Disability-based classification system for older people in integrated long-term care services: The Iso-SMAF profiles

https://doi.org/10.1016/j.archger.2005.07.001Get rights and content

Abstract

This study was conducted to develop and evaluate a disability-based classification system for management of long-term care (LTC) needs in an integrated service delivery system. We collected cross-sectional data on 29 items of the Functional Autonomy Measurement System (SMAF) from a stratified multistage sampling of 1977 elderly people with disabilities living in different environments. Disability profiles were identified using statistical clustering techniques combined with advice form a panel of experts. Their clinical meaningfulness, stability, reproducibility, homogeneity, heterogeneity and predictive validity were evaluated. The Iso-SMAF classification consisted of 14 homogeneous disability profiles characterized by a gradual progression in the severity of disabilities in instrumental activities of daily living (IADL) and activities of daily living (ADL) accompanied by predominant limitations either in mobility or mental functions. The profiles achieved a Kappa reproducibility coefficient of 0.67 through cross validation. A stable cluster structure emerged when the items were analyzed using different methods. They explained 82% of the variance in nursing care time, 80% of the variance in cost of nursing care (skilled and unskilled) and 57% of the variance in total costs including both formal and informal sources of LTC services. The conclusion recommends their use for planning, managing and predicting LTC service needs in an integrated care system.

Introduction

A range of services and resources involving many care providers and settings is frequently supplied to respond to the needs of the older person requiring long-term care (LTC) (Knickman and Snell, 2002). With the aging of the population leading to increased demand for services and resources being scarce, LTC systems are striving to match resources with subjects who are in greatest need. This trend has generated a search for better methods that systematically integrate LTC resources, services and needs.

In 1982, the Quebec province (Canada) implemented a system that attempted to accurately reflect patient care needs with a classification scheme based on the number of hours needed by residents for nursing services (Tilquin and Fournier, 1985). Required nursing time was used to justify elder access to nursing homes and to support staffing and financing processes. Although it generated major improvements, many problems remained, including important regional disparities in the allocation of services, the possibility of “manipulating the system,” the lack of a common, standardized evaluation tool over settings, the burden of the administrative process, and a large overlap in the disability level and cognitive status among subjects from different settings such as private homes, intermediate facilities and nursing homes (Bélanger et al., 1991, Hébert et al., 2001a).

This situation has led us to advocate the development of a different classification system that allows the comparison of groups of older persons who possess common characteristics, who need approximately the same combination mix and level of services, and who require similar resources. From this we get the terms iso-resource groups or case mix (Hornbrook, 1989, Dilts et al., 1995). However, as our LTC delivery systems began establishing Integrated Service Delivery Systems to improve the continuity among settings, it became crucial to provide a more useful classification system for this process (Hébert et al., 2003).

To aggregate the elderly into groups, only personal characteristics which reflect underlying medical or functional conditions and which best measure the field of interest resulting in the provision of services must be used (Hornbrook, 1989, Weissert and Musliner, 1992, Hair, 1998). It is common knowledge that functional status or disability explains a significant percentage of the variance of nursing care intensity. Activities of daily living (ADL) form the core of most LTC classification systems (Hornbrook, 1989, Weissert and Musliner, 1992). Weissert and Musliner (1992) noted that subgroups should be defined by clinical relevance, not cost of care, so that the providers recognize similarities within clinical groups. The variables should be valid, reliable, easily verifiable, difficult to manipulate by the provider, reflect only personal attributes (not the provider's) and should not be direct measures of patterns of services (Hornbrook, 1989, Weissert and Musliner, 1992). Although their primary use was for reimbursement purposes, measurement of iso-resource groups could allow comparisons of outcome, quality of care and resource use (Fries et al., 1994).

