We examined the use of residential services for people with intellectual disabilities in Finland from a regional perspective. Our aims were to investigate how service supply and need, as well as municipal factors, impact the use of residential services and to analyze whether the available information in registers and databases can be effectively utilized in studies of this kind. According to the model based on the register data, after age adjustment the most important factors determining service use were service supply, the prevalence of intellectual disabilities, the municipality's level of urbanization, the number of persons granted long-term care as a percentage of all care recipients, and the percentage of middle-class and managerial households in a given municipality.
People with intellectual disabilities require appropriate care and services in tune with their needs. However, services are typically not very systematically organized and differ from country to country. Moreover, within countries there can be regional variations in the prevalence of intellectual disabilities. Such variations may cause regional differences in the need for services not fully considered in the service organization, which may often reflect more general disability policy models prevailing in the country that do not meet the needs of all clients.
In this study we examined the impact of service supply, need, and municipal factors on the use of residential services for people with intellectual disabilities in Finland. Our specific aims were to (a) find possible explanations or clarifications for the regional variation in residential service use for people with intellectual disabilities and (b) analyze whether the available information in administrative registers and databases can be effectively utilized in studies of this kind.
In Finland, disability policy—especially services for people with intellectual disabilities—is based on the general Finnish policy of social and health care and on the Nordic social policy model (Kautto, Frizell, Hvinden, Kvist, & Uusitalo, 2001; Kautto, Heikkilä, Hvinden, Marklund, & Ploug, 2000; Vedung, 1999). The social policy debate in Scandinavia has undergone ideological changes. The traditionally strong support for the distinctive features of the Scandinavian welfare states, such as centralization and universal social programs, is gradually declining and is being replaced by values that emphasize individual freedom and diversity, community and networking, and the decentralization of public services. There is growing political support for private and voluntary arrangements as alternatives to the public sector. These ideological changes are not merely Scandinavian phenomena. Decentralization and privatization were common features of most Western countries in the 1980s, and the renaissance of the idea of the Civil Society (i.e., the set of institutions, organizations, and behavior situated between the state, the business world, and the family) seems to be one of the dominant ideologies in the 1990s and will probably continue to be so in the first decade of this millennium (Mansell & Ericsson, 1996).
A general interest in social well-being emerged at the end of the 19th century in Finland. At the same time there arose the first interest in services for people with intellectual disabilities. The first institution for these individuals was founded in 1907, but on a larger scale these institutions were established in the late 1920s. In 1927, the rights of people with intellectual disabilities were established by law in Finland (Mansell & Ericsson, 1996; Tössebrø, Gustavsson, & Dyrendahl, 1996).
However, it was not until the postwar period that services for people with intellectual disabilities started to develop along with expanding Finnish social policy. This led to increased institutionalization in the 1960s. At that time institutional care was seen to be the most appropriate service for people with intellectual disabilities. The organization of services was delegated to so-called “intellectual disability care districts,” which maintained the institutions. The catchment areas of institutions approximately defined the borders of each intellectual disability care district.
Persons with intellectual disabilities became entitled to special services through a law enacted in 1977. The law assigned the organization of services for people with intellectual disabilities to special municipal federations, which replaced the previous intellectual disability care districts, although the change was more cosmetic than radical. At that time, the first official statistical systems were also developed in order to monitor the use of services that were mainly intended for people with intellectual disabilities.
A few laws enacted in the 1980s emphasized the role of municipalities—local administrative units—in the organization of services. However, not until the 1990s did the organization of services for people with intellectual disabilities move towards to a system based predominantly on municipal service provision. The severe economic depression in Finland at the beginning of the 1990s decreased the municipalities' willingness to buy services from the special municipal federations. The simultaneous large-scale change in 1993 for the state subsidy system in health and social welfare care, designed to equalize economic disparities between the municipalities, increased possibilities for municipalities to make choices concerning the organization of services for people with intellectual disabilities (Nouko-Juvonen, 2000a). At the same time many municipalities reorganized their administrative structures. In spite of the plans to cut down the number of beds in institutional care beginning in the 1980s, the rapid deinstitutionalization of people with intellectual disabilities did not begin until the early 1990s. The decreased use of institutional care was compensated for by implementing more outpatient and assisted residential services. These changes caused serious problems for the compilation of statistics. For example, until 1995, there were no systematic data available on residential services other than institutional care (Nouko-Juvonen, 1999).
