Construct Validity of Activities of Daily Living Scale* - CHEST ...
Construct Validity of Activities of Daily Living Scale* - CHEST ...
Construct Validity of Activities of Daily Living Scale* - CHEST ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Construct</strong> <strong>Validity</strong> <strong>of</strong> <strong>Activities</strong> <strong>of</strong> <strong>Daily</strong><br />
<strong>Living</strong> <strong>Scale*</strong><br />
A Clue To Distinguish the Disabling Effects <strong>of</strong><br />
COPD and Congestive Heart Failure<br />
Raffaele Antonelli Incalzi, MD; Andrea Corsonello, MD; Claudio Pedone, MD;<br />
Francesco Corica, MD; Pierugo Carbonin, MD; Roberto Bernabei, MD; on<br />
Behalf <strong>of</strong> the GIFA Investigators†<br />
Study objectives: To assess differences, if any, in the pattern <strong>of</strong> disability measured using basic<br />
activities <strong>of</strong> daily living (BADL) and instrumental activities <strong>of</strong> daily living (IADL) in COPD and<br />
congestive heart failure (CHF), using diabetes mellitus as a reference noncardiorespiratory<br />
disabling condition.<br />
Design: Multicenter survey.<br />
Setting: General medicine or geriatric wards in tertiary hospitals throughout Italy.<br />
Patients: Patients admitted because <strong>of</strong> CHF (n 432), COPD (n 305), and diabetes mellitus<br />
(n 534).<br />
Measurements and results: <strong>Construct</strong> validity <strong>of</strong> self-reported preadmission BADL-IADL was<br />
assessed for each group by main component analysis. The three populations had a comparable<br />
average degree <strong>of</strong> dependency in BADL-IADL. In both CHF and diabetes mellitus patients, three<br />
components cumulatively explained most <strong>of</strong> variance in BADL-IADL: the BADL, 10 IADL, and<br />
4 housework-related IADL. In COPD, a four-factor solution was generated, with factor 4 having<br />
loading with IADL items assessing mobility and outdoor moving, and factor 3 with selected IADL<br />
requiring both physical and mental capabilities such as managing money, taking medicine, and<br />
traveling. Correlates <strong>of</strong> dependency in IADL related to factor 4 in COPD were older age,<br />
cognitive impairment, widowhood, and comorbidity. Both factors 3 and 4 were associated with<br />
longer stay (factor 3: 13.9 9.5 days vs 11.5 7.6 days, p < 0.05; factor 4: 14.2 8.8 days vs<br />
11.0 5.5 days, p < 0.05) <strong>of</strong> COPD patients (mean SD).<br />
Conclusion: COPD was associated with a distinctive pattern <strong>of</strong> disability expressed by loss <strong>of</strong><br />
selected BADL-IADL but not by the crude number <strong>of</strong> lost BADL-IADL.<br />
(<strong>CHEST</strong> 2005; 127:830–838)<br />
Key words: activities <strong>of</strong> daily living; congestive heart failure; construct validity; COPD; diabetes mellitus; disability<br />
Abbreviations: ADL activities <strong>of</strong> daily living; AMT abbreviated mental test; BADL basic activities <strong>of</strong> daily<br />
living; CHF congestive heart failure; GDS geriatric depression scale; GIFA Gruppo Italiano di Farmacovigilanza<br />
nell’Anzianol; IADL instrumental activities <strong>of</strong> daily living; ICD-9 CM International Classification <strong>of</strong> Diseases,<br />
Clinical Modification, Ninth Revision<br />
<strong>Activities</strong> <strong>of</strong> daily living (ADL) include basic ADL<br />
(BADL), which explore the basic capacity <strong>of</strong><br />
persons to care for themselves, and instrumental<br />
ADL (IADL), which refer to higher levels <strong>of</strong> performance.<br />
1 Both BADL and IADL should be evaluated<br />
*From the Centro di Medicina dell’Invecchiamento (Drs. Incalzi,<br />
Pedone, Carbonin, and Bernabei), Policlinico A. Gemelli, Università<br />
Cattolica del Sacro Cuore, Roma, Italy; Divisione di<br />
Medicina Geriatrica (Dr. Corsonello), Istituto Nazionale di<br />
Ricovero e Cura per Anziani, Cosenza, Italy; and Dipartimento di<br />
Medicina Interna (Dr. Corica), Università degli Studi di Messina,<br />
Messina, Italy.<br />
†The GIFA is a research group <strong>of</strong> the Italian Society <strong>of</strong> Gerontology<br />
and Geriatrics—Fondazione Italiana per la Ricerca<br />
sull’Invecchiamento.<br />
to capture the full spectrum <strong>of</strong> disability. 2 Acute<br />
disabling conditions such as stroke or hip fracture<br />
have obvious and dramatic effects on ADL, whereas<br />
chronic conditions that do not cause a segmental<br />
motor deficit have a more complex and less easily<br />
The GIFA is partially supported by a grant from the Italian<br />
National Research Council (No. 94000402).<br />
Manuscript received February 6, 2004; revision accepted October<br />
26, 2004.<br />
Reproduction <strong>of</strong> this article is prohibited without written permission<br />
from the American College <strong>of</strong> Chest Physicians (e-mail:<br />
permissions@chestnet.org).<br />
Correspondence to: Andrea Corsonello, MD, Via D. Frugiuele,<br />
39, I-87100 Cosenza, Italy; e-mail: andrea corsonello@tin.it<br />
830 Clinical Investigations<br />
Downloaded From: http://chestioumal.chestpubs.org/ on 07/03/2013
predictable effect. Among these conditions congestive<br />
heart failure (CHF) and COPD are prime<br />
examples. Ischemic heart disease, mainly as a cause<br />
<strong>of</strong> CHF, and COPD are expected to become the first<br />
and fifth cause <strong>of</strong> disability worldwide by the year<br />
2020. 3 Both these conditions are highly prevalent in<br />
the elderly, with CHF affecting 1 in 10 people 74<br />
years <strong>of</strong> age, and COPD patients needing oxygen<br />
therapy having a mean age <strong>of</strong> approximately 70<br />
years. 4,5 Disability complicates mainly CHF in New<br />
York Heart Association class III-IV and COPD with<br />
