Characteristics of Nursing Home Residents

A National Study Showed that Diagnoses Varied by Age Group in Nursing Home Residents Under Age 65

April 15, 2004

Word count : 4397

Author Page

Brant E. Fries, Ph.D. - Professor and Research Professor, Institute of Gerontology and School of Public Health, University of Michigan, Ann Arbor, Michigan, and Chief, Health Systems Research, Ann Arbor VA Healthcare System Geriatric Research, Education and Clinical Center.

Walter P. Wodchis, Ph.D. – Assistant Professor, Department of Health Policy, Management and Evaluation, University of Toronto; Scientist, Toronto Rehabilitation Institute; Adjunct Scientist, Institute for Clinical Evaluative Sciences, Toronto, Canada. At the time of writing: Research Assistant, Institute of Gerontology and Michigan Health Services Research Initiative, University of Michigan, Ann Arbor, Michigan

Caroline Blaum, M.D., M.S. - Assistant Professor, Department of Internal Medicine, University of Michigan Medical School, Ann Arbor, Michigan.

Amna Buttar, M.B.B.S., M.S. - Assistant Professor of Clinical Medicine, Indiana University School of Medicine, Indianapolis, Indiana; at the time of writing, Instructor, Department of Internal Medicine, University of Michigan Medical School, Ann Arbor, Michigan

John Drabek, Ph.D. - Economist, U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation, Washington, DC.

John N. Morris, Ph.D.- Co-Director, Research and Training Institute, Hebrew Rehabilitation Center for Aged, Boston, Massachusetts

Diagnoses and Other Characteristics Vary by Age in Nursing Home Residents Under Age 65

ABSTRACT - 208 WORDS

Objective: It is commonly held that those aged <65 in nursing homes (NHs) are substantially different from elderly residents. This study uses data gathered using the Resident Assessment Instrument’s Minimum Data Set (MDS) to describe these relatively rare residents.

STUDY DESIGN AND SETTING, RESULTS, CONCLUSION

Data: The study uses MDS assessments of close to three-quarter million residents in nine states from 1994-6. An algorithm resolved potentially incorrect ages caused by confusion between residents age < 15 from those 100-115.

Methods: Residents are described within chronological age group (0-4, 5-14, etc.). Functional status, prevalence of chronic conditions and treatments are described for each group. Co-morbidity is examined using factor analyses and cross-tabulations.

Results: Overall, pediatric residents appear substantially more physically and cognitively impaired than young adult residents. The youngest population primarily has diagnoses related to mental retardation and developmental disabilities, young adults have the highest prevalence of hemi- and quadriplegia, while older residents are typified by increasing prevalence of neurological diagnoses and more co-morbid conditions. Thirteen diagnostic factors describe nearly 85% of all NH residents and highlights differences between age groups.

Conclusions: This study offers a first description of nearly all NH residents <65. The classification demonstrates significant differences within this population and between these residents and those 65+.

1

Key Words: Nursing Homes, Disabled, Non-elderly, Diagnosis

RUNNING TITLE

(Word Count – 4397)

INTRODUCTION

Relatively little is known about the roughly 10% of nursing-home residents aged under 65. While national surveys suggest that these residents vary greatly in their diagnoses and needs for assistance,1 they provide insufficient sample sizes to describe, in any detail, subgroups within this population.

It is commonly believed that this cohort is substantially different from elderly residents. While no one would seriously question such an assertion, it is not known quantitatively how they differ or whether differences are equally applicable to the entire age spectrum under age 65. Finally, there is the challenge whether these individuals are being served in the least restrictive setting.

This study begins to address these issues by examining a very large archive of assessments describing nursing home (NH) residents in several US states. Demographic, functional and treatment characteristics are used to describe the younger residents and we demonstrate the variation in diagnosis by age group to highlight these differences.

