Algase, Donna, Antonakos, Cathy,Beattie, Elizabeth, Beel-Bates, Cynthia, & Song, Jun-Ah
(2011)
Estimates of crowding in long-term care: comparing two approaches. HERD,4(2), pp. 61-74.
This file was downloaded from: https://eprints.qut.edu.au/40226/
c
Consult author(s) regarding copyright matters
This work is covered by copyright. Unless the document is being made available under a Creative Commons Licence, you must assume that re-use is limited to personal use and that permission from the copyright owner must be obtained for all other uses. If the docu-ment is available under a Creative Commons License (or other specified license) then refer to the Licence for details of permitted re-use. It is a condition of access that users recog-nise and abide by the legal requirements associated with these rights. If you believe that this work infringes copyright please provide details by email to [email protected]
Notice:Please note that this document may not be the Version of Record (i.e. published version) of the work. Author manuscript versions (as Sub-mitted for peer review or as Accepted for publication after peer review) can be identified by an absence of publisher branding and/or typeset appear-ance. If there is any doubt, please refer to the published source.
This is the author version published as:
This is the accepted version of this article. To be published as :
This is the author’s version published as:
Catalogue from Homo Faber 2007
Algase, Donna L., Antonakos, Cathy L., Beattie, Elizabeth, Beel‐Bates,
Cynthia, & Song, Jun Ah (2011)
Estimates
of
crowding
in
long
term
care
:
comparing
two
approaches.
Health Environments Research
and Design, WINTER(2011).
1
CROWDING HAS BEEN WIDELY EXAMINED IN THE LITERATURE ON
HEALTH, DISEASE, AND HOUSING REGULATION, YET A SPECIFIC EXAMINATION
OF THIS IMPORTANT CONCEPT IS LACKING IN THE LITERATURE ON CREATING
AND EVALUATING ENVIRONMENTS FOR PEOPLE WITH DEMENTIA Crowding has
been widely examined in the literature on health, disease, and housing regulation, yet a specific
examination of this important concept is lacking in the literature on creating and evaluating
environments for people with dementia (Sloane et al., 2002; Teresi, Holmes, & Ory, 2000;
Thomas, 2004;Zeisel et al., 2003). Crowding can produce serious negative outcomes. For other
vulnerable groups where crowding has been studied (such as in low-income housing, prisons,
daycare centers, schools, and refugee camps), associations with infectious diseases, lack of sleep,
psychological distress, reduced altruism, and more negative personal interactions have been
identified (Ambrose, 1996; Parse, Lee, Estes, & Rodriguez, 2003). Additionally, there may be
gender differences in the felt stress associated with crowding, women being more susceptible
(Baldassare, 1988; Williams et al., 2008). Limited evidence suggests that resident crowding may
contribute to resident-on-resident violence in residential care (Rosen, Pillemer, & Lachs 2008).
There is widespread interest in the overall quality of life and quality of care in residential care
environments for people with dementia; this is reflected in the intentional move to create more
homelike environments in long-term and special care. Further, increased demand for such care is
predicted into the foreseeable future. During the current decade, the 65-and-over population is
growing at rates exceeding those of the total U.S. population. .If residency ratios remain
unchanged, the number of persons residing in nursing homes will double or triple by 2030.
2
care, individuals aged 85 and older, will more than double from 4.26 million in 2010 to 9.6
million in 2030, and by 2050 it will more than double again to nearly 21 million (U.S. Bureau of
the Census, 2004)1996).
Consequently, attention to the concept of crowding in these care environments is timely and
warranted. In light of these circumstances, the authors’ purposes were (1) to generate initial
estimates of crowding in nursing homes (NHs) and assisted living facilities (ALFs); and (2) to
evaluate two operational approaches to its measurement. This study of crowding was undertaken
specifically with respect to persons with dementia residing in such settings, because they are
arguably more affected by environmental influences than elderly people with intact cognition
(Lawton, 1975).
Crowding, Behavior, and Dementia
Crowding refers to a person’s psychological response to density (Gove, Hughes, & Galle, 1979).
