The multidimensionality of osteoarthritis-related impairments provides a
good model to study gender differences in musculoskeletal disease. Women have
a higher risk for the development of osteoarthritis and have higher rates of
disability attributable to
osteoarthritis1-5.
Furthermore, the need for arthroplasty is three times higher in women even
when the willingness to undergo the procedure is taken into
account6-9.
Collectively, studies have suggested that women may wait to seek medical
attention until pain and severe functional decline can no longer be
tolerated10,11.
To our knowledge, only one other study has assessed gender differences in
arthroplasty candidates with use of physical performance
measures7. In that
study, female arthroplasty candidates on waiting lists in Canada reported
greater disability and also had worse walking and stair-climbing performance
as compared with male arthroplasty
candidates7. The
lack of a comparison group and the failure to adjust for salient variables of
age, height, and weight made it difficult to discern if differences were truly
related to a gender difference in disease impact. In addition, lower extremity
muscle strength was not measured. Practice patterns in Canada result in long
waiting times for surgery, and therefore the findings of Kennedy et al. might
not reflect disease severity and disability among arthroplasty candidates in
the United
States7,12.
Understanding gender differences in disease impact and
osteoarthritis-related impairment in arthroplasty candidates is necessary in
order to develop effective preoperative practice patterns and to help to guide
patient counseling and surgical decision-making. The primary aim of the
present investigation was to quantify differences in the magnitude of
impairment between men and women with knee osteoarthritis. We hypothesized
that gender differences would be evident in quadriceps strength, activation,
and physical performance but not in self-reported function and that these
differences would also be present in an age, gender, and body mass
index-matched healthy cohort.
Entire Arthroplasty Group
All individuals with knee osteoarthritis who were scheduled for total knee
arthroplasty from April 2001 to April 2006 by three orthopaedic surgeons
(including two of the authors [L.R. and A.B.]) were contacted by mail and
telephone. Classification of end-stage, unilateral knee osteoarthritis by the
treating surgeon was based on the decision to perform total knee arthroplasty
imminently. The exclusion criteria were lower extremity orthopaedic
abnormalities other than unilateral knee osteoarthritis, contralateral knee
pain with a score of >4 of 10 on a verbal rating scale, a planned total
knee arthroplasty on the contralateral knee, cardiovascular disease,
neurologic impairment, and any physical condition limiting daily activity. Two
hundred and twenty-one individuals (ninety-five men and 126 women) were
eligible and were tested at a mean of eleven days before total knee
arthroplasty. The entire sample of arthroplasty candidates had a mean age (and
standard deviation) of 65.7 ± 8.8 years, a mean height of 1.69 ±
0.11 m, a mean weight of 88.4 ± 16.2 kg, and a mean body mass index of
30.7 ± 4.6.
Control Group
Forty-four (20%) of the 221 patients in the arthroplasty cohort were
matched, according to age, gender, and body mass index, to forty-four healthy
individuals (nineteen men and twenty-five women) who were recruited through
community advertisements. The exclusion criteria for healthy individuals were
similar to those for patients with end-stage knee osteoarthritis, with the
additional exclusion criterion of a history of diagnosed knee abnormality. The
control group had a mean age of 61.3 ± 7.7 years, a mean height of 1.72
± 0.11 m, a mean weight of 79.4 ± 16.2 kg, and a mean body mass
index of 26.8 ± 4.4. The individuals in the matched arthroplasty group
had a mean age of 62.3 ± 6.8 years, a mean height of 1.71 ± 0.11
m, a mean weight of 82.3 ± 14.8 kg, and a mean body mass index of 27.9
± 3.9. The patients in the matched total knee arthroplasty subset,
therefore, were an average of 3.4 years younger (p = 0.012) and had an average
body mass index that was 2.3 points lower (p < 0.001) than those in the
entire arthroplasty group.
All subjects provided written informed consent for participation in the
study, which was approved by the University of Delaware Human Subjects Review
Committee.
