Aging, health expenditure, proximity of death and income in Finland

Transcription

1 Aging, health expenditure, proximity of death and income in Finland Unto Häkkinen* 1, Pekka Martikainen 2, Anja Noro 1, Elina Nihtilä 2 and Mikko Peltola 1 1 Centre for Health Economics at STAKES 2 Population Research Unit, Department of Sociology University of Helsinki '* Corresponding author: Centre for Health Economics at Stakes (CHESS) P.O.Box 220, Lintulahdenkuja 4,00530 Helsinki, Finland, unto.hakkinen@stakes.fi Work in progress. Please do not quote.

2 Introduction Health care expenditure have usually been seen as a function of the size of population, its age composition and age/-sex specific utilisation rates. According to this "naive" approach health expenditure will increase when population increases in size or people move from an age group of lower health expenditure to an age group of higher expenditure. This view has been widely challenged. In a seminal study Zweifel, Felder and Meier argued that main demographic driver of health-care costs may be time to death rather than age[1]. The relationship between age, time to death and health expenditure has been extensively studied in recent years, using data from different countries[2-6]. However, its precise effects are not clear and due to different methodologies of data gathering, calculation and coverage of cost the results varies significantly[7]. More over, it is currently well recognized that health expenditure is not only determined by aging or demographic factors such as but by complex series of demand and supply side factors such as health status of population, economic growth, new technologies and medical progress, organisation and financing of the health care system and health care resources[7]. For example, two recent projections of health expenditure assumed that income is the main non-demographic driver of future health expenditure [7, 8]. This is based on results of numerous studies on income elasticity of health care expenditure. But there are at least two concerns for this. First, according to the results of studies income elasticity tends to increase with level of aggregation, i.e. the higher the level of aggregation, the higher the elasticity. In the studies using individual level data income elasticities are usually small or negative. The high income elasticities (above unity) found in macro studies may result from a failure to control many important factors such as prices and health status. It can be also assumed that at aggregate level of income is closely related to use new technology and products. For example, in Finland as in other countries expenditure on pharmaceuticals has increased more rapidly during the last decade than other health expenditure or GDP. The main driver for the increase has been the introduction new and more expensive medicines. The relationship between health expenditure and income is important also from a broader perspective. Many developed countries finance the majority of essential health services from public finance sources and endorse the equity principle that these services ought to be allocated on the basis of need, and not on the basis of willingness to pay or ability to pay for services. Willingness or ability to pay is usually measured by income. This paper has two aims. First we revisit the debate on originally introduced by Zweifel, Feldner and Meier in the 'red herrings" [1], i.e. the claim that population aging will not have a significant impact on health care expenditure(hce), using a Finnish data. As in a Swiss study [5], we decompose HCE into several components and include both survivors and deceased individual into analysis. We also compare the predictions of health expenditure based on model that take into account proximity of death with predictions of a naïve model, which includes only age and gender and their interactions. Secondly, we extend our analysis to include income as an explanatory variable. This allows us to analyse at individual level the effect of income on health expenditure and relate it also to proximity of death. In addition, for non institutionalized individuals we have information on need for services (morbidity). Thus we will evaluate the equity aspects of health care utilization. The paper is organized as follow. In next section, we briefly describe the Finnish health care system. This is followed by a description of data and methods. A results section starts with analysis of individual components of health expenditure and finishes with analysis of total expenditure. 2

