вход по аккаунту


Work Disability Before and After Total Hip Arthroplasty.

код для вставкиСкачать
Assessment of Effectiveness in Reducing Disability
We evaluated the impact of total hip arthroplasty
(THA) on work disability among working age persons.
Just over one-third of those totally work disabled prior
to THA were working at 1 or at 4 years after surgery. A
significant long-term decrease in work disability after
surgery was demonstrated by estimating disability prevented by THA, but the decrease was less than suggested
by the literature. Preoperative work status and the
number of painful joints were significant predictors of
return to work.
The evaluation of medical practices, procedures, and devices should include efforts to measure
their effects on disability, particularly when chronic
diseases are involved (1-3). Because arthritis is usually neither curable nor fatal, reduction in disability is an
important aspect of treatment. According to data from
From the University of California San Francisco.
Supported by the UCSF Multipurpose Arthritis Center
(NIH grant -AM20864), The American Kheumatism Association
Medical Information System (NCHSR grant HS03802 and NIH
grant AM21393), and the Robert Wood Johnson Foundation. The
views presented herein are not necessarily those of the Johnson
Michael C. Nevitt, CPhil: Social Science Research Component, Multipurpose Arthritis Center, Department of Medicine and
Robert Wood Johnson Clinical Scholars Program, University of
California San Francisco; Wallace V . Epstein, MD: Social Science
Research Component, Multipurpose Arthritis Center, Department
of Medicine and Robert Wood Johnson Clinical Scholars Program,
University of California San Francisco; Mathias Masem, MD:
Department of Orthopedic Surgery. University of California San
Francisco; William R. Murray, MD: Department of Orthopedic
Surgery, University of California San Francisco.
Address reprint requests to Michael C. Nevitt. CPhil. 350
Parnassus Avenue, #407, San Francisco, CA 941 17.
Submitted for publication April 1 1 , 1983: accepted in revised form November 21. 1983.
Arthritis and Rheumatism, Vol. 27, No. 4 (April 1984)
the National Health Interview Survey, arthritis and
musculoskeletal disorders are leading causes of disability in the United States; in 1976, an estimated 1.3
million Americans were unable to engage in their main
activity because of arthritis (4). Disability and global
function outcome assessments are especially relevant
in situations where an intervention is effective in
achieving more narrow clinical objectives (5).
In this paper, we present results from a study of
the impact of total hip arthroplasty (THA) on severe
work disability among working age persons. THA is a
highly successful and established treatment for endstage hip disease, and is counted among the major
recent advances in surgery (6). Numerous published
reports indicate that THA greatly relieves pain and
improves joint function, both short- and long-term, for
a high proportion of recipients (7,8). It has also been
claimed that THA “restores virtually full functional
capacity for the normal activities of middle-aged and
elderly adults” in a high proportion of cases (9).
Although most persons who undergo THA are
not in the labor force due to advanced age, employment is among the normal activities of a significant
number of patients. Nearly 45% of the 80,000 to 90,000
THAs performed yearly in the U.S. involve patients
less than 65 years of age (10). It is unlikely that many
persons undergo THA solely to improve their work
performance or to return to work ( I 1). However, over
two-thirds of respondents in the study described below
claimed that work disability was a factor in their
decision to have a total hip replacement, probably
because work is such a vital dimension of the quality
of life (12,13). Finally, because benefit-cost estimates
for THA are sensitive to assumptions regarding the
work outcome of younger recipients (14,15), it is
41 1
of severe work disability in the sample would have increased
Subjects and study design. The data reported here are
from a retrospective study of a subset of THA patients likely
to have had their work status enhanced by the operation.
Participants were recruited from approximately 1,000 patients who had undergone THA between 1969 and 1977 at a
singlc university hospital. We selected persons who were 60
years old or less when they had their first THA, and who had
a primary diagnosis of degenerative, congenital. or posttraumatic disease. We excluded those who had severc,
disabling articular problems other than in the hips prior to
surgery. We also excluded patienfs with coexisting medical
conditions prior to THA that might contribute to disability,
such as heart disease, cancer, stroke, diabetes, and renal
Two hundred nine persons met all of the selection
criteria and were sent letters requesting their participation in
the study. Of these, 178 (85.2%) completed interviews, 9
(4.3%) were deceased, and 22 (10.5%) were lost to followup.