Currently in Finland, the municipalities run the services for persons with intellectual disabilities. Most of the municipalities are very small in size (median size of about 5,000 inhabitants), but the range is from about 200 residents to over 500,000 in the capital (Helsinki). Although 90% of the 452 municipalities have under 20,000 inhabitants, 58% of the Finnish population lives in municipalities with over 20,000 citizens.
From the economic point of view, services for people with intellectual disabilities represent a large portion of the health and social welfare costs of the municipalities. Small municipalities, in particular, have difficulties bearing the financial risk created by even a few resource-intensive clients (Mikkola, Sund, Linna, & Hakkinen, 2003). The funds for the municipalities' services are derived mainly from municipal taxes and state subsidies. The latter are paid as lump sums according to a capitation formula based on regional, demographic, and socioeconomic (SES) factors.
Services for people with intellectual disabilities are typically organized such that municipalities themselves implement some services, such as guided day- and work-activities as well as serviced apartment buildings. More resource-intensive services, such as institutional care, assisted residential care, and welfare clinic services are bought mainly from the special municipal federations. In addition, there are also some private service providers. Moreover, people with intellectual disabilities are also directed to use services that are not necessarily intended only for people with intellectual disabilities, but for all residents of a municipality (Sund & Nouko-Juvonen, 2000).
The organization of services for people with intellectual disabilities in Finland is very fragmented in nature. Any investigation of variation in service use is made difficult by the complexity of the service system, as reflected in the maintenance of service registers and statistical compilation as well (Häkkinen, Valtonen, Niemelä, & Laine, 2000; Valtonen, 1996). There are also difficulties in identifying the disability sector among other social welfare and health care services.
In addition, there are major regional differences in the prevalence of intellectual disabilities in Finland. For example, in 1996, the prevalence in eastern Finland was significantly higher than in western Finland. The overall prevalence of intellectual disabilities has been estimated to be about 1%, but the prevalence varies from about 0.4% to over 1.5% if municipality-specific estimates are calculated (Sund & Nouko-Juvonen, 2000). These facts make the situation in Finland special, and only a few studies have been conducted on the causes of regional variation in the use of services for this population (Nouko-Juvonen, 2000a; Sund & Nouko-Juvonen, 2000).
Instead of adopting a regional viewpoint, researchers on service use have focused on overall trends in service use (Kaunistola & Arhenius, 1998; Nouko-Juvonen, 1999). Studies have established, among other things, that the Finnish service system is more institutionally oriented than its counterparts in the other Nordic countries (Mansell & Ericsson, 1996).
In general, the international studies on regional variation in the use of services can not be directly utilized in the Finnish context. The most closely related studies concerning regional use of services are those in which researchers investigate other social welfare services, such as care for the elderly and children's day care. These investigators have examined, for instance, the impact of the age and service structure on social welfare expenditure per inhabitant or target population. In addition to examining expenditures, studies on the criteria for state subsidies have been conducted in Finland by investigators who have analyzed the links between needs and expenses (Häkkinen et al., 2000; Valtonen, 1996). According to Valtonen, variation in social service expenditure is considerable, even when differences in demand are excluded. With reference to his previous research, Valtonen (1997) suggested that municipal decisions and the organization of activities may have a major impact on expenditure.
Some investigators dealing with the field of health care have also explored regional variation in the use of hospital services in Finland. Attempts have been made to relate this variation to morbidity, SES factors in the population, service availability, and care practices. According to Nordberg and Häkkinen (1997), studies on the regional use of health care services have yielded mixed results and offered many interpretations of the significance of the various factors.
