Chronic conditions and the risk of long-term institutionalization among older people

Abstract

Background: As the public expenditure on long-term care is likely to increase with the ageing of the population, identifying chronic medical conditions associated with the risk of long-term institutionalization is of particular interest. However, there is little systematic evidence showing how chronic medical conditions, other than dementia, affect the risk of entering into institutional care in the general older population. Methods: We used population-based follow-up data on Finnish older people aged 65 and over (n = 280 722), to estimate the impact of different chronic conditions on the risk of long-term institutionalization. Furthermore, we analysed which chronic conditions were more strongly associated with the risk of institutionalization than with the risk of death without institutionalization. Cox proportional hazard regression models were used. Results: Our results showed that dementia, Parkinson's disease, stroke, depressive symptoms, other mental health problems, hip fracture and diabetes were strongly associated with increased risk of long-term institutionalization, independent of socio-demographic confounders and the presence of other chronic conditions. All these conditions raised the risk of institutionalization by 50% or more. Dementia, Parkinson's disease, stroke and mental health problems were more strongly associated with the risk of institutionalization than with the risk of death without institutionalization. Conclusions: Overall, these results show that the future demand for institutional care depends not only on the ageing of the population but also on the development of the prevalence and severity of chronic conditions associated with institutionalization.

chronic disease, follow-up study, institutionalization, older adults, population based

Introduction

As the public expenditure on long-term care is likely to increase with the ageing of the population, a better understanding of the factors related to long-term institutional care is of particular interest. Several population-based prospective studies have shown that functional disability^1–6 and cognitive impairment^1,,2 are associated with institutionalization, but systematic evidence on the effects of different chronic diseases on institutionalization is scarce. Dementia has been shown to increase the risk of institutionalization,⁷ independent of comorbid conditions⁸ and functional disabilities.^5,,6,9 Furthermore, population-based cross-sectional studies indicate that neurological diseases in general,¹⁰ and some specific neurological diseases such as Parkinson's disease¹¹ and stroke,¹² are associated with living in an institution. However, because of the cross-sectional design of these studies it is difficult to determine whether the presence of disease predates institutional entry or not. In addition, evidence on the effects of other chronic diseases on institutionalization among general older populations is rarely available, and the effects of different diseases have seldom been studied simultaneously. Furthermore, evidence of the effect of some medical conditions, such as hip fracture, is partly inconsistent.^5,,6

Using population-based survival data with continuous time scale of institutionalization, we assessed which chronic conditions were most strongly associated with long-term institutionalization. More specifically, we examined how different chronic conditions were associated with entry into institutional care, independent of socio-demographic confounders and other chronic conditions, and evaluated which chronic conditions were associated more strongly with the risk of institutionalization than with the risk of death without institutionalization.

Methods

Register-based data

The data were based on a 40% individual-level sample of the total Finnish population aged 65 and over on 31 December 1997 (301 263 persons), drawn from a population registration database at Statistics Finland using simple random sampling. These data are collected annually from different administrative records to provide Labour-Force Statistics,¹³ and they contain all persons living in Finland and detailed socio-demographic information. This baseline sample, already linked with dates of death, was linked with information on institutional care and prior hospital diagnoses provided by the National Research and Development Centre for Welfare and Health (STAKES), and with information from medication registers provided by the Social Insurance Institution. The data linkage was carried out at Statistics Finland using personal identification codes (TK 53-576-04 and TK 53-499-05). We excluded all those who were already institutionalized (5.86%) or who for some other reason did not reside in private households at baseline (0.96%). The effective study sample, representative of the total Finnish community-living older population, consisted of 280 722 persons, who were followed for first entry into long-term institutional care or death from 1 January 1998 to 30 September 2003.

Long-term institutional care

Long-term institutional care was defined as 24-hour care in nursing homes, service homes, hospitals and health centres lasting for over 90 days or confirmed by a long-term care decision. Long-term psychiatric care was included. The over-90-days criterion was met if a patient had stayed in the same institution or successively in different institutions for the time required. About 75% of first stays in long-term institutional care that started during the follow-up began in hospitals or health centres and 25% in nursing or service homes.

