вход по аккаунту


The determinants of walking velocity in the elderly.

код для вставкиСкачать
Number 3, March 1995, pp 343-350
0 1995, American College of Rheumatology
An Evaluation Using Regression Trees
Objective. To determine predictors of walking
velocity in the elderly.
Methods. Five hundred thirty-two persons from 3
cohorts of elderly persons (retirement community, ambulatory care medical clinic, or chronically homebound
population) performed a 10-foot (for the homebound
subjects) or SO-foot (for all other subjects) walk time test
and underwent a standardized interview, chart review,
and clinical examination. The 73 independent variables
that were evaluated included demographic, musculoskeletal, neurologic, psychologic, and other comorbidity
items. Least-squares and least-absolute-deviation regression tree analyses were performed to determine the
strongest predictive factors associated with walking
Results, Sampling cohort (homebound versus
non-homebound), quadriceps strength, hip flexion
strength, lumbosacral spine impairment, lower joint
impairment, and education were found to be associated
with walking velocity. Joint pain measures were not
associated with walking velocity.
Supported by Multipurpose Arthritis and Musculoskeletal
Diseases grant AR-30692 from the NIH (NIAMS), by an Arthritis
Health Professional Research grant from the Arthritis Foundation,
and by a grant from the Illinois Chapter of the Arthritis Foundation.
Rowland W. Chang, MD, MPH: Northwestern University
Medical School, and Center for Health Services and Policy Research, Chicago, Illinois; Dorothy Dunlop, PhD: Northwestern
University Center for Health Services and Policy Research; James
Gibbs, PhD: Northwestern University Medical School Center for
Health Services and Policy Research, and the Hines Veterans
Administration Hospital, Hines, Illinois; Susan Hughes, DSW:
Northwestern University Medical School, Center for Health Services and Policy Research, and Hines Veterans Administration
Address reprint requests to Rowland W. Chang, MD,
MPH, Northwestern University Multipurpose Arthritis Center, 303
E. Chicago Avenue, Ward 3-315, Mail Stop W121, Chicago, IL
Submitted for publication December 14, 1993; accepted in
revised form October 11, 1994.
Conclusion. Muscle strength variables are better
predictors of walking velocity than are joint pain variables. Thus, clinical trials and observational studies
using walking velocity as an outcome need to take into
consideration the influence of muscle strength on this
outcome variable.
Walking velocity is a commonly used objective
parameter of functional status in a variety of clinical
research settings, most notably in randomized, controlled trials of short- or long-acting anti-rheumatic
drugs (1). While there is evidence that this parameter
is associated with the severity of symptoms of lower
extremity arthritis, other impairments, such as muscle
weakness, obesity, and depression, might also influence walking velocity (2). Furthermore, it has been
found that demographic variables, particularly age and
education, affect walking velocity in the elderly (2).
Given the limited amount of previous research
into the determinants of walking velocity, this study
was undertaken to explore a wide range of musculoskeletal, pain, psychological, and demographic measures as potential predictors of this outcome variable.
The study utilized regression tree analysis on data
from a cohort of 532 persons who were enrolled in a
longitudinal study of health status and health resource
utilization of the elderly and who underwent a standardized physical examination by a study rheumatologist.
Sample. Seven hundred sixty-one persons aged 60 or
older were enrolled in a prospective longitudinal study
examining health resource utilization and the trajectory of
functional status in 3 cohorts of elderly subjects: a continuing care retirement community (n = 250), ambulatory care
clinic users (n = 261), and a chronically homebound population (n = 250). The study sample for the present investigation consisted of 532 of these persons who were examined
Table 1. Joint regions and joints examined*
Upper body
Table 2. Grading scales used in the peripheral joint examination
Lower body
Toe 1
Lumbosacral spine
Pain on motion
0 = no pain
1 = complaint of pain
2 = complaint of pain with wincing
3 = wincing and attempt to withdraw
0 = no tenderness
1 = complaint of tenderness
2 = complaint of tenderness with wincing
3 = wincing and attempt to withdraw
Limitation of motion
0 = normal range of motion (ROM)
1 = <lo% loss of ROM
2 = 10-19% loss of ROM
3 = 2049% loss of ROM
4 = 50-99% loss of ROM
5 = 100% loss of ROM (ankylosed)
0 = no deformity
1 = deformity of joint
Upper spine
Cervical spine
Thoracic spine
* DIP = distal interphalangeal; PIP = proximal interphalangeal;
MCP = metacarpophalangeal; CMC = carpometacarpal; AC =
acromioclavicular; SC = sternoclavicular; MTP = metatarsophalangeal .
by a rheumatologist: 222 from the retirement community,
247 from the ambulatory care clinic, and 63 from the
homebound population.