chronic hypoxemia. 6,7 Furthermore, disability, as expressed<br />
by the number <strong>of</strong> lost ADL, is an independent<br />
predictor <strong>of</strong> survival in CHF. 8,9<br />
The main determinants <strong>of</strong> disability are dyspnea<br />
and loss <strong>of</strong> muscle mass in both CHF and COPD,<br />
but comorbidity is an important contributor, at least<br />
in COPD. 10,11 However, mechanisms and types <strong>of</strong><br />
dyspnea distinguish CHF from COPD 12–16 : it is<br />
mainly inspiratory and due to the high respiratory<br />
work needed to overcome the loss <strong>of</strong> pulmonary<br />
compliance in CHF, while it is elicited mainly by<br />
increased airway resistances and lung hyperinflation<br />
worsening the tension-length relationship <strong>of</strong> the<br />
diaphragm in COPD. Furthermore, the recruitment<br />
<strong>of</strong> selected upper-arm and torso muscles as auxiliary<br />
respiratory muscles makes COPD patients at risk <strong>of</strong><br />
dependency in upper-arm–based ADL. 16 Finally,<br />
both CHF and COPD are frequently associated with<br />
depression and cognitive impairment, which may<br />
further contribute to limit the physical performance,<br />
especially in the IADL domain. 17,18 However, both<br />
the timing <strong>of</strong> onset <strong>of</strong> depression and the pattern <strong>of</strong><br />
cognitive impairment distinguish CHF from<br />
COPD. 19,20<br />
Physical dependency can be seen as the end result<br />
<strong>of</strong> the complex interaction among physical, cognitive,<br />
and affective factors. Such an interaction likely yields<br />
different results for CHF and COPD patients, given<br />
that factors at the basis <strong>of</strong> dependency are to some<br />
extent different in CHF and COPD. As a consequence,<br />
we can expect that the hierarchy <strong>of</strong> cumulatively<br />
considered BADL and IADL are also different<br />
in CHF and COPD patients.<br />
We designed the present study to assess the<br />
applicability and utility <strong>of</strong> the BADL-IADL scale in<br />
CHF and COPD as well as in diabetes mellitus. We<br />
selected the last condition as a comparison disorder<br />
due to the fact that the impact <strong>of</strong> diabetes mellitus<br />
on physical independence is not primarily mediated<br />
by dyspnea. 21 The objective <strong>of</strong> our study was to verify<br />
whether these three chronic conditions correspond<br />
to different internal structures <strong>of</strong> the BADL-IADL<br />
scale. Finding distinctive multiple dimensions, ie,<br />
different combinations <strong>of</strong> items, inside the BADL-<br />
IADL scale as a function <strong>of</strong> the main disease would<br />
demonstrate that CHF, COPD, and diabetes mellitus<br />
affect physical function differently.<br />
Patients<br />
Materials and Methods<br />
The present study uses data from a large collaborative observational<br />
study group, the Gruppo Italiano di Farmacovigilanza<br />
nell’Anziano (GIFA), based in community and university hospitals<br />
located throughout Italy, that periodically surveys drug<br />
consumption, occurrence <strong>of</strong> adverse drug reactions, and quality<br />
<strong>of</strong> hospital care. We used data on patients consecutively admitted<br />
to the 24 participating centers during the 4-month survey carried<br />
out in 1998. Methods <strong>of</strong> the GIFA have been previously described.<br />
22,23 Briefly, a study physician with specific training<br />
completed a questionnaire for each patient at admission to<br />
hospital and updated it daily. Data recorded included sociodemographic<br />
characteristics, medical variables, CBC count, and<br />
neuropsychological and physical function variables.<br />
Overall, 3,010 patients were enrolled in the survey period. We<br />
excluded patients who died during hospital stay (n 117) and<br />
those for whom the BADL-IADL rating was not available<br />
(n 132), with a final sample <strong>of</strong> 2,761 patients. On the basis <strong>of</strong><br />
the primary diagnosis coded according to criteria by International<br />
Classification <strong>of</strong> Diseases, Clinical Modification, Ninth Revision<br />
(ICD-9 CM), 24 we identified 432 patients (15.6%) with CHF<br />
(ICD-9 CM 428–428.9), 305 (11.0%) with COPD (ICD-9<br />
CM 490–492.8, 494, 496), and 534 (19.3%) with diabetes<br />
mellitus (ICD-9 CM 250–250.7). These three groups represented<br />
our study populations.<br />
Preadmission functional capabilities, ie, those before the acute<br />
exacerbation leading to the hospitalization, were rated by a<br />
21-item ADL scale. 25 The patients were interviewed on the day<br />
before the planned discharge because a great proportion <strong>of</strong> them<br />
were too seriously ill to be interviewed on hospital admission.<br />
Performance on individual ADL was scored according to the five<br />
levels <strong>of</strong> difficulty recommended by the World Health Organization:<br />
without difficulty, with difficulty but without help, help<br />
for only part <strong>of</strong> the activity, help for total activity, and not able to<br />
perform. 26<br />
Variables specifically considered in this study were age, gender,<br />
education, type <strong>of</strong> ward, living alone, and being widowed. The<br />
cognitive and affective status were assessed by the Hodkinson<br />
abbreviated mental test (AMT) and the 15-item geriatric depression<br />
scale (GDS), respectively. 27,28 The Charlson index <strong>of</strong> comorbidity<br />
was calculated. 29 Drugs were coded by anatomic and<br />
therapeutic classification. 30 Procedures conformed to guidelines<br />
provided by the Catholic University Ethical Committee.<br />
Statistical Analysis<br />
Characteristics <strong>of</strong> groups were compared by the 2 test for<br />
categorical variables and the analysis <strong>of</strong> variance for continuous<br />