The information on long-term institutional care was based on the Client Censuses of Health Care (including hospitals and health centres), and on the Client Censuses of Social Care (including nursing and service homes), both of which were carried out at the end of every year from 1997 to 2003, and on the annual discharge data containing information on stays which were completed. The registers of Health Care have been collected since 1967¹⁴ and are regarded as very accurate, while those of Social Care are known to be less complete. Approximately 9% of the nursing and service homes providing 24-hour care did not participate in the Client Census of Social Care in 2003.¹⁵ However, it is very likely that the proportion of care episodes that were undetected from the Client Census is much smaller, as institutions not participating in the Census are likely to be small. It is also likely that the nursing and service home stays are somewhat better covered in the censuses than in the discharge data,¹⁶ which may have underestimated the number of short stays. We used both Client Censuses and discharge data to minimize under coverage in nursing homes, but recognize that the absolute level of institutionalization may be a slight underestimate in our study.

Chronic conditions

This study used 18 dichotomous indicators of chronic medical conditions, including cancer, diabetes, dementia, psychosis, depressive symptoms, other mental health disorders, Parkinson's disease, other neurological diseases, heart disease, stroke, chronic asthma or other similar chronic obstructive pulmonary diseases, other respiratory diseases, arthritis, osteoarthritis, hip fracture, other conditions related to accident or violence, other hospital diagnoses and other chronic diseases that give the right to reimbursement for drug costs (Appendix). We mainly used three register sources to assess chronic medical conditions: (i) the principal cause of hospitalization during 1996–97, (ii) the right to reimbursement for drug costs under the Special Refund Categories due to certain diagnosed chronic medical conditions during 1997 and (iii) purchase of prescription medication during 1996–97. The persons studied were categorized as having a chronic condition if they had it according to at least one of these sources.

The principal cause of hospitalization was based on the Tenth Revision of the International classification of diseases (ICD10),¹⁷ the right to reimbursement for drug costs under the Special Refund Categories was based on the Finnish disease classification of the Social Insurance Institution,¹⁸ and purchases of prescription medication were based on the Anatomical Therapeutic Chemical Classification (ATC).^19,,20 The Finnish disease classification was based on the drug reimbursement system of the Social Insurance Institution. Certain chronic medical conditions were reimbursed under the Special Refund Categories which covered 75% or 100% of the costs of a single drug purchase exceeding a fixed deductible of 4.20 euros.^18,,21 To receive reimbursement, the patient had to submit to the Social Insurance Institution a doctor's certificate stating the illness, its severity and the medication required to treat it. The patient's wealth, age or affiliation to other special groups did not affect reimbursement, but the severity of the illness did.²¹

Control variables

We used age, living arrangements, socio-economic measures, housing conditions, region of residence and urbanicity as control variables which were measured at baseline. Age and living arrangements are associated with both institutionalization¹ and health,^22,,23 and socio-economic characteristics and housing conditions were associated with institutionalization in our data. Region of residence and urbanicity were adjusted for to control for differences in the supply of and access to institutional care between the areas.

Living arrangement categories were: living with a spouse or partner, living alone and being married, living alone and being widowed, living alone and being divorced, living alone and being never married and living with others. The three educational categories were: tertiary education, intermediate education, and basic education or unknown. Disposable income of the household was adjusted for the number of persons in the household according to the OECD equivalence scale,²⁴ with the exception of children who were weighted as adults because of the data restrictions. Home ownership categories were: owners, renters, and others and unknown. The possession of a car was categorized: yes, no and missing. The house type categories were: detached house, semi-detached house, apartment house with lift, apartment house without lift and other. The level of equipment in a dwelling was categorized into three categories: well-equipped, poorly-equipped and very poorly-equipped. A dwelling was regarded as well-equipped if it had piped water, sewer, hot water, flush toilet, washing facilities (shower/bath/sauna) and central or fixed electric heating, as poorly-equipped if it lacked washing facilities or central or fixed electric heating, and as very poorly-equipped if it lacked piped water, sewer, hot water or flush toilet.

