Accepted Manuscript Life-Space Assessment and Physical Activity Scale for the Elderly: Validity of Proxy Informant Responses James T. Cavanaugh, PT, PhD Kelley Crawford, PT, DPT PII:
S0003-9993(14)00268-8
DOI:
10.1016/j.apmr.2014.03.027
Reference:
YAPMR 55798
To appear in:
ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION
Received Date: 21 October 2013 Revised Date:
7 March 2014
Accepted Date: 25 March 2014
Please cite this article as: Cavanaugh JT, Crawford K, Life-Space Assessment and Physical Activity Scale for the Elderly: Validity of Proxy Informant Responses, ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION (2014), doi: 10.1016/j.apmr.2014.03.027. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Running Head: Validity of Proxy Informant Responses
Title: Life-Space Assessment and Physical Activity Scale for the Elderly: Validity of Proxy
RI PT
Informant Responses
Authors: James T. Cavanaugh, PT, PhD1 and Kelley Crawford, PT, DPT2
Department of Physical Therapy, University of New England, Portland, ME (USA)
2
Department of Rehabilitation Medicine, Maine Medical Center, Portland, ME (USA)
M AN U
SC
1
Study Location: University of New England, Portland, ME (USA)
Acknowledgements:
The work was sponsored by the Department of Physical Therapy of the University of
TE D
•
New England, Portland, ME (USA).
The authors declare no competing financial interests.
EP
•
Corresponding Author Contact Information:
AC C
J.T. Cavanaugh, PT, PhD, University of New England / Department of Physical Therapy, 716 Stevens Ave., Portland, ME 04103. Tel: 1-207-221-4595. [email protected]. Reprints not available.
ACCEPTED MANUSCRIPT
1
Title: Life-Space Assessment and Physical Activity Scale for the Elderly: Validity of
2
Proxy Informant Responses
3 ABSTRACT
5
Objective: To validate the administration of the Life-space Assessment (LSA) and
6
Physical Activity Scale for the Elderly (PASE) surveys to proxy informants, as would be
7
necessary when measuring long-term outcomes in acutely ill, hospitalized older adults
8
who are initially incapacitated but eventually return to the community.
9
Design: Cross-sectional study.
10
Setting: General community.
11
Participants: Convenience sample of 40 dyads comprised of an ambulatory older adult
12
and a familiar companion.
13
Interventions: Dyads completed the LSA and PASE surveys on one occasion.
14
Companions based their responses on the recent mobility and physical activity of the
15
older adult.
16
Main Outcome Measures: Paired total scores for each instrument.
17
Results: At a group level, the difference between older adult and companion mean scores
18
for each instrument was not significant (p > 0.05). Standardized mean difference values
19
were small (d < 0.1). Paired scores were significantly yet moderately associated (ICC (1,
20
1) = 0.84 - 0.88, p < 0.01). Difference in scores was not associated with time spent
21
together (p > 0.05) or older adult gait speed (p > 0.05). At an individual level, older
22
adults and companions agreed more closely on the LSA than the PASE. However,
AC C
EP
TE D
M AN U
SC
RI PT
4
1
ACCEPTED MANUSCRIPT
disagreement in excess of estimated measurement error occurred in 40% of dyads for the
24
LSA and in none of the dyads for the PASE.
25
Conclusions: Older adults and companions collectively provided similar responses on
26
each instrument. Nonetheless, varying levels of agreement within individual dyads
27
suggested that proxy responses should be considered carefully. Implications for clinical
28
research and practice research are discussed.