Among classification systems available in LTC services (Manton et al., 1995, Urquhart et al., 1999, Coutton, 2000, Rosewarne, 2001), the Resources Utilization Groups (RUG-III) system (Fries et al., 1994), developed in the U.S. specifically for funding purposes, is now in widespread international use (Ikegami et al., 1994, Carpenter et al., 1995, Björkgren et al., 1999). RUG-III is composed of 44 groups and is based on information in the Resident Assessment Instrument (RAI) and its Minimum Data Set (MDS). In spite of its undeniable advantages, RUG-III presents some limitations that may compromise its application in LTC systems that have different characteristics. For example, the system includes resident and service variables, but fails to inform why the person is being cared for. As regards resource evaluation, RUG-III focuses on the care or therapy the patients actually receive rather than what they need (Turner et al., 1995). Furthermore, this system is restricted to institutional settings and was developed in nursing homes that accommodate post-acute, rehabilitation and long-stay patients, and may be less suitable for nursing homes composed primarily of long-stay wards as is the case in Quebec (Fries et al., 1994). To evaluate a person in home health settings, the designers of the RUG-III system have already used a modified version (RUG-III/HCBS) within the framework of a study comparing home and institutional care. They recognized that this system had not been validated for such settings (Shugarman et al., 1999). Recently, another questionnaire, the MDS-Home Care was designed (Morris et al., 2001). To our knowledge, this tool has not yet led to a case-mix system.

To evaluate people at home, Home Health Resource Groups (HHRGs) were created (St-Pierre and Dumbi, 2000). Implemented in 2000, the 80 HHRGs are based on the Outcome and Assessment Information Set (OASIS) (Shaughnessy et al., 1997), which classifies patients by the clinical severity of their medical condition, functional limitations and use of services. As with RUG-III, the HHRGs have some limits. For instance, OASIS focuses on post-acute home health outcomes and is less suited for the frail elderly home care population (Weissert et al., 2003). Although the instrumental activities of daily living (IADL) are included in OASIS, they were not considered in HHRG. Other items considered important by clinicians such as variables related to communication functions or cognitive aspects were not retained (Elias et al., 2000). Counterpart systems are available but they are mostly restricted to home- and community-based services (HCBS) (Mykyta et al., 1997).

Despite great strides in the right direction, the multiplicity of assessments and systems that the elderly currently face for demand of services compromises both LTC accessibility and efficiency. It is essential to provide clinicians and managers with a reliable assessment instrument capable of recognizing the needs of the elderly, promoting the provision of the appropriate resources and providing standardized data that can match the elderly and providers across the continuum (Shortell et al., 2000). This study was conducted to develop a disability-based classification system for the elderly by the nature and level of required nursing services, which would be useful on a day-to-day basis for care providers and managers and which could be applicable in any LTC setting.

This paper reports on the development and evaluation of this system for both its validity and reliability. The following objectives were addressed: determining how many profiles form this classification; determining the characteristics of each profile; assessing property classification such as clinical meaningfulness, stability, reproducibility, homogeneity within profiles and heterogeneity among profiles; and determining predictive validity in term of nursing care time, nursing care cost and total cost (care, infrastructure, functioning and administrative support).

Section snippets

Design, sampling and settings

From data that came from a cross-sectional study, we developed the disability-based classification system by using the 29 items of the Functional Autonomy Measurement System (SMAF) and by identifying profiles (subtypes of disability) with statistical clustering techniques. Those techniques selected for this study provided many classifications. We assessed their validity and reliability using several criteria (described elsewhere) and submitted only those with good properties to a panel of

Results

Among the remaining solutions and balanced against the six criteria listed above, the panel of experts chose the solution containing 14 profiles generated by the combination of the Ward method and the K-means method, excluding the option of running means. One solution that had an excellent Kappa coefficient (=0.94) was not retained by the panel because it was not considered clinically useful. This solution enabled five subject profiles to be distinguished, but although it was quite simple, it

Discussion

Through a sequential process that included disability assessment, statistical clustering procedures, validation with six criteria and advice from a panel of experts, this study provided evidence of the reproducibility and validity of a disability-based classification system composed of 14 Iso-SMAF profiles.

