| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Address correspondence and requests for reprints to John Zeisel, PhD, Hearthstone Alzheimer Care, 271 Lincoln Street, Lexington, MA 02421. E-mail: zeisel{at}TheHearth.org
 |
Abstract |
|---|
 TOP Abstract Design and MethodsResultsDiscussionReferences |
|---|
Key Words: Special care units • Environment • Alzheimer's disease • Nonpharmacologic treatment
This paper explores environmental treatment effects, using the conceptual model used by Beck and colleagues (1998) in their study of disruptive behavior in nursing homes. Beck's model presents disruptive behaviors as an outcome influenced both by individual factors such as gender, cognitive status, and health and by what the authors call "proximal factors," intrapersonal as well as environmental. Attributes of both the physical and the social environment are included under the title of "environmental proximal factors." Beck and colleagues identified these proximal factors but did not measure them, focusing instead on the influence of resident factors. This study measures proximal physical environment factors and statistically controls for the others.
Agitation, aggression, psychotic symptoms, depression, and social withdrawal were chosen as the dependent variables in this study because they occur frequently in Alzheimer's SCUs and often lead to other difficult to manage behaviors. These challenging behaviors are often treated with costly medications that have multiple side effects. Whereas responses to control these behaviors in the past have included the use of physical restraints (Castle & Fogel, 1998), the passage of the Omnibus Budget Reconciliation Act (OBRA) and the resulting decline in the use of physical restraints has had the unintended consequence of increasing use of pharmacological restraints (Sloane et al., 1991).
For over a decade, numerous publications have discussed the treatment advantages of design features for this population (Calkins, 1988; Chafetz, & West, 1987; Coons, 1991; Hyde, 1989; Mathew, Sloane, Kirby, & Flood, 1988; Ohta & Ohta 1988; Weisman, Cohen, Ray, & Day, 1991; Zeisel, Hyde, & Levkoff, 1994). It is timely to carefully consider how environment can more humanely and cost-effectively contribute to managing and reducing these symptoms (Day, Carreon, & Stump, 2000). The literature has in fact made strong research-based arguments that proximal physical environmental characteristics of long-term care facilities significantly influence certain behaviors:
Although compelling arguments are made for the therapeutic efficacy of an appropriate physical environment, until now little systematic research has been carried out to determine whether the special design features commonly found in SCUs are, in fact, effective in reducing symptoms and enhancing the quality of life for residents with dementia (U.S. Congress, 1992).
|
Design and Methods |
|---|
|
TOPAbstractDesign and MethodsResultsDiscussionReferences |
|---|
The expert panel that played such a crucial role in the model's refinement included Margaret Calkins, PhD, IDEAS Inc.; Paul Chafetz, PhD, University of Texas; Uriel Cohen, AIA, University of Wisconsin; Betty Rose Connell, Atlanta VA Medical Center; Irving Faunce, Exeter Hospital; M. Powell Lawton, PhD, Philadelphia Geriatric Center; Nancy Mace, MA, California Pacific Medical Center; Jon Sanford, Atlanta VA Medical Center; Philip Sloane, MD, MPH, University of North Carolina at Chapel Hill; and Myra Schiff, PhD, Canadian Alzheimer's Association.
An Alzheimer-specific set of testable E–B hypotheses resulted, describing potential effects of environmental conditions on health and behavior outcomes (Table 1). Each concept in the E–B Model includes two critical dimensions identified from published research literature and responses from the expert panel.
|
|
Sample Site Selection
Self-reported SCUs vary greatly in quality, and there is no official central or regional registration file of SCUs—a major problem that earlier SCU evaluation researchers have faced. Therefore, a nearly exhaustive list of 200 self-reported SCUs in New England and eastern New York was compiled through contact with Alzheimer's Association chapters in the study area. Calls were made to all 200 to determine which met the following definitional criteria:
A letter requesting study participation was sent to the 52 sites that met the definitional criteria. Thirty SCUs agreed to participate, all of which were visited in the process of developing the E–B checklist (Zeisel et al., 1994). A comparison of the two groups of sites—participants and nonparticipants—showed no significant differences in terms of facility size, profit and nonprofit ownership, and urban–rural location. The final 15-SCU sample was purposefully selected to maximize variability among the characteristics of the independent environmental variables as defined by the E–B checklist. By the statistical aggregation of the environmental characteristics of all sampled SCUs, clusters of independent variable conditions were created in the final hierarchical linear modeling (HLM) analysis. The purposeful sampling of the final 15 SCUs is supported by the fact that no specific SCU was treated as an independent or control case in itself.