KEY WORDS: Geriatrics; Motor activity; Outcome assessment (health care)
31 32
ABBREVIATIONS
M AN U
30
SC
29
CI:
Confidence interval
34
ICU:
Intensive care unit
35
ICC:
Intraclass correlation coefficient
36
LSA: Life Space Assessment
37
MDD: Minimum detectable difference
38
PASE: Physical Activity Scale for the Elderly
39
PRO: Patient reported outcome
40
SDD: Smallest detectable difference
EP
AC C
42
TE D
33
41
RI PT
23
2
ACCEPTED MANUSCRIPT
43
The Life-Space Assessment (LSA)1 and the Physical Activity Scale for the Elderly (PASE)a, 2 represent patient-reported outcome (PRO) measures that quantify,
45
respectively, an elderly respondent’s recent mobility and physical activity. Both represent
46
the “participation” domain of the International Classification of Functioning, Disability,
47
and Health,3 and accordingly, are well-suited for capturing community-level mobility and
48
physical activity clinical outcomes. Importantly, however, the collection of baseline (i.e.,
49
pre-admission) PRO data from acutely ill, hospitalized older adults involves important
50
methodological considerations.4,5 One central issue is how to collect reliable and accurate
51
baseline data from familiar proxy informants (e.g., family member, caregiver, or close
52
companion) in circumstances when patients are mechanically ventilated, sedated, and / or
53
present with communication, cognitive, or other severe impairment that prevents them
54
from responding on their own behalf. Neither the LSA nor the PASE survey has been
55
validated in this regard.
SC
M AN U
TE D
56
RI PT
44
The broad purpose of this study was to examine the validity of administering the LSA and PASE instruments to familiar proxy informants. To do so, we administered each
58
survey to community-dwelling older adult-companion dyads, who were asked to base
59
their responses on the mobility and physical activity of the older adult participant. Our
60
primary aim was to examine the agreement between paired scores. Based on previous
61
elder-proxy studies using physical function measures,6-9 we hypothesized that older adult
62
and companion respondents would provide similar information on each survey. Our
63
secondary aim was to examine the extent to which older adult-companion agreement
64
might be biased by (a) the number of hours per week the pair spent together and (b) the
65
ambulatory capability of the older adult. Based on a previous study,8 we hypothesized
AC C
EP
57
3
ACCEPTED MANUSCRIPT
66
that closer agreement would be associated with more hours spent together. In contrast,
67
given a lack of previous research, we conservatively hypothesized that no relationship
68
would exist between the amount of agreement and older adult ambulatory capability.
RI PT
69
71
METHODS
72
Design and Participants
73
SC
70
The study employed a cross-sectional design in which dyads formed by a community-dwelling older adult and his or her close companion were recruited as a
75
sample of convenience. Older adult participants were at least 60 years of age and
76
identified their companion as “someone who they have spent at least 7 hours per week
77
with during the last month.” Study candidates were recruited by advertisement and word
78
of mouth. In promoting the study (e.g., at local retirement communities), every effort was
79
made to recruit a range of companion types; that is, those who shared a residence with the
80
older adult participant (e.g., spouse; partner; sibling) and those who did not (e.g.,
81
extended family; friend). Dyads were screened by the primary investigator (JTC) over the
82
phone and excluded if either individual (1) was unable to speak English, (2) reported
83
cognitive, memory, or communication impairments that limited his / her ability to
84
complete surveys, or (3) if the targeted older adult participant was non-ambulatory. The
85
Institutional Review Board for the Protection of Human Subjects at the University of
86
New England approved the study.
AC C
EP
TE D
M AN U
74
87 88
Measures
4
ACCEPTED MANUSCRIPT
89
Descriptive Measures: We recorded the age and gender of each participant, whether or not dyad members shared a residence, the type of relationship (e.g., spouse,
91
friend, family member, caregiver), and the reported number of hours / week spent
92
together during the previous month. To characterize older adult ambulatory mobility, we
93
collected 2 trials of self-selected gait speed over a 3m distance; assistive devices were
94
allowed as needed.10
RI PT
90
Life-Space Assessment: The LSA is a brief survey that asks respondents to reflect
96
on their mobility activities during the preceding month.1 Mobility is categorized in terms
97
of five increasingly broader “life-space” levels, ranging from within their home to
98
beyond their town. Respondents indicate (1) whether or not they have been to each life
99
space in the past four weeks, (2) if so, how many days per week, and (3) if so, whether or
M AN U
SC
95
not they used equipment or needed help from another person. A composite score is
101
calculated based on highest level of life-space achieved, frequency of attaining each
102
level, and degree of independence at each level. Scores range from 0-120, with higher
103
scores indicating greater mobility. Previous investigations have supported the construct
104
and concurrent validity of the LSA in older adult populations,1, 11 its test-retest reliability
105
(Intraclass correlation coefficient (ICC) = 0.96),1 and its utility as an outcome measure in
106
clinical trials.12-14 We administered the LSA to both dyad members, asking each to focus
107
on the recent mobility of the older adult participant and thereby generating a LSAOlder Adult
108
and LSACompanion score.
