| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Center for Research in Applied Gerontology
3 Department of Biostatistics
4 Center for Aging, University of Alabama at Birmingham.
2 Center for Mental Health and Aging, The University of Alabama, Tuscaloosa.
5 School of Nursing, University of Pittsburgh, Pennsylvania.
Address correspondence to Louis Burgio, Center for Mental Health and Aging, Box 870315, The University of Alabama, Tuscaloosa, Alabama 35487 or to David Vance, Center for Research in Applied Gerontology, HMB100, Birmingham, Alabama 35294. E-mail: lburgio{at}sw.ua.edu or devance@uab.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Behavioral agitation has been reported to occur in 43%–86% of nursing home residents (Beck, Rossby, & Baldwin, 1991
; Ryden, Bossenmaier, & McLachlan, 1991). The behaviors associated with agitation (e.g., physical aggression and disruptive vocalization) can create difficulties in caring for residents and can place an enormous strain on formal caregivers (Algase, Kupferschmid, Beel-Bartes, & Beattie, 1997; Merrian, Aronson, Gaston, Wey, & Katz, 1988).
Researchers have proposed that agitation results from an incongruent person–environment fit. Beck and colleagues (Beck & Vogelpohl, 1999; Rossby, Beck, & Heacock, 1992) proposed that when biopsychosocial needs are not met, individuals will often react with behavioral agitation. More specifically, these authors hypothesized that agitation results from the disequilibrium of homeostasis brought on by the interaction of relatively stable background factors (e.g., neurological, cognitive, and psychological) and transient proximal factors (physical and social environment). Particularly in cognitively impaired individuals, there are numerous opportunities for unmet needs because these individuals lack internal and external resources needed to meet these needs.
Several background factors have been associated with the expression of agitation. These include, but are not limited to, gender, cognitive status, sensory loss, and impairment in activities of daily living (ADL; Beck et al., 1998; Horowitz, 1997). Unfortunately, the results of many empirical studies examining these background factors have been mixed. Although some studies have demonstrated a clear association between gender and agitation (Burgio et al., 2000; Cohen-Mansfield, Werner, & Marx, 1989; Cooper, Mungas, & Weiler, 1990; Jackson et al., 1989; Levy et al., 1996; Teri, Borson, Kiyak, & Yamagishi, 1989), others have not (Aarsland, Cummings, Yenner, & Miller, 1996; Malone, Thompson, & Goodwin, 1993; Ryden et al., 1991; Wagner, Teri, & Orr-Rainey, 1995). With regard to cognitive status, most studies have reported an inverse relationship with agitation (Aronson, Post, & Guastadisegni, 1993; Beck et al., 1998; Burgio et al., 1994; Cooper et al., 1990; Levy et al., 1996; Swearer, Drachman, O'Donnell, & Mitchell, 1988). Agitation increases as residents' cognitive status worsens, although profoundly impaired residents tend to display fewer behaviors overall, including agitation. A similar pattern describes the association between ADL status and agitation (Beck et al., 1998; Burgener, Jirovec, Murrell, & Barton, 1992; Burgio et al., 1994; Levy et al., 1996), which is logical, considering that cognitive decline often precipitates ADL decline (Burton, German, Rovner, & Brant, 1992).
Sensory impairments have also been implicated as possible causative factors in agitation. Horowitz (1997), using optometric examination of nursing home residents with dementia, found impaired vision to be an independent contributor to agitation. Uhlmann and colleagues (Uhlmann, Larson, & Koepsell, 1986; Uhlmann, Larson, Koepsell, Rees, & Duckert, 1991; Uhlmann, Teri, Rees, Mozlowski, & Larson, 1989) have found that impaired auditory functioning is related to elevated rates of reported agitation. The proposed mechanism is that demented residents are disoriented by the lack of clear auditory and visual information, thus predisposing them to exhibit agitation.
