawai/Colombo 2009). New forms of liv- ing arrangements, such as shared apartments, could enable LTC recipients to assist each other as far as possible and split the use of informal or professional assistance between the remaining tasks. For home owners, re-mortgaging to fi nance LTC sup- port is becoming more popular. However, the most promising approach is the prevention or delay of primary risk factors for LTC need. Identifying infl uential health-related LTC risk factors and assessing their effects on the plasticity of total, sectoral and age-specifi c LTC de- mand is thus useful for policy planning. In this study, dementia was identifi ed as a determinant that has a particularly big impact, especially in terms of mitigating the increase in institutional LTC demand. The results of counterfactual scenarios The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 259 cannot be interpreted as realistically achievable goals because they assume the population-wide, instantaneous elimination of a specifi c disease. However, they il- lustrate the maximum attainable impact and targeting the risk factors that have the biggest impact is a reasonable strategy. A risk factor identifi ed as signifi cant in a counterfactual scenario should also have a notable, albeit smaller impact in real life, where intervention is only partially successful. In the context of our results, this indicates that dementia should be a primary target, because its impact was so pronounced that even partly successful interventions would mitigate the increase in LTC demand to a greater extent than a completely successful intervention against less severe risk factors like LEI. Lesser risk factors should not be dismissed outright as possible intervention targets, because synergistic effects between risk factors should be taken into account. The effect of concurrent LEI and dementia, which disproportionately increases LTC risk, is a reason to target the lesser risk factor LEI, if assessed individually. Without LEI, LTC dependency can be slightly reduced directly, but its removal can also be helpful in the event of a later onset of dementia as it then prevents the disproportionately greater LTC risk. Acknowledgements I would like to thank Gabriele Doblhammer for her important input in the design of this study, and the anonymous reviewers for their very helpful suggestions which have led to signifi cant improvements in the manuscript. I would also like to thank WIdO, the scientifi c research institute of the AOK for providing access to the health insurance routine data used in this paper. This study has not received any external funding. References Barth, Alexander; Doblhammer, Gabriele 2016: Physische Mobilität und Gesundheit im Alter. Ansätze zur Reduktion von Pfl egebedürftigkeit und Demenz in einer alternden Gesellschaft. In: Mayer, Tilman (Ed.): Die transformative Macht der Demografi e. Wies- baden: Springer VS: 207-244 [doi: 10.1007/978-3-658-13166-1_15]. Barth, Alexander et al. 2016: Extremity injuries and dementia disproportionately in- crease the risk for long-term care at older age in an analysis of German Health Insur- ance routine data for the years 2006 to 2010. In: European Review of Aging and Physi- cal Activity 13,1: 1168 [doi: 10.1186/s11556-016-0169-8]. Bruijn, Renée F.A.G. et al. 2013: The association between physical activity and dementia in an elderly population: the Rotterdam Study. In: European Journal of Epidemiology 28,3: 277-283 [doi: 10.1007/s10654-013-9773-3]. Bundesministerium für Arbeit und Soziales 2013: Arbeitsmarktprognose 2030. Bonn. Bundesministerium für Gesundheit 2015: Pfl egefachkräftemangel [http://www.bmg. bund.de/themen/pfl ege/pfl egekraefte/pfl egefachkraeftemangel.html, 06.06.2016]. Christensen, Kaare et al. 2009: Ageing populations: the challenges ahead. In: The Lancet 374,9696: 1196-1208 [doi: 10.1016/S0140-6736(09)61460-4]. • Alexander