109
AC C
EP
TE D
100
Physical Activity Scale for the Elderly: The PASE is a brief survey designed to
110
measure the physical activity of older adults during leisure, household, and occupational
111
activities occurring over the previous 7 day period.2 PASE item scores are calculated by
5
ACCEPTED MANUSCRIPT
multiplying either time spent (hours / week) or participation in (yes / no) each activity
113
with empirically derived item weights. The PASE total score represents the sum of
114
individual item scores and ranges from 0 to over 400. Higher scores correspond to greater
115
physical activity. The PASE has been validated for use in elderly populations,2,15 and its
116
test-retest reliability has been reported (r = 0.75).2 We administered the PASE to both
117
dyad members, asking each to focus on the recent physical activity of the older adult
118
participant and thereby generating a PASEOlder Adult and PASECompanion score.
SC
RI PT
112
119
121
Procedures
M AN U
120
Trained research personnel collected descriptive data and administered the LSA and PASE to each older adult and companion dyad in a place convenient to them (e.g. the
123
older adult’s home). Participants were sequestered from one another when completing the
124
surveys to avoid influencing each other’s responses.
TE D
122
125
127
Data Analysis
Data were analyzed using SPSS version 20.b Descriptive statistics were used to
EP
126
characterize the sample. To account for the possibility of recall error by either participant,
129
time spent together was characterized using the average reported value. Gait speed was
130
calculated as the mean of 2 trials.
131
AC C
128
Our analysis was based partly on the approach used in a previous investigation.16
132
At a group level, we analyzed agreement of scores using the dependent samples t-test (α
133
= 0.05), standardized mean difference (d),17 and ICC (1,1). At an individual level, we
134
calculated difference values between paired scores, constructed Bland-Altman plots,18
6
ACCEPTED MANUSCRIPT
and determined the 95% limits of agreement. Finally, we analyzed the relationship
136
between difference in paired scores and (a) time spent together and (b) mean gait speed
137
using appropriate parametric or non-parametric correlation coefficients (α = 0.05).
138
To facilitate the interpretation of individual-level results, we compared one
RI PT
135
instrument to the other by normalizing the raw difference scores as percent difference
140
values (= 100* [(older adult score – companion score) / mean of paired scores]). We also
141
used previous literature to provide estimates of measurement error against which the
142
agreement for each instrument could be compared. Because LSA measurement error
143
values had not been published previously, we used standard formulas19 and information
144
contained in the original LSA publication1 to calculate a Minimum Detectable Difference
145
(MDD) value based on a 95% confidence interval (MDD95% = 13.7 using ICC = 0.96 and
146
Standard Error of Measurement = 4.94.) For the PASE, we lacked sufficient information
147
from previous literature to produce an analogous measurement error estimate. Thus, we
148
selected instead the only available PASE measurement error value available, the Smallest
149
Detectable Difference (SDD) based on a 95% confidence interval from a recent study of
150
adult patients with cancer (SDD95% = 84).20
EP
TE D
M AN U
SC
139
152 153 154
AC C
151
RESULTS
Forty-five dyads were screened initially. None were rejected due to language,
155
cognitive, or physical limitations. Four dyads were unable to participate in data collection
156
due to scheduling conflicts. Data from one dyad were incomplete and therefore not
157
included in the analysis.