Although these aforementioned factors have been identified as contributing individually to agitation, their relationship to each other and their common contributions to agitation have never been examined using two different modes of measuring agitation. The purpose of this project was to develop a multivariate predictive model of agitation as assessed through two different modes of measurement: staff retrospective reports using a standardized questionnaire (Ray, Taylor, Lichtenstein, & Meador, 1992) and direct behavioral observation (Burgio et al., 1994). We performed structural equation modeling (SEM) using LISREL to evaluate the fit of models that examined four hypotheses. Hypothesis 1 asserted that sensory declines would have a direct effect on agitation and an indirect effect on agitation through its influences on ADL function (Marx, Werner, Cohen-Mansfield, & Feldman, 1992) and cognitive status (Lindenberger & Baltes, 1994; Uhlmann et al., 1991). Hypothesis 2 posited that gender would have a direct effect on agitation and an indirect relationship through its influence on ADL function. Hypothesis 3 stated that cognitive status influences agitation directly and indirectly through ADL function. Hypothesis 4 predicted that ADL function will directly influence agitation.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
Paper-and-Pen Measures
Barthel self-care rating scale (Barthel)
The Barthel (Sherwood, Morris, Mor, & Gutkin, 1977) consists of 17 items that assess the amount of assistance that is required for the resident to complete ADL. Information was obtained from the residents' primary certified nursing assistants (CNAs). Each item was rated on a 4-point Likert-type scale with lower numbers indicating less dependence on staff and higher numbers indicating a greater dependence. Sherwood and colleagues (1977) reported high alpha reliabilities (ranging from.95 to.96), suggesting that the measure is an internally consistent measure of self-care.
Severe Impairment Battery (SIB)
The SIB (Saxton, McGonigle-Gibson, Swihart, & Boller, 1993) is a 57-item instrument that assesses a wide range of cognitive and functional abilities in adults with dementia. Each item was rated on a 3-point Likert-type scale, with lower numbers indicating greater impairment. This measure possesses good interrater reliability and has been used to assess change over time (Schmitt et al., 1997). The SIB has shown excellent discrimination for adults with moderate and severe dementia (Panisset, Roudier, Saxton, & Boller, 1994; Saxton & Swihart, 1989). This instrument was completed for each resident by the senior research nurse.
Mini-Mental State Exam (MMSE)
The MMSE (Folstein, Folstein, & McHugh, 1975) is a simple and relatively quick assessment of cognitive mental status. The maximum score is 30. The MMSE was administered by trained research staff.
Clinical Dementia Rating Scale (CDR)
The CDR (Burke et al., 1988; Hughes, Berg, Danziger, Coben, & Martin, 1982) is a 5-point scale used to discriminate residents without dementia (0), those with questionable dementia (0.5), and those with various levels of dementia (mild, 1; moderate, 2; and severe, 3). The overall CDR score combines six individual categories of cognitive functioning: memory, orientation, judgment and problem solving, involvement in community affairs, involvement in home and hobbies, and personal care. Each of these categories is scored independently on a 5-point scale. The information used to form the ratings in the CDR is derived from an examination of the resident and collateral sources such as interviews with caregivers, standardized neurological and psychological tests, medical history, or a combination of these. Using the CDR, a physician rated each of the six cognitive categories to provide a more comprehensive picture of the resident's level of dementia severity. Documented interrater agreement on overall CDR scores is 80% (Burke et al., 1988).
Medical and Psychiatric Diagnosis Form (MPDF)
The MPDF is an experimenter-generated form that gathers extensive information on residents' medical and psychiatric history and current diagnoses. This information was obtained from the medical record, interviews with nursing staff, and, when possible, family members. Data were gathered by trained research assistants under the supervision of a Doctor of Science in Nursing nurse (Ann Yurick).
Medication Tracking Form
The Medication Tracking Form is an experimenter-generated form used to assess the number, types, and dosages of medications each resident was prescribed. This information was extracted from the medical record by trained research assistants.
Nursing Home Behavior Problem Scale (NHBPS)
The NHBPS (Ray et al., 1992) is a 29-item inventory of behavioral disturbances. The rater is asked to indicate the frequency of each behavior as observed in the past 3 days. The frequency of occurrence scale consists of 5 points (0 = never, 4 = always). The sum of the 29 items indicates the overall score, with higher scores representing a greater frequency of behavior problems. This instrument was administered to the resident's primary CNA in a face-to-face interview with a research assistant. Only items that directly corresponded to the operational definitions used in the computer-assisted behavioral observation system were used in the analysis (see Appendix). The interrater reliability reported by the authors ranged from 0.75 to 0.83.