On the basis of environmental assessments, 15 SCUs were selected that included as many of the environmental conditions as possible—high and low conditions for each of the environmental characteristics from the E–B checklist just described. For example, the few SCUs with poor exit control conditions were first expressly included in the sample to include this condition in the range of test conditions for this variable, no matter what their rating was on other conditions. The next SCU was selected to represent another uncommon environmental condition, and so on until all the conditions were filled with a balanced number of residents in each condition. Selecting facilities to ensure a full range of variation in the characteristics increases the statistical power of the analysis. A random selection approach for the final 15-SCU sample is not likely to have resulted in the extreme conditions being represented in the sample, and the analysis of significant associations would not have been possible. Obviously, the research team had no knowledge of the distribution of dependent variable conditions when the sample of residents was selected. The total number of residents in the 15-site sample was 427.
Analytic Approach
Teresi (1994) described the problematic nature "of studies of intervention effects when institutional units (SCUs) are assigned to experimental or control conditions, but the individual is the unit of analysis ... the resulting mixed units of analysis (SCU and individual resident) can result in attenuated standard errors for the estimates of effects" (pp. S252–S253). She suggested that "special modeling techniques may be needed in these situations" (Teresi, 1994, p. S253).
HLM is such a "special" technique (Bryk & Raudenbush, 1992; Bryk, Raudenbush, Seltzer, & Congdon, 1988). In their study of correlates of quality care in long-term care facilities, Bravo, De Wals, Dubois, and Charpentier pointed out that "the simultaneous study of resident- and facility-level variables calls for a statistical approach that accounts for the nested (or hierarchical) structure of the data. Hence the need for hierarchical models" (Bravo et al., 1999, p. 181). HLM was used in this study to assess the association between environmental design characteristics of SCUs (facility-level variables) and health outcomes measures of residents (individual-level variables) while taking into account quality-of-care variables (facility-level variables) and resident characteristics (individual-level variables). Nonenvironmental variables are controlled for in the analysis by averaging the effects of the environmental variables across all groups defined by characteristics of the nonenvironmental variables.
HLM avoids inaccurate standard errors by establishing the association between an SCU characteristic and individual effects by using the sample size appropriate to their respective level of analysis. The standard errors and significance probabilities for the estimates are adjusted according to the number of data points available at each level—the different degrees of freedom associated with SCU variables versus resident variables. This avoids the attenuated standard errors that could occur if only the larger number of individuals were the unit of analysis as well as the inflated standard errors if only the smaller number of SCUs were the unit of analysis.
Using HLM analysis thus reduces the risk of making Type II—false positive—decisions about whether variables have an influence on the outcome. If only individual units of analysis were used, this would deflate the standard error of the facility-level effect, overstate its significance, and increase the risk of concluding it had a real effect when it did not (a false-positive decision). HLM was therefore used to reduce the cumulative risk of Type II errors and to more accurately estimate the actual significance of the environmental characteristics. It was judged that alternative procedures would give less valid and more uncertain results than HLM in this situation.
Dependent Variables
Behavioral Measures
The largest in-depth national study of SCUs and dementia residents was organized and supported by the National Institute on Aging (NIA) and began in 1991. Known as the NIA Collaborative Studies on Alzheimer's SCUs, this 10-study coordinated project developed and validated a set of measures, known as the NIA Common Core "behavior, affect and activities measure and delirium assessment" Tool (NIA, 1993). To ensure validity and reliability, and to enable comparative analysis with the combined data from these studies, the present study incorporated scales from the NIA Collaborative Studies in the resident profile instrument, used to assess behavioral symptoms among study participants. Completed from medical records and professional judgment by a nurse-informant familiar with each resident, the resident profile includes scales from the Cohen-Mansfield Agitation Inventory (CMAI; Cohen-Mansfield, Marx, & Rosenthal, 1989), measuring aggressive, physically agitated, and verbally agitated behavior; the Multidimensional Observation Scale for Elderly Subjects (MOSES; Helmes, Csapo, & Short 1987), measuring depression and social withdrawal; and the BEHAVE-AD Psychotic Symptom List(Reisberg et al., 1987) to measure misidentification syndrome and paranoid delusions (Table 3).