Self-Assessment Questionnaires
Perceived functional ability was measured with use of the Medical Outcomes
Study Short Form-36 (SF-36) mental and physical component
scores13 and the
Knee Outcome Survey Activities of Daily Living Scale
score14.
The Knee Outcome Survey question "How does pain affect the function
of your knee during daily activities?" was used to quantify the
influence of pain on the patient's function. Scores ranged from 0 (pain that
prevented the patient from all activities) to 5 (pain that had no effect on
daily activities).
Knee Range of Motion
Knee range of motion in flexion and extension was measured actively in the
supine position with a long-axis
goniometer15. A
positive value indicated a position of knee flexion.
Functional Assessment
Functional performance was assessed with the six-minute walk (the distance
walked in six
minutes)16,17
and the average of two trials of the timed up-and-go test (the time required
to rise from a seated position in an armed chair [seat height, 46 cm], to walk
a distance of 3 m, and to return to a seated position in the
chair)18 and the
average of two trials of the stair-climb test (the time required to ascend and
descend twelve 17.9-cm
steps)19. The
instructions were to move as quickly and as safely as possible with or without
the use of an assistive device.
Quadriceps Strength and Activation Assessment
A burst superimposition test was used to quantify quadriceps strength
(maximal volitional isometric contraction) and volitional muscle
activation20. In
brief, participants were seated in an electromechanical dynamometer (Kin-Com
500 H; Chattex, Chattanooga, Tennessee) with the knee flexed and stabilized at
75°. Self-adhesive electrodes were placed over the rectus femoris muscle
belly proximally and the vastus medialis muscle belly distally to deliver a
supramaximal burst of an electrical stimulus (ten-pulse, 1000-µs pulse
width, 100 pulses per second train at 135 V intensity) during a maximal
volitional isometric contraction attempt.
Volitional muscle activation was quantified as a central activation ratio,
calculated by dividing the maximal volitional isometric contraction by the
electrically elicited
force21. A central
activation ratio of 1.0 signifies complete activation. Testing was repeated a
maximum of three times if complete activation was not achieved, with the
highest maximal volitional isometric contraction test being used for analysis.
A five-minute rest period was given between trials.
Statistical Methods
Quadriceps strength was normalized to the individual's body mass index
(normalized maximal volitional isometric contraction, expressed in N/body mass
index). For standardization purposes, the limbs of the subjects in the control
group were designated as involved or uninvolved to correspond with their
matched counterparts in the arthroplasty group.
Gender differences in the entire arthroplasty cohort were analyzed with use
of independent samples t tests for questionnaire scores and knee pain and with
use of analysis of covariance (with age and body mass index as covariates) for
the timed up-and-go test, stair-climb test, and six-minute walk. A 2 × 2
(gender × limb) analysis of variance was used to analyze gender
differences in normalized maximal volitional isometric contraction, central
activation ratio, and knee range of motion.
Differences between the control group and matched arthroplasty group were
analyzed with use of a 2 × 2 (group × gender) analysis of variance
for questionnaire scores and knee pain; a 2 × 2 (group × gender)
analysis of covariance (with age and body mass index as covariates) for the
timed up-and-go test, stair-climb test, and six-minute walk; and a 2 × 2
× 2 (group × gender × limb) analysis of variance for
normalized maximal volitional isometric contraction, central activation ratio,
and knee range of motion.
An alpha level of <0.05 was considered to be significant. Independent
samples t tests and analysis of covariance were performed for post hoc
analysis with use of the Bonferroni method to adjust for multiple comparisons
(p < 0.0125).