Disability leading to institutionalization is relatively rare among children. Although it is estimated that over 12% of children aged 5-17 have difficulty with learning, communication, or mobility, only 1% have age-appropriate limitations in Activities of Daily Living (ADLs)1. At the same time, Mental Retardation/Developmental Disabilities (MRDD) are most common in younger ages: nearly 60% of those with MRDD are under 17,4 and most, it is believed, are cared for at home. An epidemiological profile of disabled working-aged adults (in both the community and institutions) found two groups with very distinct characteristics: those with MR and those with only physical disabilities.5 Using the 1987 National Medical Expenditures Survey, Schoenman5 reported that 11% of the working-aged disabled were institutionalized. Of this group, 43% had MR (mostly severe or profound), 27% had mental illness, 10% had both, and only 40% had neither.

One goal of United States’ (US) LTC policy is to provide care in the least restrictive setting, with the Olmstead decision the most recent enunciation of this policy.3 The least restrictive setting is the patient’s home, where services are typically non-medical, although sophisticated medical services are increasingly available. At the other extreme are the LTC institutions, including NHs, large Intermediate Care Facilities for the Mentally Retarded (ICFMRs), and mental hospitals. A medical model drives care in these facilities. In between these two extremes are various emerging quasi-institutional settings, including assisted living facilities, board and care homes, group homes, and small ICFMRs.

We cannot estimate accurately how many non-elderly persons are receiving long-term care (LTC) services across all settings. In 1999, an average of 50,094 individuals resided in large state MRDD facilities,4 although the age distribution is not known. NH care is more common for adult populations. Spector and colleagues2 reported that 3.4 million adults aged 18-64 (2% of working-age adults) received LTC services in the community (in 1994), while 138,000 were in NHs (in 1996). Using data from the National Health Interview Survey Disability Supplement, they found that 71% of respondents aged 18-64 who received LTC, were cared for solely by unpaid caregivers while 6% received care only from paid caregivers.

Utilization of NH services varies by age and disability as well. Again, the study by Spector and colleagues found that individuals with physical disabilities had a mean age on admission of 50 and a mean stay of just less than 2 years. Most were in the final stages of a chronic illness (cancer) or in transition back to the community after trauma. However, some with conditions such as cerebral palsy without MR, quadriplegia due to rare syndromes or trauma, and multiple sclerosis (MS) were admitted to NHs at younger ages and had long lengths of stay Similarly, those with MR tended to enter the institution at a much earlier age (mean 28) and stay an average of 11 years.

While past research described the common diagnoses of non-elderly nursing home residents, their number and characteristics are not known. This project begins to explore these issues.

METHODS

Examination of younger residents in nursing homes requires a large primary database, as the prevalences of younger persons there, and especially subgroups thereof, are low. This project employs secondary analysis of a large data archive describing US NH residents: the National Resident Assessment Instrument/Minimum Data Set (RAI/MDS) that was federally mandated in 1991 for all NH residents. To meet its primary purpose of improving care planning,6 it includes 250+ items covering a broad spectrum from physical and cognitive function to the receipt of specialized services such as rehabilitation therapy. The MDS is performed on admission and annually thereafter, with a subset of items performed quarterly, on all residents (regardless of payment source) in nursing homes that qualify to receive federal funding. MDS items are completed using all sources of information: the resident, care providers, staff, family, medical record, etc. Extensive evaluation has shown the reliability and validity of the MDS and sub-scales.7,8,9 However, the MDS was developed primarily for use in geriatric NH residents and its capacity to be used for the rarer non-elderly populations is unproven.

Study Population:

The data for this study include all available MDS assessments from nine states that computerized the MDS early (i.e., prior to mandated computerization in 1998). Full-state data were available for the years 1994 through 1996 for Maine, Mississippi, Nebraska, New York, Ohio, Pennsylvania, South Dakota, Vermont, and Washington (1994-1996, except Washington for 1994 only). We included all “full” assessments (admission, annually, and significant change) and the abbreviated assessments performed quarterly. WALTER: I WANT TO BE ABLE TO SAY SOMETHING ABOUT THE NUMBERS IN EACH GROUP – WHAT % OF RESIDENTS WERE AGES 35-44, ETC. DID WE/COULD WE UNDUPLICATED BY PERSON HERE SO THAT WE COULD ADD THIS INFO TO TABLE 1 AND ADD A SENTENCE ABOUT THIS RIGHT AT THE FRONT OF THE RESULTS SECTION? Some states use “MDS+” forms that capture core MDS items and supplementary data. Though both types of assessments were included here, only items common to both assessments – virtually all of the items – were included in the study (and we refer in the following only to the “MDS”). Assessments were excluded if measures of either ADLs or cognitive functioning were unavailable. We selected one assessment per resident to identify the diagnoses and conditions of each resident: the assessment closest to the midpoint in the study period (July 1st, 1995) was selected to represent all states’ NH populations from 1994-1996.