3 After that we use our results in projecting health expenditure. This follows an analysis on the effect of income and the final section discusses the findings and concludes. Finnish health care system In its institutional structure, financing and goals, the Finnish health care system is closest to those of other Nordic countries and the UK, in that it covers the whole population and its services are mainly produced by the public sector and financed through general taxation. Finland s 432 municipalities (local government authorities) are responsible for providing municipal health services. Municipal taxes, state subsidies and user charges finance the municipal health services. Municipally provided services include primary (mostly services produced in health centres) and specialist health care. Municipalities are responsible for other basic services, such as nursing homes and other social services for the elderly, child day care, social assistance and basic education. In addition National Health Insurance (NHI) subsidises the use of specific private health services and outpatient medicines. In Finland, the hospitals (i.e. university, central and regional hospitals) owned by hospital districts (federations of municipalities) produce most of specialised (outpatient and inpatient services). However, some municipalities produce some specialist services themselves in their own health centres. In addition, some health centres gives also long term-care services. Municipalities (public hospitals and nursing homes) pay for the drug expenditures of in-patient care, while in outpatient care, both the patients and the NHI contribute to the expenditure. User charges and cost-sharing play a prominent role in funding health services. Cost-sharing is lower for municipal services than for the privately provided services and products, particularly prescribed medicines, eligible for NHI reimbursement. For example, user charges represent about 10% of the total cost of services provided by health centres, about 5% of the cost of hospital services but about 35% of the cost of drugs prescribed outside hospitals. If patients need long-term care in a health centre ward or an old peoples home, up to 80% of their income will be charged for their accommodation, provided a (low) minimum amount is available for their own use. 2. Data and methods Data As a data source, we used a 40 % sample of the Finnish population aged 65 and over at the end of 1997 (N = ). Follow-up for death, hospital and medication use was until the end of year With an individual level unique identification code we linked data from Statistics Finland Population Registration, Finnish Hospital Discharge Register, Finnish Death Register, registers of the Social Insurance Institution (SII) and data from the Finnish Hospital Benchmarking Project. [9, 10]. The data included information on all use of inpatient services (both hospital and nursing home), outpatient services for somatic (i.e. non psychiatric ) specialized care, cost and use of outpatient prescribed medicines, socioeconomic status (e.g. family income linked directly from tax files), region (hospital district/hospital area) of the individual and day and causes of death. In addition, for non institutionalized persons we had information on morbidity. This is based in a SII records of all patients entitled to specially reimbursed, free or nearly free medicines, for certain chronic (totally about 45) diseases. The utilisation information of hospital inpatient care was converted into costs using Finnish standard costs for different types of bed days[11].the somatic and other acute hospital inpatient 3

4 admissions were first grouped according to national version of Nordic DRGs (NordDRG). The groups were converted into costs using average costs per inpatient day specific to each DRG. The outpatient visits in specialist hospitals we converted using average cost per visits specific to each speciality and type (emergency /elective) of visit. Our expenditure include total cost i.e. also the share paid by patients/consumers. In addition, our total (health) expenditure includes also expenditure on services for the elderly. We divided cost into four categories : 1. Somatic specialised care - All inpatient care given in specialised hospitals (except psychiatry) - Outpatient visits in specialised hospital - Acute (i.e. under 21 day) hospital inpatient care given in health centres 2. Health centre and psychiatric inpatient care - Health centre inpatient care (21 days-3 months), psychiatric inpatient care and inpatient rehabilitation given in specific institutions 3. Long-term care - All long-term care given in health centres, psychiatric institutions, nursing homes and other institutions giving 24-hour service 4. Prescribed medicines - all outpatient prescribed medicines reimbursed by NHI Altogether our data included information on 80 % of total expenditure of health care and services for the elderly. The most important excluded categories were visits to primary care as well as home care and home services. For these services we do not have national wide register data. Methods Most of earlier studies [1-4] on effects of proximity of death to health expenditure have used a data, in which the explanatory variable has been cost per a specific time interval (i.e. one month, two months, three months or a year) before the death (or end of follow up). The data have been based on several cost observations from the same person. The regressions have included many dummy variables representing time to death. This approach has raised methodological concerns about the multicolinearity between and endogenity of the explanatory variables [2, 12]. For example, the endogenity arises from the fact that time to death might be influenced by current and previous use of services (expenditure). Much of the methodological critique can be taken into account using an approach applied in two previous studies[5, 6], in which health expenditure in one year is studied and time to death is measured from that year as single explanatory variable. This approach can be extended to include also surviving patients, since concern has also been that the effect of age may be different for survivors than deceased. In addition, using the approach it is quite easy to do make health care expenditure projections, since the explanatory variable (annual health expenditure person years) is the same as available in routine statistics. In this study we calculated for each person health care expenditure for the year 1998 (Table 1). Since most of our standard cost information was from the year 2000 or 2001, they were deflated to 1998 prices using municipal health care price index. The costs of prescribed medicines were based on information on actual reimbursements at prevailing prices. Since we followed individuals until the end of the year 2002, we can evaluate how annual health expenditure can be explained by time of death during the next four years. 4