Walking velocity. Walking velocity was based on a
test of the time (measured to the nearest 0.1 seconds)
required to walk 10 feet (for the homebound group) or 50 feet
(for all others). Subjects were instructed to walk from a
standing start at their normal pace while using any assistive
device they might normally use to walk the tested distance.
These walk times were converted into walking velocities
expressed in feetkecond. A walking velocity of 0 was
assigned to the 20 subjects who were non-ambulatory.
Standardized musculoskeletal examination. A standardized physical examination was performed by 1 of 4
study rheumatologists. The examination focused on joint,
muscular, and neurologic abnormalities. Seventy peripheral
joints grouped into 16 joint regions (Table 1) were examined
for pain on motion, tenderness, limitation of motion, and
deformity, using the scales summarized by the Glossary
Committee of the American College of Rheumatology (formerly, the American Rheumatism Association) (1) (Table 2).
The spine was divided into upper spine (cervical and thoracic) and lower spine (lumbosacral) regions. The cervical
spine was assessed for pain on motion, bone and muscular
tenderness, and limitation of motion (flexion, extension,
rotation, lateral bending). The thoracic spine was assessed
for bone and muscle tenderness and deformity (kyphosis,
scoliosis). The lumbosacral spine was assessed for tenderness (bone, muscle, and sacroiliac), limitation of motion
(flexion and extension), and deformity (loss of lordosis,
scoliosis). Scaling for the spine findings is shown in Table 3.
Data on pain on motion for individual joints were
used to calculate joint region pain in two ways. First, a
maximum pain score for a joint region was determined,
reflecting the highest pain-on-motion score attributed to any
joint in the region. Thus, if the pain-on-motion score was 3 in
1 or more of the left foot joints, the maximum pain score for
the left foot would be 3. Maximum pain scores for sided
peripheral joints were added. Thus, the maximum pain score
for the feet could range from 0 to 6. The second joint region
Table 3. Grading scales used in the spine examination
Pain on motion of cervical spine
0 = no pain
1 = complaint of pain
2 = complaint of pain with wincing and/or withdrawal
Bone and paraspinal tenderness at cervical, thoracic, and
lumbosacral spine and sacroiliac joints
0 = no tenderness
1 = complaint of tenderness
2 = complaint of tenderness with wincing and/or withdrawal
Limitation of motion of cervical and lumbosacral spine
0 = normal range of motion (ROM)
1 = 4 0 % loss of ROM
2 = 50-100% loss of ROM
Deformity (thoracic kyphosis and scoliosis, loss of lumbar
lordosis, and lumbar scoliosis)
0 = no deformity
1 = deformity of spine
Table 4.