Region of residence was categorized into 20 official regions (NUTS3), with the exception of the region of Uusimaa which was divided into three parts (Helsinki, the metropolitan area, and the rest of Uusimaa), and the Åland Islands which were combined with Southwest Finland. The urbanicity was based on the proportion of people living in different built-up areas and the population of the largest built-up area. The municipality was categorized as urban if at least 90% of the population lived in built-up areas and the largest built-up area had at least 15 000 residents, as semi-urban if 60–90% lived in built-up-areas and the largest built-up area had 4000–15 000 residents, and as rural if under 60% lived in built-up-areas and the largest built-up area had under 15 000 residents or if 60–90% lived in built-up-areas and the largest built-up area had under 4000 residents. A built-up area was defined as a group of houses with at least 200 residents and where the distance between the houses did not normally exceed 200 m. The distribution of the study cohort by the control variables (expect NUTS3) is presented in table 1.

Table 1

The distribution of Finnish community-living older women and men by socio-demographic characteristics

	Women Distribution (%)	Men Distribution (%)
Mean age (SD)	74.2 (6.7)	72.6 (6.1)
Living arrangements
Living with spouse or partner	36.4	72.0
Living alone/married	0.9	1.7
Living alone/widowed	34.6	10.3
Living alone/divorced	6.6	4.7
Living alone/never married	7.9	5.2
Living with others	13.6	6.1
Income
5. Quintile (highest)	17.5	23.9
4. Quintile	18.3	23.1
3. Quintile	20.1	21.0
2. Quintile	20.8	17.3
1. Quintile (lowest)	23.4	14.7
Education
Tertiary	8.0	13.6
Intermediate	13.6	12.3
Basic or unknown	78.4	74.1
Home ownership
Owner	78.1	83.8
Renter	18.0	12.7
Other or unknown	3.8	3.5
Possession of car
Yes	8.6	58.6
No	90.9	40.8
Missing	0.5	0.6
House type
Detached house	42.7	55.7
Semi-detached house	11.9	10.7
Apartment house with lift	23.6	16.9
Apartment house without lift	19.7	14.5
Other	2.2	2.2
Level of equipment in dwelling
Well equipped	81.6	79.0
Poorly equipped	8.3	8.9
Very poorly equipped	10.1	12.1
Urbanicity
Urban	55.8	51.2
Semi-urban	15.9	17.1
Rural	28.2	31.8
All	100.0	100.0
N	172 248	108 474

Note: Finnish regions excluded from the table

Statistical methods

The Cox proportional hazards regression models were used to estimate the determinants of entry into institutional care. Time to first entry was measured in days. A study person was censored at the time of death or at the end of the follow-up. Separate Cox regression models were fitted to estimate the determinants of death without entering into institutional care. A study person was censored at the time of institutionalization. All statistical analyses were performed with StataSE 8²⁵ separately for men and women, as diabetes, psychoses, depressive symptoms, other neurological diseases, stroke, chronic asthma, arthritis, and conditions related to accidents or violence were differently associated with the risk of institutionalization for the sexes.

Results

The impact of chronic conditions on institutionalization

Almost 15% of older women and 10% of older men entered into long-term institutional care during the follow-up (table 2). Entering institutional care was associated with several chronic conditions. Older people with dementia, Parkinson's disease and hip fracture had the highest rates of institutionalization (women: 70%, 40% and 39%, men: 55%, 30% and 32%).

Table 2

Prevalence of chronic conditions, proportions entering into institutional care and dying without institutionalisation (January 1998 to September 2003), and relative institutionalization rates^a (and their 95 confidence intervals) by chronic condition, Finnish community-living women and men aged 65 and over