5 A two-part model (logit/porbit+ OLS) is a usual way to analyse health expenditure. However, in our case the sample included both long-term care (LTC) and non long-term care (non LTC) individuals. It can be assumed that age and time of death affects the expenditure of the two groups in a different way. In addition, we wanted to analyse separately the effect of proximity of dead to main components of health expenditure. Thus we ended up to estimating eight different models (Figure 1). The first model analyses the likelihood of being a LTC patient and the rest of models are estimated separately for LTC (model 2) patients and for non LTC individuals (models 3-8). For latter group 1 (non LTC) we applied a two part model, in which firsts parts (models 3,5 and 7) describes the likelihoods of the use the three main categories of health expenditure and second parts (models 4, 6 and 8) expenditure of those individuals who have used the services. It can be assumed that the three categories are to some extent related to each other (either complements or supplements). This can taken into account by applying SUR (seemingly unrelated) estimation. Since in this cases the samples are unbalanced (i.e. the equations have an unequal number of observations) we applied a procedure developed for Stata statistical software[13]. We used a linear functional form of expenditure, because it allows a simple calculation of expected costs 2. In the econometric analysis (N= ) we excluded those who died in 1998, since we want to have expenditure data for whole year. We estimated three (basic, naive and extended) specifications for each of the eight models. The basic specifications include four independent variables (age, gender (=1, if female), time to death (TTD, days from to death, maximum 1460 if survived until end of 2002), death (=1, if an individual died prior to the end of the year 2002). We do not have complete theoretical guidance on functional forms (square, cubic) of the four variables and their interactions. Thus we performed a specific-to-general specification search with the aim of finding a model that fits the data well and in where the parameters are significant. We started from a specification using the four variables in their original forms. We ended up to different specifications for each eight models. The naive specifications use the same variables than the basic specifications except those which are related to death. The extended specifications uses all variables included in basic model as well as regional variables (25 dummy variables describing the hospital district where the individual lives) and for non LTC individuals morbidity variables (18 dummy variables describing chronic illness). In addition, the extended specifications include family income (divided by OECD consumption unit scale) and its significant transformations and interactions with the four factors (age, gender, time to death and death). The estimated model coefficients are very difficult to interpret, since the same factor is included in many variables and in different parts of the models (Tables 2-6). Thus we illustrated the results by calculating the expected expenditure for the individuals died in different years according to age. In these calculations the time to death was fixed to average time for each year of death (for example to 177 for those died in 1999). The expected expenditure for both genders were calculated from gender specific estimates by weighting the share of gender in each age groups. 1 We did not analyse separately the expenditure categories for LTC patients, since over 80 % of their expenditure were allocated to long -term care services. 2 A Box-Cox test rejected both linear and logarithmic functional form for all expenditures categories (models 2,4, 6 and 8). However, use of other than linear functional forms creates the problem of retransformation back to raw-scale expenditure [14]. An alternative will be to use generalised linear model [15] but this will complicate application of SUR estimation. Since our sample is large we can assume that linear functional form will produce robust results. 5

6 Results Long-term care (LTC) patients Although the share of LTC patients of population over 65 was 7% they used 55% of total expenditure. Age has an important positive and increasing effect on the probability of being a LTC user (Table 2). But the share is also related to time to death: those died in 1999 had a 10 percentage points higher probability to be LTC user than those survived. The relative difference between the two groups increases to 30 percentage points among those aged 90. After this age the difference reduces slightly. As can be seen from figure 2 the naïve model overestimates the age relationships: its curve increases considerable faster than the curves where time of death is fixed. The females had a higher risk to use LTC than males and the difference between genders widens as age increases. Annual average total cost of LTC patients were euros i.e euros per month. When time of death was fixed these costs were in U shaped relation with age minimum being at the age of 75 (Figure 3). For example, annual expenditure of a 90 years old LTC patient died in 2001 was about 1800 euros (7%) higher than expenditure of a 75 year old LTC patient who died in the same year. However, the again the time to death as well as gender have greater effect of expenditure than aging. The annual expenditure of an LTC patient died in 1999 was 9000 euros higher than those survived. When time of death was taken into account females expenditure was about 2400 euros higher than males, which 7-11% higher depending the age as well as time of death of a LTC patient (Figure 4). Non long-term care (non LTC )individuals In somatic specialized care the proximity of death determines very strongly the relationship between age and expenditure. The expenditure is highest among persons died in 1999 (Figure 5). In this group the expenditure also most clearly decline with age. The proximity of dead affects also to expenditure of those persons, who have lived two years (died in 2002). The expenditure of individuals lived 3 and 4 years are clearly lower than those who have died before them but even among these groups expenditure reduces when age increases. Only among the survivors expenditure increases with respect to age until to age 85. The expenditure of a 65 year old person died in 1999 is about 6.5 times higher than expenditure of a person with same age but who have survived. When age increases this share decreases and is 2.5 times among 90 years old persons. In somatic specialized care the naïve model gives a totally different picture on relations ship between age and expenditure than models which take into account the proximity of death. In inpatient care of health centres the expenditure increases with age but again the inclusion of proximity of death diminishes the relationship (Figure 6). Age reduce most clearly the expenditure on of prescribed medicines (Figure 7). This effect is strongest among those who have died in the following (1999) year. On the other hand, among the survivors expenditure on prescribed medicine increase along with age until age of 80 years and after that age start to reduce. Total expenditure 6