Computer-Assisted Behavioral Observational System (CABOS)
The CABOS was used to record the occurrence and duration of resident and staff behaviors (Burgio et al., 1994). Kappa reliabilities for behavioral categories ranged from 0.78 to 1.00, with a mean of 0.95 across categories. The observation system sampled staff and resident behaviors on a second-by-second basis throughout the day on the nursing units. Residents were the main target of the observation sessions. Twelve 30-min observations were scheduled for each resident during a 3-week period. Each hourly interval was sampled between 8:00 a.m. and 8:00 p.m. The schedule of observations was generated through a stratified random time sampling method with the condition that not more than one observation could be conducted on an individual resident on any day. This condition was imposed in an attempt to distribute the observations evenly over the 3-week period. Consequently, the CABOS sampled 6 hours of resident time, during which agitation (verbal disruptions, restlessness, and physical aggression) could be recorded. The percentage of observation time that agitation was observed was calculated, and this percentage was analyzed as the primary dependent variable in this study. In past studies, this method and schedule of observation have been sufficiently sensitive to detect changes in agitation in response to psychosocial interventions in the nursing home (Burgio et al., 2001; Burgio et al., 2002).
Procedure
Before the observation sessions, the observer entered identifying information in the residents' computer file (e.g., subject number, date, and time of day). Only one resident was observed at a time. The observer then keyed the location, activity, sound, social environment, and restraint use of the target resident. Other categories were keyed when they occurred. Observers were instructed to keep the resident in view at all times during the observational period. Observers were present in the setting 2–3 weeks before observation to reduce observer reactivity. Haynes (1978) indicated that lengthy preexposure has been found to decrease reactivity. Observers attempted to position themselves at the maximum distance from the target resident, and they were instructed to limit eye contact with the resident. If a resident was aware of the observer, the observer greeted the resident and then informed him or her that the observer would be doing some work on the other side of the room and to ignore him or her. It was the impression of the observers that residents quickly habituated to their presence.
A research assistant completed the paper-and-pen assessments by abstracting information from the residents' medical records or from face-to-face interviews with the residents' primary CNA. All paper-and-pen assessments were administered during the 3-week direct observational period. The specific timing of the assessments was determined by the availability of the CNAs and were completed throughout the 3-week period.
Data Analysis
Correlations among study variables were computed by means of SPSS-X (Noru
is, 1993) and LISREL (Jöreskog & Sörbom, 1993). A latent variable was extracted for cognitive status, and a causal model was constructed to determine the effects of sensory impairment, cognitive status, and ADL status on agitation. Standard fit indices were used to evaluate and compare the fit of models. Standard absolute fit indices, such as the goodness-of-fit index and the adjusted goodness-of-fit index, were used because the models were to be compared with a saturated model instead of a nested one. The chi-square test was also used as a conventional overall test of fit; this test is based on the discrepancy between the sample and fitted covariance matrices and is often overly sensitive.
It was proposed that each model would be constructed with agitation as the dependent variable and the proposed latent and observed variables serving as the predictor variables (see Figure 1). Cognitive status and ADL status were modeled as possible mediators of the effects of sensory impairment and gender on agitation. In addition to tests of overall model fit, t tests of the significance of each estimated path were examined, and nonsignificant paths were subsequently fixed to zero to improve model parsimony. Finally, modification indices were examined to identify paths that could be added to improve overall model fit.
|
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
Table 2 displays the correlation matrix for the variables used in this analysis. Standard Pearson correlations were estimated between pairs of continuous variables, whereas polychoric correlations were estimated when one or both variables were measured on an ordinal or dichotomous scale. Violations of multivariate normality were explored using bivariate graphs and Mahalanobis distance. Although outliers were located, none exceeded four standard deviations; therefore, these analyses report findings from all residents. Given the skewness and kurtosis that were observed for some variables, maximum likelihood estimation was used because of its robustness in the presence of nonnormality (Hoyle & Panter, 1995; Jöreskog & Sörbom, 1993; Kelloway, 1998). In an effort to be thorough, data were transformed and compared with all of the models that used nontransformed data. Model fit was similar using two methods, and the same paths remained statistically significant.
|
have suggested that SEM analysis should follow a sequential procedure whereby different models are compared in a logical manner. In our analysis, we first constructed a full or baseline model with seven conceptual variables—gender, hearing impairment, vision impairment, cognitive status, ADL impairment, and two separate variables for agitation (CABOS and NHBPS). Cognitive status was a latent variable with MMSE, SIB, and CDR as indicators. Vision and hearing impairment were examined as predictors of cognitive status, ADL status, and agitation, whereas gender was hypothesized to be related to ADL status and agitation only. Cognitive status and ADL status were examined as possible mediators of the effects of sensory impairment on agitation.