Selected clinical and demographic variables analyzed*
Joint painhmpairment
Neurologic findings
Maximum pain (0-6)
Proportional pain (0-1)
Impairment (0-6)
Maximum pain (0-6)
Proportional pain (0-1)
Impairment (0-6)
Maximum pain ( 0 4 )
Proportional pain (0-1)
Impairment (0-1)
Maximum pain (0-6)
Proportional pain (0-1)
Impairment (0-1)
Maximum pain (0-6)
Proportional pain (0-1)
Impairment (0-6)
Maximum pain (0-6)
Proportional pain (0-1)
Impairment (0-6)
Maximum pain (0-6)
Proportional pain (0-1)
Impairment (0-6)
Maximum pain (0-6)
Proportional pain (0-1)
Impairment (0-6)
Lumbosacral spine
Maximum pain (0-2)
Proportional pain (0-1)
Impairment (0-3)
Cervicothoracic spine
Maximum pain (0-2)
Proportional pain (0-1)
Impairment (0-3)
Lower body joints
Maximum pain (0-6)
Proportional pain (0-1)
Impairment (0-27)
Upper body joints
Maximum pain (0-6)
Proportional pain (0-1)
Impairment (0-33)
Total joints
Maximum pain (0-6)
Proportional pain (0-1)
Impairment (0-60)
Motor strength (0-10)
Shoulder abductors
Hip flexors
Reflexes (0-6)
Babinski’s (0-1)
Rigidity (0-1)
Tremor (0-1 )
Romberg’s (0-1)
Demographic, comorbidity,
and GERI-AIMS variables
Marital status
Sampling group
Number of comorbidities
Overall total
Central nervous system
Peripheral nervous system
Psychological status (GERI-AIMS)
Anxiety score (0-10)
Depression score (0-10)
Anxiety/Depression mean
Self-reported joint pain (GERI-AIMS)
Individual items ( 1 4 )
Pain severity
Pain frequency
Length of A M stiffness
Frequent polyarthralgias
Summary pain score (0-10)
* Ranges of scores are shown in parentheses. GERI-AIMS = Arthritis Impact Measurement Scales
modified for use in an elderly population (ref. 4).
pain score was calculated by adding all the pain-on-motion
scores of the joints in a region (both left and right) and
dividing this by the total possible pain-on-motion score.
Thus, if the only pain on motion detected was localized to
both first metatarsophalangeal joints and graded as 2 , the
proportion of pain score for the feet would be 4/66 or 0.06,
and possible scores could range from 0 to 1 .
An impairment score was calculated for each joint
region based on the total number of abnormal signs associated with physical disruption of the joint (tenderness, de-
Table 5. Demographic data on the study samples
Study sample
Retirement Ambulatory
community care clinic Homebound
(n = 247)
(n = 63)
(n = 222)
Mean -t S D age
% female
% white
% married
% with high school degree
83 t 6.0
70 2 6.8
formity, or limitation of motion) found in 1 or more joints in
the joint region. Thus, the impairment score could range
from 0 to 3 for each joint region. Left and right joint region
scores were summed to obtain an overall joint impairment
score; the impairment score for the feet, for example, could
range from 0 to 6.
Maximum pain, proportion of pain, and impairment
scores were calculated for the upper body and lower body
joints overall, as well as for all joints.
A limited neurologic examination was performed to
investigate neuromuscular function. Motor strength of the
shoulder abductors, hip flexors, and quadriceps muscles was
measured using the standard manual muscle testing scale,
which records strength as 0 (no contraction), 1 (trace contraction), 2 (strength sufficient to move joint but not against
gravity), 3 (strength sufficient to move joint against gravity
but not against resistance of examiner), 4 (strength sufficient
to overcome modest resistance of examiner), or 5 (normal
strength). Left and right scores were added to create summary scores for each of the 3 muscle groups, which could
range from 0 to 10.
Reflexes at 5 sites were elicited and recorded as 0
(none), 1 (reduced), 2 (normal), or 3 (hyperreflexic). Again,
left and right scores were added to create summary scores
for each pair of reflex sites, which could range from 0 to 6.
Babinski’s sign, rigidity, tremor, and Romberg’s sign
were tested and record as 0 (absent) or I (present).
Standardized interview and chart review. Basic demographic data (age, sex, marital status, education, and race)
were obtained in a standard manner by a research assistant.
Nonmusculoskeletal diagnoses were determined by standardized interview and chart review done by study rheumatologists, using diagnostic categories used in the National
Health and Nutrition Survey (3). A comorbidity score was
derived by counting the total number of probable/definite
nonmusculoskeletal diagnoses. The number of comorbid
conditions within selected organ systems (e.g., cardiac,
vascular, central nervous system, peripheral nervous system, hearing and vision, and diabetes) was also computed as
an alternative comorbidity index.
