| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a University of North Carolina, Chapel Hill
b University of Wisconsin, Milwaukee
c Bowman Gray School of Medicine, Wake Forest University, Winston-Salem, North Carolina
d I.D.E.A.S. Incorporated, Kirtland, Ohio
e Philadelphia Geriatric Center, Pennsylvania
f Hebrew Home for the Aged, Riverdale, New York, and Columbia University Stroud Center, New York
g University of Minnesota, Minneapolis
h Mather Institute on Aging, Evanston, Illinois, and Rush Institute on Aging, Chicago, Illinois
i University of Kansas, Lawrence
Philip D. Sloane, Goodwin Distinguished Professor, Department of Family Medicine, Cecil G. Sheps Center for Health Services Research, University of North Carolina, 725 Airport Rd. CB 7590, Chapel Hill, NC 27599-7590 E-mail: psloane{at}med.unc.edu.
Decision Editor: Fredric D. Wolinsky, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. A National Institute on Aging workgroup identified and subsequently revised items that evaluated exit control, maintenance, cleanliness, safety, orientation/cueing, privacy, unit autonomy, outdoor access, lighting, noise, visual/tactile stimulation, space/seating, and familiarity/homelikeness. The final instrument contains 84 discrete items and one global rating. A summary scale, the Special Care Unit Environmental Quality Scale (SCUEQS), consists of 18 items. Lighting items were validated using portable light meters. Concurrent criterion validation compared SCUEQS scores with the Professional Environmental Assessment Protocol (PEAP).
Results. Interrater kappa statistics for 74% of items were above .60. For another 10% of items, kappas could not be calculated due to empty cells, but interrater agreement was above 80%. The SCUEQS demonstrated an interrater reliability of .93, a test–retest reliability of .88, and an internal consistency of .81–.83. Light meter ratings correlated significantly with the Therapeutic Environment Screening Survey for Nursing Homes (TESS-NH) lighting items (r = .29–.38, p = .01–.04), and the SCUEQS correlated significantly with global PEAP ratings (r = .52, p < .01).
Discussion. The TESS-NH efficiently assesses discrete elements of the physical environment and has strong reliability and validity. The SCUEQS provides a quantitative measure of environmental quality in institutional settings.
THE physical environment is emerging as an important determinant of psychosocial and health outcomes for older persons with Alzheimer's disease and related dementias. Among demented residents of long-term care facilities, environmental factors have been linked to agitation (Nelson 1995
; Sloane et al. 1998), intellectual deterioration (Annerstedt 1994), orientation (Namazi and Johnson 1991; Namazi, Rosner, and Rechlin 1991), sleep patterns (Cruise, Schnelle, Alessi, Simmons, and Ouslander 1998), and influenza A attack rates (Drinka et al. 1996). The increased sensitivity of persons with dementia to environmental conditions occurs because the illness reduces the individual's ability to understand the implications of sensory experiences. As a result, agitated behaviors, increased confusion, delusions, and other psychiatric disturbances are readily triggered by environmental stimuli (Lawton and Nahemow 1973; Sloane et al. 1998).
Because long-term care settings are increasingly serving persons with dementia (Eppig and Poisal 1997; U.S. Congress, 1987), objective assessment of the physical environment of these care settings has become increasingly important (Teresi, Lawton, Ory, and Holmes 1994). This is heightened by the fact that physical environments are more accessible to change than many other factors that relate to outcomes among older persons with dementia, such as age, diagnosis, disease stage, and comorbid conditions (Carp 1994). Little consensus is present, however, about what environmental parameters are most salient for persons with dementia (Grant 1994), and no widely accepted measure yet exists for dementia care environments (Carp 1994).