Barth260 Comas-Herrera et al. 2006: Future long-term care expenditure in Germany, Spain, Italy and the United Kingdom. In: Ageing and Society 26,2: 285-302 [doi: 10.1017/ S0144686X05004289]. Doblhammer, Gabriele 2012: Entwicklung des Pfl egebedarfs und der häuslichen Pfl ege in Deutschland. In: Becker-Stoll, Fabienne; Klös, Hans-Peter; Thüsing, Gregor (Eds.): Expertisen zum Achten Familienbericht. München: Ifo-Institut für Wirtschafts- forschung: 355-394. Doblhammer, Gabriele et al. 2012: Demografi e der Demenz. Bern: Verlag Hans Huber. Doblhammer, Gabriele; Scholz, Rembrandt D. (Eds.) 2010: Ageing, care need and qual- ity of life. The perspective of care givers and people in need of care. VS research. 1st ed. Wiesbaden: VS, Verl. für Sozialwiss [doi: 10.1007/978-3-531-92335-2]. Doblhammer, Gabriele; Ziegler, Uta 2010: Care Need Projections by Marital Status and Childlessness for Germany 2000-2030 based on the FELICIE Project. In: Doblhammer, Gabriele; Scholz, Rembrandt D. (Eds.): Ageing, care need and quality of life. The per- spective of care givers and people in need of care. VS research. 1st ed. Wiesbaden: VS, Verl. für Sozialwiss: 42-60. Dudel, Christian 2015: Vorausberechnung des Pfl egepotentials von erwachsenen Kindern für ihre pfl egebedürftigen Eltern. In: Sozialer Fortschritt 64,1-2: 14-26 [doi: 10.3790/sfo.64.1-2.14]. Dukhovnov, Denys; Zagheni, Emilio 2015: Who Takes Care of Whom in the U.S.? Time Transfers by Age and Sex. In: Population and development review 41,2: 183-206 [doi: 10.1111/j.1728-4457.2015.00044.x]. Ehing, Daniel; Hagist, Christian; Saal, Tobias 2015: Pfl egeverläufe im Spiegel von Rou- tinedaten der GKV: Eine Analyse für die Jahre 2003 bis 2010. In: Zeitschrift für die gesamte Versicherungswissenschaft 104,2: 179-210 [doi: 10.1007/s12297-015-0298-6]. Fink, Anne; Doblhammer, Gabriele 2015: Risk of Long-Term Care Dependence for De- mentia Patients is Associated with Type of Physician. An Analysis of German Health Claims Data for the Years 2006 to 2010. In: Journal of Alzheimer’s disease: JAD 47,2: 443-452 [doi: 10.3233/JAD-142082]. Friedman, Susan M. et al. 2010: Dementia and Hip Fractures: Development of a Patho- genic Framework for Understanding and Studying Risk. In: Geriatric Orthopaedic Sur- gery & Rehabilitation 1,2: 52-62 [doi: 10.1177/2151458510389463]. Fujisawai, Rie; Colombo, Francesca 2009: The Long-Term Care Workforce: Overview and Strategies to Adapt Supply to a Growing Demand. OECD Health Working Papers 44. Geerts, Joanna; Willemé, Peter; Mot, Esther 2012: Long-Term Care Use and Supply in Europe: Projections for Germany, The Netherlands, Spain and Poland. ENEPRI Re- search Reports. Guralnik, Jack M. et al. 1995: Lower-Extremity Function in Persons Over the Age of 70 Years as a Predictor of Subsequent Disability. In: The New England Journal of Medi- cine 332,9: 556-561. Hamer, Mark; Chida, Yoichi 2009: Physical activity and risk of neurodegenerative dis- ease: a systematic review of prospective evidence. In: Psychological Medicine 39,01: 3-11 [doi: 10.1017/S0033291708003681]. Hoffmann, Falk; Icks, Andrea 2012: Unterschiede in der Versichertenstruktur von Krank- enkassen und deren Auswirkungen für die Versorgungsforschung: Ergebnisse des Bertelsmann-Gesundheitsmonitors. In: Gesundheitswesen (Bundesverband der Arzte des Offentlichen Gesundheitsdienstes (Germany) 74,5: 291-297 [doi: 10.1055/s-0031- 1275711]. The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 261 Inagawa, Toshimitsu et al. 2013: Decreased activity of daily living produced by the combination of Alzheimer’s disease and lower limb fracture in elderly requiring nurs- ing care. In: Environmental Health and Preventive Medicine 18,1: 16-23 [doi: 10.1007/ s12199-012-0283-9]. International Institute for Applied Systems Analysis (IIASA) 1997: PDE Population Pro- jection Software [http://webarchive.iiasa.ac.at/Research/POP/pub/software/pde/, 