In close environments, unwanted interactions and psychological stress produce a feeling or sense
of being crowded. Density, on the other hand, is a more objective measure generally referred to
in the literature as the number of people in a given space, from which statistical measures such as
occupancy rate can be calculated (Gray, 2001). Density occurring at certain prespecified levels is
generally accepted as a statistical estimate of crowding; however, other important features of an
environment, such as physical proximity to others, the size and purpose of the space involved,
3
In animals, responses to crowded conditions vary somewhat by species. For some, invasion of an
animal’s hunting ground or territory by a member of one’s own species produces an aggressive
response when food is scarce. Freedman (1979) reports that, in experiments that increased
density, animals (mammals) displayed greater activity levels, engaged in more social interaction,
and had larger adrenal glands, but they were not stressed. Only when density interfered with nest
building or leading a normal life did animals display disorientation and a breakdown in social
behavior.
In humans, both the amount of space and the distance between people are important. Comfort
with proximity to others varies by (1) the nature of one’s relationship, closer distances being
more comfortable among relatives and friends; (2) the degree of formality or setting, with formal
settings and hierarchical social structures leading to greater distances between people, even
among those who are otherwise related or friendly; and (3) culture and ethnicity, with distinct
ethnic or cultural groups having specific comfort zones (Freedman, 1979). Intrusions into, or
invasions of, one’s personal space can increase discomfort, but they do not necessarily result in
aggressive behavior or other ill effects. Research on people reveals that their reactions to density
and proximity are complex. According to Freedman (1979), density and its concomitant sense of
crowding alone are not injurious. Rather, greater density intensifies the reaction to a situation in
the direction (positive or negative) that one would react if the situation were to occur under less
dense conditions; reactions to neutral situations remain neutral, even as density increases.
In the context of dementia, a variety of human behaviors and behavioral patterns are affected
4
function declines. Consistent with Lawton’s (1975) docility hypothesis, environment plays a
greater role in affecting behavior under conditions of impaired cognition. Consequently,
behaviors such as aggression or wandering, which can occur in response to environmental
conditions, may be heightened when conditions are crowded. In long-term care settings, where
density exceeds that of a normal residence, increased social interaction may be sought or
imposed and perhaps unwelcome; unfamiliar people are close by, other cognitively impaired
residents may ignore social norms and staff may invade a residents personal space to perform
care procedures, and normal life surely has been disrupted.
In the absence of prior studies investigating the role of crowding on the behavior of residents in
long-term care settings, the authors first sought to investigate an approach to quantify it called
the Long Term Care Crowding Index (LTC-CI), which accounted for density and proximity in
relation to a given individual’s physical position in space and time. Then they sought to examine
the LTC-CI performance in relation to alternative quantifications of density and other measures
of the physical environment. The research questions were:
1. What is the estimated crowding level in NHs and ALFs for locations common to these
settings?
2. How are estimates of crowding affected by location and time?
3. How do two approaches to estimating crowding (LTC-CI and people counts) compare to
other measures of the overall physical environment?
Methods
5
The authors used a cross-sectional approach to evaluate multiple observations of crowding
within the context of a larger study to test the Need-driven, Dementia-compromised Behavior
model). In the parent study, ambulatory people with dementia from 22 NHs and 6 ALFs were
observed 12 times in their natural surroundings. Participants were assigned randomly to one of
four observation schedules such that each participant was observed once every hour between
8:00 a.m. and 8:00 p.m.; observation schedules were structured so that half occurred on each of
two nonconsecutive days during a 72-hour interval. Study procedures were approved by the
institutional review boards of two participating universities. Participants (if able) or their proxies
provided written informed consent. This article reports on measures of crowding and other
environmental conditions as encountered by each participant, but it does not measure the
characteristics of participants themselves.
Sample
The sample consisted of 2,166 observation periods for 185 participants who averaged 11.81
observation periods each (range = 3-12). The length of observations averaged 19.85 minutes (SD
= 0.44; range: 18.85 -20.27). The measurements used in this analysis were taken at the
beginning, after 10 minutes, and again at the end of each observation period (except where noted
later), yielding 6,455 observations for analysis.