Ages were separated by group: 0-4, 5-14, etc. to age 65+. Some birth dates in MDS computerized records mis-specify the century – e.g., 1996, when the true birth year is 1896. As a result, the ages of individuals 100-115 are confused with those under age 15. We thus developed an algorithm to differentiate these individuals (available from authors). While originally it appeared that 13,111 residents were aged <15, only 200 were verified (using conditions such as MRDD or low weight/height). Most others (12,607) were deemed centenarians, identified by conditions including Parkinson’s or Alzheimer’s diseases. For a small number of individuals (n=304), we had insufficient information to classify them as either centenarians or children and these observations were dropped.

Measures:

Selected MDS assessment items and scales were used to describe NH residents. Of particular use were the following:

MDS-ADL Hierarchy :10A seven-level hierarchical scale, using individual scores for bed mobility, transferring, locomotion, eating, and toileting. The scale was collapsed into three levels: self-performance, need for assistance, and total dependence in ADLs.

Cognitive Performance Scale (CPS) :11 The CPS uses five MDS items for a composite measure of cognitive functioning. Field trials found the CPS to be comparable to the Folstein Mini-Mental State Exam.11,12 Again, three levels of the scale were used: intact, impaired, and severely impaired.

Resource Utilization Groups, Version III : 13 Using 100+ MDS items, RUG-III divides all NH residents into 44 distinct case-mix categories that are both clinically relevant and descriptive of resource use. We use here the seven major clinical “Hierarchy” groups, descriptive of the primary type of resident, and the nursing cost weights (case-mix index, orCMI) associated with each RUG-III group, a relative measure of the cost of a resident’s care.

Resident Assessment Protocol (RAP) Triggers: 11,14 The 18 RAP Triggers indicate whether care planning in particular major problem domains is indicated. (Whenever possible, the new triggering logic for the MDS Version 2 was used, even though all data were the earlier MDS Version 1 or MDS+.)

The MDS data also provided information on physical functioning, diagnoses and treatments, mental health, and problem behaviors. We used both MDS-specified diagnoses (indicated by an associated checkbox) and up to six ICD-9 codes entered on the assessment form.

Descriptive Analysis

The first analytic phase was descriptive, examining the prevalence of residents’ conditions and characteristics, by age. Each age group was characterized using the MDS measures listed earlier. Significant patterns across the age groups informed the derivation of a diagnostic typology. The prevalences of added ICD-9 codes were examined overall and within each age group. We observed ICD-9 codes for other MR, schizophrenia, other psychiatric conditions, brain syndromes, and non-medically related accidents. Eventually, only schizophrenia was retained as it exclusively had a prevalence above 1% in the under 65 population.

Development of the Diagnostic Clusters

While NH admission has been associated with physical or cognitive functional decline, such decline is likely caused by an underlying chronic disease. Unfortunately, the MDS does not identify directly a resident’s primary diagnosis. As well, multiple co-morbid conditions may be contributing factors for both the need for institutionalization and the care resources used. We thus explored the patterns of (co-morbid) diagnoses and conditions using both statistical analysis and clinical knowledge of nursing home residents, with the goal of finding a parsimonious set of diagnostic clusters that would represent major diagnoses of nursing home residents likely to those primary responsible for the stay. The literature reviewed in the introduction identifies different diagnoses associated with residents of different ages; here we focused on those diseases prevalent in residents under age 65. Some diseases are attributable to congenital and perinatal conditions in infants and children, while other syndromes have an adult onset and may combine with or replace earlier childhood conditions where present. We therefore took advantage of the epidemiology of disabling neurological diseases to reduce the potentially large number of chronic conditions in the young NH resident into a more limited set of clusters of diagnoses and conditions.