Data were obtained from two sources. The surgeons prospectively recorded a modified Harris hip rating (16) for each
patient at 3-6 months prior to surgery and at approximately
8-12 months after surgery. as well a s information on othcr
joint involvement, previous surgery, and operative details.
The second source of data was a long-term followup patient
questionnaire that contained items related to respondents’
work history, onset of disabilities, and nature of their
employment and work environment before and since surgery, as well as questions pertaining to their health, functional status, and joint problems at the time of the study.
Numerous studies attest to the validity of long-term recall
data when it is anchored in such salient life events as major
surgery or the onset of severe disability (17).
Using data from the questionnaire. we further narrowed the study population by selecting patients with a
history of consistent attachment to the work force, or whose
employment history suggested that they would be working in
the absence of their hip dysfunction. Patients were classified
as disabled if they considered themselves unable to work for
reasons related to their health, in this case primarily joint
disease. Our analyses focus on the effect of surgery on work
status since total cessation of employment is one of the most
severe and costly of the disability impacts of joint disease,
while acknowledging that other dimensions of functional
outcome also contribute significantly to the joint implant
recipient’s overall quality of life.
We first tested for differences in the proportions of
palients who were working or disabled before surgery compared with I and 4 years after THA ( P statistically significant
at 0.05). The fact that respondents underwent THA during
different years eliminates the possibility of aggregate
changes in work activity pre- and post-surgery being due
primarily to fluctuations in the unemploymcnt rate. However, a before and after comparison assumes that the proportion of patients working or disabled would remain unchanged
from the observed preoperative level if these patients were
unablc to undergo THA. It is more likely that the frequency
beyond what was reported for the period just prior to
surgery, if THA had not occurred or if a different surgical
procedure were used (18).
Therefore, we projected the amount of disability that
would have occurred without surgery, using 3 different
approaches that involve different assumptions about the
continuation of preoperative disability trends if these patients had not had THA. In the first approach we assumed
that if THA had not occurred, the rate of dropout from the
work force in the preoperative period for different functional
status and sex groupings of respondents would have continued, and thal none of those already disabled would have
returned to work. In the second approach we estimated a
logistic regression equation to explain each patient’s risk of
disability prior to THA as a function of his or her medical,
surgical, and social characteristics. This equation was then
used to predict the number of disablcd persons in the sample
if THA had not occurred. In the third method, we asked
respondents to assess, retrospectively, how likely they were
to have been unable to continue working had they not had
surgery, assuming no worsening of their hip problems. Using
these projections we computed the expected percent of
patients working in the absence of THA, for comparison
With the percent actually working after surgery.
Finally, we also used multiple logistic regression
techniques to identify patient characteristics that affected
the actual work outcome and to assess the effect of surgical
result on work outcome.
Statistical methods.* In the comparison of work status before and after THA, each patient was his or her own
(matched) control. McNemar’s chi-square was used to analyze thc effect of surgery on the matched sample (19). The
Work status of subgroups of respondents was compared
using 5 stratifying variables. McNemar’s chi-square was
used to test for the effect of surgery on work (presurgery
compared with I and 4 years postsurgery in each matched
In the first alternative approach, disability projections were made as follows. To estimate the number of
respondents, in addition to those already totally disabled,
who would have become disabled after I and 4 more years
without surgery, monthly dropout rates specific to sex and
functional status subgroups were applied to the number of
persons remaining at work in each group just prior LO
surgery. Functional status groupings based on preoperative
hip ratings were further stratified by sex, since sex is often
related to the strength of labor force attachment and being a
primary wage earner (20).
The 2-way stratification provided
a crude adjustment for differences in disability risk among
individuals remaining at work. Survival analysis techniques
were used to compute disability hazard (dropout) rate (21).
In the second alternative approach, disability projections were made using logistic regression techniques to
calculate the probability of any individual without THA
being disabled at any point in time, as a function of multiple
respondent characteristics. Logistic regression is commonly
used to estimate probabilities for dichotomous dependent
-* Additional detail on statistical methods is available from
the author (MCN)upon request.
variables such as work status (22,231. l:irst, we estimated
logistic Coefficients for variables potentially related to the
risk for preoperative disability (age. sex, education, being
the primary household wage earner. severity of hip dysfunction, duration of hip disease and of disability, previous
surgery, bilateral hip and multiple joint involvement) (20.24)
by regressing work status at 1 month presurgery on these
factors. Next, values for the age variable and for the
duration and severity of hip dysfunction variables were
increased to reflect the passage of time. and then used to
estimate a new probability of disability for each individual
respondent. We estimated the deterioration in hip function
that would have occurred over time without surgery. based
on the relationship observed for each patient between the
level of hip dysfunction recorded just before THA and the
number of years since the onset of dysfunction.