Entire Arthroplasty Group
Women in the arthroplasty group had lower SF-36 physical component scores,
SF-36 mental component scores, and Knee Outcome Survey scores in comparison
with men (p < 0.05), indicating greater reported disability. With the
numbers available, Knee Outcome Survey scores for pain were not different
between men and women (p = 0.784). Female arthroplasty candidates had longer
timed up-and-go test and stair-climb test times and shorter six-minute walk
distances (p < 0.001). Differences between male and female arthroplasty
candidates in knee flexion (p = 0.652) and knee extension (p = 0.464) and the
gender-by-limb interaction term (differences between genders according to
limb) (p > 0.100) for range of motion were not significant, with the
numbers available. The gender-by-limb interaction term for normalized maximal
volitional isometric contraction was not significant (p = 0.768); women had
lower quadriceps normalized maximal volitional isometric contraction in
comparison with men (p < 0.001), and the involved leg had lower normalized
maximal volitional isometric contraction in comparison with the uninvolved leg
(p < 0.001). With the numbers available, differences in the central
activation ratio between men and women were not significant (p = 0.326), nor
was the gender-by-limb interaction term (p = 0.895); the central activation
ratio was lower in the involved limb in both men and women (p < 0.001)
(Table I).
Matched Arthroplasty Group Compared with Control Group
Lower SF-36 physical component scores and Knee Outcome Survey scores were
reported in the matched arthroplasty group as compared with the control group.
The matched arthroplasty group had longer timed up-and-go test and stair-climb
test times, shorter six-minute walk distances, lower quadriceps normalized
maximal volitional isometric contraction values, and less range of motion of
the knee in flexion and extension (p < 0.001). The SF-36 mental component
score (p = 0.118) and the central activation ratio (p = 0.365) were not
different between the control and matched arthroplasty groups, with the
numbers available (Fig. 1).
Women in both the control group and the arthroplasty group had significantly
lower quadriceps normalized maximal volitional isometric contraction values
and shorter six-minute walk distances in comparison with men (p < 0.001),
even when accounting for age and body size.
The group-by-gender interaction terms were significant (that is,
differences between groups differed according to gender) for the timed
up-and-go test (p = 0.012), the stair-climb test (p = 0.002), and the involved
central activation ratio (p < 0.001). Post hoc analysis of covariance with
use of age and body mass index as covariates revealed no difference in the
timed up-and-go test time between male and female controls (F = 1.61, p =
0.212); however, women in the arthroplasty group had longer timed up-and-go
test times in comparison with men (F = 8.59, p = 0.006). The mean stair-climb
test time for female controls was 1.79 seconds (22%) longer than that for male
controls (F = 8.02, p = 0.007). The gender difference in the stair-climb test
time was magnified in the arthroplasty group; the mean stair-climb test time
for women in the arthroplasty group was 7.95 seconds (59%) longer than that
for men (F = 11.22, p = 0.002). Central activation ratio values for the
matched limb were significantly higher for female as compared with male
controls (t = 2.65, p = 0.011). The central activation ratios in the involved
limb were significantly lower in female as compared with male arthroplasty
candidates (t = 2.06, p = 0.045); however, the central activation ratios in
the uninvolved limb were not significantly different between men and women in
the arthroplasty group (t = 1.57, p = 0.124)
(Table II).
Differences in normalized strength, stair-climb test times, and six-minute
walk performance were noted between healthy men and women; however, these
gender differences were magnified in arthroplasty candidates and gender
differences also emerged in timed up-and-go performance and voluntary muscle
activation. These data support the concept that gender differences in strength
and function are inherent. On the other hand, the greater degree of difference
among arthroplasty candidates suggests that women are more adversely affected
by osteoarthritis than men are and that women undergo arthroplasty at a more
advanced disease state than men do. The mechanisms underlying these gender
differences remain unclear but could be related to the activity level of the
patient, the duration of symptoms and disability, and the timing of
surgery.
Our data support the concept that women report greater disability at the
time of arthroplasty and provide insight into the degree of functional impact
attributable to osteoarthritis. Similar to the findings of Katz et
al.11, female
arthroplasty candidates in the present study had worse self-reported function.
Women with osteoarthritis in our study cohort also exhibited worse functional
performance on walking tests, supporting the findings of Kennedy et
al.7. The larger
differences in functional performance between genders in the arthroplasty
group as compared with the control group substantiated the disproportionate
impact of osteoarthritis-related impairments on women.
Results from the immobilization model used by Yasuda et
al.22 support our
findings of the gender-specific impact on quadriceps strength and activation.