	Women			Institutionalization		Men			Institutionalization
	Prevalence (%)	Entering institution (%)	Dying without institutionalization (%)	Hazard ratio	95% confidence interval	Prevalence (%)	Entering institution (%)	Dying without institutionalization (%)	Hazard ratio	95% confidence interval
Cancer	4.1	17.0	31.8	1.24	(1.17–1.32)	5.1	13.6	48.5	1.35	(1.25–1.45)
Diabetes	10.1	21.7	25.0	1.52	(1.46–1.57)	10.1	15.2	33.1	1.66	(1.57–1.75)
Dementia	0.6	70.2	22.4	4.38	(4.06–4.72)	0.6	55.3	39.1	4.20	(3.73–4.72)
Psychosis	2.9	28.6	17.3	1.95	(1.84–2.07)	1.8	19.4	28.8	1.40	(1.26–1.56)
Depressive symptoms	10.7	27.8	18.6	1.59	(1.54–1.64)	6.5	21.3	31.1	1.48	(1.39–1.57)
Other mental health disorders	5.4	33.4	21.3	1.67	(1.61–1.74)	4.3	26.3	36.5	1.74	(1.63–1.86)
Parkinson's disease	1.7	40.0	20.1	2.15	(2.02–2.28)	1.9	29.5	34.9	2.40	(2.20–2.62)
Other neurological diseases	4.0	24.6	19.9	1.30	(1.23–1.36)	4.7	17.9	31.3	1.40	(1.30–1.50)
Heart disease	26.1	20.2	23.4	1.08	(1.05–1.11)	29.7	11.7	31.6	1.05	(1.01–1.09)
Stroke	1.6	35.2	26.0	1.93	(1.80–2.06)	2.5	26.6	32.8	2.23	(2.06–2.41)
Chronic asthma and COPD	6.6	13.8	18.5	1.00	(0.95–1.05)	7.7	10.5	34.4	1.09	(1.02–1.17)
Other respiratory diseases	2.9	26.0	31.3	1.23	(1.16–1.30)	4.1	18.4	44.8	1.33	(1.24–1.44)
Arthritis	4.8	19.0	20.3	1.39	(1.32–1.47)	2.5	11.0	31.2	1.16	(1.04–1.31)
Osteoarthritis	2.9	17.2	12.7	1.07	(1.00–1.14)	1.9	11.6	19.2	1.06	(0.93–1.21)
Hip fracture	1.0	38.6	27.3	1.52	(1.41–1.65)	0.5	31.5	36.7	1.83	(1.56–2.15)
Other accident or violence	4.0	29.0	21.9	1.46	(1.40–1.53)	3.5	17.5	31.9	1.28	(1.18–1.38)
Other hospital diagnoses	26.1	22.2	20.8	1.30	(1.26–1.33)	25.5	14.5	30.9	1.26	(1.21–1.32)
Other diseases	41.9	16.5	17.4	1.05	(1.02–1.07)	33.9	10.9	25.9	1.07	(1.02–1.11)
At least one of the diseases	73.7	17.1	16.9			71.7	11.6	26.4
None of the diseases	26.3	7.6	7.0			28.3	5.8	13.7
All	100.0	14.6	14.3			100.0	10.0	22.8
N	172 248					108 474

a: Adjusted for other chronic conditions, age, living arrangements, education, income, home ownership, possession of a car, house type, level of equipment in a dwelling, region of residence and urbanicity

After controlling for socio-demographic confounders and other chronic conditions, several chronic conditions were still associated with institutionalization (table 2). In both men and women, dementia and Parkinson's disease raised the risk of institutionalization the most. Furthermore, psychosis, stroke, other mental health problems, depressive symptoms, hip fracture and diabetes were strongly associated with institutionalization. All these conditions raised the risk of institutionalization by 50% or more, except for psychoses in men which raised the risk by 40%. Also other conditions related to accidents or violence, arthritis, other neurological diseases, cancer, respiratory diseases other than asthma, other hospital diagnoses, heart disease and other chronic diseases raised the risk of institutionalization. Chronic asthma was associated with institutionalization only among men. Osteoarthritis was not associated with institutionalization.

We further tested for proportionality of hazards over time. Parkinson's disease was the only chronic condition for which the relative risk of institutionalization rose during the follow-up. In contrast, for depression and other hospital diagnoses the relative risk of institutionalization declined during the follow-up among both men and women.

Comparing relative institutionalization and mortality rates by chronic condition

In both men and women, dementia, Parkinson's disease, psychosis, depressive symptoms, other mental health disorders and stroke were more strongly associated with the relative risk of institutionalization than with the relative risk of death without institutionalization, after controlling for socio-demographic confounders and the presence of comorbid conditions (figure 1). Our results, not shown here, indicated that the institutionalized older adults with dementia, Parkinson's disease, mental health problems or stroke stayed in institutions for longer periods than those with other conditions measured in our study. In contrast, cancer and heart diseases were more strongly associated with the risk of death without institutionalization than with the risk of institutionalization.