06.06.2016]. Ismail, Abbas A. et al. 2002: Incidence of limb fracture across Europe: results from the European Prospective Osteoporosis Study (EPOS). In: Osteoporosis International 13,7: 565-571 [doi: 10.1007/s001980200074]. Kharicha, Kalpa et al. 2007: Health risk appraisal in older people 1: are older people living alone an “at-risk” group? In: The British journal of general practice: the journal of the Royal College of General Practitioners 57,537: 271-276. Koller, Daniela et al. 2014: Multimorbidity and long-term care dependency – a fi ve-year follow-up. In: BMC Geriatrics 14: 70 [doi: 10.1186/1471-2318-14-70]. Krogseth, Maria et al. 2011: Delirium is an important predictor of incident dementia among elderly hip fracture patients. In: Dementia and Geriatric Cognitive Disorders 31,1: 63-70 [doi: 10.1159/000322591]. Lautenschlager, Nicola T. et al. 2008: Effect of Physical Activity on Cognitive Function in Older Adults at Risk for Alzheimer Disease. A Randomized Trial. In: Journal of the American Medical Association 300,9: 1027-1037 [doi: 10.1001/jama.300.9.1027]. Maretzke, Steffen 2016: Der demografi sche Wandel läuft und läuft. Ohne regionale An- passungsstrategien geht es nicht. In: Mayer, Tilman (Ed.): Die transformative Macht der Demografi e. Wiesbaden: Springer VS: 531-553 [doi: 10.1007/978-3-658-13166- 1_32]. Neubauer, Sarah et al. 2017: Access, use, and challenges of claims data analyses in Germany. In: The European journal of health economics: HEPAC: health economics in prevention and care 18,5: 533-536 [doi: 10.1007/s10198-016-0849-3]. Neuman, Mark D. et al. 2014: Survival and functional outcomes after hip fracture among nursing home residents. In: JAMA Internal Medicine 174,8: 1273-1280 [doi: 10.1001/ jamainternmed.2014.2362.]. Nowossadeck, Sonja; Engstler, Heribert; Klaus, Daniela 2016: Pfl ege und Unterstützung durch Angehörige. Report Altersdaten. Berlin: Deutsches Zentrum für Altersfragen. OECD 2017: Health at a Glance 2017: OECD. Pfaff, Heiko 2010: People in Need of Long-term Care: The Present and the Future. In: Doblhammer, Gabriele; Scholz, Rembrandt D. (Eds.): Ageing, care need and quality of life. The perspective of care givers and people in need of care. VS research. 1st ed. Wiesbaden: VS, Verl. für Sozialwiss: 14-28 [doi: 10.1007/978-3-531-92335-2_1]. Pickard, Linda et al. 2007: Modelling an entitlement to long-term care services for older people in Europe: projections for long-term care expenditure to 2050. In: Journal of European Social Policy 17,1: 33-48 [doi: 10.1177/0958928707071879]. Pötzsch, Olga 2011: Entwicklung der Privathaushalte bis 2030: Ende des ansteigenden Trends. In: Wirtschaft und Statistik 3: 205-218. Prince, Martin J. et al. 2015: The burden of disease in older people and implications for health policy and practice. In: The Lancet 385,9967: 549-562 [doi: 10.1016/S0140- 6736(14)61347-7]. • Alexander Barth262 Riedel, Monika; Kraus, Markus 2011: The Organisation of Formal Long-Term Care for the Elderly. Results From the 21 European Country Studies in the ANCIEN Project. ENEPRI Research Reports. Rothgang, Heinz et al. 2017: Ambulantisierung stationärer Einrichtungen und innovative ambulante Wohnformen. Endbericht. Studie im Auftrag des Bundesministeriums für Gesundheit. Bonn. Rothgang, Heinz et al. 2012a: Themenreport “Pfl ege 2030”. Was ist zu erwarten – was ist zu tun? Gütersloh. Rothgang, Heinz et al. 2012b: Barmer GEK Pfl egereport 2012. Schwerpunktthema: Kos- ten bei Pfl egebedürftigkeit. Schriftenreihe zur Gesundheitsanalyse 17. Siegburg: As- gard-Verl.-Service. Schulz, Erika 2010: Projection of Care Need and Family Resources in Germany. In: Dobl- hammer, Gabriele; Scholz, Rembrandt D. (Eds.): Ageing, care need and quality of life. The perspective of care givers and people in need of care. VS research. 