We then computed the ratio of the actual percent
working postsurgery to the expected percents working without surgery. This ratio is related to the relative risk of
disability. or the probability of disability with surgery divided by the expected probability without surgery. We used this
iation, rather than McNemar’s chi-square.
on methods were used to predict a level of
disability in the sample, rather than to assign an expected
work status to specific individuals. Confidence intervals
were derived using the odds ratio to approximate the upper
and lower confidence limits for the relative risk (19).
h g i s t i c regression was also used to analyze factors
affecting the actual work outcome. First, we determined
patient background and baseline characteristics associated
with the probability of working after THA. A stepwise
elimination procedure was performed by regressing these
background variables on actual work status. The purpose
was to develop the strongest possible explanatory models
using only selected background characteristics. by eliminating variables from the overall model that failed to add
significantly to the explanatory power in this sample (i.e.,
P > 0.1).
Using 3 groups of variables, we then tested for any
additional effect on the probability of working. The first
group consisted of variables representing subdimensions of
the hip rating obtained before surgery. The second group
consisted of the same variables based on the hip rating at
approximately 8-12 months after THA. The third group of
variables measured aspects of the procedure and its sequelae, and included different variables for the analysis of 1- and
4-year work outcomes. For each additional group of explanatory variables, we tested the null hypothesis (using a
likelihood ratio test) that all of the variable coefficients were
simultaneously 0, with a test size of 0.10, in the context of a
model that already included background characteristics.
Equations were estimated for work status at 1 year and 4
years after surgery.
Independent variables included sociodemographic
characteristics thought to account for much labor market
behavior (20). and a variable measuring the extent to which
the preoperative job required such physical activities as
walking, standing, stooping, lifting, and stair climbing, a
better measure of a job’s physical requirements than is
obtained from broad occupational groupings.
Preoperative medical and surgical background char-
acteristics included diagnosis, failed prior reconstructive
surgery. bilateral hip pain. pain in joints other than the hips,
duration of disease. and duration of limitations. Also included was a dummy variable for whether the first THA was
performed between 1975 and 1977, to control for large
increases in the unemployment rate during those years.
advances in surgical technique, and a changing patient
Separate items from the hip rating instrument were
combined to form composite measures of pre- and postoperative hip dysfunction. Included was a variable combining 3
dimensions of hip pain (pain at rest, pain with motion/
weight-bearing, and pain at night: reliability coefficient [standardized item alpha] = 0.84). a variable for type and
frequency of use of pain medications, the Harris range of hip
motion index (flexion, abduction, adduction, external rotation; reliability coefficient - 0.83). and a variable combining
4 dimensions of ambulatory dysfunction (limp, use of supports, walking distance. and stair climbing: reliability coeficient = 0.71).
Surgical and postoperative sequelae variables included a variable for whether a Mueller prosthesis was used. I:or
the I-year work outcome there were also variables for ’I‘IIA
of the other hip within 18 months of the first and for
biomechanical, wound, and systemic complications during
the perioperative period. For work outcome at 4 years postsurgery, we included variables for THA of the other hip
within 4 years of the first, and for revision due to Pl’l
J ure or
infection. There was also a variable for whether the respondent reported health problems in addition to hip disease in
the study year.
We report the individual regression coefficients for
the final models, the model chi-square (a measure of the
combined effects of the independent variables on work
status), the D-statistic (a measure of the goodness offit of the
model analogous to the R’ in multiple linear regression). and
the percent of cases for which the final models predicted
work status correctly (25).
Table I displays medical and demographic characteristics of patients who completed interviews and
of those not interviewed due to death or loss to
followup. Compared with THA recipients nationally,
this sample includes a relatively high percentage with
degenerative or anatomic abnormalities subsequent to
congenital defects and trauma, or with failed prior
reconstructive surgery (15). This probably results from
regional referral patterns and the comparatively young
age of the sample.