Self-reports. Psychological status and the extent of
overall arthritis pain were measured by administration, by a
research assistant, of the Anxiety, Depression, and Joint
Pain subscales of the GERI-AIMS (Arthritis Impact Measurement Scales modified for use in an elderly population)
(4). The Anxiety score, the Depression score, and an average of the Anxiety and Depression scores were used to
measure psychological status. Each of the 4 arthritis pain
items on the GERI-AIMS Joint Pain scale was evaluated
using the &point Likert scale inherent in the individual
items. The overall GERI-AIMS Joint Pain score (which is
normalized from 0 to 10) was also used in the following
The 73 potential predictor variables that were submitted for analysis are listed in Table 4.
Data analysis. In order to better understand the
relationship between walking velocity and potential predictor variables, a regression tree analysis was done (5). This
nonparametric approach screens a large pool of variables
and indicates the specific values at which a predictor is
informative. Thus, the method is not limited by non-normal
distribution of data. The resulting regression tree diagram
identifies strong predictors of walking velocity, which can
point to interventions that might improve ambulatory function.
Regression tree construction begins by using a
computer-intensive algorithm that recursively searches over
all predictor variables to produce a sequence of optimal
binary splits. The splitting continues until all cases in a
subgroup have the same walking velocity or there are too
few cases to split. Once this very large tree is grown, a
sequence of smaller trees is obtained by systematically
pruning off branches to reduce the overall error. The error of
a least-squares (LS) regression tree is the total squared
difference between each walking velocity value and its
subgroup mean. The error of a least absolute deviation
(LAD) regression tree is the total of the absolute difference
between each walking velocity value and its subgroup median. When there are sufficient data, the sample is divided
into a learning set used to grow the large tree and test set
used for pruning. Alternatively, if the sample is smaller as in
these analyses, the entire sample can be used to grow the
large tree and a computer-intensive technique of crossvalidation is used to prune. The final tree selected from the
pruned trees is the smallest tree with a low degree of error in
walking velocity data (for a complete explanation, see ref.
5). Both LS and LAD trees were grown using the CART
module from SYSTAT software (6).
Patient characteristics. Table 5 summarizes the
demographic data from the study samples. The mean
age was 76.6 years (SD 9.1). Seventy-four percent of
Table 6. Joint impairment by study sample
Study sample
Retirement Ambulatory
community care clinic Homebound
(n = 222)
(n = 247)
(n = 63)
% with hip impairment
% with lower back
% with knee impairment
% with ankle
% with foot impairment
Walking Velocity (feet/second)
Figure 1. Distribution of walking velocities among the total study
the subjects were female, 66% were single, and 87%
were white. In general, the subjects were well educated.
Table 6 shows the frequency of clinically detectable musculoskeletal impairments found by phys-
ical examination of the study subjects. Lower extremity impairments were present in 71% of the subjects.
Figure 1 shows the distribution of walking velocities for the combined samples. The mean walking
velocity was 2.9 feetlsec (SD 1.3).
Regression tree. An LS regression tree was
initially grown to identify variables that are most
informative for predicting walking velocity. This regression tree, shown in Figure 2, used only 6 of the 73
potential predictor variables: sampling cohort, quadriceps strength, hip flexion strength, lumbar spine impairment, lower body joint impairment, and education.
Of note, self-report and physical examination measures assessing pain were not selected in this regression tree.
The LS regression tree in Figure 2 accounted
for 56% of the walking velocity variance in these data
(relative error 0.44). The top split in Figure 2 separates
off the homebound elderly population, which represented the most impaired sampling cohort. Within the
homebound group, the strongest predictor was quadriceps strength. Normal quadriceps strength in both
legs (labeled “Quadriceps Strength = 10” in Figure 2)
separated 32 homebound people (51% of the 63 homebound) with faster walking velocities (mean 1.50 feet/
second) from 31 people (49% of the 63 homebound)
Hip Flexion Strength
Strength = 1 0
x = 2.9ft/s
x = 1.7ft/s
x = 1.5ft/s
Figure 2. Least-squares regression tree for walking velocity. HS = high school.