Measures
Crowding
Crowding was assessed in two ways. The LTC-CI, developed for this study, was used to capture
6
specific point in time. On a set of concentric circles indicating radii of 2 feet (Zone 1), 4 feet
(Zone 2), and more than 4 feet (Zone 3) from the participant, a trained observer marked the
location of all individuals in the room (see Figure 1). The LTC-CI was recorded at the beginning,
10 minutes into, and at the end of each observation period. To give greater salience to individuals
relative to their proximity to the participant, the LTC-CI was calculated as a weighted sum using
the following formula:
Σ = 9 (no. of people in Zone 1) + 3 (no. of people in Zone 2) + no. of people in Zone 3
Accordingly, the LTC-CI value for Figure 1 is 15, or
Σ = 9 (1 person in Zone 1) + 3 (1 person in Zone 2) + 3 people in Zone 3
[Insert Figure 1 about here.]
Figure 1. LTC-CI and people counts by study site.
The alternative approach was a simple count of people present with the participant in the same
room (e.g., dining room) or area (e.g., hallway); this approach made no allowance for proximity
or for density as a function of area.
Sound Level
Sound measurements were taken at the beginning, 10 minutes into, and at the end of each
20-minute observation using the Quest Technologies Model 2400 Sound Level Meter. According to
Quest Technologies (1999), the accuracy of the sound measurement is within 0.5 decibel (dB) at
25OC; within 1.0 dB over the temperature range of -10OC to 50OC. The meter was calibrated
weekly and research assistants (RAs) inspected the meter microphone daily for damage. All RAs
were trained by the same trainer and observed monthly to ensure protocol accuracy. Interand
7
participant at ear level with the microphone on top of the meter pointed toward the object or area
the subject was facing. The meter ran for 5 seconds and then the value displayed digitally was
recorded.
Ambiance
The Ambiance Scale (AS) yields a rating of the emotional effect or impact of the immediate
environment or surroundings as a gestalt (Algase et al, 2007). It consists of nine pairs of polar
adjectives rated on a 5-point semantic differential scale (-2 to +2) grouped into two subscales.
The engaging subscale (6 items), which includes adjectives such as embellished-stark,
welcoming-impersonal, and novel-boring, has a Cronbach’s alpha of .91–.92; the soothing
subscale (3 items), which includes pairs such as peaceful-chaotic and informal-formal, has a
Cronbach’s alpha of .69 to .78. Subscale scores are calculated as item means; higher scores
indicate greater engaging and soothing qualities. Construct validity was established through
factor analysis; by comparisons of AS subscale ratings for NHs and ALFs, where significantly
higher ratings for ALFs were found; and via comparisons of various locations (dining rooms,
hallways, and residents’ rooms) within facility type, where ratings on both AS subscales differed
significantly by location. Inter- and intra-rater reliability were also established. Ratings were
generated by trained observers who recorded their immediate impressions of the environment in
which participants were situated at the end of each observation period.
Data Analysis
Crowding, location, time of day, and sound were analyzed at the observation level (three
8
present were calculated. LTC-CI and number of people present were analyzed by setting using a
t-test and by location using a one-way analysis of variance (ANOVA). For the dining room, day
room, hallway, and resident room locations, analysis of LTC-CI and number of people were
calculated by place and time-of-day cluster. LTC-CI and number of people within 8 feet were
also analyzed by location using one-way ANOVA. Associations between crowding and sound
were estimated using Pearson’s correlations.
Ambiance was measured once at the end of each observation period, so measures of crowding
included in the analysis were restricted to locations matching those where ambiance was
measured. Crowding measures were aggregated by averaging at the observation period level.
Associations between ambiance and crowding measures were estimated using Pearson’s
correlations.
Results
Crowding Estimates by Setting and Location
Across observations, the LTC-CI generated higher values, a wider variance, and a larger range
compared to a simple people count. In NHs, both the LTC-CI and the people count were higher
than in ALFs. These differences were significant in the expected direction. (See Table 1.)
Both the LTC-CI and people counts varied considerably across sites; graphs depicting this
variation are shown in Figure 2.
[Insert Table 1 and Figure 2 about here.]