In 1991, a National Institute on Aging (NIA) cooperative agreement funded 10 investigative teams to study dementia special care units (SCUs). A number of work groups were formed to recommend common data collection instruments and procedures, one of which was to focus on the physical environment. This article describes the instrument that resulted from these efforts, the Therapeutic Environment Screening Survey for Nursing Homes (TESS-NH). This new instrument, a substantial revision of previous versions (Sloane and Mathew 1990; Sloane, Mitchell, Long, and Lynn 1995), was given the suffix NH in recognition that its development and psychometric evaluation were conducted in the nursing home setting. In contrast to the earlier instrument, the TESS-NH is a collection of descriptive items (hence the term survey rather than scale, which was used in the previous versions). Embedded within the TESS-NH is the Special Care Unit Environmental Quality Scale (SCUEQS), which is also described in this article. Finally, preliminary reports are provided to indicate that TESS-NH items and the SCUEQS are useful in the evaluation of units other than those specializing in the care of persons with dementia.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). A poor fit between the environment and an individual's needs and desires was postulated to lead to adverse outcomes, such as negative behavior, decline in physical health, and depression (Kahana 1982; Lawton and Nahemow 1973). The physical setting was one dimension of such models; others were personal, interpersonal, and sociopolitical elements (Lawton 1986). Although early social ecological assessments often viewed the environment globally, Weisman, Lawton, Sloane, Norris-Baker, and Calkins 1994 proposed that, due to marked disagreement about the critical elements for persons with dementia, the individual dimensions should be assessed separately. In the case of SCU research, Ohta and Ohta 1988 called for "a clear taxonomy of unit characteristics" and Grant 1994 called for assessment at the unit level, to avoid confounding by facility factors.
On the basis of the above studies and reports, the NIA SCU environment workgroup recommended developing an instrument that (a) contained a limited inventory of physical environmental features, (b) addressed consensus environmental goals (described below), and (c) built on the existing Therapeutic Environment Screening Scale, a 12-item instrument developed by Sloane and Mathew 1990. The workgroup recommended that the new instrument employ direct observation, using, as much as possible, discrete indicators of sensory/spatial properties and focusing on the unit (Lawton, Fulcomer, and Kleban 1984; Weisman 1981). On the basis of its review of existing instruments (Calkins 1988; Cohen and Weisman 1991; Lawton 1980; Moos 1980), the workgroup identified six consensus goals of the physical environment in long-term care that were to be evaluated by the instrument: provision of safety, security, and physical health; orientation; provision of privacy, control, and autonomy; stimulation (both positive and negative); enhancement of socialization (social milieu); and personalization/familiarity. (A seventh conceptual goal, competence/function/negotiability, was initially identified; however, its elements were subsumed under the other domains.)
With the help of an architectural/design consultant team, the workgroup reviewed existing literature on assessment of long-term care environments (Calkins 1988; Cohen and Weisman 1991; Grant 1996; Lawton et al. 1984; Moos and Lemke 1980, Moos and Lemke 1984; Nehrke et al. 1981; Sloane and Mathew 1990) to identify items that addressed the above goals. These were grouped into domains, based on the work of Moos and Lemke 1984 and Weisman 1991, and were further modified based on review by workgroup members and by field testing. The resulting provisional instrument, the TESS-2+, was used to collect data in the NIA SCU studies. Although face-validated by numerous experts, the TESS-2+ was fielded with only modest pilot testing, without reliability studies, and without scale development. Consequently, a goal of the SCU studies was to revise the instrument and to create one or more scales for global evaluation of SCU quality.
Each of the NIA SCU study sites collected TESS-2+ data in the daytime, excluding meal times, during a 30–45-min structured walk through a unit. A training manual and a set of training slides were created and distributed to all study sites, and representatives at each site were trained by Margaret Calkins. Training involved approximately 4 hr of viewing slides with guided discussion on individual items and response categories and approximately 4 hr of onsite practice and discussion. The unit of observation was defined as a geographically distinct area of a long-term care facility or a whole facility, containing spaces for sleeping and public use. An SCU was defined as an area within a facility or an entire facility that represented itself to the public as primarily serving persons with dementia, contained a majority of persons with dementia, and was separated from the remainder of a facility by closed doors.
For these analyses, 7 NIA SCU sites contributed data on 204 SCUs and 59 non-SCU nursing home units in 10 states. Cooperating sites were the University of North Carolina at Chapel Hill (53 SCUs in Maine, Mississippi, Kansas, and South Dakota); University of California, Davis (12 SCUs and 5 non-SCUs in California); University of Kansas (48 SCUs and 33 non-SCUs in Michigan, North Carolina, and Washington); University of Minnesota (65 SCUs in Minnesota); Hebrew Home in Riverdale, New York (7 SCUs); Hebrew Rehabilitation Center (19 SCUs in Massachusetts); and the University of Maryland at Baltimore (21 non-SCU units). The 53 SCUs from the North Carolina site served as the developmental sample, and the remaining sites provided cross-validation data. The methods by which facilities were selected differed slightly across studies; however, each site attempted to achieve a random or proportional sample of nursing homes and SCUs within its study state(s).