variables. The construct validity <strong>of</strong> the BADL-IADL scale within<br />
individual groups was assessed by the main component analysis. 31<br />
As a first step, we tested the appropriateness <strong>of</strong> the main<br />
component analysis by the Kaiser-Meyer-Olkin measure <strong>of</strong> sampling<br />
adequacy and the Bartlett test <strong>of</strong> sphericity. 32 Commonalities<br />
(the squared multiple correlation coefficients between a<br />
variable and all other variables) were assumed to reflect the<br />
strength <strong>of</strong> the linear association among the variables.<br />
The basic assumption <strong>of</strong> factor analysis is that complex phenomena,<br />
such as the BADL-IADL based pr<strong>of</strong>ile <strong>of</strong> functional<br />
www.chestjournal.org <strong>CHEST</strong> / 127 /3/MARCH, 2005 831<br />
Downloaded From: http://chestioumal.chestpubs.org/ on 07/03/2013
independence, can be explained by underlying dimensions or<br />
factors. Factors corresponding to a set <strong>of</strong> closely interrelated<br />
BADL-IADL items were identified by the main component<br />
analysis (factor extraction). The component matrix was obtained,<br />
and then a rotated component matrix was derived through an<br />
oblique rotation. The component matrix represents the crude<br />
relationship between the factors and the individual BADL-IADL<br />
items, but it fails to disclose an easily interpretable pattern <strong>of</strong><br />
correlation because many items have moderate-size correlations<br />
with most factors. On the contrary, the oblique rotation simplifies<br />
the relationship between individual items and factors by minimizing<br />
the number <strong>of</strong> BADL-IADL items having loadings on a<br />
factor. The rotation phase makes the initial matrix easier to<br />
interpret, grouping BADL-IADL items to identify factors corresponding<br />
to well-defined dimensions <strong>of</strong> physical independence.<br />
The strength <strong>of</strong> the relationship between factor and item was<br />
assumed to be directly proportional to the magnitude <strong>of</strong> the<br />
coefficient relating individual items to the extracted factor.<br />
Furthermore, the oblique rotation provides a measure <strong>of</strong> correlations<br />
among factors.<br />
We identified factors to be retained in the final model on a<br />
graphical plot <strong>of</strong> explained variance by corresponding factor:<br />
retained factors can be recognized as those preceding the<br />
boundary between factors explaining large and small proportions<br />
<strong>of</strong> variance. 31 A scale that explains most <strong>of</strong> the observed variance<br />
by a few factors has a solid internal structure, ie, chance variations<br />
are unlikely to confound the interpretation <strong>of</strong> the differences<br />
observed both within and between groups. The main component<br />
analysis was repeated on subgroups defined by sex to verify to<br />
which extent gender contributed to the loading <strong>of</strong> BADL//IADL<br />
items on retained factors. All analyses were performed using<br />
statistical s<strong>of</strong>tware (SPSS V10.0; SPSS; Chicago, IL).<br />
Results<br />
Sociodemographic, neuropsychological, and clinical<br />
characteristics <strong>of</strong> CHF, COPD, and diabetes<br />
Table 1—General Characteristics <strong>of</strong> the Groups Studied*<br />
mellitus patients are reported in Table 1. CHF<br />
patients were characterized by older age and higher<br />
prevalence <strong>of</strong> widowhood than COPD and diabetes<br />
mellitus patients, whereas male gender prevailed in<br />
the COPD group. Admission to a geriatric ward was<br />
more frequent in both CHF and COPD groups. The<br />
cognitive status, as reflected by the mean AMT<br />
score, was fairly normal, whereas the mean GDS<br />
score was consistent with a highly prevalent depressive<br />
trait in all groups. CHF was recorded as a<br />
secondary diagnosis in 24.9% <strong>of</strong> COPD patients and<br />
in 19.5% <strong>of</strong> diabetes mellitus patients; 17.6% <strong>of</strong><br />
CHF patients and 8.6% <strong>of</strong> diabetics had a secondary<br />
diagnosis <strong>of</strong> COPD. Finally, a secondary diagnosis <strong>of</strong><br />
diabetes mellitus was present in 24.1% <strong>of</strong> CHF<br />
patients and 15.1% <strong>of</strong> COPD patients.<br />
The functional status <strong>of</strong> groups is summarized in<br />
Table 2. The observation <strong>of</strong> crude data did not reveal<br />
any distinctive clustering <strong>of</strong> BADL/IADL within<br />
individual groups.<br />
The main component analysis was considered<br />
feasible in all the three groups. Indeed, the high<br />
Kaiser-Meyer-Olkin measures <strong>of</strong> sampling adequacy<br />
(CHF 0.931, COPD 0.923, diabetes mellitus<br />
0.943) confirm the hypothesis that the correlations<br />
between pairs <strong>of</strong> variables can be explained<br />
by the other variables. The highly significant Bartlett<br />
test <strong>of</strong> sphericity (CHF 8680.904, p 0.001;<br />
COPD 6674.681, p 0.001; diabetes mellitus<br />
12208.377, p 0.001) denies that the item<br />
correlation matrix is an identity matrix.<br />
Three main components explained most <strong>of</strong> the<br />
Characteristics CHF (n 432) COPD (n 305) Diabetes Mellitus (n 534) p Value†<br />
Age, yr 77.0 10.6 74.7 11.8 71.0 12.7 0.008<br />
Male gender 225 (52.1) 197 (64.6) 282 (52.8) 0.047<br />
Education, yr 5.8 4.1 5.6 3.8 6.3 4.0 0.089<br />
<strong>Living</strong> alone 70 (16.2) 36 (11.8) 77 (14.4) 0.288<br />
Being widowed 161 (37.3) 93 (30.5) 164 (30.7) 0.146<br />
Type <strong>of</strong> ward 0.046<br />
Geriatric 295 (68.3) 219 (71.8) 331 (62.0)<br />
Medicine 137 (31.7) 86 (28.2) 203 (38.0)<br />