A LTC classification system should define clientele by elderly characteristics, not provider characteristics (Hornbrook, 1989, Weissert and Musliner, 1992). The Iso-SMAF classification is

Acknowledgements

The authors would like to thank Alan M. Jette, Ph.D., from Boston University for his judicious advice during the review of this manuscript. This study was funded by Quebec's Department of Health and Social Services and Health Evidence Application and Linkage Network (HEALNet). Nicole Dubuc hold a doctorate award from the Quebec Health Research Funds and the Quebec Geronto-Geriatric Research Network.

References (50)

  • J.J. Macnab et al.

    The search for subtypes of DCD: is cluster analysis the answer?

    Hum. Mov. Sci.

    (2001)
  • W.P. Wodchis

    Applying activity-based costing in long-term care

    Health Manage. Forum

    (1998)
  • J.H. Beitchman et al.

    Comorbidity of psychiatric and substance use disorders in late adolescence: a cluster analytic approach

    Am. J. Drug Alcohol Abuse

    (2001)
  • Bélanger, L., Bolduc, M., Trahan, L., 1991. Les mécanismes régionaux d’orientation et d’admission: Description de la...
  • M.A. Björkgren et al.

    Validity and reliability of Resource Utilization Groups (RUG-III) in Finnish long-term care facilities

    Scand. J. Public Health

    (1999)
  • G.I. Carpenter et al.

    Casemix for the elderly inpatient: resource utilization groups (RUGs) validation project

    Age Ageing

    (1995)
  • J. Cheah et al.

    Case mix for better or for worse?

    Singapore Med. J.

    (1999)
  • A. Coenen et al.

    Using nursing diagnoses to explain utilization in a community nursing center

    Res. Nurs. Health

    (1996)
  • Coutton, V., 2000. Les mécanismes de la grille AGGIR. Convention d’étude no 99/13/0019, Ministère de l’Emploi et de la...
  • K.E. Covinsky et al.

    Measuring prognosis and case mix in hospitalized elders: the importance of functional status

    J. Gen. Med.

    (1997)
  • D. Dilts et al.

    Using cluster analysis for medical resource decision making

    Med. Decis. Making

    (1995)
  • N. Dubuc et al.

    Fidélité de la détermination des services requis en terme d’heures par le CTMSP modifié à des fins de recherche épidémiologique

    Année Gérontologique

    (1996)
  • N. Dubuc et al.

    Assessing disability in older adults: the effects of asking questions with and without health attribution

    J. Rehabil. Med.

    (2004)
  • J.W. Elias et al.

    When world views collide: a commentary on home health care casemix and patient outcomes

    Exp. Aging Res.

    (2000)
  • B.S. Everitt et al.

    Cluster Analysis

    (2001)
  • B.E. Fries et al.

    Refining a case-mix measure for nursing homes: resource utilization groups (RUG-III)

    Med. Care

    (1994)
  • J.F. Hair

    Multivariate Data Analysis with Readings

    (1998)
  • R. Hébert et al.

    Resources and costs associated with disabilities of elderly people living at home and in institutions

    Can. J. Aging

    (2001)
  • R. Hébert et al.

    The functional autonomy measurement system (SMAF): a clinical-based instrument for measuring disabilities and handicaps in older people

    J. Can. Geriatr. Soc.

    (2001)
  • R. Hébert et al.

    Prisma: a new model of integrated services delivery for the frail older people in Canada

    Int. J. Integr. Care

    (2003)
  • C.A. Hernandez et al.

    Validity and reliability of nursing workload measurement systems: review of validity and reliability theory

    Can. J. Nurs. Adm.

    (1996)
  • M.C. Hornbrook

    Nursing home case-mix measurement

  • N. Ikegami et al.

    Applying RUG-III in Japanese long-term care facilities

    Gerontologist

    (1994)
  • M.W. Isken et al.

    Data mining to support simulation modeling of patient flow in hospitals

    J. Med. Syst.

    (2002)
  • J.R. Knickman et al.

    The 2030 problem: caring for aging baby boomers

    Health Serv. Res.

    (2002)
  • Cited by (62)

    View all citing articles on Scopus
    View full text