In our analysis we approached the organization of services for people with intellectual disabilities from a sociological point of view. There is no global, national model for the organization of services. Instead, the local definition and implementation of services and the need for them are social constructions created by decision-makers working under the restrictions raised by the history of institutional welfare system and by the major regional differences in the prevalence of intellectual disabilities (Nouko-Juvonen, 2000b). In other words, we assumed that the definition of service need is fundamentally a political matter—a compromise made by the local authorities.
On the basis of a number of Finnish registers and databases, we compiled data on the use of residential services for people with intellectual disabilities, on the supply of and need for services, and on municipal factors (see Table 1). Data obtained from different sources were linked and aggregated in such a way that the final data set contained cross-sectional municipality-specific variables for 1996. The data were restricted to residential services because the administrative databases contain no data of good quality concerning other types of care. Residential services include institutional inpatient care and assisted residential services. The term assisted residential services refers to group apartments with around-the-clock presence of personnel. The province of Åland was omitted from the analyses, due to inadequate information.
In the study, service use was determined by variables indicating supply, need, and general municipal factors. The number of resident days accumulated in 1996 was selected as the indicator of service use because it reflects the burden on municipal resources more accurately than, for example, the number of clients.
The supply was measured as regionally available beds in residential services. Because data concerning the number of beds were available only at the special municipal federation level, the municipality-specific estimates were calculated on the basis of the data on special municipal federations by assigning beds to each municipality in proportion to the number of people in institutional care on December 31, 1996.
Service need was indicated by variables denoting the prevalence of intellectual disabilities, the number of patients granted long-term care as a percentage of all care recipients, and the age structure. A decision to grant long-term care is made when a person is considered to be in need of institutionalized care for a period of more than 3 months (Section 15 of the Finnish Decree on client fees in social and health services).
Other analyzed factors affecting municipal services included variables indicating the municipal service structure, municipality-specific morbidity, SES structure, and regional profile. Variables indicating the service structure consisted of the administrative model in social and health services and information concerning service providers. Here service provider means the party producing the service (municipality, special municipal federation, private service provider). Variables measuring morbidity included standardized indices concerning the recipients of sickness allowance, medicines subject to special reimbursement, and disability pensions. The SES structure was described by means of variables indicating the municipality's SES status, educational status, and the life situation of the families. The level of urbanization was used as a variable indicating the general regional profile.
The variables reflecting the age structure of the population with intellectual disabilities are important factors in explaining service use. It was necessary to eliminate these self-evident factors in order to reveal the impact of other variables, and, thus, inpatient days were age-adjusted by direct standardization using eight age groups (0 to 9, 10 to 19, 20 to 29, 30 to 39, 40 to 49, 50 to 59, 60 to 69, and 70+. The service users' age structure represented standard population (Wunsch & Termote, 1978, pp. 51–60).
It can be presumed that service supply is simultaneously determined on the basis of need and use. In order to eliminate the effects of supply on use in this study, a prediction indicating supply was created by means of a linear regression analysis, thus explaining the observed supply (regionally available beds in residential services) on the basis of general factors in the municipality.
As variables indicating general municipal factors, we used the distance from the central care facility within the special municipal federation, the proportion of service-sector employees in the employed labor force, the proportion of families with children (ages 0 to 17 years), and the purchasing power of households. Distance information was determined on the basis of the Euclidean distance between the municipality's geometric center and the special municipal federation's central institution. If the municipality had such an institution, the distance assigned was zero. Purchasing power means the amount of income subject to state tax that households have at their disposal after paying taxes.
For administrative reasons, one can presume that the number of beds available varies at the level of special municipal federations. In the model, this effect was taken into account by creating the corresponding dummy variables.