7 Figures 8 and 9 describes the calculations for expected total expenditure including both LTC and non LTC individuals. They are based on models 1-8 by estimating first gender specific costs of the expenditure categories for person died in different years according to age. After that we calculated gender specific total cost and finally the total cost for both genders by weighting their share in the sample. Since over half of total expenditure is devoted to LTC patient the figure 8 is quite similar as the figure 2 describing the probability of use of LTC. The probability for LTC was higher among females and thus their total costs were also higher than males. The difference between the genders widens as age increases but also this reflects the gender differences in probability of being a LTC user. Projections for health expenditure We combined our gender specific age-expenditure profiles and population forecast (by Statistics of Finland) to estimate the purely demographic impact on health-per capita expenditure (over 65 population) for the years 2016 and As a first step, we calculated gender specific projections (expenditure per individual) for each expenditure category according to survival status (i.e. died in first year, second year, third year fourth year or survived in the end of forth year) by one-year age groups. The population forecast was also made gender specific by one-year age groups and thus enables us to estimate how many persons in each age group will die within each following years. Thus the projection could be derived by multiplying the estimated number of persons in each age and gender specific survival status categories by their respective expenditure and summing all first according to each expenditure category and finally adding all to total expenditure together. Our econometric was based on individual who lived in the beginning of the year 1999 i.e. we excluded those who died during the year However this group consumed about 14 % of total expenditure (Table 1). In order to take this account we calculated from our sample also for them gender specific expenditures (per individual) by expenditure categories and age. We multiplied these (annual average expenditure per individual) by number of person who is projected to die in next year in each age and gender group. Thus the expenditure for those died in the same year was calculated by summing up these age and gender specific expenditures. We made projections using the naive 3 approach and the basic specification, which takes in account the proximinity of death. In previous section we found that our total expenditure is strongly depending on the probability for the use of long term care (LTC) services. Thus we made also a projection which (in addition to proximity of death) assumes that the probability to use LTC will change later by three years. We assume, for example that probability of a 70 year old female (in a each survival category) in 2016 and 2036 is same as respective probability of a 67 year old female in Table 6 shows the results of the three projections for the years 2016 and According to the naive projection expenditure will increase annually 2.2 percent by the year The projection based on basic specification (taking into account the proximity of death) reduce the average annual growth into 1.9 % and gives 12% lower projection for total expenditure than the naive approach. The difference between the two projections is highest in the expenditure on inpatient care of health centres and psychiatry (23%) and in somatic specialized care (14%) and smallest in prescribed medicines (3%). The assumption concerning the probability of the use of LTC seem to be very 3 The naive projection was based on simple age, gender specific expenditure (per person) estimates derived form our sample. We calculated the naive projections also using estimation results and age, gender specific expenditures for those who died in These both approaches give similar projections, 7