Figure 1

Relative institutionalization and mortality rates without institutionalization by chronic condition and their 95% confidence intervals, Finnish community-living women and men aged 65 and over (Adjusted for other chronic conditions, age, living arrangements, education, income, home ownership, possession of a car, house type, level of equipment in a dwelling, region of residence and urbanicity)

Discussion

The impact of chronic conditions on institutionalization

Our results indicate that dementia and Parkinson's disease were the strongest determinants of institutionalization in both genders, after adjustment for socio-demographic confounders and for other diseases. Furthermore, stroke, depressive symptoms, other mental health problems, hip fracture and diabetes were strongly associated with institutionalization. All these conditions raised the risk of institutionalization by 50% or more. Because of the large number of study subjects followed for institutionalization, many other chronic conditions were also statistically significant predictors of institutionalization despite the relative weakness of some effects.

As most previous studies on institutionalization include only older adults with certain specific disabilities, results on the effects of chronic conditions in the general older population are scarce. However, our results are consistent with the few earlier population-based findings showing that dementia is a strong predictor of institutionalization,^5–7,,9 and confirms findings of Banaszak-Holl et al.⁹ indicating that stroke, mental health problems and diabetes are associated with institutionalization, independent of various socio-demographic confounders. Furthermore, Valiyeva et al.²⁶ found that diabetes increased the risk of institutionalization, independently of socio-demographic confounders and baseline medical conditions, especially when combined with lifestyle-related risk factors such as smoking, obesity or physical inactivity. However, Banaszak-Holl et al.⁹ measured date of admission into nursing home with surveys carried out every 2 or 3 year, so that nursing home stays that were completed between the surveys were not included. This could have underestimated the effects of certain diseases, such as cancer, that cause institutionalization for shorter periods at the proximity of death. However, in a continuous follow-up study of older Canadians living in Manitoba, Tomiak et al.⁶ were able to detect shorter nursing home stays more accurately. Their results showed that Alzheimer's disease and dementia, and other mental disorders raised the risk of nursing home admission among both men and women, and stroke and musculoskeletal disorders only among men. However, Tomiak et al.⁶ analysed nursing-home admission after adjustment for functional disability which could have underestimated the effects of conditions (e.g. stroke, diabetes, hip fracture) which are likely to cause functional disability. For example, Banaszak-Holl et al.⁹ showed that the effects of stroke and diabetes on institutionalization disappear after controlling for functional disability.⁹ This indicates that the effects of stroke and diabetes are largely mediated through functional disability. In a longitudinal study of Swedes over 75 years of age living in an urban district of Stockholm, Aguero-Torres et al.⁵ showed that dementia and hip fracture raised the probability of institutionalization, independent of functional disability. This could mean that dementia and hip fracture have effects that go beyond functional disability, such as heavy care giving burden to relatives. Our study supports the Swedish result⁵ that hip fracture markedly increases the risk of institutionalization.

Our follow-up study supports previous cross-sectional evidence of the association between Parkinson's disease and living in an institution that has been reported for older Canadians,²⁷ Europeans¹¹ and Hong Kong Chinese.²⁸ Few previous studies have examined whether cancer,^5,,6,9 heart disease^5,,6,9 and arthritis^6,,9 are associated with institutionalization. Overall, previous studies have suggested that cancer and heart disease are not associated with institutionalization, after adjustment for socio-demographic confounders⁹ and functional disability.^5,,6 In contrast to previous studies, we found that cancer raised the risk of institutionalization, after adjustment for socio-demographic confounders. In addition to different modelling strategies, these differing results could be related mainly to two factors. First, we included institutional care provided in general hospitals. Cancer patients who need intensive treatment or palliative care before death are more likely to move into general hospitals than nursing homes or mental hospitals. In Finland, publicly provided palliative care is mainly given in hospitals. Second, our study identified only those cancer cases that were recently treated; untreated or cured cancers were excluded. Furthermore, in our exceptionally large data set, heart disease was a statistically significant, albeit a very weak predictor of institutionalization.