Table 1. Crowding by Location
9
Values for the LTC-CI and people counts across locations are compared in Table 2; those for the
LTC-CI are also displayed in Figure 3. By either measure, crowding was greatest in the dining
and activity rooms, lowest in residents’ rooms and shower/baths, and similar across all other
locations. However, a lower people count does not necessarily mean less crowding (see
activities room versus dining room in Table 2). Similarly, approximately equal people counts
may produce differing estimates of crowding (see day room vs. outdoors), and similar crowding
estimates may result from dissimilar people counts (see shower/bath versus other resident’s
room).
For another view of such comparisons, see Figure 4, in which graphs show the number of people
in four locations (dining room, day room, hallway, and resident’s room) as they equate to
selected values for the LTC-CI. Similar values of the LTC-CI can result from varying numbers
of people distributed across various distances from participants. For example, while an LTC-CI
value of 5 looks nearly the same in the number and distance of people in each location, LTC-CI
values of 30 or 45 reveal quite dissimilar distributions of people for each distance at different
locations.
[Insert Table 2 and Figures 3 and 4 about here.]
Table 2. Crowding by Location
Figure 3. LTC-CI by location.
Figure 4. Number of people by location for selected values of the LTC-CI.
10
One-way ANOVA showed that location had a strong effect on the LTC-CI, accounting for
44.39% of the obtained variance in crowding (N = 6,341, df (10, 6330), F = 485.23, p < .0001),
with greater variation between than within locations. The intraclass correlation was moderate (r
= .494); 95% CI = 0.15–0.84, reliability = .998. Results were similar for the people count (N =
6340, df (10, 6330), F = 523.67, p < .0001), also with greater variation between than within
locations. The intraclass correlation was moderate (r = .512); 95% CI = 0.17–0.86, reliability =
.998.
Crowding also varied by time of day across settings and locations. The distribution of LTC-CI
values for the hours from 8:00 a.m. through 7:00 p.m. is shown in Figure 5. Crowding is highest
at mealtimes and higher in the morning than in the afternoon or evening.
[Insert Figure 5 about here.]
Figure 5. Crowding by hour.
Given that some locations may be frequented more often at specific times of day, the interaction
of time and location was also examined. Overall, the interaction effect was significant (N =
5,559, df (47, 511), F = 105.69, p < .0001). Converse to location alone, the interaction had
greater effect within location-by-hour than between location-by-hour, but the effect explained
slightly less variance in LTC-CI estimates (47.41%) than location alone. Intraclass correlation
was moderate (r = .50); 95% CI is slightly narrower (0.36–0.59), reliability = 0.991. Again,
results were similar for the people count (N = 5,559, df (47, 511), F = 116.34, p < .0001) and
explained 49.8% of the variance. There was also greater variation within location-by-hour than
between location-by-hour, and the effect explained slightly less variance in people count
(49.81%) than location alone. Intraclass correlation was moderate (r = .50); 95% CI is slightly
11
Time effects on the LTC-CI were also examined for one location with high crowding estimates
(dining room), one with moderate crowding but frequent occurrence (hallway), and one with low
crowding (resident’s room). For 1,942 observations occurring in the dining room, the effect was
significant (df [11, 1,930], F = 10.36; p < .0001) and greater within than between hours. Time
explained only 5.58% of the variance in LTC-CI estimates in the dining room. Time effects on
the LTC-CI in the hallway were not significant. Time effects on 1,779 observations of crowding
in the resident’s room were significant but small (N = 1,779, df [11, 1767], F = 1.89, p < .05) and
greater within than between hours. Time explained only 1.16% of the variance in crowding in
this location.
Relationship of Crowding to Sound and Ambiance
The LTC-CI and people counts were evaluated for association to sound levels and ambiance
ratings, both expected to vary with crowding. Small, highly significant, and nearly identical
correlations in the expected direction were found for each crowding measure. For the LTC-CI,
these were r = .16, p < .0001, n = 5,597 for sound; r = .08, p < .001, n = 2,027 for the AS
engaging subscale; and r = -0.13, p < .0001 for the AS soothing subscale. For the people count,
correlations were identical with the exception of the AS engaging subscale (r = .10, p < .0001).
Discussion
This paper presents the first known description of crowding in long-term care facilities. The data
illuminate differences in crowding between NHs and ALFs, with the former being more
12
rooms are the most crowded, and private and personal spaces such as showers and resident
rooms are the least crowded. Crowding values were somewhat high, averaging nearly nine
people present across all daytime observations in NHs and closer to seven in ALFs, excluding
the individuals who were the focus of observations. However, substantial variation was also
evident across facilities of both types and across locations within them.