|
For each of the 13 items from the MOSES, nurse-informants indicated for each resident the frequency of these behaviors: depression (looked or said something depressed or sad, or made sad, depressed, gloomy, or mournful sounds, but not bored ones), being worried, tense, and anxious (looked or said something worried, and cried), being pessimistic (saying the future was hopeless or unbearable), being in good spirits (happy, smiling, cheerful), and being socially withdrawn (initiating interactions, responding to social contacts, paying attention to things around the patient, keeping self-occupied, and helping other residents). Cronbach's alpha for the two MOSES subscales was.88 and.79.
For each of eight items on two BEHAVE-AD Psychotic Symptom List subscales, nurse-informants indicated per resident the frequency of paranoid delusions (being stolen from or being harmed) and misidentification syndromes (saw someone else in the mirror, saw an imposter, or saw TV actors in the room). Cronbach's alpha for the two Psychotic Symptom subscales was.64 and.63.
Controlling for Nonenvironmental Variables
For the physical environmental correlates of the behavioral health variables to be isolated accurately, the HLM model was used to control for potential resident-level and nonenvironmental facility-level variables—both of which were entered into the model. The first are characteristics of each resident, including cognition, activity of daily living (ADL) skills, and length of stay in the SCU. One resident-level characteristic particularly attended to in the analysis is prescription drug use. This characteristic, if overlooked, could completely distort all findings.
The nonenvironmental facility-level characteristics controlled for in the HLM analysis include staff ratio, facility size, organization, and degree of dementia friendliness. Both sets of variables, listed in Table 4, are explained in the paragraphs that follow. There were no missing data in the HLM analysis.
|
Information on residents' ability to perform ADLs as well as their cognitive status was included from data in the Minimum Data Set (MDS) that the SCUs maintained for each resident. Where no MDS data were maintained, SCU nurses generated these data for each resident in the SCU. The Activities of Daily Living scale ranges between 1 and 28 and includes such items as bed mobility, walking, dressing, eating, toileting, bathing, balance, and task segmentation. Higher values indicate greater ADL impairment. Cronbach's alpha for the 28 items in the ADL scale was.78. The MDS Cognition scale from the NIA Common Core Assessment Tool was used to determine each resident's cognitive performance. The MDS Cognition scale has 13 items relating to short- and long-term memory, and to decision-making ability. Cronbach's alpha for the 13 MDS Cognition scale items was.73
Pharmacological agents can independently affect residents' agitation, anxiety, depression, and other behaviors. These were therefore specifically controlled for statistically. Pharmacological-therapeutic drugs were classified into three categories: minor tranquilizers (benzodiazepines, anxiolytics, sedatives, or hypnotics), antidepressants, and other antipsychotic agents. Each patient's drug record was obtained and coded according to this typology, and the patient was given a score for each type of drug taken. If any drug in each of the three categories was administered, the patient received a score of 1, and if none of this type was taken, the patient's score was 0 for that drug category.
Facility characteristics: As a way to control for the possible effects of nonenvironmental characteristics of the facilities themselves, several variables were constructed to describe each facility. The number of residents per SCU was used to take into account possible effects that facility size might have on health and behaviors (Bravo et al., 1999; Castle & Fogel, 1998; Leon & Ory, 1999). Because Castle and Fogel (1998) found that health and behaviors were influenced by organizational status, a dummy variable was created to indicate profit or nonprofit status.
A final scale, developed by rating each SCU's policy and procedures manual, was created to reflect the overall Alzheimer's appropriateness of the facility. Ratings were assigned to each SCU's policy and procedures manual on the basis of their Alzheimer specificity, and on the ratio of key dementia features (assessment, mission, staff empowerment, family support, community involvement, staff appropriateness, management or financial issues, and activities) to nondementia features (quality assurance, housekeeping or maintenance, nursing procedures, emergency procedures, research, clothing, incidents, record keeping, and physician services). Two researchers independently reviewed and ranked the documents on Likert scales. When 2 or fewer points apart, the two scores were averaged. When more than 2 points apart, a third rater mediated a discussion of the scores until consensus was reached.