There is evidence that the early Mueller prostheses, frequently used in this sample prior to 1974,
are associated with elevated rates of loosening and
failure (26). Nearly all procedures were performed in
the lateral decubitus position via a postcrolateral approach. Trochanteric osteotomy was performed on
38% of interviewed patients, in most instances to
facilitate exposure in cases of previous reconstructive
surgery (27). Fifty-four interviewed patients (30.3%)
had reported bilateral hip pain at preoperative evaluation, and 40 (22.5%) underwent replacement of the
other hip within 18 months of the first THA. It should
also be noted that many patients had reported some
pain in joints other than the hips both prior to and 8-12
months after THA.
Work histories for all patients who completed
interviews are summarized in Table 2. Ninety-four
respondents (52.8%) were employed at I month prior
to surgery. Thirteen of those who claimed employment
also said that they were not able to work at 1 month
prior to THA due to their hip problems. Forty-five of
the 84 respondents not employed at 1 month presurgery had stopped working because of problems with
thcir hip(s), bringing the total number of work-disabled
persons to 58 (32.6%).
Postoperative work status was an appropriate
outcome criterion for the 139 persons (78.1%) who
were either employed at the time of THA or who had
previously worked regularly but were totally disabled
when they underwent THA. Questionnaire data revealed a strong attachment to work in the work status
group. For instance, only 9% worked less than 30
hours per week in their last presurgery job, and 80%
worked 40 hours or more. Though nearly one-half said
that other members of their household also had jobs
prior to their surgery, over two-thirds still considered
themselves the primary wage earner in the household.
Ninety percent had been employed for 10 years or
more presurgery, and 50% for 25 years or more. Fortyseven percent of those in the work status group were
Work outcome assessments. Table 3 describes
the range of work disability impacts experienced
among the 139 persons in the work status group. While
only 19% were totally disabled at 12 months presurgery, this figure more than doubled to 42% by 1 month
before surgery. In addition, a high proportion of
persons who were still working at 1 month before THA
experienced limitations in their work activities, thus
placing them at risk for total work disability.
The ability of THA to reduce total work disability results from a combination of successful return to
work among the disabled and the enhanced capacity of
those working to stay on the job. The work status of
respondents before and after surgery is illustrated in
Figure 1. Among the 58 persons unable to work at 1
month presurgery, 20 (34.5%) attempted to return to
work within 1 year. Not all of these attempts were
Table 1. Medical and demographic characteristics of total hip
arthroplasty (THA) study patients
Primary diagnosis
Congenital dislocation
Sequelae of trauma
Post-traumatic arthritis
Old fracture
Prosthesis for primary THA
Hip surgery prior to THA
Previous hip surgery
Cup arthroplasty
Internal fixation
Correctional osteotomy
Hanging hip
Trochanteric osteotomy
Presurgery bilateral
hip pain
Bilateral THA
Within 1.5 years
of 1st THA
At any time
after 1st THA
Other joint pain
Any joint
Age at THA
5 1-55
% male
% female
Presurgery marital status
Never married
Mean age at THA
Mean age at study year
Mean disease duration
Not interviewed
(n = 31)
16. I
I .7
58. I
53. I
successful in the long run, however, since half the
disabled persons (10 persons) returning to work were
again disabled by year 4 postsurgery. Another 12 of
Table 2. Preoperative work history, total hip arthroplasty (THA)
Work history
NO. (5%)
1. Employed prior to THA
l a . Employed regularly at
any time prior to THA
Ib. Employed regularly during
10 years prior to THA
Ic. Employed at 1 month presurgery
2. Employed at 1 month presurgery
2a. Actually working
2b. Employed but not working
due to hip(s)
3. Not employed at I month presurgery
3a. Not working due to hip(s)
3b. Never worked regularly
3c. Worked regularly but stopped
for non-health reasons
155 (87.1)
145 (81.5)
94 (52.8)
81 (45.5)
13 ( 7.3)
45 (25.3)
23 (12.9)
16 ( 9.0)
Appropriate outcome criteria: I-work status (2a + 2b + 3a), n =
139 (78.1%); 11-household activities (3b + 3c), n = 39 (21.9%).
those disabled presurgery attempted to return to work
between 1 and 4 years after their first THA. Altogether, 32 (55.2%) of those disabled before surgery attempted to return to work within 4 years of THA.