To this end, we used statistical factor analysis, while working interactively with the clinical input of a small physician panel knowledgeable about NH residents and the etiology of their diseases. We began by selecting chronic conditions and diagnoses (for simplicity, we will refer to these as “diagnoses” in the following) in the relatively comprehensive MDS, the epidemiological literature, preliminary results, and panel recommendations. Both prevalence in among residents under age 65 and the “permanence” of the chronic condition were considered. For example, temporary conditions like pneumonia were not selected, while chronic conditions such as diabetes and cerebrovascular accident (CVA) were. As well, only a limited number of diagnoses representative of the over-65 population were included. The 17 chosen diagnoses included:

-Mental Retardation/Developmental Disability (MRDD - including cerebral palsy, epilepsy, autism, Down's syndrome and other organic conditions),

-Parkinson’s disease

-schizophrenia,

-Multiple Sclerosis,

-seizure disorder,

-bipolar disease,

-Alzheimer’s disease,

-other dementia,

-cerebrovascular accidents,

-congestive heart failure (CHF),

-arteriosclerotic heart disease (ASHD),

-peripheral vascular disease (PVD),

-other cardiac conditions,

-diabetes,

-cancer,

-quadriplegia

-hemiplegia.

Some groupings in this list were necessary due to relatively low prevalence (e.g., epilepsy, by itself, was not sufficiently prevalent to include as a separate diagnosis, so it was combined with other diagnoses into “MRDD”).

Many of these diagnoses are disjoint and rarely coexist (e.g., MS and Alzheimers Disease), while others represent related conditions (e.g., epilepsy and seizure disorder). To identify pairs or multiple diagnoses and conditions that usually coexist across the age spectrum, the selected items listed earlier were entered into factor analysis that wasrun both for all ages (the overall sample) and within each age category. As the diagnostic variables are dichotomous, polychoric correlations were employed and factor analyses were run following both Promax and Varimax rotations.15

From the results of these analyses, factors were determined and interpreted in the traditional approach: a diagnosis was identified with that particular factor on which it “loaded high.” Two or more diagnoses that had substantial association with a single factor were considered related and “paired.”. Factors resulting from the analyses were evaluated according to two criteria: that they provided the most consistent factor results across age groups and in the full sample, and that they had reasonable clinical interpretation. The final list of diagnostic clusters was then examined for prevalence across the age groups of NH residents and for comorbidities within the specified age groups.

RESULTS

Development of Diagnostic Clusters

Our attempt to develop diagnostic cluster began with the 17 diagnoses and chronic conditions listed earlier. These diagnoses and chronic conditions were examined in separate exploratory factor analyses in each of the age groups, with one additional model run for the entire sample (all ages). With heterogeneity in the prevalence of chronic conditions between age groups, not all factors (and thus associated pairs) were identically represented in each age group; nevertheless, we sought diagnoses that were consistently related within all age strata where they substantial prevalence. Factors from age-specific analyses provided information for developing additional factors or modifying the preliminary ones.

The factor analysis identified three domains for clustering diagnoses (results available from authors). First, MRDD combined with seizures was the only diagnosis pair that factored together consistently in every age group. Given that residents are in a NH, these diagnoses likely reflect a neuro-developmental syndrome that is more complex than would be represented solely by a developmental diagnosis or a seizure disorder. CVA with hemiplegia similarly factored together in nearly every age group, and seemed clinically appropriate because a CVA that is serious enough to be associated with NH placement will typically manifest paralysis.

Hemiplegia and quadriplegia are syndromes associated with multiple possible diagnoses, including CVA (pre- or perinatal, arterosclerotic in older age), cerebral palsy (included within MRDD), and spinal cord or brain trauma or injury (motor vehicle accidents, anoxia, infection). In our sample, these syndromes associated with different conditions in different age groups, and these associations made clinical sense. While an association of hemiplegia with CVA is present in most age groups, this was not observed in the 0-4 or 15-24 age groups. In the 5-14 and 25-34 age groups, quadriplegia factored with MRDD and seizures and in the oldest group with MS. In those 15-24, hemiplegia stood by itself without comorbidities, which could represent an undiagnosed neurological disease or the spinal cord or brain injury. In a similar vein, we noted that hemi-/quadriplegia with MRDD is very different than the latter syndromes without this condition.