7
ACCEPTED MANUSCRIPT
158
The final sample contained 22 female and 18 male older adult participants (mean age ± SD of 80.7 ± 7.5 years.) Mean gait speed ± SD was 0.82 ± 0.26 m/s. Companions
160
included 33 females and 7 males (mean age ± SD of 69.2 ± 14.4 years.) Two companions
161
did not report their age. Twenty-five companions reported living with the older adult.
162
Mean time spent per week together ± SD was 90.7 ± 62.3 hours, with 19 dyads reporting
163
spending more than 100 hours per week together. Companions included 21 spouses, 11
164
children, 6 friends, and 2 unrelated caregivers. Additional characteristics of the sample
165
have been reported elsewhere.21
167 168
M AN U
166
SC
RI PT
159
Life-Space Assessment
Older adults as a group reported a moderate level of life-space mobility (mean LSAOlder Adult = 71.1, 95% confidence interval (CI) = 61.8 – 80.3), similar to the collective
170
report provided by their companions (mean LSACompanion = 68.5, 95% CI = 59.8 – 77.2).
171
The difference between group mean scores was not significant (mean difference = 2.54,
172
95% CI = - 1.77 to 6.84, t (39) = 1.19, p = 0.24), and the standardized mean difference
173
was small (d = 0.09). The 2 sets of scores were significantly yet moderately associated
174
(ICC (1,1) = 0.88, 95% CI = 0.79 - 0.94, p < 0.001). Difference in LSA scores was not
175
associated with time spent together (Spearman rho = -0.26, p = 0.11) or older adult gait
176
speed (Pearson r = 0.06, p = 0.70). Family members (n = 32) were relatively more
177
accurate in their characterization of older adult life-space mobility than friends or
178
caregivers (n = 8) (Figure 1). At an individual level, the mean percent difference between
179
paired LSA scores was 18.1% (95% CI = 13.5 – 22.7). Seven dyads had ≤ 5% difference
180
in scores, 15 dyads had ≤ 10% difference in scores, and 22 dyads had ≤ 20% difference in
AC C
EP
TE D
169
8
ACCEPTED MANUSCRIPT
scores (Figure 2). For all dyads, the percent difference in paired LSA scores was ≤ 60%.
182
Twenty-four dyads had raw difference values less than MDD95% = 13.7.
183
INSERT FIGURE 1 ABOUT HERE
184
INSERT FIGURE 2 ABOUT HERE
RI PT
181
Figure 3A features a Bland-Altman plot,18 in which the mean of the paired scores
186
(x-axis) was plotted against difference in paired LSA scores (y-axis). There was 1 case of
187
exact agreement. Positive differences (n = 23) indicated cases in which older adults
188
reported greater life-space mobility than their companion (mean difference = 11.8, 95%
189
CI = 8.1 – 15.4). Negative differences (n = 16) indicated cases in which companions
190
reported greater life-space mobility of the older adult than the older adult reported about
191
his or herself (mean difference = - 10.6, 95% CI = - 14.4 to - 6.8). The maximum paired
192
difference (=26.0) was less than 26.9, the 95% limits of agreement.