Those authors reported that women have greater weakness following an
immobilization paradigm and suggested that changes in neural activation play a
greater role than muscle atrophy does in explaining the differences between
genders. Our novel finding of reduced muscle activation among female
arthroplasty candidates supports their hypothesis. Similar findings were not
evident in the control group. Healthy women had greater activation levels than
healthy men did. We further hypothesize that larger activation impairments
compounded by osteoarthritis in women may magnify preexisting strength
differences between genders.
Reduced muscle activation levels were related to functional performance.
Post hoc analyses revealed significant correlations between the central
activation ratio of the involved quadriceps and function as measured with the
timed up-and-go test (r = 0.33, p < 0.001), the stair-climb test (r = 0.22,
p = 0.001), and the six-minute walk (r = 0.21, p = 0.006). These data support
the argument that physical activity influences voluntary muscle
activation23,24.
The relationship between activation and function provides a strong explanation
for the lower activation levels measured in the female candidates and holds
important implications when counseling female knee arthroplasty
candidates.
The timing of total knee arthroplasty may be crucial for optimizing
postoperative outcomes. Preoperative function and quadriceps strength are the
best predictors of postoperative
outcomes19,25,26.
The presence of comorbidities, poor mental status, severe pain, and disability
adversely affect
outcomes26. Common
advice is to delay knee replacement until pain and disability are no longer
tolerable. Delaying surgery may decrease the need for future revision
arthroplasty; however, substantial evidence favors earlier surgical
intervention to optimize postoperative
outcomes26-28.
In light of the recent reports regarding gender issues related to knee
replacement, it may be prudent to target potential female arthroplasty
candidates to inform them of the risks associated with delaying surgical
intervention. In addition, impairment-based rehabilitation should target women
earlier in the disease process to try to mitigate declines in function,
strength, and activation, with the overall aim of improving preoperative
function and postoperative quality of life.
The present study had some limitations. First, contralateral knee
osteoarthritis was not measured radiographically; however, none of the
patients were candidates for bilateral or staged total knee arthroplasty, and
all patients reported the "worst pain" as being <3 of 10 on a
verbal rating scale. Therefore, we hypothesize that the weakness of the
uninvolved limb is likely the consequence of reduced activity levels due to
severe osteoarthritis in the involved knee. Second, the onset and duration of
symptoms in the involved knee were not recorded. All arthroplasty candidates
sought medical attention and were scheduled for total knee arthroplasty by
experienced surgeons. Evidence in the literature suggests that women wait
longer to undergo total knee arthroplasty than men
do10,11,
giving us reason to believe that the studied female arthroplasty cohort had a
more advanced disease state despite the absence of gender differences in terms
of age, radiographic severity, or knee range of motion. Future research should
longitudinally assess gender differences from the onset of symptoms to the
time of total knee arthroplasty.
Diminished muscle activation is a hallmark of osteoarthritis and is by far
the largest impairment after total knee
arthroplasty29.
These data suggest that osteoarthritis has a differential impact on muscle
activation in women. Targeting activation failure with use of electrical
stimulation, biofeedback, and volitional exercise to improve strength is an
important preoperative strategy in both genders but may be more critical in
women. Many strength and functional differences between men and women worsen
with knee osteoarthritis, and even more emerge (for example, quadriceps
voluntary muscle activation). Quadriceps activation is related to functional
status and helps to explain the gender differences in rising from a chair,
stair-climbing, and walking ability in the arthroplasty cohort.
In conclusion, while some impairments are simply inherent differences
between men and women, others reflect gender differences in disease impact,
indicating that osteoarthritis has a larger impact on function in women,
especially at the time of surgery. Preoperative interventions in patients with
osteoarthritis should be impairment-based, regardless of gender; however,
close monitoring of women with osteoarthritis is warranted on the basis of
these data. Knowing the substantial impact of preoperative strength and
function and that declines in strength and function provide a good indication
of disease impact, earlier surgical consideration may enhance outcomes in
women. ?
Note: The authors acknowledge the expertise of the physical
therapists at the University of Delaware Physical Therapy Clinic and the
assistance of William Newcomb, MD.