Charlson index <strong>of</strong> comorbidity 2.5 1.5 2.5 1.5 2.6 1.6 0.348<br />
AMT score 7.9 2.1 7.9 2.2 8.0 2.1 0.568<br />
GDS score 5.7 3.7 5.4 3.5 5.2 3.8 0.267<br />
Coronary artery disease 139 (32.2) 82 (26.9) 191 (35.8) 0.091<br />
Congestive heart failure 76 (24.9) 104 (19.5) 0.114<br />
Peripheral vascular disease 27 (6.3) 21 (6.9) 24 (4.5) 0.283<br />
Cerebral vascular disease 56 (13.0) 53 (17.4) 79 (14.8) 0.298<br />
Stroke 18 (4.2) 20 (6.6) 23 (4.3) 0.323<br />
Dementia 27 (6.3) 26 (8.5) 43 (8.1) 0.425<br />
COPD 76 (17.6) 46 (8.6) 0.001<br />
Chronic renal failure 93 (21.5) 38 (12.5) 84 (15.7) 0.010<br />
Malignancies 26 (6.0) 22 (7.2) 37 (6.9) 0.763<br />
Diabetes 104 (24.1) 46 (15.1) 0.010<br />
*Data are expressed as mean SD or No. (%). An AMT score 6 and GDS score 6 are considered normal values.<br />
†Values refer to the three (groups) by two (level) 2 test for categorical variables or one-way analysis <strong>of</strong> variance for continuous variables.<br />
832 Clinical Investigations<br />
Downloaded From: http://chestioumal.chestpubs.org/ on 07/03/2013
Table 2—ADL and IADL Performance in the Groups Studied*<br />
Variables CHF (n 432) COPD (n 305) Diabetes Mellitus (n 534) p Value†<br />
1 Transferring 37 (8.6) 28 (9.2) 58 (10.9) 0.516<br />
2 Ambulation 37 (8.6) 32 (10.5) 61 (11.4) 0.406<br />
3 Dressing 36 (8.3) 28 (9.2) 58 (10.9) 0.447<br />
4 Eating 25 (5.8) 20 (6.6) 42 (7.9) 0.391<br />
5 Toileting 31 (7.2) 27 (8.9) 50 (9.4) 0.456<br />
6 Bathing 35 (8.1) 29 (9.5) 57 (10.7) 0.454<br />
7 Continence 34 (7.9) 27 (8.9) 58 (10.9) 0.313<br />
Dependent in 3 BADL 37 (8.6) 30 (9.8) 60 (11.2) 0.409<br />
8 Getting around outside 76 (17.6) 54 (17.7) 96 (18.0) 0.990<br />
9 Going up or down the stairs 85 (19.7) 56 (18.4) 108 (20.2) 0.839<br />
10 Walking for at least 400 m 87 (20.1) 58 (19.0) 111 (20.8) 0.861<br />
11 Grocery shopping 135 (31.3) 83 (27.2) 145 (27.2) 0.413<br />
12 Taking a bath or shower 82 (19.0) 54 (17.7) 100 (18.7) 0.896<br />
13 Preparing meals 94 (21.8) 59 (19.3) 102 (19.1) 0.650<br />
14 Light houseworking 103 (23.8) 65 (21.3) 111 (20.8) 0.594<br />
15 Heavy houseworking 195 (45.1) 120 (39.3) 236 (44.2) 0.463<br />
16 Cutting toe nails 113 (26.2) 76 (24.9) 125 (23.4) 0.698<br />
17 Telephoning 51 (11.8) 34 (11.1) 71 (13.3) 0.705<br />
18 Laundry 128 (29.6) 85 (27.9) 141 (26.4) 0.619<br />
19 Traveling 141 (32.6) 86 (28.2) 150 (28.1) 0.368<br />
20 Taking medicine 70 (16.2) 52 (17.0) 81 (15.2) 0.823<br />
21 Managing money 92 (21.3) 54 (17.7) 108 (20.2) 0.541<br />
Dependent 3 IADL 161 (37.3) 97 (31.8) 167 (31.3) 0.216<br />
*Data are presented as absolute No. (% <strong>of</strong> patients dependent in individual BADL-IADL items).<br />
†Values refer to the three (groups) by two (level) 2 test.<br />
observed variance in BADL/IADL in CHF patients<br />
(69% <strong>of</strong> variance) and diabetes mellitus patients<br />
(73%); a four-component solution explained 75% <strong>of</strong><br />
variance in COPD group (Table 3).<br />
Table 4 shows the rotated component matrix and<br />
total variance explained by factors 1 to 3 in CHF<br />
patients and in diabetes mellitus patients. Both<br />
models were based on the loading <strong>of</strong> BADL items,<br />
telephoning and, in diabetes mellitus, taking medicine<br />
on the factor 1. Of the remaining IADL,<br />
housework-related had loadings on factor 2, and the<br />
others on factor 3. In both CHF and diabetes<br />
mellitus populations, the main component analysis<br />
limited to female gender gave a simpler solution<br />
based on two factors having loadings with BADL and<br />
IADL, respectively.<br />
Table 5 shows the rotated component matrix and<br />
total variance explained by factors 1 to 4 in COPD.<br />
Factor 1 had loading with BADL and three IADL<br />
(getting around outside, taking a bath or shower, and<br />
Table 3—Main Components Identified and Percentage <strong>of</strong> Variance Explained in CHF, COPD, and Diabetes Mellitus<br />
Groups*<br />
Component<br />
Initial Eigenvalues Extraction Sums <strong>of</strong> Squared Loadings Rotation Sums <strong>of</strong> Squared Loadings<br />
Total % <strong>of</strong> Variance Cumulative % Total Cumulative % % <strong>of</strong> Variance Total Cumulative % % <strong>of</strong> Variance<br />
CHF<br />
1 10.057 47.891 47.891 10.057 47.891 47.891 7.190 34.237 34.237<br />
2 2.889 13.757 61.647 2.889 13.757 61.647 4.366 20.792 55.029<br />
3<br />
COPD<br />
1.572 7.484 69.132 1.572 7.484 69.132 2.962 14.103 69.132<br />
1 10.132 48.250 48.250 10.132 48.250 48.250 7.304 34.781 34.781<br />
2 3.124 14.876 63.126 3.124 14.876 63.126 3.021 14.386 49.167<br />
3 1.411 6.721 69.847 1.411 6.721 69.847 2.938 13.991 63.158<br />
4<br />
Diabetes Mellitus<br />
1.051 5.005 74.851 1.051 5.005 74.851 2.456 11.693 74.851<br />
1 11.221 53.431 53.431 11.221 53.431 53.431 8.075 38.453 38.453<br />
2 2.853 13.583 67.014 2.853 13.583 67.014 4.155 19.787 58.240<br />
3 1.267 6.033 73.047 1.267 6.033 73.047 3.110 14.808 73.047<br />
*Only factors preceding the boundary between factors explaining large and small fractions <strong>of</strong> the global variance in BADL-IADL are reported.<br />
www.chestjournal.org <strong>CHEST</strong> / 127 /3/MARCH, 2005 833<br />
Downloaded From: http://chestioumal.chestpubs.org/ on 07/03/2013
BADL-IADL Item<br />
Table 4—Rotated Structure Matrix in Patients Affected by CHF and Diabetes Mellitus*<br />
telephoning). Factor 2 had loading with houseworkrelated<br />
IADL. Two IADL strictly related to physical<br />
mobility (going up or down the stairs, walking for at<br />
Congestive Heart Failure Diabetes Mellitus<br />
Factor 1 Factor 2 Factor 3 Factor 1 Factor 2 Factor 3<br />
1 Transferring 0.951† 0.106 0.536 0.951† 0.184 0.650<br />
2 Ambulation 0.936† 0.117 0.562 0.935† 0.193 0.651<br />
3 Dressing 0.948† 0.144 0.592 0.952† 0.194 0.661<br />
4 Eating 0.917† 0.090 0.434 0.926† 0.140 0.512<br />
5 Toileting 0.952† 0.116 0.501 0.950† 0.153 0.585<br />