In the diagnostic analyses, the residuals showed heteroscedasticity in terms of the population in a municipality. In larger municipalities, the regional variables are based on a higher number of observations, and, thus, the variance is in an inverse relationship with the variable denoting size (Haining, 1990, p. 49). The model's heteroscedasticity was corrected by using information concerning the average population in the municipality as a weight variable in the regression analysis (Jørgensen, 1997, p. 199).
The model predicting supply meets the normal expectations placed on regression analysis, and its coefficient of determination is an R² of .39. The standardized regression coefficients and their significance are shown in Table 2.
The impact of service supply, need, and municipal factors on the use of residential services for people with intellectual disabilities was modeled using a linear regression model. Figure 1 shows the structure of the connections of this model. The left-hand column and the bold-print headings categorize phenomena presumably affecting service use, whereas the regular-print text denotes variables associated with the phenomena. When the model was being constructed, some variables turned out to be statistically nonsignificant. The results of the final regression model are presented in Table 3. The variables remaining in the final model are also underlined in Figure 1.
Several diagnostic analyses of the regression model were conducted in order to verify that the specification of the model was correct and find possible multicollinearity (Cook & Weisberg, 1994). According to residual plots, the model, on average, yielded the right result with the value combinations of determining variables occurring in the material, and no systematic bias was apparent. The Breush–Pagan test, which measures heteroscedasticity, gave a p value of .38 (i.e., the model did not have statistically significant heteroscedasticity). According to the Anderson–Darling test, the residuals have a normal distribution, P = .235. The graphic analyses corroborated this.
Finnish municipalities cannot be regarded as mutually independent. Therefore, Moran's I was employed to analyze the regional correlation between the observations (Haining, 1990, p. 230). Roughly speaking, Moran's I can be interpreted as the correlation coefficient between neighboring municipalities, indicated by the proximity matrix W (Bailey & Gatrell, 1995, p. 270). Neighboring municipality refers to all municipalities sharing a border with the municipality studied. With the original dependent variable, Moran's I is .17 (i.e., moderate regional correlation is apparent). According to Moran's I (I = .10) as calculated from the residuals, regional correlation does not result in systematic error in the model established.
The correlation matrix does not have variables showing strong mutual correlation, and the eigenvalues of the product moment matrix deviate from zero. Calculated from the predictor matrix, normalized by each column, the condition value (the relationship between the highest and lowest singular value) is 14.9. The condition index (CI) of 1.60 is at its highest in the case of the variable indicating middle-class households. The analysis, thus, shows that no disturbing degree of multicollinearity is apparent.
On the basis of the regression model devised in the study, statistically significant associations were found between the use of residential services and each of the following: modeled supply, prevalence of intellectual disabilities, proportion of long-term care decisions, municipality's urbanization level, and percentage of middle-class and managerial households in terms of all households in a municipality (Table 3). The coefficient of determination of the final model is 47%.
As can be seen in Table 3, of these variables, urbanization and the prevalence of intellectual disabilities are the factors that most significantly determine service use. A high prevalence of intellectual disabilities increases the use of institutional and residential services, as was expected. In addition, a high proportion of long-term care decisions increased service use. Service use is also determined by the modeled supply of services: When services are available, they are also used. In addition to the above factors, service use is determined by the occupational structure of a municipality's population. When a municipality has a high number of managerial households, service use is lower, whereas a high number of middle-class households increases service use.
In Finland the organization of services for people with intellectual disabilities has been regarded as such an important part of the welfare system that official statistical systems have been developed in order to monitor the use of these services. Currently, the local administrative units (i.e., the municipalities) operate these services. About 90% of the 452 municipalities have populations of under 20,000 inhabitants. The great number of municipalities implies that there might be large-scale variations in the organization of services in Finland. In this study a “picture” of the use of these services was created by examining the official statistics and administrative registers to find possible explanations or clarifications for the regional variation in residential service use for people with intellectual disabilities and to analyze whether the available information in registers and databases can be effectively utilized in studies of this kind.