Table 7. Walking velocity characteristics from least-squares regression tree*
Fastest walking velocities
Mean 3.78 feetkecond (n = 190)
Hip flexor strength = 9 or 10
Lumbar spine impairment = 0 or 1
Education = beyond high school
Mean 2.99 feet/second (n = 42)
Hip flexor strength = 9 or 10
Lumbar spine impairment = 0 or 1
Education = high school or less
Intermediate walking velocities
Mean 2.95 feetkecond (n = 146)
Hip flexor strength = 9 or 10
Lumbar spine impairment = 2 or 3
Mean 2.85 feethecond (n = 40)
Non- homebound
Hip flexor strength = 0-8
Lower body joint impairment = 0-4
Mean 1.65 feetkecond (n = 51)
Hip flexor strength = 0-8
Lower body joint impairment
Slowest walking velocities
Mean 1.50 feetkecond (n = 32)
Quadriceps strength = 10 (normal)
Mean 0.46 feetkecond (n = 31)
Quadriceps strength = 1-9
* Numbers in parentheses are the number of subjects.
with less strength and slower velocities (mean 0.45
feetkecond). Among the non-homebound elderly, the
strongest predictor variable was hip flexion strength.
Normal hip flexion strength in 1 or both hips (labeled
“Hip Flexion Strength = 9 or 10” in Figure 2) separated 378 people (80.6% of the 469 non-homebound
elderly) with faster velocities (mean 3.35 feetkecond)
from 91 people (19.4% of the 469 non-homebound)
with slower velocities (mean 2.15 feethecond). Additional splits using lumbar spine impairment, lower
body joint impairment, and education were informative in further identifying subgroups of people with
similar walking velocities.
Since the decision rules were tailored to the
sample data, it would be expected that applying these
rules to a new data set would have a higher degree of
error. The cross-validation error, a gauge of the predictive ability of the tree for an independent data set,
indicated that this tree would account for 50% of the
walking velocity variance (cross-validation relative
error = 0.50).
Table 7 summarizes the final subgroups of the
regression tree shown in Figure 2. It is evident that the
subjects with the fastest walking velocities were from
the least impaired or disabled sampling cohorts (retirement community or ambulatory care clinic), had minimal lumbar spine impairment, and had satisfactory
hip flexion muscle strength. Greater lumbar spine
impairment or compromised muscle strength resulted
in slower walking velocities. As a confirmatory analysis, an LAD regression tree, which is less sensitive to
extreme outcome values, was grown. The LAD tree
duplicated the top 3 layers of the LS tree with the
exception that lumbar spine impairment was replaced
by a split on self-reported arthritis pain. Again, the
prominent role of muscle strength in relation to walking velocity was demonstrated.
To determine the impact on the regression tree
of the 20 subjects who had walking velocity values of
0 assigned because of inability to complete the walking
test, an LS tree omitting these cases was grown. This
tree was a pruned version of the tree shown in Figure
2. Specifically, the split among the homebound population according to quadriceps strength was omitted.
This analysis suggested that the function of this split
was primarily to identify those 20 extremely impaired
Recognizing the possibility that there were too
many unimpaired people in this sample for any of the
physical examination measures of pain to be selected
as important predictors, another LS regression tree
analysis was done using only those subjects (n = 380)
who had signs of lower body joint impairment. The
resulting regression tree duplicated the top 3 layers of
the tree in Figure 2; differences in the pruning of the
lower layers resulted in some additional lower
branches created by splits on rigidity, psychological
status, and age. The similarity of this tree, based on
subjects with lower body joint impairment, to the tree
in Figure 2 lends further support to the predictive
relationship between muscle strength and walking
It is notable that demographic variables were
not selected in the LS tree in Figure 2. However, the
data in Table 5 suggest that the sampling groups had
different demographic profiles, (e.g., the ambulatory
care clinic group was younger), indicating that sociodemographic features may be masked by group membership. An LS tree was grown to investigate whether
demographic features would emerge if group were
suppressed. This tree used quadriceps strength for the
top split, and lower splits were created using lower
extremity joint impairment and education. This analysis again confirmed the prominent role of muscle
strength in the prediction of walking velocity. The
lower body joint impairment and education variables,
appearing when group was suppressed, acted to distinguish the homebound people, the more impaired,
and the less educated from the rest of the sample.