8 crucial for the results of the projections. If the user of LTC services can be changed three years later (for example by decreasing their dependency) the projected expenditures will decrease about 25% compared to naive model. Compared with basic specification the assumed change will decrease the expenditure on LTC by 672 million euros (i.e. 12%) and increase the expenditure on other cost categories totally by 55 million euros (2%). Effect of income Family income turn out to be significant all (except model 6) of the estimated extended models. Also the coefficients of income are very difficult to interpret, since the variable is included in many variables and in different part of models. In order to illustrate income effects we calculated expected expenditures for three type of individuals: i) for a person whose family income is at the value of first decile in income distribution (low income), ii) for a person whose family income is at the value of 9th decile (high income); and iii) for a person who had a average income (mean income). The expected expenditures are calculated from expanded specifications in a similar way as for person died in different years and survivors (as explained earlier in method section) by fixing income to three alternatives 4 values. The expected expenditure for both genders were calculated from gender specific estimates by weighting the share of genders in each age and income groups. In addition, we need to fix also the regional and morbidity (for non LTC) variables. We fixed all regional variables to zero. Thus we calculated expected expenditure to a person who lived in southwest part (Hospital district of "Varsinais-Suomi") of Finland (the reference region). For non LTC individuals we fixed all except one (coronary heart disease) morbidity variable into zero i.e the expenditures have been calculated to a person who has coronary heart disease but no other chronic illness. The total expenditure of LTC patients is concentrated to low income patients. As can be seen from figure 10 the absolute differences (in term of euros) between high and low income individuals even increases as age increases both among survival and those died in It should be noted that we did not have possibility to adjust these figures by need (such dependency, health status). Almost all persons who had used LTC services are interpreted as living "alone" and thus their family income equals their own income. This reflects to a great extent their early work position and work history. The poorest group included persons who had been in early retirement (because of disability) and thus it is quite possible that among LTC individuals income is highly and negatively correlated with high dependency and poor health. Among non LTC individuals income has a positive effect of expenditure in somatic specialised care and use prescribed medicines (Figures 11 and 12). However, in somatic care effect is quite small: for example expenditure for an 75 years old high income person who died in 1999 is about 200 euros (8,5%) higher than a low income person in same age died in Among the survived persons with same age the relative difference between the extreme income groups was the same but absolute the difference smaller (80 euros). Only among the use of prescribed medicines we found differences on the effect of income between deceased and survived. For an 75 years old person the correspondence difference between the two extreme groups was about 149 euros (19%) among those died in 1999 and 46 euros (9%) among those survived 5. 4 The value at first and ninth deciles mean income was calculated for LTC patients (Figure 10) from whole sample where as for non LTC patients (Figures 11 12) we used the non LTC sample 5 Relatively differences are higher when they are estimated for individuals with no chronic illness. For a 75 year old persons with no chronic heart disease (all morbidity variable fixed at zero) the correspondent difference in expenditure 8

9 Conclusion and discussion In this study we have analysed the relationship between health expenditure, age, and time of death in a representative and much larger sample than have been available in previous studies. In addition, we were able to analyse the expenditure on whole range of services in health care as well as care of the elderly. According to our results total expenditure on health care and care of the elderly increases somewhat with age but the relationship is not so clear as usually assumed when a naïve model is used in projections. Our conclusion that age still has an effect on expenditure is due to the fact that we have included use of LTC services in our analysis. Among non LTC persons we found a clear positive relationship between expenditure and age only in health centre and psychiatric inpatient care and even in these services proximity of death is strongly positively related to individual's expenditure. On the other hand, in somatic care and prescribed drugs the expenditure among the deceased clearly decreased with age among deceased individuals. Compared with the naive approach the projection that takes into account the proximity of death gives 12 % lower projection for total expenditure by the year This is somewhat lower difference than found in previous studies in US[4] and German [16], although the differences in results may be due to differences in age structure and population projection between the countries. However, one explanation for thee differences is the relatively strong importance long-term care services in Finland. In 1998 the share of LTC patients of population over 65 was 7% and they used 55% of total expenditure. Our projections indicate clearly that future health policy should concentrate to actions which maintain activities of daily living of elderly people and prevent longterm institutional care. From point of view of public expenditure the future is not so alarming if current financing system in long-term care prevails. User charges of LTC patients are income related and richer patients pay about 80% of their income to municipalities (providers). This can to some extent explain our results that long-term care is clearly correlated with low income, since there are economic incentives for high income individuals (and their next in kin) to avoid public funded nursing homes. As the income level among future patients will be higher there would be more those people for whom it might be cheaper to take a private nursing home (in stead of public) with no public subsidise. We do not find any strong positive association between income and expenditure which can be interpreted that equity targets are realised at least satisfactory among the care of the elderly. However, income was related to expenditure use prescribed medicines, in which cost sharing is relatively high. This can be seen an indication of inequity in use, specially if the higher expenditure is associated with use of more effective medicines. In addition, it also may indicate that income is related to use new of technologies and products. According to our projection aging and demographic factors will increase use of prescribed medicines less than other expenditure categories. On the other hand, during the years the expenditure on prescribed medicines have annually been in average 7 % (at constant prices ), of which annually only 0.7% (10 % per cent) is related to changes in demographic factors (age and gender structure) [17]. All these clearly emphasise the fact that even in future health care expenditure might be driven more by medical technology than demographic factors as often claimed in public discussion. on prescribed medicines between the two extreme (high and low income) is 139 euros ( 33 %) among those died in 1999 and 46 euros (23 %) among those survived. 9