The independence model, which tested the hypothesis that all variables are uncorrelated, was tested first and rejected, 
2(36, N = 123) = 341.78, p <.001. The baseline model was tested next, and fit was much improved, though not perfect, 2(14, N = 123) = 51.73, p <.001. A chi-square difference test indicated a significant improvement in fit between the independence model and the baseline model, 2diff(22, N = 123) = 290.05, p <.001. Many of the model estimates were nonsignificant, so a trimmed model was constructed by sequentially removing nonsignificant paths and reevaluating model fit. The least significant paths (based on the lowest t value) were dropped one at a time; estimates were then recalculated. This continued until only statistically significant predictive paths remained (p <.05).
Figure 1 shows the fully trimmed model, including coefficients in standardized form. This model also provided good fit to the observed data, 2(24, N = 123) = 61.62, p <.001. A chi-square difference test between the baseline model and the trimmed model revealed no significant difference, 2diff(10, N = 123) = 9.89, p >.10, indicating that the trimmed model fit as well as the baseline model. Standard fit indices for the baseline model and the trimmed model are displayed in Table 3. Because post hoc model modifications were performed, a correlation was calculated between the baseline model estimates and the trimmed model estimates (Ullman, 1996). A very high correlation was observed (r = 0.99), indicating that the parameter estimates for the statistically significant paths were unchanged after deleting several nonsignificant paths from the model.
|
2 (23, N = 123) = 40.54, p =.013. The final trimmed model shows that gender, vision impairment, hearing impairment, and cognitive status directly affected agitation as measured by the CABOS. However, only cognitive status had a significant effect on agitation as measured by the staff report NHBPS. Hearing impairment was strongly predictive of greater agitation as measured by CABOS (standardized coefficient =.42), and vision impairment was negatively predictive of greater agitation (standardized coefficient = -.21) after controlling for the other predictors. Gender was predictive of greater agitation as measured by CABOS (standardized coefficient =.21), with higher percentages observed for women compared with men. Cognitive impairment was strongly predictive of higher levels of ADL impairment (standardized coefficient = -.64) and agitation as measured by both the CABOS and NHBPS (standardized coefficients = -.35 and -.28, respectively). However, neither cognitive status nor ADL impairment were effects of sensory impairment. Consequently, significant indirect or mediated effects were not observed.
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Rossby et al., 1992) as contributing to behavioral agitation in individuals suffering from dementia. Through the use of SEM analytical techniques, we were able to investigate both the independent contributions of these variables to agitation and the common contributions among these factors. Specifically, support for Hypothesis 1 was found in that sensory function was directly related to agitation; however, its influence on ADL was not found. Examination of Hypothesis 2 revealed that gender does have a relationship to agitation but not to ADL as posited. In Hypothesis 3, cognitive status was found to be strongly related to agitation and also related to ADL function. However, ADL function did not mediate the effect of cognitive status on agitation because ADL function was not found to contribute to agitation, as we anticipated it would in Hypothesis 4. Collectively, the results showed that hearing and vision impairment, gender, and cognitive status were all independent predictors of agitation as measured by CABOS. However, none of the mediated or indirect effects that we hypothesized was supported by these data. After deleting nonsignificant paths from our baseline model, the trimmed model provided a relatively parsimonious multivariate predictive model for agitation. Excellent model fit was obtained when we allowed a correlated residual between the SIB and hearing impairment. Although this path was added as a post hoc adjustment to the model, it may reflect the greater sensitivity of the SIB relative to the other measures of cognitive status for assessing individual differences in cognitive functioning at lower ability levels. Consequently, although hearing impairment was not directly related to cognitive status as indexed by our latent variable, accounting for some specific residual correlation between the SIB and hearing impairment significantly improved model fit to acceptable levels.