Environmental characteristics: Each environmental factor, already explained in earlier paragraphs, was coded as two dummy variables: excellent environmental condition (yes or no) and poor environmental condition (yes or no), with the medium environmental condition as the referent category.
Descriptive Statistics
Resident Characteristics
The sample included 308 women and 119 men, ranging in age from 53 to 102 (M = 81.14; SD = 7.8). The majority (58%) were widowed, with 29% married, 6% divorced, and the remaining single or unspecified. The largest percentage (46%) of the sample was moderately to severely dependent in ADLs, with 39% experiencing mild to moderate dependency, and 15% who required supervision only or were independent. Twenty percent had cognitive impairment rated moderate to severe, 67% rated moderate, and 13% rated mild. Prescription medication use for minor tranquilizers was 37%, for antidepressants 30%, and for other antipsychotic agents 33%, with 65% of the sample on at least one psychotherapeutic prescription drug. Twenty-seven percent had experienced a fall in the past 3 months. The average length of institutionalization was 26.63 months (SD = 20.24), with a range of less than 1 month to 151 months for a resident 95 years old.
Facility Characteristics
The sample included 15 facilities. The size of the facilities ranged from 20 to 50 residents (M = 33.67). The clinical, 24-hour staff-to-resident ratio ranged from 0.91 to 2.50 (M = 1.54). Nine facilities were for profit and six were not for profit (one religious and two government facilities are included). On the basis of a systematic two-person analysis of their mission statement, training protocol, policies and procedures, and activities programming, the level of Alzheimer's capability of each facility was ranked. Seven facilities were characterized as highly Alzheimer's capable, three as only Alzheimer's friendly, and five as dementia unfriendly. The intraclass correlation, expressing the variability in the outcomes from residents sampled from the same facility, was 0.36. This weak commonality of characteristics by facility is possibly due to institutional policies that exclude residents inappropriate to the care available or that provide special services benefiting a particular population.
Facility Environmental Characteristics
No single facility received consistent high or low scores on all its environmental characteristics. Facilities given a high rating on some characteristics received only medium scores on the remaining characteristics. Facilities scoring medium on some characteristics scored low on the remaining characteristics. Only one facility received a range of low, medium, and high scores.
Six facilities, housing 37% of the residents in the study, scored high on exit control, whereas seven scored medium (49% of the residents) and two scored low (14% of the residents). For walking paths, three facilities (18% of the residents) were high, eight facilities (50% of the residents) were medium, and four facilities, with 32% of the residents, were low. There were three facilities (16% of the residents) with excellent privacy–personalization, and six facilities each with medium and low scores with 34% and 50% of the residents, respectively. Five facilities with 27% of the residents had excellent common space, seven with 55% of the residents had medium common space, and the remaining three facilities with 18% of the residents had poor common space. Ten percent of the residents living in two facilities ranked high on residential, versus institutional, quality; the remaining eight medium-rated and five low-rated facilities each had 45% of the residents.
Five facilities with 31% of the residents scored high on autonomy support and 10 scored medium (69% of the residents). No sites scored low on autonomy support. For sensory comprehension, six facilities (36% of the residents) were high, seven facilities (46% of the residents) were medium, and two facilities with 18% of the residents scored low. There were eight facilities (54% of the residents) with excellent scores for physical access to a garden, four facilities (27% of residents) with medium access scores, and three facilities with low access scores (19% of the residents).
Although data were collected on the physical accessibility—or absence—of gardens in each of the participating facilities, data were not systematically collected on residents' actual access to these spaces. In other words, the data do not clearly indicate if doors to gardens were or were not kept unlocked during the day in good weather, whether staff only permitted residents to take walks when accompanied by staff, or whether residents had continual access. Because these attributes are so critical to the definition of "outdoor access," this variable was unfortunately omitted from the analysis. Future research must gather more extensive data on this critical environmental variable to determine the impact of therapeutic gardens on behavioral health outcomes.