6 Bathing 0.962† 0.104 0.520 0.960† 0.153 0.604<br />
7 Continence 0.860† 0.030 0.412 0.897† 0.140 0.493<br />
8 Getting around outside 0.687 0.140 0.827† 0.741 0.215 0.841†<br />
9 Going up or down the stairs 0.656 0.126 0.743† 0.598 0.215 0.850†<br />
10 Walking for at least 400 m 0.658 0.127 0.741† 0.611 0.266 0.855†<br />
11 Grocery shopping 0.143 0.700† 0.376 0.232 0.674† 0.513<br />
12 Taking a bath or shower 0.674 0.176 0.811† 0.761 0.288 0.808†<br />
13 Preparing meals 0.145 0.780† 0.215 0.198 0.771† 0.264<br />
14 Light houseworking 0.210 0.796† 0.211 0.281 0.782† 0.286<br />
15 Heavy houseworking 0.011 0.794† 0.087 0.099 0.829† 0.233<br />
16 Cutting toe nails 0.457 0.216 0.662† 0.644 0.245 0.725†<br />
17 Telephoning 0.633† 0.190 0.619 0.851† 0.157 0.577<br />
18 Laundry 0.061 0.732† 0.049 0.016 0.772† 0.114<br />
19 Traveling 0.281 0.041 0.745† 0.439 0.198 0.790†<br />
20 Taking medicine 0.584 0.184 0.692† 0.793† 0.201 0.640<br />
21 Managing money 0.286 0.242 0.661† 0.381 0.395 0.567†<br />
*The magnitude <strong>of</strong> coefficients relating individual items to the extracted factor is a measure <strong>of</strong> the strength <strong>of</strong> this association. Extraction<br />
method main component analysis; rotation method oblimin with Kaiser normalization.<br />
†The greatest loading is reported to show the relationship with the main components.<br />
least 400 m) had loading on factor 4, whereas five<br />
IADL requiring highly efficient physical and mental<br />
status had loading on factor 3. After stratification for<br />
Table 5—Rotated Structure Matrix in Patients Affected by COPD*<br />
BADL-IADL Item Factor 1 Factor 2 Factor 3 Factor 4<br />
1 Transferring 0.926† 0.063 0.372 0.496<br />
2 Ambulation 0.924† 0.079 0.389 0.506<br />
3 Dressing 0.957† 0.071 0.440 0.461<br />
4 Eating 0.932† 0.045 0.322 0.345<br />
5 Toileting 0.951† 0.064 0.362 0.395<br />
6 Bathing 0.962† 0.039 0.422 0.414<br />
7 Continence 0.823† 0.023 0.376 0.184<br />
8 Getting around outside 0.749† 0.101 0.565 0.700<br />
9 Going up or down the stairs 0.616 0.105 0.434 0.880†<br />
10 Walking for at least 400 m 0.611 0.105 0.446 0.879†<br />
11 Grocery shopping 0.117 0.577† 0.480 0.266<br />
12 Taking a bath or shower 0.717† 0.166 0.691† 0.576<br />
13 Preparing meals 0.077 0.806† 0.236 0.022<br />
14 Light houseworking 0.126 0.819† 0.240 0.098<br />
15 Heavy houseworking 0.025 0.824† 0.159 0.074<br />
16 Cutting toe nails 0.525 0.238 0.655† 0.413<br />
17 Telephoning 0.678† 0.138 0.487 0.086<br />
18 Laundry 0.050 0.815† 0.102 0.020<br />
19 Traveling 0.305 0.133 0.733† 0.408<br />
20 Taking medicine 0.649 0.209 0.762† 0.160<br />
21 Managing money 0.257 0.237 0.715† 0.076<br />
*The magnitude <strong>of</strong> coefficients relating individual items to the extracted factor is a measure <strong>of</strong> the strength <strong>of</strong> this association. Extraction<br />
method main component analysis; rotation method oblimin with Kaiser normalization.<br />
†The greatest loading is reported to show the relationship with the main components.<br />
834 Clinical Investigations<br />
Downloaded From: http://chestioumal.chestpubs.org/ on 07/03/2013
gender, a simpler three-factor solution was identified<br />
in COPD female patients, whereas the four-factor<br />
solution identified in male patients was comparable<br />
to that generated in the whole COPD population.<br />
In all the statistical models, the ratio <strong>of</strong> the<br />
variance explained by factor 1 to that explained by<br />
factor 2 was inferior to the cut <strong>of</strong>f <strong>of</strong> 3.5 that defines<br />
a test as unidimensional. 18 This finding is consistent<br />
with the partitioning <strong>of</strong> ADL in two groups <strong>of</strong> items<br />
(BADL and IADL) assessing different levels <strong>of</strong><br />
physical capacity.<br />
Table 6 shows the correlations among factors in<br />
individual groups <strong>of</strong> patients. In both CHF and<br />
diabetes mellitus groups, the strongest correlation<br />
was found between factors 1 and 3; both these<br />
factors were weakly correlated with factor 2, ie, to<br />
housework-related IADL. In COPD group, correlations<br />
among factors were weaker than in CHF and<br />
diabetes mellitus groups.<br />
In COPD patients, dependency in mobility-related<br />
IADL (going up or down the stairs, walking for<br />
at least 400 m) was associated with older age (82 8<br />
years vs 74 12 years, p 0.001 [mean SD]),<br />
male gender (67.4% vs 43.6%, p 0.01), lower<br />
cognitive performance (AMT 6.0 2.7 vs<br />
AMT 8.0 2.1, p 0.001), and greater comorbidity<br />
(Charlson index 3.3 1.9 vs 2.4 1.5, p 0.01),<br />
but not with depression (GDS 5.8 4.5 vs<br />
5.3 3.4, p 0.522). Arterial hypoxemia requiring<br />
long-term oxygen therapy was not significantly less<br />
prevalent in patients identified by factor 2 (7.5%) than<br />
in those identified by factor 1 (11.1%), factor 3 (11.7%),<br />
and factor 4 (12.3%).<br />
Diabetics dependent in at least three IADL were<br />
older (77.7 10.0 years vs 68.0 12.6 years,<br />
p 0.001), more depressed (GDS 6.7 4.1 vs<br />
4.7 3.5, p 0.001), and cognitively impaired<br />
Table 6—Components Correlation Matrix in Patients<br />
Affected by CHF, Diabetes Mellitus, and COPD*<br />
Component 1 2 3 4<br />
CHF<br />
1 1.000 0.103 0.533<br />
2 0.103 1.000 0.191<br />
3<br />
Diabetes mellitus<br />
0.533 0.191 1.000<br />
1 1.000 0.165 0.620<br />
2 0.165 1.000 0.303<br />
3<br />
COPD<br />
0.620 0.303 1.000<br />
1 1.000 0.060 0.426 0.348<br />
2 0.060 1.000 0.260 0.057<br />
3 0.426 0.260 1.000 0.255<br />
4 0.348 0.057 0.255 1.000<br />
*Extraction method main component analysis; rotation<br />
method oblimin with Kaiser normalization.<br />
(AMT 6.8 2.5 vs 8.4 1.8, p 0.001), and had<br />
greater comorbidity (Charlson index 3.2 1.7 vs<br />
2.3 1.5, p 0.01) and prevalence <strong>of</strong> chronic renal<br />