Perhaps the most significant finding in this study was that residents with significant hearing impairment were more likely to engage in agitation than less impaired residents. Uhlmann and colleagues (1991) proposed that lack of clear auditory input increases resident disorientation, which, in turn, results in agitation. Burgio and colleagues (1994) suggested that hearing impairment in dementia patients may result in a state of sensory deprivation. Thus, it is possible that engaging in disruptive vocalization (a major component of agitation) may function as a form of self-stimulation in an effort to reestablish homeostasis. Perhaps the most parsimonious explanation is that, due to the reduction in auditory input, residents speak loudly or display other vocalizations that are coded as disruptive.
Researchers have used audiotaped environmental sounds (Burgio et al., 1994) and music (Gerdner, 2000) to successfully reduce agitation in nursing home residents. It is unknown whether these stimuli served as auditory enrichment in a barren environment or whether they provided relaxing auditory stimuli for agitated individuals. If agitation is related to disorientation or sensory deprivation, merely providing hearing aids to hearing-impaired residents could reduce agitation. However, Burgio and Lewis (1999) reported that providing an audio augmentation device to residents failed to reliably reduce disruptive vocalization.
Although there is only one study in the literature that reported increased agitation in visually impaired residents (Horowitz, 1997), our findings suggest that visually impaired residents were less likely to display agitation after controlling for the effects of gender, hearing impairment, and cognitive status. It is important to clarify that the raw association between visual impairment and agitation was not statistically significant (r =.03; see Table 2), but after controlling for other predictive relationships, visual impairment was found to have a protective effect. As with hearing impairment, one could argue that a visual impairment could similarly result in disorientation and sensory deprivation. One possible explanation for our finding is that residents with dementia and hearing impairment are less likely to call out to staff and other residents if they are unable to locate them visually in the environment. Also, it should be noted that Horowitz (1997) only included residents who ranged from moderately cognitively impaired to no impairment. The residents in our study were severely cognitively impaired (mean MMSE = 7.72). Nevertheless, we must conclude from our results that visual impairment might actually provide a protective effect, especially for severely cognitively impaired nursing home residents. However, this result awaits replication.
Our finding that women showed higher agitation, particularly disruptive vocalization, than men corroborates the findings of several previous studies (Burgio et al., 2000; Jackson et al., 1989; Levy et al., 1996; Teri et al., 1989). In fact, Burgio and colleagues (2000) found that female residents exhibit almost three times as much agitation as male residents. This gender difference may result from women being socialized to be more verbally expressive than men. It should be noted, however, that this gender difference was detected using direct behavioral observation (CABOS) and not with the staff report measure (NHBPS). This may partially explain the mixed findings in prior studies examining gender and agitation.
Cognitive status was conceptualized as a latent variable with significant loadings on three observed variables. All three observed variables were measured with well-documented and standardized instruments. The factor loadings and residual errors showed minimal variability across baseline and trimmed models. In every scenario, decreased cognitive status was associated significantly with increased agitation. As discussed earlier, this association has been reported in numerous studies (e.g., Aronson et al., 1993; Beck et al., 1998; Burgio et al., 1994; Everitt, Fields, Soumerai, & Avory, 1991; Levy et al., 1996; Spector, 1991; Swearer et al., 1998). The present study uniquely demonstrates the role and strength of this association in the context of a latent variable measurement model and a structural equation modeling approach.
Measurement of Agitation
Creating a reliable and valid measure of agitation is essential for examining potential predictors. The NHBPS and CABOS were originally conceptualized to be alternative indicators of the same phenomena. We hypothesized that because the NHBPS measure was standardized, and because the CABOS displayed high interrater reliability, the two measures would be highly correlated and load on the latent variable of agitation. This hypothesis was suggested by findings from prior research that test–retest scores from staff report measures are stable up to 1 month between assessments (Koss et al., 1997; Patterson et al., 1997). However, the correlation between the two measures was unexpectedly low, and this contraindicated the extraction of a single latent variable. The low correlation between the two measures strongly suggests that they are capturing different aspects of agitation. Consequently, they were viewed as separate factors on our SEM analyses.