Among the 81 respondents still working presurgery, 6
did not successfully return to work and considered
themselvcs disabled by I year postsurgery. Another 10
had become disabled by 4 years postsurgery.
Several observations should be made here.
First, successful return to work by persons previously
disabled was partially offset by the onset of total
disability among persons who were working. There
was a + 10.0 percentage point change from the presurgery situation in the proportion working at year 1, and
a +4.3 percentage point change at year 4 postsurgery.
Only the change at 1 year postsurgery achieved statistical significance.
Table 3.
Impact of hip disease on respondents’ work prior to
n (%)
1. Entire work status group (n = 139)
38 (27.3)
Change of job or occupation
23 (16.5)
Reduced hours of work
23 (16.5)
Missed work more than I d a y h o n t h
104 (74.8)
Limited in physical activities of work
27 (19.4)
Unable to work at 1 year presurgery
58 (41.7)
Unable to work at I month presurgery
Receiving disability benefits at 1
month presurgery
34 (24.5)
2. Persons still working at 1 month presurgcry (n = 81)
21 (25.9)
a. Change of job or occupation
6 ( 7.4)
b. Reduced hours
10 (12.3)
c. Missed work 2 1 daylmonth
55 (67.9)
d. Limited in physical activities of work
36 (44.4)
e . Limited in kind or amount of work
Figure 1. Work status before and after total hip arthroplasty
(THA). The P values are based on McNemar’s chi-square test for
difference in proportions presurgery versus postsurgery.
Second, having made the break with work
substantially decreased the probability of working
after surgery. Only 35 and 38% of those disabled
before surgery, compared with 93 and 80% of those
working presurgery, were working at 1 and 4 years
after THA, respectively (Table 4). In addition, the
duration of disability prior to surgery affected work
status 1 year after surgery (Table 4).
Third, the work status of respondents was
unstable at 1 year after THA, suggesting that this
interval is inadequate to fully evaluate the impact of
the procedure, and that short-term work outcome may
not be projected confidently into future years.
Finally, among the 107 persons who worked at
all after surgcry, 35% experienced limitations in the
physical activities of their postsurgery work, and 41%
were limited in the kind or amount of work they could
do, compared with 68 and 44%, respectively, among
persons working at I month presurgery.
Table 5 compares the preoperative work status,
and the relationship of surgery to work status, for
subgroups of patients defined by medical characteristics and by the year in which they first underwent
THA. Only the strata defined by the presence or
absence of pain in joints other than the hips showed
significantly different levels of preoperative work activity. The effect of surgery on work status aftcr I year
was significant for those with unilateral THA, with no
prior hip surgery, and for those with no other joints
Table 6 presents estimates of the impact of
surgery on work status based on the 3 alternative
methods of projecting an increase in the number of
Table 4.
Work outcome by preoperative work status
Preoperative work
Working at I month
presurgery (n = 81)
Disabled at 1 month
presurgery (n = 58)
Disabled < 6 months
presurgery (n = 22)
Disabled 2 6 months
presurgery (n = 36)
Working at 1
n (%)*
Working at 4
postsurgery ,
n (%)t
75 (92.6)
65 (80.2)
20 (34.5)
22 (37.9)
13 (59.1)
10 (45.4)
7 (19.4)
12 (33.3)
* P < 0.001. working at 1 month presurgery versus disabled at I
month presurgery; p < 0.01, disabled < 6 months presurgery versus
disabled 2 6 months presurgery.
i P < 0.001. working at 1 month presurgery versus disabled at 1
month presurgery; P > 0.5, disabled < 6 months presurgery versus
disabled 2 6 months presurgery.
totally disabled persons had respondents not had
THA. Row 1 shows the results of applying sex- and
function-specific preoperative dropout rates to those
remaining at work. After 1 year without THA, a
projected 16 respondcnts (in addition to the 58 persons
already disabled before THA) would have become
disabled, and after 4 years, 46 additional disability
cases were projected. This corresponds to estimated
increases of 21.6 and 37.4 in the percent actually
working at 1 and 4 ycars after surgery, respectively,
compared with the expected percent working without
Row 2 in Table 6 shows the results when hip
dysfunction and time-related disability risk factors
were increased in a logistic regression model. By this
approach, 9 more persons were projected to have
become disabled after 1 year, and 21 persons after 4
more years without surgery, corresponding to estimated increases of 16.6 and 19.4 in the percent working
after surgery compared with the cxpected percent.