10 REFERENCES 1. Zweifel P, Felder S, Meiers M.Aging of population and health care expenditure: A red herring? Health Economics 1999; 8;6: Seshamani M, Gray AM.Aging and health-care expenditure: the red herring argument revisited. Health Economics 2004; 13;4: Seshamani M, Gray AM.A longitudinal study of the effects of age and time to death on hospital costs. Journal of Health Economics 2004; 23: Stearns SC, Norton EC.Time to include time to death? The future of health care expenditure? Health Economics 2004; 13;4: Werblow A, Felder S, Zweifel P. Population aging and health care expenditure: a school of 'red herrings'? vol. Working Paper No.11 June 2005: Faculty of Economics and Management Magdeburg; Zweifel P, Felder S, Werblow A.Population aging and health care expenditure: New evidence on the "red herring". Geneva Papers on Risk and Insurance: Issues and Practices 2004; 29;4: Comission EPCaE. The impact of aging on pubic policy for 25EU Member states on pensions, health care, long term-care, education and unemployment transfers ( ); OECD. Projecting OECD health and long-term expenditures: What are the main drivers? vol. working papers No.47; Järvelin J, Linna M, Häkkinen U.The Hospital Benchmarking project: Productivity Information for Hospital Comparison. Dialogi 2003;1B,: 22-28,. 10. Junnila M (ed.): Sairaaloiden tuottavuus. Benchmarking-tietojen käyttö erikoissairaanhoidon toiminnan suunnitelussa, seurannassa ja arvioinnissa. Helsinki: Stakes; Hujanen T. Terveydenhuollon yksikkökustannukset Suomessa vuonna vol. 1/2003. Helsinki: Stakes; Salas C, Raftery J.Econometric issues in testing the age neutrality of health care expenditure. Health Economics Letters 2001; 10: McDowell A.From the help desk: Seemingly unrelated regression with unbalanced equations. The Stata Journal 2004; 4;4: Manning WG.The logged dependent variable,heteroscedasticity and retransformation problem. Journal of Health Economics 1998; 17: Manning WG, Mullahy J.Estimating log models:to transform or not to transfom? Journal of Health Economics 2001; 20: Breyer F, Felder S. Life expectancy and health care expenditures: a new calculations using the cost of dying. In.; Hujanen T, Pekurinen M, Häkkinen U. Terveydenhuollon ja vanhustenhuollon alueellinen tarve ja menot : Stakes;

11 Long-term care individuals Model 2. OLS, N=21717 Explanatory variable: total expenditure Model 1, Logit N= likelihood of use of long term care (LTC) Pr(LTC>0). -Non long-term care individual Model 3. logit N= likelihood of use of somatic care PR(SOM>O) Model 4.SUR N= (Those who had used somatic care) Explanatory variable : expenditure on somatic care Model 5, Logit N= likelihood of use of health centre or psychiatric inpatient care PR (HCP>0) Model 6. SUR N=15977, ( Those who had used health centre or psychiatric inpatient care ) Explanatory variable: expenditure on health centre or psychiatric care Model 7, Logit N=260951, likelihood of use of prescribed medicines PR (PRE>0) Model 8. SUR N= (Those who had used prescribed medicines), Explanatory variable: expenditure on prescribed medicines Figure 1. The structure of the eight part model Figure 2. Probabilty of use of LTC of deceased and survived as a funtion of age, both genders (based on model 1, basic and naive specificatinos) Probabilty of use of LTC Died in 1999 Died in 2000 Died in 2001 Died in 2002 Survivors Naive Age 11

12 Figure 3. Expected total expenditure of deceased and survived LTC patients as a function of age, both genders (based on model 2, basic and naive specifications) Annual expenditure per individual ( ) Died in 1999 Died in 2000 Died in 2001 Died in 2002 Survivors Naive AGE Figure 4. Expected total expenditure of deceased (in 1999) and survived LTC patients as funtion of age, males and females (based on model 2, basic and naive spefications) Annual expenditure per individual ( ) Died in 1999 male Died in 1999 male Survivors male Survivors female Naive female Naive female Age 12

13 Figure 5. Expected expenditure on somatic specialised care of deceased and survived non LTC individuals as function of age, both genders (based on models 3 and 4, basic and naive specifications) Annual expenditure per individual ( ) Age Died in 1999 Died in 2000 Died in 2001 Died in 2002 Survivors Naive 13