1st ed. Wies- baden: VS, Verl. für Sozialwiss: 61-81 [doi: 10.1007/978-3-531-92335-2_4]. Statistisches Bundesamt 2015a: Daten zu Geburten, Familien und Kinderlosigkeit. Ergebnisse des Mikrozensus 2012. Wiesbaden. Statistisches Bundesamt 2015b: Pfl egestatistik. Wiesbaden. Statistisches Bundesamt 2017: Pfl egestatistik 2015. Pfl ege im Rahmen der Pfl egeversi- cherung. Deutschlandergebnisse. Wiesbaden. Stevens, Judy A.; Sogolow, Ellen D. 2005: Gender differences for non-fatal unintentional fall related injuries among older adults. In: Injury prevention: journal of the Interna- tional Society for Child and Adolescent Injury Prevention 11,2: 115-119 [doi: 10.1136/ ip.2004.005835]. Straub, Christoph 2018: Pfl egereform hat viele Verbesserungen gebracht. In: Gesund- heitsökonomie & Qualitätsmanagement 23,01: 5-6 [doi: 10.1055/s-0043-124822]. Unger, Rainer; Schulze, Alexander 2013: Can We Really (All) Work Longer? Trends in Healthy Life Expectancy According to Social Stratum in Germany. In: Comparative Population Studies 38,3: 565-582 [doi: 10.4232/10.CPoS-2013-03en]. United Nations, Department of Economic and Social Affairs, Population Division 2017: World Population Ageing 2017. United Nations New York, 2017 Vereinigung der Bayerischen Wirtschaft 2012: Pfl egelandschaft 2030. Vos, Theo et al. 2012: Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. In: The Lancet 380,9859: 2163-2196 [doi: 10.1016/S0140-6736(12)61729-2]. Wang, Hui-Xin et al. 2002: Late-Life Engagement in Social and Leisure Activities Is As- sociated with a Decreased Risk of Dementia: A Longitudinal Study from the Kungshol- men Project. In: American journal of epidemiology 155,12: 1081-1087 [doi: 10.1093/ aje/155.12.1081]. Zhou, Ying; Putter, Hein; Doblhammer, Gabriele 2016: Years of life lost due to lower extremity injury in association with dementia, and care need: a 6-year follow-up pop- ulation-based study using a multi-state approach among German elderly. In: BMC geriatrics 16,1: 9 [doi: 10.1186/s12877-016-0184-7]. The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 263 Date of submission: 14.10.2018 Date of acceptance: 07.10.2019 Alexander Barth (). University of Rostock, Institute for Sociology and Demography. Rostock, Germany. E-mail: alexander.barth2@uni-rostock.de URL: https://www.isd.uni-rostock.de/en/isd/lehrstuhl/esf/staff-members/ • Alexander Barth264 Appendix Fig. A1: Base year transition probabilities for status quo scenario 1A 0.001 0.010 0.100 1.000 75 77 79 81 83 85 87 89 91 93 95+ Age No care —> Home care No care —> Institutional care No care —> Death No care —> No care 0.000 Probability Male 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Home care —> Institutional care Home care —> Death Home care —> Home care Probability 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Institutional care —> Death Institutional care —> Institutional care Probability 0.001 0.010 0.100 1.000 75 77 79 81 83 85 87 89 91 93 95+ Age No care —> Home care No care —> Institutional care No care —> Death No care —> No care 0.000 Probability Female 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Home care —> Institutional care Home care —> Death Home care —> Home care Probability 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Institutional care —> Death Institutional care —> Institutional care Probability The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 265 Fig. A2: Base year transition probabilities for dementia-free counterfactual scenario 2A 0.001 0.010 0.100 1.000 75 77 79 81 83 85 87 89 91 93 95+ Age No care —> Home care No care —> Institutional care No care —> Death No care —> No care 0.000 Probability Male 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Home care —> Institutional care Home care —> Death Home care —> Home care Probability 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Institutional care —> Death Institutional care —> Institutional care Probability 0.001 