On the basis of the distribution of responses in the data collected by the NIA SCU studies, a number of modifications were made in the instrument. Categorical items for which more than 85% of responses fell into a single response option were eliminated for lack of variability. Also, items with one or more options with fewer than 5% of responses were simplified, usually to dichotomous items. The resulting instrument was designated the TESS-NH.
Characteristics of the TESS-NH
To capture the six consensus goals of the NIA SCU workgroup, the TESS-NH contains 84 discrete items plus 1 global item that cover 13 domains. These domains include exit control, maintenance, cleanliness, safety, orientation/cueing, privacy, unit autonomy, outdoor access, lighting, noise, visual/tactile stimulation, space/seating, and familiarity/homelikeness. Good exit control provides for safer wandering (Day, Carreon, and Stump 2000). Peeling paint, loose handrails, cracked tile, wheelchairs with missing parts, broken handles, and other maintenance problems are hypothesized to constitute both a hazard and an aesthetic liability. Lack of cleanliness is hypothesized to be associated with increased infection rates, poor resident care, and resident and family dissatisfaction. Safe environments reduce injury and mobility restriction (Day et al. 2000). Although there is controversy over what methods are most effective, proper cueing has been demonstrated to reduce confusion, undesired exiting, and incontinence among persons with dementia (Day et al. 2000; Namazi and Johnson 1991; Namazi et al. 1991; Weisman 1987). Privacy is hypothesized to reduce agitation (Morgan and Stewart 1998) and increase family and resident satisfaction. Greater unit autonomy is hypothesized to enhance resident behavior, mood, and social function; reduce interruptions and noise; and facilitate staff function (Mace 1991). Outdoor access is associated with increased vitamin D levels (Brawley 1997) and reduced resident agitation (Mooney and Nicell 1992). Bright, even, glare-free illumination has been associated with improved wayfinding (Netten 1989) and reduced depression in persons with dementia (Hanger, Ancoli-Israel, Kripke, Jones, and Almendarez 1992). Noise has been associated with poor sleep (Cruise et al. 1998; Schnelle et al. 1998), reduced ability to perform tasks (Elm, Warren, and Madill 1998), and agitation (Sloane et al. 1998) in persons with dementia. Varied stimuli are characteristic of homelike, noninstitutional environments, although studies are mixed as to the optimal degree and type of stimulation in dementia (Cohen and Weisman 1991; Hall, Kirschling, and Todd 1986; Lawton et al. 1984). The absence of long corridors reduces agitation (Lawton et al. 1984; Morgan and Stewart 1998), and appropriately configured seating enhances social opportunities (Day et al. 2000). Links to the familiar have been associated with increased satisfaction, reduced agitation and exit seeking, and improved functionality (Day et al. 2000; Reed, Payton, and Bond 1998).
All of the observed items on the TESS-NH include a response format such that the higher number is hypothesized to represent a more favorable attribute of the physical environment. All items are categorical, except for the global measure of physical environment, which is in Likert format with responses ranging from 1 (low, distinctly unpleasant, negative, and nonfunctional) to 10 (high, quite pleasant, positive, and functional). The TESS-NH instrument and instructional manual are available at http://www.unc.edu/depts/tessnh.
Reliability Studies
Interrater reliability.
Two research assistants visited 12 SCUs concurrently and completed the items simultaneously and independently. Three of these facilities were in Maine, 4 in Mississippi, 4 in Kansas, and 1 in South Dakota. During each visit, the research assistants also conducted multiple observations on a subset of 17 TESS-NH items thought most likely to vary over the course of the day (e.g., lighting and cleanliness); each of these items was evaluated concurrently and independently on three occasions over the course of an 8-h data collection visit. Confirmatory interrater reliability data for binary TESS-NH items were gathered by two independent raters in 29 New York units.
Percentage of agreement and weighted kappa statistics were calculated to examine interrater reliability for categorical items; the Pearson correlation coefficient (equivalent to the intraclass correlation coefficient in the case of two raters and parallel measures) was used for continuous items.