M AN U
SC
185
193
TE D
INSERT FIGURE 3 ABOUT HERE
194
196
Physical Activity Scale for the Elderly
Physical activity levels reported by older adults as a group were relatively low
EP
195
(mean PASEOlder Adult score = 81.7, 95% CI = 63.4 – 100.0) and slightly less than the
198
collective report of the companions about the older adults (mean PASECompanion = 82.4,
199
95% CI = 64.6 – 100.1). The difference between group mean scores was not significant
200
(mean difference = - 0.70, 95% CI = - 11.77 to 9.71, t (39) = 0.14, p = 0.89), and the
201
standardized mean difference between scores was small (d = 0.01). The 2 sets of scores
202
were significantly yet moderately associated (ICC (1,1) = 0.84, 95% CI = 0.71 - 0.91, p
S0003-9993(14)00268-8
DOI:
10.1016/j.apmr.2014.03.027
Reference:
YAPMR 55798
To appear in:
ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION
Received Date: 21 October 2013 Revised Date:
7 March 2014
Accepted Date: 25 March 2014
Please cite this article as: Cavanaugh JT, Crawford K, Life-Space Assessment and Physical Activity Scale for the Elderly: Validity of Proxy Informant Responses, ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION (2014), doi: 10.1016/j.apmr.2014.03.027. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Running Head: Validity of Proxy Informant Responses
Title: Life-Space Assessment and Physical Activity Scale for the Elderly: Validity of Proxy
RI PT
Informant Responses
Authors: James T. Cavanaugh, PT, PhD1 and Kelley Crawford, PT, DPT2
Department of Physical Therapy, University of New England, Portland, ME (USA)
2
Department of Rehabilitation Medicine, Maine Medical Center, Portland, ME (USA)
M AN U
SC
1
Study Location: University of New England, Portland, ME (USA)
Acknowledgements:
The work was sponsored by the Department of Physical Therapy of the University of
TE D
•
New England, Portland, ME (USA).
The authors declare no competing financial interests.
EP
•
Corresponding Author Contact Information:
AC C
J.T. Cavanaugh, PT, PhD, University of New England / Department of Physical Therapy, 716 Stevens Ave., Portland, ME 04103. Tel: 1-207-221-4595. [email protected]. Reprints not available.
ACCEPTED MANUSCRIPT
1
Title: Life-Space Assessment and Physical Activity Scale for the Elderly: Validity of
2
Proxy Informant Responses
3 ABSTRACT
5
Objective: To validate the administration of the Life-space Assessment (LSA) and
6
Physical Activity Scale for the Elderly (PASE) surveys to proxy informants, as would be
7
necessary when measuring long-term outcomes in acutely ill, hospitalized older adults
8
who are initially incapacitated but eventually return to the community.
9
Design: Cross-sectional study.
10
Setting: General community.
11
Participants: Convenience sample of 40 dyads comprised of an ambulatory older adult
12
and a familiar companion.
13
Interventions: Dyads completed the LSA and PASE surveys on one occasion.
14
Companions based their responses on the recent mobility and physical activity of the
15
older adult.
16
Main Outcome Measures: Paired total scores for each instrument.
17
Results: At a group level, the difference between older adult and companion mean scores
18
for each instrument was not significant (p > 0.05). Standardized mean difference values
19
were small (d < 0.1). Paired scores were significantly yet moderately associated (ICC (1,
20
1) = 0.84 - 0.88, p < 0.01). Difference in scores was not associated with time spent
21
together (p > 0.05) or older adult gait speed (p > 0.05). At an individual level, older
22
adults and companions agreed more closely on the LSA than the PASE. However,
AC C
EP
TE D
M AN U
SC
RI PT
4
1
ACCEPTED MANUSCRIPT
disagreement in excess of estimated measurement error occurred in 40% of dyads for the
24
LSA and in none of the dyads for the PASE.
25
Conclusions: Older adults and companions collectively provided similar responses on
26
each instrument. Nonetheless, varying levels of agreement within individual dyads
27
suggested that proxy responses should be considered carefully. Implications for clinical
28
research and practice research are discussed.