Both location and the interaction of location and time had significant effects on measures of
crowding. Some locations were significantly more crowded than others, and crowding within a
location was likely to vary by time of day. This study thus documents that crowding fluctuates
consistent with routine activities such as meals in long-term care settings. Additionally, this
study confirms a relationship between crowding and other physical characteristics of the
environment in these settings, namely sound level and the engaging quality of an environment,
which vary positively with crowding, and the soothing quality of an environment, which varies
inversely with crowding.
Also of interest are the approaches used to generate crowding estimates in this study: the LTC-CI
and a simple people count. The LTC-CI yields estimates from the vantage point of a given
individual, taking both proximity and density into account> The people count is simply the
number of individuals within a defined space; it quantifies neither proximity nor density. Both
approaches performed similarly when differentiating between NHs and ALFs, among locations
13
However, each measure captured somewhat different information to describe crowding. By
accounting for proximity and density, the LTC-CI generated a wider range of values and greater
variation in all instances where it was assessed. It also afforded a more nuanced estimate of
crowding relative to a given individual. This is reflected, for example, in crowding estimates of
dining and activity rooms, where the people count shows the dining room to be the most
crowded location, and the LTC-CI identifies the activity room as such. People in the dining room
are more evenly spaced by virtue of furniture placement, whereas in the activity room, people are
more likely to be bunched together within a large space by virtue of their involvement in a
common activity. From the vantage point of an individual, this may feel more crowded, even if
not unpleasantly so. Even at the low end of crowding, in private and personal spaces, the
shower/bath has the highest LTC-CI, but the lowest people count, consistent with the necessary
proximity of care providers in these intimate spaces. Consequently, it seems that the LTC-CI is
likely to be more sensitive than people counts when evaluating the effects of crowding on the
behavior of elders—particularly those with dementia—in long-term care settings.
Interaction of Design and Practice Consideration
This study portrayed crowding in long-term care settings in relation to specific individuals and as
a function of the number of people present, the space available, and the dispersion of people
within that space, dispersion being associated with uses of the space and time of day. Any effort
to affect crowding should take all these parameters into account. For a given number of people,
simply adding square footage will not necessarily reduce crowding. For example, increasing the
area of a dining room could allow greater distance between tables, thereby improving dispersion;
14
given resident could actually increase. Similarly, enlarging the activity room is not likely to
reduce crowding there, because people are drawn more closely together to participate in an
activity regardless of the overall space available.
Where crowding is regarded as a problem and adding space is not feasible, the author’s approach
suggests altering other parameters, such as time of day or uses of the space. For example,
crowding in the dining room might be alleviated by lengthening the time frame over which meals
are offered or assigning residents to a first or second seating for each meal, thereby spreading
people out in the dining room over a greater period of time.
Further, this study demonstrates that various locations within long-term care settings can be
grouped according to the average amount of crowding that occurs within them. Dining and
activity rooms can be expected to be the most crowded locations; however, the scope of this
study did not extend to identifying problematic consequences in these crowded areas.
Nonetheless, given that social conventions are a major driver of behavior in these highly public
spaces, crowding there is not likely to produce dramatic reactions, unless some highly personal
affront occurs. However, attention must be paid to other potentially negative consequences of
crowding in such spaces, e.g., infection. On the other hand, the least crowded areas are not free
of crowding-related consequences. In bedrooms and the shower/bath, where privacy is expected,
residents may exhibit negative behaviors because of the presence and proximity of others.
Research Implications
This study compared two approaches to assessing crowding, simple people counts and the
15
performed equally well in distinguishing differences between NHs and ALFs and locations
within them, indicating reliability. Additionally, both showed evidence of validity in relation to
other environmental variables: higher sound levels and the engaging qualities of the environment
were positively associated with crowding; less soothing qualities were associated with crowding.
The advantage of the LTC-CI is its specificity to individuals, which allows for assessment of the
impact of crowding on a personal level. This advantage enables studies that examine inter- and
intra-individual variation in response to crowding that would appear objectively similar viewed
from a group perspective.