Test–retest reliability.
Test–retest reliability was assessed in 21 non-SCU units in six nursing homes in Maryland. One evaluator conducted two observations on each unit, with the interobservation interval ranging from 103 to 150 days (M = 129 days).
Item test–retest reliability was evaluated using percentage of agreement, correlation coefficients (for continuous items), and weighted kappa statistics (for noncontinuous items). For the global rating and the SCUEQS (see below), scaled scores were compared and intraclass correlation coefficients were calculated using Kish's roh (Kish 1965).
Evaluation of Scalability of Domains and Development of a Global Quality Scale
Evaluation of scalability of domains.
Items within each of the 13 domains were evaluated for scalability by computing Cronbach's alpha (Cronbach 1951). To improve internal consistency, items that were negatively correlated or had a corrected item-to-total correlation of less than .20 were deleted, and the remaining items evaluated for scalability.
Development of the SCUEQS.
One goal of the NIA studies was to construct a single standardized and operationalized measure of overall environmental quality that would be equivalent to a global rating by an expert. To construct the SCUEQS, we considered all TESS-NH items that had a correlation above .20 with global ratings of environmental quality, adequate interrater reliability, and adequate item variability. The scale was further refined by deleting items that reduced the overall alpha. The SCUEQS was developed using data from 53 SCUs in Kansas, Maine, Mississippi, and South Dakota; validation was conducted using data from 96 SCUs in California, Michigan, Minnesota, North Carolina, and Washington.
The SCUEQS is a summary scale consisting of 18 TESS-NH items; it is computed by adding the observed value for each item. Items reflect measures of maintenance, cleanliness, safety, lighting, physical appearance/homelikeness, orientation/cueing, and noise. SCUEQS items are footnoted i or k in Table 1 . Possible scores range from 0 to 41. Use of various methods of weighting items, including z scores, did not significantly alter scale properties; thus, it is recommended that the SCUEQS be constructed by adding the items or by creating z scores.
|
Validation
Concurrent criterion validity.
Validation was conducted using the Professional Environmental Assessment Protocol (PEAP), a standardized method of expert evaluation of the physical environment of dementia SCUs (Lawton et al. 2000; Norris-Baker, Weisman, Lawton, Sloane, and Kaup 1999; Weisman et al. 1994). The PEAP involves a several-hour site visit by an expert in environmental design and aging, during which eight dimensions are rated: safety and security, awareness and orientation, support of functional abilities, facilitation of social contact, privacy, opportunities for personal control, regulation and quality of stimulation, and encouragement for continuity of self. For the validation, 44 nursing home SCUs were jointly site-visited by an expert in environmental design, who completed the PEAP, and a trained research assistant, who completed the TESS-NH.
Because the PEAP involves global evaluations of the physical environment, PEAP results were compared with SCUEQS scores for the units studied. The relationship between scores on the two instruments was assessed using Pearson's product–moment correlations. In addition, the SCUEQS items and the global rating of the overall physical environment were compared with independent PEAP dimensions.
Concurrent criterion validity of lighting items was evaluated in 52 SCUs by using a portable light meter. On the same day that TESS-NH evaluations were conducted, light meter readings were taken at four locations in activity areas. Intensity was measured in foot-candles, and evenness was computed by dividing the reading in the darkest area of a room by the reading in the brightest area. The Spearman correlation was used to examine the association between estimates of light intensity using the TESS-NH and those obtained using the light meter.
Concurrent outcome validity.
The ability of the SCUEQS to explain cross-sectional nursing home resident outcomes was evaluated in a study of factors associated with resident agitation during 3,723 observations of residents in 53 SCUs (Sloane et al. 1998). Analyses evaluated associations between TESS-NH variables and the outcomes of interest.