0.010 0.100 1.000 75 77 79 81 83 85 87 89 91 93 95+ Age No care —> Home care No care —> Institutional care No care —> Death No care —> No care 0.000 Probability Female 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Home care —> Institutional care Home care —> Death Home care —> Home care Probability 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Institutional care —> Death Institutional care —> Institutional care Probability • Alexander Barth266 Fig. A3: Base year transition probabilities for lower extremity injury-free counterfactual scenario 3A 0.001 0.010 0.100 1.000 75 77 79 81 83 85 87 89 91 93 95+ Age No care —> Home care No care —> Institutional care No care —> Death No care —> No care 0.000 Probability Male 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Home care —> Institutional care Home care —> Death Home care —> Home care Probability 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Institutional care —> Death Institutional care —> Institutional care Probability 0.001 0.010 0.100 1.000 75 77 79 81 83 85 87 89 91 93 95+ Age No care —> Home care No care —> Institutional care No care —> Death No care —> No care 0.000 Probability Female 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Home care —> Institutional care Home care —> Death Home care —> Home care Probability 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Institutional care —> Death Institutional care —> Institutional care Probability The Impact of Dementia and Extremity Injuries on the Plasticity of Long-term Care Demand • 267 Fig. A4: Base year transition probabilities for dementia and lower-extremity injury-free counterfactual scenario 4A 0.0001 0.0010 0.0100 0.1000 1.0000 75 77 79 81 83 85 87 89 91 93 95+ Age No care —> Home care No care —> Institutional care No care —> Death No care —> No care 0.0000 Probability Male 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Home care —> Institutional care Home care —> Death Home care —> Home care Probability 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Institutional care —> Death Institutional care —> Institutional care Probability 0.0001 0.0010 0.0100 0.1000 1.0000 75 77 79 81 83 85 87 89 91 93 95+ Age No care —> Home care No care —> Institutional care No care —> Death No care —> No care 0.0000 Probability Female 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Home care —> Institutional care Home care —> Death Home care —> Home care Probability 0.00 0.01 0.10 1.00 75 77 79 81 83 85 87 89 91 93 95+ Age Institutional care —> Death Institutional care —> Institutional care Probability Published by Prof. Dr. Norbert F. Schneider Federal Institute for Population Research D-65180 Wiesbaden / Germany 2019 Managing Editor Prof. Philip Rees Dr. Katrin Schiefer Copy Editor Julia Luther Editorial Assistant Beatriz Feiler-Fuchs Wiebke Hamann Layout Beatriz Feiler-Fuchs E-mail: cpos@bib.bund.de Scientifi c Advisory Board Karsten Hank (Cologne) Michaela Kreyenfeld (Berlin) Marc Luy (Vienna) Natalie Nitsche (Vienna) Peter Preisendörfer (Mainz) Zsolt Spéder (Budapest) Rainer Wehrhahn (Kiel) Comparative Population Studies www.comparativepopulationstudies.de ISSN: 1869-8980 (Print) – 1869-8999 (Internet) Board of Reviewers Martin Abraham (Erlangen) Laura Bernardi (Lausanne) Hansjörg Bucher (Bonn) Claudia Diehl (Konstanz) Andreas Diekmann (Zurich) Gabriele Doblhammer-Reiter (Rostock) Jürgen Dorbritz (Wiesbaden) Anette Eva Fasang (Berlin) E.-Jürgen Flöthmann (Bielefeld) Alexia Fürnkranz-Prskawetz (Vienna) Beat Fux (Salzburg) Joshua Goldstein (Berkeley) Sonja Haug (Regensburg) Hill Kulu (Liverpool) Aart C. Liefbroer (The Hague) Kurt Lüscher (Konstanz) Emma Lundholm (Umeå) Nadja Milewski (Rostock) Dimiter Philipov (Vienna) Roland Rau (Rostock) Tomáš Sobotka (Vienna) Jeroen Spijker (Barcelona) Olivier Thévenon (Paris) Helga de Valk (Brussels) Heike Trappe (Rostock) Michael Wagner (Cologne)