The findings of this regression tree analysis
indicate that in the elderly, the strongest predictor of
walking velocity, after controlling for sampling group,
is muscle strength. This conclusion also pertained to the
subgroup who exhibited evidence of lower body joint
arthritis, i.e., those with evidence of lower body joint
impairment found by a rheumatologist on a standardized physical examination. In the retirement community and ambulatory care clinic groups, hip flexion
strength was the strongest determinant of walking
velocity, while quadriceps strength was the strongest
determinant in the homebound group. Other determinants in the retirement community and ambulatory
care clinic groups included lumbar spine impairment,
lower body joint impairment, and education level. Of
note was the absence of a substantial effect of lower
extremity peripheral joint pain, as assessed by physical examination or self-report, on walking velocity.
A regression tree is an informative analytic tool
for identifying and screening variables that strongly
predict walking velocity. The final LS regression tree
focused on only 6 variables, which were simultaneously considered among 73 factors. In addition, the
model revealed both the level at which each factor
becomes an important predictor and the interrelationships among predictor variables. For example, the
slowest walking velocity among the homebound elderly was associated with reduced quadriceps strength
(scores 1-9), whereas the slowest walking velocity
among the non-homebound elderly was associated
with reduced hip flexion strength (scores &S) and increased lower body joint impairment (scores 5-27).
Although regression tree methodology is informative, it is computer intensive and thus timeconsuming, which can lead to high computer costs. It
is also data intensive and generally requires more
cases than does classic regression. A further practical
limitation is that the method is new to many clinicians
and requires some reorientation in thought.
As with any statistical method, the results depend on the quality of the data collected. It is possible
for a variable not to be chosen because it was inadequately measured rather than because it is inherently
uninformative. To increase measurement sensitivity,
we utilized a detailed joint examination protocol similar to one used in modern studies of rheumatic disease
drugs, as well as 2 pain scales. Thus, we believe lower
extremity joint pain (either in the aggregate or by
specificjoint regions) was not chosen by the regression
tree analysis because it did not predict walking velocity, and not because it was inadequately measured.
From a theoretical perspective, the findings of
this study are not surprising. Muscles are responsible
for generating the force that results in joint motion and
ambulation. It follows that the weaker the hip or knee
muscles are, the slower one would walk. The mechanism by which joint pain and/or inflammation influence
walking velocity may be via their secondary effects on
muscle function. Indeed, inactivity related to other
factors, such as sedentary lifestyle and/or some other
chronic disease, would presumably have the same
effect on walking velocity. In order to further examine
the effects of other variables, in particular joint impairment and pain, on muscle strength, surrogates of
muscle strength variables were identified in the regression tree analysis. These included total joint impairment, foot impairment, number of neurologic comorbidities, and age, all of which make theoretical sense.
Lumbosacral spine impairment as well as lower
body joint impairment were also important in predicting walking velocity. The mechanical aspects of the
lumbar spine in particular, and lower body joints in
general, are known to be important in maintaining a
normal gait pattern, perhaps explaining their apparent
influence on walking velocity (7). The demonstrated
effect of education might be mediated through general
health, since better educated subjects tend to be
healthier (8,9). While our data support all of the
above-mentioned theories, the cross-sectional nature
of this inquiry limits our ability to equate causality
with association.
Very few studies reported in the literature address the determinants of walking velocity. Aniansson
and colleagues evaluated functional capacity in activities of daily living among 419 70-year-old Swedish
men and women and found a correlation between
quadriceps isokinetic muscle strength, as measured
with an isokinetic dynamometer, and walking velocity
in healthy women (r = 0.40-0.45, P < 0.05), but not in
the total group of women or in men (10). No joint
examination was done, Ihowever, and multiple comparisons were performed, which can potentially limit
the statistical significance of the finding. The authors
postulated that several factors potentially influence
walking velocity, includiing balance, posture, and coordination disorders, joint motion sensation and perception, muscle function, locomotor disorders (such
as arthritis, osteoporosis, and other bone abnormalities, and amputations), reduced aerobic capacity, and
failing vision. Our study is one of the very few that
have empirically assessed several of these variables in
a multivariate manner.
Platto and colleagues reported a significant correlation (r = -0.49, P < 0.005) between foot pain and
walking velocity in 31 patients with rheumatoid arthritis (11). Muscle strength was not tested in that gait
study. Moreover, the findlings in patients with rheumatoid arthritis cannot be generalized to the rest of the
elderly population given the multijoint and inflammatory nature of this disorder.