As many chronic diseases tend to cause decline in functional status, the effects of chronic diseases on entering into institutional care are likely to be mediated by the onset of functional disability. Several prospective studies have shown that stroke or transient ischaemic attack, diabetes, arthritis, hip or other fractures and hypertension are associated with decline in functional status,²⁹ as well as dementia,³⁰ and Parkinson's disease.³¹ At older ages, admission to institutional care is in most cases preceded by a professional assessment of functional disability and thus routine adjustment for functional disability in the analyses of institutionalization tends to underestimate the effects of medical conditions that cause functional disability. As our register-based data did not contain information on functional disability, we were not able to examine the mechanisms, such as different aspects of functional disability, through which chronic conditions affect institutionalization.

Our results showed that diabetes, one of the most common chronic diseases worldwide, was strongly associated with institutionalization. As the prevalence of diabetes is estimated to increase in the future,^32,,33 the costs of long-term institutional care related to diabetes could also increase. The population attributable risk percentage of institutionalization due to diabetes was even higher than due to Parkinson's disease (5.3% vs. 2.1%) because diabetes is more common. Lifestyle changes that affect obesity and physical inactivity, the two major risk factors for type 2 diabetes,^33,,34 are required to level off the potential future increase in long-term care expenses associated with diabetes. In addition, further research is needed to assess the effect of chronic diseases on institutionalization in different population subgroups.

Comparing relative institutionalization and mortality rates by chronic condition

Comparison of the relative institutionalization and mortality rates by chronic condition showed that mental health problems (psychosis, depressive symptoms and other mental health problems) and specific neurological disorders (e.g. dementia, Parkinson's disease, stroke) were more strongly associated with the risk of institutionalization than with the risk of death without institutionalization. That these conditions affect institutionalization more than mortality is related to two related processes. First, these conditions tend to a have long-term disability impact on older individuals’ lives, and older people institutionalized with the preceding neurological or mental problems stay in institutions for longer periods than older people with other chronic conditions. Second, because of this disability impact, some of the neurological conditions, such as dementia³⁵ and stroke,³⁶ are known to be highly burdensome to the caregiver, which could strengthen families’ intentions to seek institutional care for the older person.

Considerations on data and measurement of disease prevalence

Our data on disease prevalence are based on registration data of in-patient hospital care over a period of two years prior to baseline and on information from two different medication registers. In general, the prevalence rates obtained in this study were quite close to those derived from population-based clinical examinations and other sources (e.g. diabetes,³⁷ heart disease,³⁷ Parkinson's disease,³⁸ stroke,³⁷ depression,^39,,40 psychosis,⁴¹ cancer⁴²). The two notable exceptions were dementia and osteoarthritis. Our study appears to underestimate the prevalence of dementia, covering only about 10% of the prevalence obtained in clinical data sets in Finland⁴³ and elsewhere in Europe⁴⁴ but provided similar estimates of the prevalence of dementia that Tomiak et al.⁶ reported for older Canadians. We were only able to identify those persons with dementia who had received hospital care due to dementia, and these cases are likely to represent the most severe forms of dementia. Furthermore, our study appears to cover only about 15% of the prevalence of osteoarthritis among Finnish older people.³⁷ Common musculoskeletal disorders that seldom lead to hospitalization or specific medicinal treatment cannot be covered using register-based data sets. The osteoarthritis cases we were able to identify were based on hospital diagnoses. In our study, osteoarthritis was not associated with the risk of institutionalization. This could be related to the fact that those older adults hospitalized for knee or hip osteoarthritis can undergo endoprosthesis operations, which improve their functional capacity and decrease their need for institutional care.

However, nationally representative data that link different administrative registers provided several empirical and methodological advantages, as missing information and loss due to follow-up are minimal. This is unique as prospective studies on institutionalization based on questionnaires can suffer from lack of complete follow-up due to attrition related to severe disability or due to long periods between the surveys. The latter problem can easily overestimate the effects of certain conditions that cause very long periods of institutional care, such as dementia, and underestimate the effects of other important conditions, such as diabetes or cancer, that cause shorter periods of institutional care. Furthermore, our data covered several types of institutions that provided long-term care, including nursing homes, services homes, hospitals and health centres. In addition, in Finland, information on socio-economic factors, such as disposable income, originating from the Tax Administration, is more reliable than self-reported income based upon questionnaires, especially in very old age.