|
Results |
|---|
|
TOPAbstractDesign and MethodsResultsDiscussionReferences |
|---|
|
.10 to be trivial—regardless of significance, in Cohen's (1969) effect size taxonomy. Those associations that are large enough (r 
.10) and significant at p .05 are shown in boldface in Table 5. The result of the multivariate model analysis, based on introducing potential confounders as controls, is the final test of the research model. The variables from the bivariate analysis were therefore entered in the HLM program to assess which variables remain significantly related after other factors were controlled for. The bivariate associations hold when this statistical model is used, and the exit control association is reversed once the hierarchy of variables is taken into account. A mixed HLM model was used where environmental variables were fixed, and the individual correlates were treated as random effects. The assumption of the individual variables as random effects implies that they are assumed to vary across Level 2 variables. The individual correlates were centered on the grand mean. Our research indicates that these variables, independently—and possibly interactively—are correlated with the resident behavioral health characteristics measured in this study. Seven environmental and nine resident and nonenvironmental facility factors were entered in the final model, with one environmental variable—autonomy support—dropping out in the final HLM analysis. Table 6 presents the final variables and equations used in the HLM model. This contains the variables that remained significant after the other independent variables were controlled for. For environmental factors, 49 independent significant tests were carried out. For the resident and nonenvironmental facility factors, 63 tests were carried out. Seventeen of the relationships were statistically significant, whereas only six false positives would have been expected by chance.
|
|
The degree of privacy–personalization in the SCUs studied was negatively correlated with patient scores on the Cohen-Mansfield total aggression scale. Residents in facilities with more privacy—more rooms that are individual and more opportunities for personalization—generally scored lower on this scale, representing less anxiety and aggression.
The amount of variability among common spaces in a facility was negatively correlated with patient social withdrawal scores. The degree of social withdrawal among residents decreased as the variability among the common spaces in a facility increased. Depression was negatively correlated with another environmental factor—exit design. Residents in facilities whose exits were well camouflaged and had silent electronic locks rather than alarms tended to be less depressed. A hypothesis to explain this correlation is that residents try to elope less in such settings and that caregivers—tending to consider such environments safer—afford residents greater independence of movement. Residents who experience this greater freedom, and hence have less conflict about trying to leave the SCU, feel a greater sense of control and empowerment, leading in turn to less depression. Until further research is carried out measuring personal state-of-mind variables that might be implicated in such a process, this explanation remains only a hypothesis.
Aggression and its various expressions were also correlated with characteristics of the physical environment. Persons living in SCUs with a more residential, less institutional environment expressed lower levels of overall aggression than those living in more institutional settings. Physical aggression, scored separately, did not appear to be associated with environmental design after individual and facility characteristics were controlled for, but verbal aggression appears to be correlated with environment. In facilities where sensory input is more understandable and where such input is more controlled, residents tended to be less verbally aggressive.
Finally, environmental design also was correlated with resident psychotic problems. Those living in environments scoring high on privacy–personalization tended to have lower scores on the psychotic problem scale. The same was true for those living in facilities with higher scores on sensory comprehension scores.
It should be noted that privacy–personalization and sensory comprehension were related to lower scores for more than one negative characteristic. In several of this study's HLM models where these two factors did not reach statistical significance, they were among the last variables dropped from the models. This suggests that privacy–personalization and sensory comprehension are likely to be particularly important environmental design features contributing to improved behavioral health outcomes.
Taken with other correlates of the behavioral health characteristics discussed in the paragraphs that follow, this analysis appears to provide evidence that environmental design is related to behavioral health outcomes among residents of SCUs with dementia-related health problems. Because the present research did not include measures of subjective control such as alienation and anomie, while it is clear that there are relationships between environment and behavioral health characteristics, we can only hypothesize what the specific links are which may relate them. Elaboration and exploration of the nature of these relationships must await future research.
Resident-Level Influences
Certain resident-level characteristics are shown through the HLM analysis to correlate with residents' behavioral health. The more a resident has problems performing ADLs, the more likely he or she is to have a lower total anxiety and aggression score on the Cohen-Mansfield scale. The likelihood of a resident falling, in contrast, was positively related to expressions of anxiety and aggression.
The longer a person has lived in a caregiving institution, the more highly correlated her or his social withdrawal score is likely to be. This may reflect the fact that those living longer in SCUs are also more likely to be confounded with progression of dementia or other medical conditions with a developmental process.
Gender is associated with depression. Our study shows depression to be more prevalent among Alzheimer's residents who are women than among those who are men. This is also the case for those taking antipsychotic medications—the more such medications, the greater the correlation with depression, according to our findings.