The results showed that several factors (gender, vision impairment, hearing impairment, and cognitive status) were significantly related to the CABOS measure; only cognitive status was a significant predictor of agitation as measured by both the CABOS and the NHBPS. The NHBPS is a subjective, retrospective survey instrument that was completed by CNAs. It is possible that those residents who exhibit extreme or memorable behaviors such as physical aggression might be perceived by staff to be more problematic even though the actual frequency of their behavior problems might not be that different from other residents who exhibit less extreme, less salient forms of aggression, such as repetitive questions. This severity bias could then be generalized to other behavior problem categories; consequently, a subjective evaluation like the NHBPS might be unduly affected. McCann, Gilley, Hebert, Beckett, and Evans (1997) discussed the possibility of this severity-to-frequency attribution in their paper, in which they also reported discordant findings between direct observation and staff ratings of behavior in nursing home residents with dementia. Their discussion extends to other rating instruments such as the Cohen-Mansfield Agitation Inventory (CMAI; Werner, Cohen-Mansfield, Koroknay, & Braun, 1994) and the Nurse Oriented Scale for Inpatient Evaluation (NOSIE; Ray et al., 1992). In this regard, Ray and colleagues (1992) reported correlations between the full NHBPS, the CMAI, and the NOSIE as.91 and -.75, respectively.
Inconsistencies in the measurement of agitation and its impact might be resolved by assessing the level of burden associated with behavioral disturbances in the nursing home. However, in a recently completed study from our lab, we found that CNAs rarely report behavioral burden (Allen et al., 2002). It is our suspicion that CNAs might consider it socially undesirable to admit burden in the workplace.
This study has several limitations. First, our sample of residents was almost exclusively White; thus, our findings cannot be generalized to the increasingly diverse nursing home population. Second, due to fiscal constraints, the CABOS system sampled behaviors from 8:00 a.m. to 8:00 p.m. (See Burgio, 1996, for a discussion of the advantages and disadvantages of using direct observation in behavioral research.) Nighttime behavioral disturbances, reported to be quite common (Bliwise, Carroll, Lee, Nekich, & Dement, 1993; Bliwise et al., 1995), were not represented in these data. However, it is possible that day shift CNAs completing the NHBPS may have been aware of residents' nighttime behavioral disturbances. This awareness could have influenced their ratings on this measure and could have, in turn, contributed to the relatively low correlation between the CABOS and NHBPS. Finally, although the prevalence of hearing and visual impairment detected through our ordinal scaling of medical record information is similar to national norms (Dey, 1997; Hing, Sekscenski, & Strahan, 1989), data from more sensitive formal auditory and visual assessments may have rendered different results. We recommend that future studies use audiologists and optometrists to establish the extent of these sensory limitations.
In conclusion, this study has established that gender, hearing impairment, and cognitive status are independent predictors of agitation in the nursing home. Our results are striking in that these relationships are found only when agitation is measured through direct observational methodology. Only cognitive status was found to be related to agitation as measured by retrospective staff report NHBPS. This methodological finding may help explain the often contradictory results of prior studies investigating predictors of agitation in nursing home residents.
|
Acknowledgments |
|---|
Received for publication October 18, 2001. Accepted for publication July 15, 2002.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
is, M. J. (1993). SPSS: SPSS for Windows based systems user's guide, Release 6.0. Chicago: SPSS.This article has been cited by other articles:
 |
|
 |
|
  L. D. Burgio, N. S. Park, J. M. Hardin, and F. Sun A Longitudinal Examination of Agitation and Resident Characteristics in the Nursing Home Gerontologist, October 1, 2007; 47(5): 642 - 649. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. S. Allen, L. D. Burgio, S. E. Fisher, J. Michael Hardin, and J. L. Shuster Jr., Behavioral Characteristics of Agitated Nursing Home Residents With Dementia at the End of Life Gerontologist, October 1, 2005; 45(5): 661 - 666. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
0.05); broken lines indicate proposed nonsignificant paths. NHBPS = Nursing Home Behavior Problem Scale; CABOS = Computer-Assisted Behavioral Observational System; ADL = Activities of Daily Living; MMSE = Mini-Mental State Exam; CDR = Clinical Dementia Rating Scale; SIB = Severe Impairment Battery