Based on each patient’s retrospective assessment of
his Or her work prognosis (row in the
would have been 16 additional disability cases after 1
year and 31 after 4 ycars, corresponding to 21.6 and
26*6percentage point
Figure 2 converts the results from Table 6 into
the ratios of the percent of respondents actually working after surgery to the expected percent working
without THA, and the 95 percent confidence intervals
for each ratio. The figure also includcs the ratios of the
percents actually working after surgery to the percent
working before THA (approach I). The figure indicates
that estimates of the long-term effect of surgery on
work status derived from the alternative methods of
projecting disability without THA (approaches 11. 111,
and IV) are statistically significant.
Table 7 re-expresses the high (II), middle (111),
and low (I) estimates of the effect of surgery on total
Table 5. Percent working before and after total hip arthroplasty (THA) by medical characteristics
and year of surgery
Change in 76 working
Characteristic (n)
7% working I
month presurgery
I year postsurgery
Osteoarthritis (70)
Sequelae of trauma (30)
Congenital dislocation (32)
Osteonecrosis (7)
Unilateral T H A (93)
Bilateral THA (46)
+ 14.3
+ 10.9
Prior hip surgery (57)
No prior hip surgery (82)
+ 12.2*
Other joint pain (56)
No other joint pain (83)
Year of T H A
1969-1971 (42)
1972-1973 (33)
1974-1975 (30)
1976-1977 (34)
+ 16.7
- 14.3
t 9.6
+ 15.2
* Change in
4 years post-
percent working significant (P< 0.05) by Mch’emar’s chi-square.
Significant difference (P< 0.05) in percents working presurgery.
Table 6. Estimated effect of surgery on percent working, based on projections of increased total
disability and expected percent working if patients did not have total hip arthroplasty
I year postsurgery
Basis for projection
increase in
Actual % expected %
increase in
Actual % expected 96
1. Function/sex-specific
+ 16.6
+ 19.4
presurgery dropout
2. Multivariate logistic
prediction of disability
3. Patient’s own
assessment of
probable work status
work disability as the number of work-years gained
among the 139 respondents. These figures can be
multiplied by average respondent earnings to provide
one type of measure of the economic benefit of THA
(14). Estimates of work-years gained per person after 4
years range from 0.3 to 1.0.
Since we are frequently interested in the longterm benefits of chronic disease treatments, we also
%Working post op
4 years postsurgery
Expected % working
without THA
1 Year
post - op
4 Years
post - op
Figure 2. Ratio of percent working postoperatively to expected
percent working without total hip arthroplasty (THA), and 95%
confidence intervals. Expected percents are based on: 1, percent
working presurgery; 11, presurgery dropout rates: Ill, multiple
logistic regression; IV, patient’s own assessment. Ratios (1 year
postsurgery): 1, 1.17; 11, 1.46; 111, 1.09; I V , 1.19; (4 years postsurgery): I , 1.07; 11, 2.49; 111, 1.45; I V , 1.74.
estimated the average work-years that would be
gained through 8 years of followup. Not all patients
were followed for 8 years, so we assumed that 75% of
those working at 4 years after surgery continued to
work for 4 more years, while 25% retired due to age or
disability during this interval. (This was the rate actually observed among the 76 respondents followed for 8
years.) Expected levels of work activity after 8 years
without THA were determined using the previously
described methods for projecting additional disability
cases. Estimates of between 0.4 and 2.5 work-years
gained per person after 8 years of followup in this
sample were obtained (Table 7).
The effect of surgical result on postoperative
work status. First we analyzed the effect of selected
patient background characteristics on the probability
of actually working after surgery. At both 1 and 4
years after THA, females were significantly less likely
than males to be working, and those with more education significantly more likely to be working. Having
reported some pain in joints besides the hips prior to
surgery significantly reduced the chances of working
at both followup points, while a failed prior procedure
lessened the probability of working at 1 year, and
bilateral hip pain lessened the probability at 4 years
after THA. (Diagnosis dummy variables as a group did
not significantly affect the probability of working.)