In contrast, gender is correlated with overall aggression scores in the opposite direction. Male residents are more likely to score higher than female residents on overall aggression—both physical and verbal. Ability to perform ADLs is negatively correlated with all patient scores on physical aggression, whereas a patient's likelihood of falling down, level of anxiety, age, and taking of antipsychotic medications all appear to be positively correlated with the expression of verbal aggression by residents.
Nonenvironmental Facility-Level Influences
Only three nonenvironmental facility-level characteristics appear to correlate with the behavioral health measures in this study. The larger the facility—the more residents there are in the SCU—the lower the social withdrawal scores tend to be. The higher the staff ratio, the lower are the verbal aggression scores among residents. In addition, the higher the facility scores on Alzheimer capability, the fewer reported psychotic problems residents tend to exhibit.
|
Discussion |
|---|
|
TOPAbstractDesign and MethodsResultsDiscussionReferences |
|---|
Environments conventionally designed for the cognitively able appear to put stress on the cognitive abilities of those with Alzheimer's. One explanation for the results presented here may be that the design characteristics discussed relieve residents' cognitive stress, thus reducing their anxiety and related aggressive acts. Another is that the design features, by providing residents with greater control over their own lives, empower them and thus reduce their tendency to withdraw and even to be situationally depressed. The present research can only indicate the direction for such interpretations. More extensive research will be required to be more definitive.
The findings support and expand on the Intervention model proposed by Beck and colleagues (1998). Each of the behavioral health characteristics measured in the present study—in addition to correlating with at least one environmental factor—was also associated with resident characteristics, such as gender, cognitive status, and medications and to nonphysical facility factors such as staff ratio and Alzheimer's friendliness of the mission. This would appear to indicate a dynamic, interactive relationship between resident characteristics and environmental features on one hand and the behavioral health of residents on the other.
For design and construction, these findings have clear implications. Even though SCU design has to meet stringent fire safety requirements and construction standards that may seem rigid and imply institutional design responses, SCUs should strive to model their interior environments after homelike settings to reduce aggressive and other symptoms. Applied in design of SCUs, these findings will lead to more private and less shared rooms, variation in common room design within an SCU, common rooms for activities located at ends of hallways, and doors located along side walls whenever possible instead of at the end of hallways where they act as "attractive nuisances." Alzheimer Association chapters are already advising consumers to take environmental design issues such as these into account when determining the quality of nursing homes, SCUs, and assisted living special care programs they are considering for their loved ones (Alzheimer's Association, 2000; Raia, Zeisel, Cacciapuoti, Stout, & Rodman, 2003).
Funding for major renovation of heavily institutional nursing home SCUs must be weighed against the potential of decreased costs for medications and increases in quality of life, not merely as cosmetic alterations.
Limitations
It is possible that the significant associations found between behavioral health characteristics and environmental factors are spurious and result from their correlation with some other unspecified or poorly specified factor. If this were the case, however, it is highly unlikely that each of the environmental factors would be significant and in the direction hypothesized.
Nevertheless, the data presented here may be limited by the questions asked and the variables studied. This presents limitations that have to be considered when the results are interpreted.
There may be unspecified factors not included in the model that also correlate with behavioral health characteristics. Some of these may be environmental variables such as whether or not the SCUs had gardens that residents used, whether the environments reflected the cultural background of residents, or if they were maintained well or were cluttered and messy. In our analysis, we assumed random distribution of such characteristics, but these may have independent effects on the behavioral health characteristics measured.
One or more of the factors studied may have been only partly specified. For example, the study team identified two dimensions of exit control design: the degree to which the exit is camouflaged to reduce the likelihood of elopement, and the degree to which the doors were actually locked (magnetically or by key) versus unlocked but alarmed. Other dimensions of exit control that were overlooked may still be important for the analysis.
Limited variability may also be influencing the data. For example, although the research team attempted to have high and low independent variable scores for each factor, secure exits were a criterion for inclusion in the sample. One SCU with a large enough group of residents was found to have only a "black line on the floor" as its barrier to the rest of the facility, providing only one SCU in the cell representing poor exit control.