Data used in the study were collected from a wide range of administrative registers and databases. On the one hand, these data sources potentially lack some sound variables concerning the use of services for people with intellectual disabilities. On the other hand, the databases include all the information that is available, reducing the need for expensive and time-consuming collection of new data. To find a suitable compromise between the actual problem solving and data quality, we restricted the analyses to residential care.
The model used in the study was based on the assumption that service use in a municipality is determined by supply, need, and general municipal factors. This kind of model has been found to be well-suited for modeling the use of health services as well as for modeling care for the elderly and children's day care in Finland (Nordberg & Häkkinen, 1997; Valtonen, 1996). However, this study is the first in which this kind of model has been applied to the analysis of the use of services for persons with intellectual disabilities. On the one hand, the benefits statistics of the Finnish Social Insurance Institution allowed the identification of persons with intellectual disabilities, which is, in general, exceptional for models of this kind because it makes it possible to quite accurately determine the need, independent of the supply and use. On the other hand, diagnostic-based identification of intellectual disabilities has been criticized as medicalistic in nature (Zola, 1993). However, it is unlikely in Finland that individuals who are not receiving any benefits for intellectual disabilities use or need residential care services.
The structure of the model is characterized in Figure 1. We controlled for several confounding factors by using statistical techniques. The regression model constructed is specified correctly with regard to diagnostic analyses, and it has a reasonably good coefficient of determination, R² = .47. The regression coefficients based on the cross-validation analyses conducted on the model are reasonably stable. In other words, the statistical associations between variables exist in this data set.
Of these associations, prevalence of intellectual disabilities, modeled supply, and proportion of long-term care decisions reflect the effects of need and supply on use. These findings validate that the basic structure of the model makes sense also in terms of the data set based on official registers and statistics. The need variables in the model seem to have a rather strong association with service use because the effects remain even after adjusting for age.
However, it is somewhat controversial that these associations would hold for the whole country. There are large-scale variations in the prevalence of intellectual disabilities in Finland, suggesting that there should also be large-scale variations in the use of services for people with intellectual disabilities. Further, the utilization of residential services is resource intensive, and long-term care decisions may be a burden for the economy of a small municipality. How is it possible that municipalities with a high prevalence of intellectual disabilities can finance services equivalently compared to municipalities with a low prevalence? Does the current municipality-based financing system systematically emphasize polarization in local disability policies and in the organization of services? Is the quality of services equal for everybody? Unfortunately, these important questions cannot be answered based on results of the present study. Further research on the topic is required.
Although long-term care decision may work as a crude proxy for the degree of disability in the model, it may also be the case that this association is a reflection of the Finnish social policy debate from the years preceding 1990; the waiting lists for institutions for people with intellectual disabilities had a central role in this debate. In other words, some decisions concerning long-term care may have been made “too quickly,” without adequate consideration of other possible forms of treatment.
Supply increases use, a result that is consistent with health-care research, suggesting that a service provider may increase the demand for services such as hospital care: “a built bed is a filled bed,” a phenomenon known as Roemer's law (Roemer, 2001). However, it is questionable whether health-care studies are applicable to services for persons with intellectual disabilities. It is not even easy to judge whether there are enough appropriate services available. In fact, these kinds of definitions are a more general political matter because the decisions must be made under many competing interests and limited (financial) resources.
Our study also revealed that variables indicating the service structure, morbidity, and financial state of a municipality did not have direct statistically significant associations with the use of residential services for people with intellectual disabilities. In addition to the possibility that these factors are not associated with service use, the quality and accuracy of the variables may be poor or the effect may occur indirectly through, for example, supply.
However, the more urbanized the municipality, the greater the number of resident days—a fact that is probably attributable to the municipality's economic structure. In small and rural municipalities, people tend to stay home in the daytime and have more space in their homes to care for their relatives. In other words, rural communities better integrate people with intellectual disabilities (Patja, 2001, p. 49). Because the impact of supply on use and need was controlled in the model, the centralization of services in urban areas is not an explanation for this finding. However, it is possible that in urban areas people have more information concerning services, and this could be a reason for increased use of such services.