Conclusions

Our study is unique in that we have systematically assessed the effects of a broad range of diseases on admission into institutional care in a nationally representative follow-up study of over 280 000 older men and women. Our study was the first population-based follow-up study to show that Parkinson's disease, with dementia, is one of the strongest predictors of institutionalization. Furthermore, Parkinson's disease was the only chronic condition for which the relative risk of institutionalization rose during the follow-up. This is fully consistent with the nature of Parkinson's disease as a progressive disease that is likely to cause gradual decline in functional status over time.⁴⁵ These results may have important implications for targeting home help services for older adults with chronic diseases to delay or prevent their institutionalization. Our results indicate that the future demand for institutional care depends not only on the ageing of the population, but also on the development of the prevalence and severity of chronic diseases among older people.

Acknowledgements

We thank Statistics Finland, the National Research and Development Centre for Welfare and Health (STAKES), and the Social Insurance Institution for making the data available to us. This study was supported by the Finnish Post-Graduate School in Social Sciences (SOVAKO) and by the Academy of Finland (210752 and 205631). This research was part of an EU-funded research programme on the Future Elderly Living Conditions In Europe (FELICIE). Earlier versions of these results were presented at the 25th International Population Conference of the International Union for the Scientific Study of Population, in Tours, France, on 8–23 July 2005.

Key points

• Systematic evidence showing how chronic conditions, other than dementia, affect the risk of entering into institutional care in the general older population is meagre.

• We assessed the effects of a broad range of diseases on admission into long-term institutional care in a nationally representative follow-up study of over 280 000 Finnish older adults.

• Parkinson's disease was, after dementia, the strongest predictor of institutionalization, followed by stroke, mental health problems, hip fracture, and diabetes.

• The results have important implications for targeting home help services for older people to delay or prevent institutionalization

• The future demand for institutional care depends not only on the ageing of the population, but on the development of the prevalence of chronic diseases.

Appendix

Classification of chronic medical conditions used in this study

Classification of chronic conditions	Hospital diagnoses (ICD 10 codes in parentheses)	Special refund category (Finnish codes in parentheses)	Prescription medication (ATC codes in parentheses)
Cancer	(C00-C97)	Breast cancer (115)	Cancer medication (L)
		Prostatic cancer (116)
		Gynecological cancers (128)
		Other malignant tumors (130)
		Melanoma and renal cancer (180)
Diabetes	(E10–E14)	Diabetes (103)	Diabetes medication (A10)
Dementia	(F00–F03, G30)	–	–
Psychosis	(F20–F29, F30.2, F31.2, F31.5, F32.3, F33.3)	Psychosis (112)	–
Depressive symptoms	(F31.3, F31.4, F31.6, F32, F33, F34.1, F38.10, F41.2, excluding F32.3, F33.3)	–	Anti-depressants (N06A)
Other mental health disorders	(Other F00–F99)	–	Psychosis medication (N05A) if not information on diagnosis of psychosis (112)
Parkinson's disease	(G20)	Parkinson's disease (110)	Parkinsonism medication (N04)
Other neurological diseases	(Other G00–G99)	Epilepsy (111)	Epilepsy medication (N03)
		Multiple sclerosis (109)
		Some apoplectic symptoms (108)
		Trigeminusneuralgia or glossofaryngikusneuralgia (119)
Heart disease	(I00–I09, I20–I52)	Cardiac insufficiency (201)	–
		Coronary heart disease (206)
		Arrhythmia (207)
Stroke	(I60–I69)	–	–
Chronic asthma and COPDa	(J40–J45)	Chronic asthma and COPD (203)	–
Other respiratory diseases	(Other J00–J99)	–	–
Arthritis	(M05–M06)	Arthritis (202)	–
Osteoarthritis	(M15–M19)	–	–
Hip fracture	(S72)	–	–
Other accidents or violence	(Other S00–T98)	–	–
Other hospital diagnoses	(Other A00–Z99)	–	–
Other diseases		Other chronic conditions (Other 101–601)

a: Other chronic obstructive pulmonary diseases