14 1400 Figure 6. Expected expenditure on health centre and psychiatric inpatient care of deceased and survived non LTC individuals as funtion of age, both genders (based on models 5 and 6, basic and naive specifications) Annual expenditure per individual ( ) Age Died in 1999 Died in 2000 Died in 2001 Died in 2002 Survivors Naive Figure 7. Expected expenditure on prescribed medicines for deceased and survived non LTC individuals as function of age, both genders (based on models 7 and 8, basic and naive specifications) Annual expenditure per individual ( ) Age Died in 1999 Died in 2000 Died in 2001 Died in 2002 Survivors Naive 14

15 Figure 8. Expected total expenditure of deceased and survived individuals (both LTC and non LTC) as funtion of age, both genders( based on modes 1-8, basic and naive spesifications) Annual expenditure per individual ( ) Age Died in 1999 Died in 2000 Died in 2001 Died in 2002 Survivors Naive 15

16 Figure 9. Expected total expenditure of deceased (in 1999) and survived individuals (both LTC and non LTC) as function of age, males and females (based on models 1-8, basic and naive specifications) Annual expenditure per individual ( ) Died in 1999 male Died in 1999 male Survivors male Survivors female Naive female Naive female Age Figure 10. Expected total expenditure on LTC (and other expenditure items used by LTC patients) of deceased in 1999 and survived patients as function of age according to income level (based on models 1 and 2, extended specifications) Annual expenditure per indivdual ( ) low income survivors mean income survivors high income survivors low income died in 1999 mean income died in 1999 high income died in Age 16

17 Figure 11. Expected expenditure on somatic specialised care of deceased in 1999 and survived individuals as function of age according to income level (based on models 3 and 4, extended specifications and estimated for coronary heart disease patients) 4500 Annual expenditure per indivdual ( ) low income survivors mean income survivors high income survivors low income died in 1999 mean income died in 1999 high income died in Age Figure 12. Expected expenditure on prescribed medicines of deceased in 1999 and survived individuals as function of age according to income level (based on models 8 and 9, extended specifications and estimated for coronary heart disease patients) 1200 Annual expenditure per individual low income survivors mean income survivors high income survivors low income died in 1999 mean income died in 1999 high income died in Age 17

18 Table 1 Expenditure on health care and care of the elderly among population over 65 in 1998 according to year of death (based on the sample and calculated at whole country level) Expenditure item Expenditure of all individuals Died in 1998 Died in 1999 Died in 2000 Died in 2001 Died in 2003 Survivours (after 2002) Euro million % of total expenditure % expenditure % expenditure % expenditure % expenditure % expenditure % expenditure Non-long-term care indviduals Total expenditure, of which: Somatic specialised care Health centre and psychiatric inpatien Prescribed medicines Long-term care individuals Total expenditure, of which: long-term care Somatic specialised care Health centre and psychiatric inpatien Prescribed medicines Total expenditure of all invividuals

19 Table.2. Two part estimation of total expenditure of LTC patients Model 1. Logit (N= ),Pr(LRC>0) Model 2. OLS (N=21717), Total expenditure among LTC users Variable Basic spefication Naive specification Extended specification Basic spefication Naive specification Extended specification coeff z-score coeff z-score coeff z-score coeff t-score coeff t-score coeff t-score CONSTANT AGE AGE2/ AGE3 / FEMALE FEMALE*AGE FEMALE*AGE2/ DEATH DEATH*AGE TTD TTD TTD*FEMALE INCOME INCOME INCOME*AGE INCOME*FEMALE INCOME*DEATH Regional variables (reference: Varsinais- Suomi) Satakunta Kanta-Häme Pirkanmaa Päijät-Häme Kymeenlaakso Elelä-Karjala Etelä-Savo Itä-Savo Pohjois-Karjala Keski-Suomi Etelä-Pohjanmaa Vaasa Keski-Pohjanmaa Pohjois-Pohjanmaa Lappi Kainuu Länsi-Pohja Helsinki Hyvinkää region Porvoo region Länsi-Uusimaa Jorvi Peijas Lohja region Pohjois-Savo R 2 /Psedo R