failure (20.4% vs 10.1%, p 0.001) and cerebrovascular<br />
disease (21% vs 12%, p 0.01) than the<br />
remaining ones. Except for chronic renal failure and<br />
cerebrovascular diseases, the same conditions were<br />
found to be associated with dependency in three or<br />
more IADL in CHF and COPD patients.<br />
The relationship between factors and length <strong>of</strong><br />
stay is represented in Figure 1. Both factors 3 and 4<br />
were associated with a longer length <strong>of</strong> stay in the<br />
COPD group, whereas such an association was found<br />
for factor 2 only in the CHF group. The latter<br />
finding reflects the longer stay <strong>of</strong> female patients<br />
(14.5 9.9 days vs 12.0 6.7 days, p 0.05) but<br />
not <strong>of</strong> male patients (12.6 8.4 days vs 12.9 6.9<br />
days, p 0.864) identified by factor 2. No factor was<br />
associated with a longer stay in the diabetes mellitus<br />
group. Conventional measures <strong>of</strong> physical dependency,<br />
such as dependency in at least half <strong>of</strong> BADL<br />
or IADL, were unrelated to the length <strong>of</strong> stay.<br />
Discussion<br />
The main finding from this study is that BADL-<br />
IADL cluster in a very similar manner in two<br />
populations with different chronic conditions such as<br />
CHF and diabetes mellitus. A quite different hierarchy<br />
<strong>of</strong> IADL characterizes COPD, with a factor<br />
being expression <strong>of</strong> IADL related to outdoor mobility<br />
and another summarizing selected highly demanding<br />
IADL. These findings demonstrate that the<br />
crude computation <strong>of</strong> lost IADL may not be fully<br />
representative <strong>of</strong> the impact the loss has on personal<br />
independence. Furthermore, the clustering <strong>of</strong> selected<br />
IADL within a given population may be a<br />
distinctive feature <strong>of</strong> the pr<strong>of</strong>ile <strong>of</strong> personal independence<br />
<strong>of</strong> subjects belonging to that population.<br />
The pattern <strong>of</strong> segregation <strong>of</strong> IADL in COPD<br />
patients deserves consideration. The two components<br />
respectively based on IADL exploring outdoor<br />
mobility and selected highly demanding IADL accounted<br />
for about one third <strong>of</strong> the cumulatively<br />
explained variance in BADL-IADL. The former<br />
component confirms that difficulty in moving outdoor<br />
is a distinctive feature <strong>of</strong> COPD-related disability<br />
in the elderly. 33 Both these components are<br />
related to neuropsychological status and comorbidity,<br />
but also to older age and widowhood. The<br />
relationship with age might reflect either the negative<br />
effect <strong>of</strong> aging per se on cognition or the role <strong>of</strong><br />
age as a surrogate marker <strong>of</strong> COPD duration. Indeed,<br />
both self-medication and managing money<br />
require an intact cognition, whereas traveling implies<br />
www.chestjournal.org <strong>CHEST</strong> / 127 /3/MARCH, 2005 835<br />
Downloaded From: http://chestioumal.chestpubs.org/ on 07/03/2013
Figure 1. Length <strong>of</strong> hospital stay in relation to factors identified by main component analysis in the<br />
three groups studied. See Tables 4, 5 for definition <strong>of</strong> factors. NIDDM non–insulin-dependent<br />
diabetes mellitus.<br />
that both physical and mental capabilities are preserved.<br />
In agreement with our findings, dependency<br />
in taking medications and handling finances is distinctively<br />
associated with cognitive impairment in the<br />
broad population <strong>of</strong> elderly community residents. 34<br />
These findings might have implication for the management<br />
<strong>of</strong> COPD patients because they identify<br />
clusters <strong>of</strong> functions that are lost or retained together.<br />
Theoretically, selected patients might regain<br />
a cluster <strong>of</strong> functions by a dedicated intervention.<br />
For example, outdoor mobility could be regained by<br />
using a portable device for oxygen delivery. However,<br />
measures <strong>of</strong> nonphysical capabilities might<br />
help to recognize patients with distinctive problems<br />
and, then, need <strong>of</strong> care; for example, scoring 6at<br />
the AMT might target patients at risk <strong>of</strong> low compliance<br />
with pharmacologic and oxygen therapy,<br />
which is very common among COPD patients with<br />
hypoxemia. 35 Collaterally, our results confirm that<br />
dependency is highly prevalent among older COPD<br />
patients. 36 However, COPD patients have been reported<br />
to receive less formal support (home care,<br />
district nurse input, physiotherapy) than patients<br />
having a similar degree <strong>of</strong> dependency due to neurologic<br />
or orthopedic problems, as if the health-care<br />
pr<strong>of</strong>essionals did not fully appreciate the disabling<br />
effect <strong>of</strong> COPD. 37<br />
The slightly higher prevalence <strong>of</strong> important limitations<br />
in BADL and IADL in diabetic than in CHF<br />
and COPD population is only to some extent unexpected.<br />
Indeed, home-dwelling Mexican Americans<br />
60 years old with non–insulin-dependent diabetes<br />
mellitus were found to be more impaired than<br />
nondiabetics in both BADL and IADL, with an<br />
average loss <strong>of</strong> eight IADL. 38 Interestingly, lower<br />
functional status was associated with diabetic complications<br />
and was predictive or more rapid further<br />
decline. 38 Also, in a more healthy population 65<br />
years old, diabetes mellitus was associated with lower<br />
functional status. 21 Thus, our findings confirm that<br />
diabetes mellitus carries a high risk <strong>of</strong> dependency in<br />
the elderly. The association between dependency in<br />