KEY WORDS: Geriatrics; Motor activity; Outcome assessment (health care)
31 32
ABBREVIATIONS
M AN U
30
SC
29
CI:
Confidence interval
34
ICU:
Intensive care unit
35
ICC:
Intraclass correlation coefficient
36
LSA: Life Space Assessment
37
MDD: Minimum detectable difference
38
PASE: Physical Activity Scale for the Elderly
39
PRO: Patient reported outcome
40
SDD: Smallest detectable difference
EP
AC C
42
TE D
33
41
RI PT
23
2
ACCEPTED MANUSCRIPT
43
The Life-Space Assessment (LSA)1 and the Physical Activity Scale for the Elderly (PASE)a, 2 represent patient-reported outcome (PRO) measures that quantify,
45
respectively, an elderly respondent’s recent mobility and physical activity. Both represent
46
the “participation” domain of the International Classification of Functioning, Disability,
47
and Health,3 and accordingly, are well-suited for capturing community-level mobility and
48
physical activity clinical outcomes. Importantly, however, the collection of baseline (i.e.,
49
pre-admission) PRO data from acutely ill, hospitalized older adults involves important
50
methodological considerations.4,5 One central issue is how to collect reliable and accurate
51
baseline data from familiar proxy informants (e.g., family member, caregiver, or close
52
companion) in circumstances when patients are mechanically ventilated, sedated, and / or
53
present with communication, cognitive, or other severe impairment that prevents them
54
from responding on their own behalf. Neither the LSA nor the PASE survey has been
55
validated in this regard.
SC
M AN U
TE D
56
RI PT
44
The broad purpose of this study was to examine the validity of administering the LSA and PASE instruments to familiar proxy informants. To do so, we administered each
58
survey to community-dwelling older adult-companion dyads, who were asked to base
59
their responses on the mobility and physical activity of the older adult participant. Our
60
primary aim was to examine the agreement between paired scores. Based on previous
61
elder-proxy studies using physical function measures,6-9 we hypothesized that older adult
62
and companion respondents would provide similar information on each survey. Our
63
secondary aim was to examine the extent to which older adult-companion agreement
64
might be biased by (a) the number of hours per week the pair spent together and (b) the
65
ambulatory capability of the older adult. Based on a previous study,8 we hypothesized
AC C
EP
57
3
ACCEPTED MANUSCRIPT
66
that closer agreement would be associated with more hours spent together. In contrast,
67
given a lack of previous research, we conservatively hypothesized that no relationship
68
would exist between the amount of agreement and older adult ambulatory capability.
RI PT
69
71
METHODS
72
Design and Participants
73
SC
70
The study employed a cross-sectional design in which dyads formed by a community-dwelling older adult and his or her close companion were recruited as a
75
sample of convenience. Older adult participants were at least 60 years of age and
76
identified their companion as “someone who they have spent at least 7 hours per week
77
with during the last month.” Study candidates were recruited by advertisement and word
78
of mouth. In promoting the study (e.g., at local retirement communities), every effort was
79
made to recruit a range of companion types; that is, those who shared a residence with the
80
older adult participant (e.g., spouse; partner; sibling) and those who did not (e.g.,
81
extended family; friend). Dyads were screened by the primary investigator (JTC) over the
82
phone and excluded if either individual (1) was unable to speak English, (2) reported
83
cognitive, memory, or communication impairments that limited his / her ability to
84
complete surveys, or (3) if the targeted older adult participant was non-ambulatory. The
85
Institutional Review Board for the Protection of Human Subjects at the University of
86
New England approved the study.
AC C
EP
TE D
M AN U
74
87 88
Measures
4
ACCEPTED MANUSCRIPT
89
Descriptive Measures: We recorded the age and gender of each participant, whether or not dyad members shared a residence, the type of relationship (e.g., spouse,
91
friend, family member, caregiver), and the reported number of hours / week spent
92
together during the previous month. To characterize older adult ambulatory mobility, we
93
collected 2 trials of self-selected gait speed over a 3m distance; assistive devices were
94
allowed as needed.10
RI PT
90
Life-Space Assessment: The LSA is a brief survey that asks respondents to reflect
96
on their mobility activities during the preceding month.1 Mobility is categorized in terms
97
of five increasingly broader “life-space” levels, ranging from within their home to
98
beyond their town. Respondents indicate (1) whether or not they have been to each life
99
space in the past four weeks, (2) if so, how many days per week, and (3) if so, whether or
M AN U
SC
95
not they used equipment or needed help from another person. A composite score is
101
calculated based on highest level of life-space achieved, frequency of attaining each
102
level, and degree of independence at each level. Scores range from 0-120, with higher
103
scores indicating greater mobility. Previous investigations have supported the construct
104
and concurrent validity of the LSA in older adult populations,1, 11 its test-retest reliability
105
(Intraclass correlation coefficient (ICC) = 0.96),1 and its utility as an outcome measure in
106
clinical trials.12-14 We administered the LSA to both dyad members, asking each to focus
107
on the recent mobility of the older adult participant and thereby generating a LSAOlder Adult
108
and LSACompanion score.