In addition to the above factors, the model suggests that service use is determined by the occupational structure of a municipality's population. These results may indicate a connection between high SES position and decreased prevalence of intellectual disabilities (cf. Gissler, 1998). Other research suggests that low social class predisposes populations to mild intellectual disability (Strömme & Magnus, 2000). However, it is possible that these variables describe more complicated connections that are not appropriately recorded in register-based data. The combined interpretation could be that the general structure of the model does not hold for the whole country and that variables remain in the “mean model” because the model is, in fact, a combination of two or more models of significantly different structure.
In the compilation of the data used in this study, we paid special attention to the reliability of the data sources. No exceptional inaccuracies were observed in the data, but the lack of additional, more informative variables resulted in a complicated model. The problem here is that a more complicated model is more difficult to interpret, and the potential to use register data in studies concerning the use of services for people with intellectual disabilities still seems to be quite limited, even though the quality and coverage of the databases have improved over the decades.
All in all, it is doubtful whether this specific model constructed in the study can be used for other areas or times, and the applicability of the predictions yielded from it must be treated with reservation. Moreover, long-term care clients have a key impact on service use, and a cross-sectional data set cannot adequately describe decisions made decades earlier. It may be that some variables reflect changes in the service system or that the SES structure of a municipality has “changed” more quickly than have the clients of residential care.
Studies on the use of services for people with intellectual disabilities should also incorporate services provided for people with intellectual disabilities through other service sectors (elder care, service housing and supported accommodation, and psychiatric care). So far, attempts based on register information have failed to identify a specific client group. A comprehensive analysis should also include outpatient services, but the register data concerning them are still unreliable. However, the coverage of outpatient service data is likely to improve in the near future in Finland because new administrative information systems incorporating outpatient data are under construction.
The data in the Finnish administrative registers are based on uniform compilation criteria and, thus, the analysis can be further expanded by incorporating data from other registers, too. The use of personal identification codes to link discrete chunks of data on individuals makes it possible to use these data in longitudinal studies or in comparisons of special municipal federations operating in the field of specialized services.
Statistical methods more flexible than standard linear regression analysis, such as multilevel models or geographically weighted regression, could also be beneficial in the analysis of local variations in the use of services for people with intellectual disabilities.
In spite of the potential limitations in the actual results of the constructed model, we accomplished our aims in this study. We were able to use administrative data to construct a reasonable model with several highly statistically significant associations. The ideas raised by the model were discussed and possible explanations and clarifications for the regional variation in residential service use for people with intellectual disabilities were given. Moreover, in the study we introduce the Finnish perspective on service use and organization for the international research community, which may raise some internationally important issues in the discussion about services for people with intellectual disabilities.
Our results also show that the use of administrative data to model the use of services for people with intellectual disabilities is made difficult by the complexity of the material and by the interpretative ambiguity associated with the phenomenon, which make it hard to establish appropriate and reliable variables or the “correct” model structure. It would, therefore, be appropriate to bring to light more phenomena affecting the use of services for people with intellectual disabilities and explore new research designs. Even though data in administrative registers and in official statistics are not a “cure-for-all” key to the “black box” of complex phenomena studied, in this case they were useful to stimulate the discussion on important issues and introduce further research objectives.
The authors thank the anonymous reviewers and senior researchers Ilmo Keskimäki (National Research and Development Centre for Welfare and Health, Helsinki, Finland), Kristiina Patja (National Public Health Institute, Helsinki) and Kristiina Härkäpää (Rehabilitation foundation, Helsinki) for their useful comments concerning the earlier versions of the manuscript.
Authors: Susanna Nouko, DSocSc, Senior Researcher, Rehabilitation Foundation, Helsinki, Finland. Reijo Sund, MSocSc ( firstname.lastname@example.org), Statistician, National Research and Development Centre for Welfare and Health (STAKES), PO Box 220, fin-00531 Helsinki, Finland. The senior author passed away in November 2002