the IADL and both renal failure and cerebrovascular<br />
disease was particular to diabetic patients, and seems<br />
to confirm that microvascular and macrovascular<br />
complications dramatically affect the functional status<br />
in diabetics. However, the analogy in the patterning<br />
<strong>of</strong> BADL-IADL between CHF and diabetes<br />
mellitus demonstrates that differences in the correlates<br />
<strong>of</strong> disability are not automatically associated<br />
with different patterns <strong>of</strong> disability.<br />
The fact that IADL related to outdoor mobility<br />
and selected highly demanding IADL identified<br />
COPD patients with a longer length <strong>of</strong> stay testifies<br />
836 Clinical Investigations<br />
Downloaded From: http://chestioumal.chestpubs.org/ on 07/03/2013
that selected patterns <strong>of</strong> disability might be relevant<br />
to explain the need <strong>of</strong> care. We could not clarify<br />
whether these patterns qualified as generic indexes<br />
<strong>of</strong> COPD severity or identified a distinctive effect <strong>of</strong><br />
COPD on the health status. However, the association<br />
between longer stay and housework-related<br />
IADL in CHF was mainly due to the longer stay <strong>of</strong><br />
female patients and is therefore unlikely to reflect a<br />
distinctive effect <strong>of</strong> CHF on physical independence<br />
<strong>of</strong> selected patients. Instead, it reflects the greater<br />
reliability <strong>of</strong> housework-related IADL as indicators<br />
<strong>of</strong> physical capabilities in CHF female patients.<br />
Limitations <strong>of</strong> this study deserve considerations.<br />
First, we could not assess the relationship between<br />
disease severity and physical limitation because the<br />
database did not include any index <strong>of</strong> COPD and<br />
CHF severity, except for the use <strong>of</strong> long-term oxygen<br />
therapy by COPD patients. Indeed, a recent spirometry,<br />
ie, a spirometry performed in the 6 months<br />
before hospital admission, was available only for a<br />
minority <strong>of</strong> COPD patients, whereas approximately<br />
two thirds <strong>of</strong> them had an old spirometry. Similarly,<br />
a recent echocardiogram was available for a minority<br />
<strong>of</strong> CHF patients. Limiting the analysis to these<br />
subgroups could have introduced a selection bias by<br />
focusing on patients cared for or posing special<br />
diagnostic problems. Thus, we can draw conclusions<br />
about the pattern <strong>of</strong> disability, but not about its<br />
determinants. Second, a variable proportion <strong>of</strong> patients<br />
grouped according to an index main disease<br />
had one or both <strong>of</strong> the other two conditions as<br />
comorbid diseases. The corresponding BADL-IADL<br />
pattern to some extent reflects the effect <strong>of</strong> the other<br />
two chronic diseases. However, CHF, COPD, and<br />
diabetes mellitus are highly prevalent and frequently<br />
coexisting conditions. As a consequence, a pure CHF<br />
or COPD or diabetes mellitus population would be<br />
poorly representative <strong>of</strong> the reality.<br />
In conclusion, this study identifies a relatively<br />
uniform modality <strong>of</strong> BADL-IADL partitioning in<br />
main components within two older populations having<br />
CHF and diabetes mellitus as the main disease.<br />
However, the identification <strong>of</strong> two distinctive main<br />
components <strong>of</strong> BADL-IADL in COPD cautions<br />
health-care pr<strong>of</strong>essionals against oversimplifying the<br />
interpretation <strong>of</strong> the individual performance in terms<br />
<strong>of</strong> the number <strong>of</strong> activities lost or retained. Testing<br />
the prognostic implications <strong>of</strong> individual components<br />
<strong>of</strong> BADL-IADL scale in COPD might help to clarify<br />
both meaning and usefulness <strong>of</strong> present findings.<br />
References<br />
1 Guralnik JM, Simonsick EM. Physical disability in older<br />
Americans. J Gerontol 1993; 48:3–10<br />
2 Thomas VS, Rockwood K, McDowell I. Multidimensionality<br />
in instrumental and basic activities <strong>of</strong> daily living. J Clin<br />
Epidemiol 1998; 51:315–321<br />
3 Murray CJL, Lopez AD. Alternative projections <strong>of</strong> mortality<br />
and disability by cause 1990–2020: Global Burden <strong>of</strong> Disease<br />
Study. Lancet 1997; 349:1498–1504<br />
4 Ho KK, Pinsky JL, Kannel WB, et al. The epidemiology <strong>of</strong><br />
heart failure: the Framingham Study. J Am Coll Cardiol 1993;<br />
22:6A–13A<br />
5 Pelletier-Fleury N, Lanoe JL, Fleury B, et al. The cost <strong>of</strong><br />
treating COPD patients with long-term oxygen therapy in a<br />
French population. Chest 1996; 110:411–416<br />
6 Hobbs FD, Kenkre JE, Roalfe AK, et al. Impact <strong>of</strong> heart<br />
failure and left ventricular systolic dysfunction on quality <strong>of</strong><br />
life: a cross-sectional study comparing common chronic cardiac<br />
and medical disorders and a representative adult population.<br />
Eur Heart J 2002; 23:1867–1876<br />
7 Okubadejo AA, O’Shea L, Jones PW, et al. Home assessment<br />
<strong>of</strong> activities <strong>of</strong> daily living in patients with severe chronic<br />
obstructive pulmonary disease on long-term oxygen therapy.<br />
Eur Respir J 1997; 10:1572–1575<br />
8 Konstam V, Salem D, Pouler H, et al, for the SOLVD<br />
Investigators. Baseline quality <strong>of</strong> life as a predictor <strong>of</strong> mortality<br />
and hospitalization in 5,025 patients with congestive<br />
heart failure. Am J Cardiol 1996; 78:890–895<br />
9 Teno JM, Harrell FE, Knaus W, et al. Prediction <strong>of</strong> survival<br />