109
AC C
EP
TE D
100
Physical Activity Scale for the Elderly: The PASE is a brief survey designed to
110
measure the physical activity of older adults during leisure, household, and occupational
111
activities occurring over the previous 7 day period.2 PASE item scores are calculated by
5
ACCEPTED MANUSCRIPT
multiplying either time spent (hours / week) or participation in (yes / no) each activity
113
with empirically derived item weights. The PASE total score represents the sum of
114
individual item scores and ranges from 0 to over 400. Higher scores correspond to greater
115
physical activity. The PASE has been validated for use in elderly populations,2,15 and its
116
test-retest reliability has been reported (r = 0.75).2 We administered the PASE to both
117
dyad members, asking each to focus on the recent physical activity of the older adult
118
participant and thereby generating a PASEOlder Adult and PASECompanion score.
SC
RI PT
112
119
121
Procedures
M AN U
120
Trained research personnel collected descriptive data and administered the LSA and PASE to each older adult and companion dyad in a place convenient to them (e.g. the
123
older adult’s home). Participants were sequestered from one another when completing the
124
surveys to avoid influencing each other’s responses.
TE D
122
125
127
Data Analysis
Data were analyzed using SPSS version 20.b Descriptive statistics were used to
EP
126
characterize the sample. To account for the possibility of recall error by either participant,
129
time spent together was characterized using the average reported value. Gait speed was
130
calculated as the mean of 2 trials.
131
AC C
128
Our analysis was based partly on the approach used in a previous investigation.16
132
At a group level, we analyzed agreement of scores using the dependent samples t-test (α
133
= 0.05), standardized mean difference (d),17 and ICC (1,1). At an individual level, we
134
calculated difference values between paired scores, constructed Bland-Altman plots,18
6
ACCEPTED MANUSCRIPT
and determined the 95% limits of agreement. Finally, we analyzed the relationship
136
between difference in paired scores and (a) time spent together and (b) mean gait speed
137
using appropriate parametric or non-parametric correlation coefficients (α = 0.05).
138
To facilitate the interpretation of individual-level results, we compared one
RI PT
135
instrument to the other by normalizing the raw difference scores as percent difference
140
values (= 100* [(older adult score – companion score) / mean of paired scores]). We also
141
used previous literature to provide estimates of measurement error against which the
142
agreement for each instrument could be compared. Because LSA measurement error
143
values had not been published previously, we used standard formulas19 and information
144
contained in the original LSA publication1 to calculate a Minimum Detectable Difference
145
(MDD) value based on a 95% confidence interval (MDD95% = 13.7 using ICC = 0.96 and
146
Standard Error of Measurement = 4.94.) For the PASE, we lacked sufficient information
147
from previous literature to produce an analogous measurement error estimate. Thus, we
148
selected instead the only available PASE measurement error value available, the Smallest
149
Detectable Difference (SDD) based on a 95% confidence interval from a recent study of
150
adult patients with cancer (SDD95% = 84).20
EP
TE D
M AN U
SC
139
152 153 154
AC C
151
RESULTS
Forty-five dyads were screened initially. None were rejected due to language,
155
cognitive, or physical limitations. Four dyads were unable to participate in data collection
156
due to scheduling conflicts. Data from one dyad were incomplete and therefore not
157
included in the analysis.