Use of the TESS-NH in Non-SCU Settings
As part of the NIA collaborative studies, several projects successfully used a preliminary version of the TESS-NH to assess non-SCU settings. The only logistical problem reported with data collection in nondementia units was in defining the boundaries of a unit, as non-SCU areas of nursing homes frequently have common areas that serve more than one nursing unit. To evaluate the applicability of the TESS-NH to non-SCU settings, Cronbach's alpha was computed for the SCUEQS using data from 80 dementia-care areas that did not meet the NIA criteria for a dementia unit (generally clusters within large wards) and 45 nondementia units.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
A separate study of interrater reliability of selected TESS-NH items, conducted at the New York site of the NIA studies, using two raters and 29 units, yielded kappas ranging from .75 to 1.0 for 17 out of 24 items examined. Two additional items had kappas that could not be computed due to empty cells; however, agreement was perfect. Only 2 items had low kappas: .43 for light evenness in activity areas and .45 for status of television in public areas.
Test–Retest Reliability
Item analysis indicated that environmental factors that are the most fixed, such as the nature of the floor surface, demonstrated high levels of test–retest reliability (agreement above .8). Those that reflect resident or staff behavior, such as adequacy/evenness of lighting, the amount of dirt/litter, the level of maintenance, and the presence of odors, demonstrated moderate to substantial agreement. The global subjective rating item demonstrated an intraclass correlation coefficient of .81; the test–retest reliability (intraclass correlation coefficient) of the SCUEQS was .88.
Scalability of Items Within TESS-NH Domains
When items within each of the 13 TESS-NH domains were evaluated for scalability, the following domains could not be scaled (reasons are given in parentheses): unit autonomy (dispersion problems in Item 3, inadequate number of observations of Items 3 and 4 due to item revision), exit control (items mutually exclusive), space/seating (only 2 items; inadequate observations of Item 15 due to revision; Item 16 not ordered), access to outdoors (only 2 items; inadequate observations of Item 27 due to revision), orientation/cueing (inadequate number of observations due to item revision), and privacy (only 1 item). The means, standard deviations, and Cronbach's alphas for the other domains were as follows: for maintenance (4 items), the mean was 6.6 (SD = 1.6.) and the alpha was .83; for cleanliness (6 items), the mean was 6.7 (SD = 1.4) and the alpha was .72; for safety (5 items), the mean was 6.6 (SD = 2.3) and the alpha was .71; for lighting (6 items), the mean was 7.5 (SD = 2.3) and the alpha was .62; for familiarity/homelikeness (5 items), the mean was 6.9 (SD = 2.9) and the alpha was .63; for visual/tactile stimulation (4 items), the mean was 8.4 (SD = 2.2) and the alpha was .54; and for noises (8 items), the mean was 9.2 (SD = 2.4) and the alpha was .33
Psychometric Properties of the SCUEQS
The mean SCUEQS in the development data set was 26.3 (SD = 5.6), and the range was between 13 and 38. Corrected item-to-total correlations for the SCUEQS ranged between .25 and .67. The Cronbach's alpha in the development data set was .81; in the validation data set, the Cronbach's alpha was .83. As noted above, SCUEQS scores demonstrated an intraclass correlation coefficient of .88 on test–retest data and .93 on interrater data. Table 2 summarizes the item composition and psychometric properties of the SCUEQS.
|
|
). Adequacy of the factor pattern was examined through inspection of the residuals (difference between the observed and reproduced correlations from the pattern matrix), which showed that most were less than .05; eight were greater than .12. The overall root mean square off-diagonal residual was .07, indicating adequate fit; however, there were some items that fit less well, most notably the two items related to odors. As shown in Table 3 , there is evidence of a possible second factor consisting of items related to lighting, noise, and visual and other stimulation. A number of the items loading on the second factor were not well explained by the factors extracted, a not-unexpected result given that the methodology for scale development was criterion referenced rather than attempting to define a latent attribute (Seraphine 2000).
Concurrent Validity
When SCUEQS scores were compared with independently conducted expert assessments using the PEAP in 44 SCUs (Norris-Baker et al. 1999), the correlation between the global PEAP assessment (a 5-point scale) and the SCUEQS was moderately strong (r = .52, p < .01). In addition, the correlation between the global PEAP scores and the TESS-NH global rating item was very strong (r = .68, p < .01).
Light meter levels at four locations in 53 SCU activity areas correlated significantly with TESS-NH ratings of light intensity (rs = .29–.38; ps = .01–.04). Similarly, TESS-NH ratings of light evenness of activity rooms in the same units correlated significantly with the ratio of light meter readings in midroom divided by readings at the darkest area of the room (r = .33, p < .05).