Nevertheless, the study analysis did attempt to rigorously control for resident characteristics and organizational factors that previous research has indicated might confound the data analysis. These included such resident characteristic variables as prescription drug usage, risk of falling, cognition, need for assistance with ADLs, length of stay in the facility, and age. Organizational factors included the ratio of staff to residents, the not-for-profit or for-profit status of the organization, and the dementia friendliness of the facility's mission.
Two major limitations can only be overcome by increased research funding in this area. Resource allocation in the research led the team to a final selection of 15 sites and 427 residents, instead of 30 sites with nearly 950 residents. There is clearly a need to replicate this research with the use of more facilities as well as to gather longitudinal data to examine in greater detail the impacts of the environment on outcomes inferred by the correlations found in this study.
Future Research
This study paves the way for further exploration into environmental design as one important nonpharmacologic treatment modality for people with Alzheimer's disease (Zeisel & Raia, 2000). Future research might focus on applying a similar methodology to other environmental factors the literature has indicated might be therapeutic. These could include accessible gardens, soothing colors, nondisorienting carpeting patterns, higher lighting levels, and alternative bathing settings.
A larger sample size in future research would increase the significance of findings and would permit the inclusion and control of a greater number of variables—environmental, individual, and facility characteristics. A more diverse sample would enable researchers to test the generalizability of the study findings to a greater variety of Alzheimer care settings—assisted living residences for people with Alzheimer's, foster care settings, and at-home care in addition to nursing home SCUs.
An examination in future research of the cumulative effects of environmental features might also yield useful information. For example, is the association between reduced aggression and both privacy–personalization and unique common spaces cumulative, such that planning both together could augment the reduction of aggression more than either separately? Even more challenging and rewarding would be insight into the dynamic interactive and cumulative effects of environmental conditions with organizational factors as well as pharmacologic treatment.
A Significant Direction
This research demonstrates the great opportunity systematic attention to environmental factors opens for improving Alzheimer's symptoms. The greatest likelihood for this approach to make a significant contribution is to consider environment as one of at least three modalities—pharmacologic, behavioral, and environmental—for improving the quality of life, health, and behavior of people with Alzheimer's disease. These might well be considered three "treatments" for the disease.
As conceptual and empiric research in the area of environment and health accelerates, it is becoming increasingly clear that a combination of drug treatment, supportive environments, and focused caregiving approaches provides the highest likelihood that those with Alzheimer's disease can indeed live more satisfying lives. The stage is set for interdisciplinary intervention studies to identify the optimum balance and arrangement of these treatment modalities.
|
Footnotes |
|---|
We gratefully acknowledge the contributions of Richard N. Jones to the clarity of the statistical explanations, and of research assistants Linda Shi to data collection and Helen Miltiades to data analysis. 
1 Hearthstone Alzheimer Care, Ltd., Lexington, MA.
2 Gerontology Center, University of Massachusetts, Boston.
3 Harvard Upper New England Geriatric Education Center, Harvard University, Boston, MA.
4 Philadelphia Geriatric Center, Philadelphia, PA (deceased).
5 Sociology Department, University of Massachusetts, Boston.
Decision Editor: Laurence G. Branch, PhD
Received for publication December 10, 2001. Accepted for publication September 19, 2002.
|
References |
|---|
|
TOPAbstractDesign and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. Duggan, T. Blackman, A. Martyr, and P. Van Schaik The impact of early dementia on outdoor life: A `shrinking world'? Dementia, May 1, 2008; 7(2): 191 - 204. [Abstract] [PDF]
|
|
|
|
|
|
J. M . Cioffi, A. Fleming, L. Wilkes, M. Sinfield, and J. Le Miere The effect of environmental change on residents with dementia: The perceptions of relatives and staff Dementia, May 1, 2007; 6(2): 215 - 231. [Abstract] [PDF]
|
|
|
|
 |
|
 R. Leonard, M. E. Tinetti, H. G. Allore, and M. A. Drickamer Potentially modifiable resident characteristics that are associated with physical or verbal aggression among nursing home residents with dementia. Arch Intern Med, June 26, 2006; 166(12): 1295 - 1300. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Wood, S. Harris, M. Snider, and S. A. Patchel Activity situations on an Alzheimer's disease special care unit and resident environmental interaction, time use, and affect American Journal of Alzheimer's Disease and Other Dementias, March 1, 2005; 20(2): 105 - 118. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