7
ACCEPTED MANUSCRIPT
158
The final sample contained 22 female and 18 male older adult participants (mean age ± SD of 80.7 ± 7.5 years.) Mean gait speed ± SD was 0.82 ± 0.26 m/s. Companions
160
included 33 females and 7 males (mean age ± SD of 69.2 ± 14.4 years.) Two companions
161
did not report their age. Twenty-five companions reported living with the older adult.
162
Mean time spent per week together ± SD was 90.7 ± 62.3 hours, with 19 dyads reporting
163
spending more than 100 hours per week together. Companions included 21 spouses, 11
164
children, 6 friends, and 2 unrelated caregivers. Additional characteristics of the sample
165
have been reported elsewhere.21
167 168
M AN U
166
SC
RI PT
159
Life-Space Assessment
Older adults as a group reported a moderate level of life-space mobility (mean LSAOlder Adult = 71.1, 95% confidence interval (CI) = 61.8 – 80.3), similar to the collective
170
report provided by their companions (mean LSACompanion = 68.5, 95% CI = 59.8 – 77.2).
171
The difference between group mean scores was not significant (mean difference = 2.54,
172
95% CI = - 1.77 to 6.84, t (39) = 1.19, p = 0.24), and the standardized mean difference
173
was small (d = 0.09). The 2 sets of scores were significantly yet moderately associated
174
(ICC (1,1) = 0.88, 95% CI = 0.79 - 0.94, p < 0.001). Difference in LSA scores was not
175
associated with time spent together (Spearman rho = -0.26, p = 0.11) or older adult gait
176
speed (Pearson r = 0.06, p = 0.70). Family members (n = 32) were relatively more
177
accurate in their characterization of older adult life-space mobility than friends or
178
caregivers (n = 8) (Figure 1). At an individual level, the mean percent difference between
179
paired LSA scores was 18.1% (95% CI = 13.5 – 22.7). Seven dyads had ≤ 5% difference
180
in scores, 15 dyads had ≤ 10% difference in scores, and 22 dyads had ≤ 20% difference in
AC C
EP
TE D
169
8
ACCEPTED MANUSCRIPT
scores (Figure 2). For all dyads, the percent difference in paired LSA scores was ≤ 60%.
182
Twenty-four dyads had raw difference values less than MDD95% = 13.7.
183
INSERT FIGURE 1 ABOUT HERE
184
INSERT FIGURE 2 ABOUT HERE
RI PT
181
Figure 3A features a Bland-Altman plot,18 in which the mean of the paired scores
186
(x-axis) was plotted against difference in paired LSA scores (y-axis). There was 1 case of
187
exact agreement. Positive differences (n = 23) indicated cases in which older adults
188
reported greater life-space mobility than their companion (mean difference = 11.8, 95%
189
CI = 8.1 – 15.4). Negative differences (n = 16) indicated cases in which companions
190
reported greater life-space mobility of the older adult than the older adult reported about
191
his or herself (mean difference = - 10.6, 95% CI = - 14.4 to - 6.8). The maximum paired
192
difference (=26.0) was less than 26.9, the 95% limits of agreement.
M AN U
SC
185
193
TE D
INSERT FIGURE 3 ABOUT HERE
194
196
Physical Activity Scale for the Elderly
Physical activity levels reported by older adults as a group were relatively low
EP
195
(mean PASEOlder Adult score = 81.7, 95% CI = 63.4 – 100.0) and slightly less than the
198
collective report of the companions about the older adults (mean PASECompanion = 82.4,
199
95% CI = 64.6 – 100.1). The difference between group mean scores was not significant
200
(mean difference = - 0.70, 95% CI = - 11.77 to 9.71, t (39) = 0.14, p = 0.89), and the
201
standardized mean difference between scores was small (d = 0.01). The 2 sets of scores
202
were significantly yet moderately associated (ICC (1,1) = 0.84, 95% CI = 0.71 - 0.91, p