Concurrent Outcome Validity
In a study of 53 SCUs in four states, a correlation of -.34 (p < .01) was identified between unit SCUEQS scores and the prevalence of resident agitation, averaged over multiple direct daytime observations. Individual items demonstrating significant negative correlations (p < .05) with agitation included hall cleanliness, maintenance in public areas, maintenance of bedrooms, urine/stool odor in public areas, availability of kitchen for resident use, and resident grooming. In multivariate analyses involving 35 measures of facility, unit structure, staffing, and treatment procedures, the SCUEQS emerged as the strongest independent predictor of the level of observed resident agitation (Sloane et al. 1998).
Use of the TESS-NH in Non-SCU Settings
In the combined NIA SCU data set, the mean global ratings across settings were 6.46 for 204 SCUs, 5.31 for 80 non-SCU dementia units (e.g., clusters without closed doors), and 6.93 for 45 nondementia units (Sloane et al. 1995). Cronbach's alpha for the SCUEQS was .78 in the non-SCU dementia units and .63 for the non-SCU units.
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). During the 1990s, two additional instruments were developed that evaluated the environment of institutional settings for persons with dementia: the PEAP (Lawton et al. 2000; Norris-Baker et al. 1999; Weisman et al. 1994) and the Nursing Unit Rating Scale (NURS) (Grant 1996). The PEAP is limited, however, by its global nature and the fact that its ratings must be performed by an expert in environmental design (Carp 1994); the NURS deals primarily with policy and program features rather than physical environmental attributes (Lawton, Weisman, Sloane, and Calkins 1997). Thus, an objective, dementia-focused environmental assessment tool is needed.
This article describes the development, components, and characteristics of a new instrument, the TESS-NH. It can be completed in a 30–45 min walk through a unit by someone with a relatively modest level of training and experience in person–environment concepts (Lawton et al. 1997). It contains a summary scale, the SCUEQS, that appears to have strong reliability and validity and that consists of discrete, standardized, easily understood components, thereby better identifying sources of differences and, potentially, targets for interventions. Originally developed for the evaluation of nursing home SCUs, the TESS-NH has been successfully used in nonspecialized nursing home units.
The most basic use of the TESS-NH is in the description of individual elements of the physical environment of institutional settings that are considered by experts to be of potential importance in the provision of quality care for persons with Alzheimer's disease and related dementias. A further potential use may be for self-evaluation by administrative and clinical staff of nursing homes and other long-term care facilities. In addition, the SCUEQS provides a single measure of overall environmental quality that may be useful in comparative outcome studies of long-term care settings, thereby distilling environmental quality into one variable that can, for example, be entered into regression analyses. Similarly, the SCUEQS, or a modification of it, might be useful to families or others interested in making a quick assessment of the overall quality of an SCU.
The results of an exploratory factor analysis of the SCUEQS indicate that the item set may not be unidimensional in that a second (albeit minor) factor emerged; however, inspection of the pattern and structure matrices (Table 2 ) shows that several of the items load on both factors and that the loadings for numerous items are not especially high. Although definitive interpretation must await cross validation, this result is not unexpected because the scale was developed using a procedure that does not necessarily produce an item set that is highly intercorrelated or that has high correlations with a specific underlying factor. Because of the exploratory nature of the analyses, it is not possible to define the exact dimensionality of the scale; nonetheless, the results provide preliminary data for future investigation.
The results of the exploratory factor analysis should be considered in the context of different goals of scale development. One goal is to best measure a theoretical underlying latent attribute; a second is to reduce parsimoniously the number of variables in an analysis; a third is to maximize the relationship of a measure with a criterion. Exploratory factor analyses have frequently been used as a first step in pursuing the first goal (to ensure a highly intercorrelated item set that is reflective of the underlying factors). The SCUEQS was developed to address the third goal, and this method of scale development will not necessarily produce the most satisfactory factor analytic result. Nonetheless these results suggest that (a) the SCUEQS, although developed using a different philosophy, does appear to reflect a theoretical construct, (b) it can be treated as a relatively homogeneous measure of such a construct, and (c) there is some evidence suggesting that the SCUEQS may reflect more than one underlying dimension. Thus, although investigators using the SCUEQS are encouraged to conduct factor analyses to determine the dimensionality of the SCUEQS in other data sets, the current recommendation would be to use the SCUEQS as a single scale.