Our findings have several implications for clinical studies using walking velocity. First, since lower
extremity muscle strength is virtually never measured
in studies of short- or long-acting antirheumatic drugs
in which walk time is an outcome, inferences about the
association between drug effect and walking velocity
must be made with caution. The conclusion that a
particular drug is associated with an increased walk
time assumes that the mean muscle strength is equal in
the groups studied. Perhaps more importantly, in such
drug studies a finding that joint pain relief is not
associated with improved walking velocity can be
explained by the fact that 1 here was not sufficient time
for improvement in muscle function to occur, even
though the joints were less inflamed and painful. These
findings are also consistenl with observations by other
investigators that strengthening of lower extremity
muscles can improve one’s ability to ambulate efficiently (12,13).
We conclude that walking velocity, an observed
measure of lower extremity function, is linked most
closely to muscle function. Intervention or observational studies that attempt to show associations between walking velocity and various pharmacologic,
surgical, or rehabilitation interventions or other organ
impairments must take this into account if this parameter is to be used as an outcome variable.
We would like to thank all study participants, including the residents and staff of the Presbyterian Homes of
Evanston, IL and the clients and staff of the Loyola Nursing
and the Five Hospital Homebound Elderly Programs (both
in Chicago) and the Chicago and Illinois Departments on
1 . American Rheumatism Association Glossary Committee: Dictionary of the Rheumatic Diseases. Vol. 1: Signs and Symptoms. New York, Contact Associates International, 1982
2. Gibbs J, Hughes S, Dunlop D, Edelman P, Singer RH, Chang
RW: Joint impairment and ambulation in the elderly. J Am
Geriatr SOC41:1205-1211, 1993
3. Plan and Operation of the Second National Health and Nutrition
Examination Survey 1976-1980: Vital and Health Statistics,
series 1 , no. 15. DHHS publication no. PHS 81-1317. Washington, DC: National Center for Health Statistics, 1981
4. Hughes SL, Edelman P, Chang RW, Singer RH, Schuette P:
The GERI-AIMS: reliability and validity of the Arthritis Impact
Measurement Scales adapted for elderly respondents. Arthritis
Rheum 34:85&865, 1991
5 . Breiman L, Fnedman J, Olshen RA, Stone CJ: Classification
and Regression Trees. Belmont, CA, Wadsworth, 1984
6. Steinberg D, Colla P: CART: A Supplementary Module for
SYSTAT. Evanston, IL, SYSTAT, Inc., 1992
7. Inman VT, Ralston HJ, Todd F: Human Walking. Baltimore
MD, Williams and Wilkins, 1981
8. Winkleby MA, Jatulis DE, Frank E, Fortman S: Socioeconomic
status and health: how education, income, and occupation
contribute to risk factors for cardiovascular disease. Am J
Public Health 82:816-820, 1992
9. Pincus T, Callahan L: Formal education as a marker for
increased mortality and morbidity in rheumatoid arthritis. J
Chronic Dis 38:973-984, 1985
10. Aniansson A, Rundgren A, Sperling L: Evaluation of functional
capacity in activities of daily living in 70-year-old men and
women. Scand J Rehabil Med 12:145-154, 1980
11. Platto MJ, O’Connell PG, Hicks JE, Gerber LH: The relationship of pain and deformity of the rheumatoid foot to gait and an
index of functional ambulation. J Rheumatol 18:38-43, 1991
12. Fisher NM, Pendergast DR, Gresham GE, Calkins E: Muscle
rehabilitation: its effect on muscular and functional performance
of patients with knee osteoarthntis. Arch Phys Med Rehabil
72:367-374, 1991
13. Fiatarone MA, O’Neill EF, Ryan ND, Clements KM, Solares
GR, Nelson ME, Roberts SB, Kehayias JJ, Lipsitz LA, Evans
WJ: Exercise training and nutritional supplementation for physical frailty in very elderly people. N Engl J Med 330: 1769-1775,
Без категории
Размер файла
709 Кб
velocity, determinants, walking, elderly
Пожаловаться на содержимое документа