20 Table 3:Two part estimation of somatic spesialised hospital care (not LTC individuals) Model 3, Logit (N= ) Model 4,SUR (N= ) Variable Basic spefication Naive specification Extended specification Basic spefication Naive specification Extended specification coeff z-score coeff z-score coeff z-score coeff z-score coeff z-score coeff z-score CONSTANT AGE AGE2/ AGE3 / FEMALE FEMALE*AGE FEMALE*AGE2/10 00 DEATH DEATH*AGE TTD TTD2 7.40E E TTD*FEMALE INCOME 3.75E INCOME2-2.99E INCOME*AGE INCOME*FEMALE -5.11E-06-3 INCOME*DEATH ÍNCOME*TTD -1.06E Chronic illness variables: Diabetes Thyroid insuffiency Anemia Parkinson's disease Epilepsy Severe mental disorder Mental retardation Claucoma Breast cancer Prostatic cancer Leukaemia Cardiac insuffiency Reumathoid arthritis Asthma Hypertension Coronary heart disease Arrhythmias Ulcerative colitis and Cronh's disese Regional variables (reference: Varsinais-Suomi) Satakunta Kanta-Häme Pirkanmaa Päijät-Häme Kymeenlaakso Elelä-Karjala Etelä-Savo Itä-Savo Pohjois-Karjala Keski-Suomi Etelä-Pohjanmaa Vaasa Keski-Pohjanmaa Pohjois-Pohjanmaa Lappi Kainuu Länsi-Pohja Helsinki Hyvinkää region Porvoo region Länsi-Uusimaa Jorvi Peijas Lohja region Pohjois-Savo Psedo R

21 Table 4.Two part estimation of health centre and psychiatric inpatient care(non LTC individuals) Model 5,Logit (N=260951) Model 6.SUR (N=15 977) Variable Basic spefication Naive specification Extended specification Basic spefication Naive specification Extended specification coeff z-score coeff z-score coeff z-score coeff z-score coeff z-score coeff z-score CONSTANT AGE AGE2/ AGE3 /1000 FEMALE FEMALE*AGE FEMALE*AGE2/ DEATH DEATH*AGE TTD TTD2 1.04E E TTD*FEMALE INCOME INCOME2 INCOME*AGE 5.79E INCOME*FEMALE INCOME*DEATH ÍNCOME*TTD Chronic illness variables: Diabetes Thyroid insuffiency Anemia Parkinson's disease Epilepsy Severe mental disorder Mental retardation Claucoma Breast cancer Prostatic cancer Leukaemia Cardiac insuffiency Reumathoid arthritis Asthma Hypertension Coronary heart disease Arrhythmias Ulcerative colitis and Cronh's disese Regional variables (reference: Varsinais- Suomi) Satakunta Kanta-Häme Pirkanmaa Päijät-Häme Kymeenlaakso Elelä-Karjala Etelä-Savo Itä-Savo Pohjois-Karjala Keski-Suomi Etelä-Pohjanmaa Vaasa Keski-Pohjanmaa Pohjois-Pohjanmaa Lappi Kainuu Länsi-Pohja Helsinki Hyvinkää region Porvoo region Länsi-Uusimaa Jorvi Peijas Lohja region Pohjois-Savo Psedo R

22 Table 5.Two part estimation on prescribed medicine (non LTC individuals) Model 7, Logit (N=260951) Model 8,SUR (N= ) Variable Basic spefication Naive specification Extended specification Basic spefication Naive specification Extended specificatio coeff z-score coeff z-score coeff z-score coeff z-score coeff z-score coeff z-score CONSTANT AGE AGE2/ AGE3 / FEMALE FEMALE*AGE FEMALE*AGE2/ DEATH DEATH*AGE TTD TTD E TTD*FEMALE INCOME 5.07E INCOME2-6.22E E INCOME*AGE INCOME*FEMALE INCOME*DEATH ÍNCOME*TTD -8.90E Chronic illness variables: Diabetes Thyroid insuffiency Anemia Parkinson's disease Epilepsy Severe mental disorder Mental retardation Claucoma Breast cancer Prostatic cancer Leukaemia Cardiac insuffiency Reumathoid arthritis Asthma Hypertension Coronary heart disease Arrhythmias Ulcerative colitis and Cronh's disese Regional variables (reference: Varsinais- Suomi) Satakunta -1.71E Kanta-Häme Pirkanmaa Päijät-Häme Kymeenlaakso Elelä-Karjala Etelä-Savo Itä-Savo Pohjois-Karjala Keski-Suomi Etelä-Pohjanmaa Vaasa Keski-Pohjanmaa Pohjois-Pohjanmaa Lappi Kainuu Länsi-Pohja Helsinki Hyvinkää region Porvoo region Länsi-Uusimaa Jorvi Peijas Lohja region Pohjois-Savo Psedo R