Estimates of the interrater and test–retest reliability of the TESS-NH were generally high. However, caution must be taken because few raters were involved. The typical study of test–retest reliability examines only a subset of cases rated by two or more raters, with the remaining cases rated by only one rater. When only one rater rates cases, the interrater reliability may be overestimated. Moreover, examining interrater reliability when only two raters are involved limits decomposition of the variance due to raters and may therefore not adequately represent the variability in ratings (Fan and Chen 2000). Furthermore, because the item base rate influences reliability coefficients, interpretations of the magnitude of coefficients must be made cautiously. Landis and Koch 1977 and Fleiss 1981 presented widely cited, although somewhat controversial, guidelines for interpreting coefficient kappa (see, e.g., Dunn 1989; Roberts and McNamee 1998). Landis and Koch considered .40 as moderate agreement and .60–.80 as substantial.
Not all items in the final TESS-NH meet the criteria for moderate or substantial agreement. Additional items were retained for one or more of the following reasons: (a) Items 7a, 7b, 13c, 15, and 28a2 were revised after primary data collection, with expectation for greater reliability with increased rater training and experience; (b) skewed distributions reduced or invalidated the kappa of Items 4c, 5a, 11a, 11b, 28d1, 28e1, 28g1, and 29a; (c) rater bias was noted during reliability testing of Items 8a, 8b, 8c, 9a, and 9b, thereby lowering the percentage of agreement; and (d) Items 6e, 6i, 10c, 27a, and 27b were believed to be the best available measure of an important environmental domain. Regardless of reported reliabilities, investigators using the TESS-NH are encouraged to conduct reliability studies on all items in their own settings.
Several additional caveats are noted. Because few principles of environmental quality are backed by controlled trials, both the TESS-NH and the SCUEQS rest largely on untested empirical grounds. Also, some resident-specific goals of physical design are difficult to evaluate on a unit level, and therefore are poorly represented by the TESS-NH. The most noteworthy example is the capacity of the environment to maximize resident function, in that environmental interventions may require evaluation at the resident level rather that at the unit level (Lawton et al. 1997; Namazi and Johnson 1992; Norris-Baker et al. 1999). Furthermore, because the development of the TESS-NH favored retention of items that had some dispersion across the nursing home SCUs in the NIA data set, rare design elements or innovations from other sectors (e.g., assisted living) would not be detected by this instrument. Thus, the TESS-NH provides estimates that are rooted in the current state of the field of environmental design and health. Future work will better establish the role of the TESS-NH and its variables in evaluating quality and will identify areas where additional or different measures are appropriate.
|
Acknowledgments |
|---|
This is dedicated to the memory of Powell Lawton, our colleague and mentor.
Received for publication September 10, 1999. Accepted for publication May 23, 2001.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. Hyde, R. Perez, and B. Forester Dementia and Assisted Living Gerontologist, December 1, 2007; 47(suppl_1): 51 - 67. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. L. Chappell, R. C. Reid, and J. A. Gish Staff-based measures of individualized care for persons with dementia in long-term care facilities Dementia, November 1, 2007; 6(4): 527 - 547. [Abstract] [PDF]
|
|
|
|
|
|
M. Boustani, S. Zimmerman, C. S. Williams, A. L. Gruber-Baldini, L. Watson, P. S. Reed, and P. D. Sloane Characteristics Associated With Behavioral Symptoms Related to Dementia in Long-Term Care Residents Gerontologist, October 1, 2005; 45(suppl_1): 56 - 61. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Zimmerman, P. D. Sloane, C. S. Williams, P. S. Reed, J. S. Preisser, J. K. Eckert, M. Boustani, and D. Dobbs Dementia Care and Quality of Life in Assisted Living and Nursing Homes Gerontologist, October 1, 2005; 45(suppl_1): 133 - 146. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. L. Howard, P. D. Sloane, S. Zimmerman, J. K. Eckert, J. F. Walsh, V. C. Buie, P. J. Taylor, and G. G. Koch Distribution of African Americans in Residential Care/Assisted Living and Nursing Homes: More Evidence of Racial Disparity? Am J Public Health, August 1, 2002; 92(8): 1272 - 1277. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