for older hospitalized patients: the HELP survival model.<br />
J Am Geriatr Soc 2000; 48:S16–S24<br />
10 Burns RB, McCarthy EP, Moskowitz MA, et al. Outcomes for<br />
older men and women with congestive heart failure. J Am<br />
Geriatr Soc 1997; 45:276–280<br />
11 Ferrer M, Alonso J, Morera J, et al. Chronic obstructive<br />
pulmonary disease stage and health-related quality <strong>of</strong> life:<br />
The Quality <strong>of</strong> Life <strong>of</strong> Chronic Obstructive Pulmonary Disease<br />
Study Group. Ann Intern Med 1997; 127:1072–1079<br />
12 Stassijns G, Lysens R, Decramer M. Peripheral and respiratory<br />
muscles in chronic heart failure. Eur Respir J 1996;<br />
9:2161–2167<br />
13 American Thoracic Society. Dyspnea: mechanisms, assessment,<br />
and management; a consensus statement. Am J Respir<br />
Crit Care Med 1999; 159:321–340<br />
14 Laghi F, Tobin MJ. Disorders <strong>of</strong> respiratory muscles. Am J<br />
Respir Crit Care Med 2003; 168:10–48<br />
15 Harver A, Mahler DA, Schwartzstein RM, et al. Descriptors<br />
<strong>of</strong> breathlessness in healthy individuals: distinct and separable<br />
constructs. Chest 2000; 118:679–690<br />
16 Lebzelter J, Klainman E, Yarmolovsky A, et al. Relationship<br />
between pulmonary function and unsupported arm exercise<br />
in patients with COPD. Monaldi Arch Chest Dis 2001;<br />
56:309–314<br />
17 Antonelli Incalzi R, Gemma A, Marra C, et al. Chronic<br />
obstructive pulmonary disease: an original model <strong>of</strong> cognitive<br />
decline. Am Rev Respir Dis 1993; 148:418–420<br />
18 Antonelli Incalzi R, Trojano L, Acanfora D, et al. Verbal<br />
memory impairment in congestive heart failure. J Clin Exp<br />
Neuropsychol 2003; 25:14–23<br />
19 Trojano L, Antonelli Incalzi R, Acanfora D, et al. Cognitive<br />
impairment: a key feature <strong>of</strong> congestive heart failure in the<br />
elderly. J Neurol 2003; 250:1456–1463<br />
20 Antonelli Incalzi R, Chiappini F, Fuso L, et al. Predicting<br />
cognitive decline in patients with hypoxemic chronic obstructive<br />
pulmonary disease. Respir Med 1998; 92:527–533<br />
21 Bourdel-Marchasson I, Dubroca B, Manciet G, et al. Prevalence<br />
<strong>of</strong> diabetes and effect on quality <strong>of</strong> life in older French<br />
living in the community: the PAQUID Epidemiologic Survey.<br />
J Am Geriatr Soc 1997; 45:295–301<br />
22 Carbonin PU, Pahor M, Bernabei R, et al. Is age an independent<br />
risk factor <strong>of</strong> adverse drug reactions in hospitalized<br />
www.chestjournal.org <strong>CHEST</strong> / 127 /3/MARCH, 2005 837<br />
Downloaded From: http://chestioumal.chestpubs.org/ on 07/03/2013
medical patients? J Am Geriatr Soc 1991; 39:1093–1099<br />
23 Carosella L, Pahor M, Pedone C, et al. Pharmacosurveillance<br />
in hospitalized patients in Italy: study design <strong>of</strong> the Gruppo<br />
Italiano di Farmacovigilanza nell’Anziano’ (GIFA). Pharmacol<br />
Res 1999; 40:287–295<br />
24 PHS-HCF. International classification <strong>of</strong> diseases, ninth revision,<br />
clinical modifications. Washington, DC: Public Health<br />
Service, Health Care Financing Administration, 1980<br />
25 Katz S, Ford AB, Moskowitz RW, et al. Studies <strong>of</strong> illness in<br />
aged: the index <strong>of</strong> ADL, a standardized measure <strong>of</strong> biological<br />
and psychosocial function. JAMA 1963; 185:94–106<br />
26 Heikkinen E, Waters WE, Brzezinski ZJ. The elderly in<br />
eleven countries: a sociodemographic survey. Copenhagen,<br />
Denmark: World Health Organization, 1983<br />
27 Jitapunkul S, Pillay I, Ebrahim S. The abbreviated mental<br />
test: its use and validity. Age Ageing 1991; 20:332–336<br />
28 Lesher EL, Berryhill JS. Validation <strong>of</strong> the Geriatric Depression<br />
Scale-Short Form among inpatients. J Clin Psychol 1994;<br />
50:256–260<br />
29 Charlson ME, Pompei P, Ales KL, et al. A new method <strong>of</strong><br />
classifying prognostic comorbidity in longitudinal studies:<br />
development and validation. J Chronic Dis 1987; 40:373–383<br />
30 Pahor M, Chrischilles EA, Guralnik JM, et al. Drug data<br />
coding and analysis in epidemiologic studies. Eur J Epidemiol<br />
1994; 10:405–411<br />
31 Gruijter DNM, Van der Kamp LJT. Statistical models in<br />
psychological and educational testing. Lisse, the Netherlands:<br />
Swets & Zeitlinger, 1984<br />
32 Zwick W, Velicer W. Comparison <strong>of</strong> five rules for determining<br />
the number <strong>of</strong> components to retain. Psychol Bull 1986;<br />
99:432–442<br />
33 Isohao R, Puolijoki H, Huhti E, et al. Chronic obstructive<br />
pulmonary disease and self-maintaining functions in the<br />
elderly: a population-based study. Scand J Prim Health Care<br />
1995; 13:122–127<br />
34 Cromwell DA, Eagar K, Poulos RG. The performance <strong>of</strong><br />
instrumental activities <strong>of</strong> daily living scale in screening for<br />
cognitive impairment in elderly community residents. J Clin<br />
Epidemiol 2003; 56:131–137<br />
35 Antonelli Incalzi R, Gemma A, Marra C, et al. Verbal memory<br />
impairment in COPD: its mechanisms and clinical relevance.<br />
Chest 1997; 112:1506–1513<br />
36 Antonelli Incalzi R, Bellia V, Catalano F, et al. Evaluation <strong>of</strong><br />
health outcomes in elderly patients with asthma and COPD<br />
by using disease-specific and generic instruments: the Sa.R.A.<br />
study. Chest 2001; 120:734–742<br />
37 Yohannes AM, Roomi J, Connolly MJ. Elderly people at<br />
home disabled by chronic obstructive pulmonary disease. Age<br />
Ageing 1998; 27:523–525<br />
38 Wu JH, Haan MN, Liang J, et al. Diabetes as a predictor <strong>of</strong><br />
change in functional status among Mexican Americans: a<br />
population-based cohort study. Diabetes Care 2003; 26:314–<br />
319<br />
838 Clinical Investigations<br />
Downloaded From: http://chestioumal.chestpubs.org/ on 07/03/2013