We then tested for the further contribution of 3
additional groups of variables to the probability of
working. Preoperative hip rating variables as a group
significantly affected the probability of working at both
1 and 4 years after THA ( P < 0.05). Among individual
variables, frequent use of pain medications and severe
ambulatory dysfunction decreased the probability of
working after THA. In contrast, there was no evidence
that postoperative hip rating variables, as a group or
Table 7.
Estimated work-vears gained 4 and 8 vears after total hit, arthrodastv (THA)
Work-years gained after:
Hasis of expected o/c working
without THA
4 years
8 years
Per person
I. Presurgery work status
11. Presurgery dropout rate
111. Multivariate logistic prediction
of disability
1. 0
Work-years remaining in
the absence of any disability
individually, affected the probability of working after
surgery when background and presurgery hip rating
variables were included in the model ( P > 0. I), nor did
postsurgery hip rating variables significantly affect the
probability of working when they were the only group
in the equation, or in any combination with other
groups of variables, including preoperative work status. Surgical and postoperative scquclae variables as a
group did not add significantly to the predicted probability of working at 1 year followup ( P > 0. I ) , but did
at 4 years after THA ( P = 0.08). Persons reporting
other health problems in the study year were significantly less likely to have been working after 4 years.
Contingency table analyses indicated that postoperative hip rating variables were more strongly
related to work status in certain subgroups of this
sample (data not shown). For example, those with
good ambulatory function both before and after surgery were the most likely to be working after surgery,
and those with poor ambulatory function before and
after surgery were the least likely to be working. Also,
postsurgery hip rating variables were more strongly
relatcd to work outcome among respondents who had
undergone THA in later years of the series. We
included interaction terms representing these effects in
our multiple logistic regression modcls, but still found
that postoperative hip rating variables plus the interaction tcrms did not significantly affcct the probability of
working after surgery.
Finally, we added a variable for work status at 1
month before surgery to obtain the final models reported in Table 8. This variable had a large and highly
significant effect on work outcome at both 1 and 4
years postsurgery, with persons working just prior to
THA more likely to be working afterward. Equally
important, with this variable in the model the coefficients for sex, education, hip involvement, pain medications, and ambulation dysfunction decreased in size
and were no longer significant ( P > 0.05). This sug-
Per person
gests that these factors affectcd work outcome, in part,
through their association with work status prior to
THA. Both of the final models showed that the combined effects of the independent variables on work
status were highly significant (model chi-square), and
that the models contributed substantially to the explanation of outcome variance (D-statistic, accuracy of
predicted work status).
We used the regression coefficients from the
final models to compute the probability of working for
representative patients with different combinations of
selected preoperative characteristics (Table 9). For
example, a patient with good hip function prior to
surgery (defined in Table 9) and with only 1 hip
involved, no prior hip surgery, and no other painful
joints had a 0.90 probability of working 4 years after
surgery provided he or she was still working presurgery, and a 0.85 chance if he or she was disabled
presurgery. This compares with a 0.35 probability of
working 4 years after THA for the patient who was
working prior to surgery but had poor hip function,
bilateral hip pain, prior surgery, and 2 other joints
involved, and a 0.11 chance for a similar individual
who was disabled presurgery.
The data reported here indicate that total hip
arthroplasty was moderately effective in reducing the
burden of total work disability in a group of persons
likely to so benefit. Over one-third of respondents for
whom work status was an appropriate outcome, and
who were not able to work prior to THA, were
working at 1 or at 4 years aftcr surgery. About half of
disabled patients attempted a return to work after
surgery, including one-third of those disabled for 6
months or longer. However, within 4 years of THA,
one-third of all disabled persons’ return to work attempts ended in total disability. Respondents working
before surgery also continued to drop out of work after
surgery in significant numbers.
As a result, a finding of a significant long-term
increase in work activity depends on our ability to
estimate disability prevented by the procedure among
persons at work. Since these findings pertain to a
select subset of patients most likely to have their work
status enhanced by the operation, our estimates of
reduced total disability should bc considered an upper
bound. Our results may also have been influenced by
the 15% of patients deceased or lost to followup, for
whom work outcome data were not available. However, the preoperative medical and social characteristics
of these patients were not indicative of a radically
different work outcome in this group compared with
interviewed patients.
Table 8. Logistic regression analysis of factors affecting the probability of working 1 and 4 years
after total hip arthroplasty (THA)
Logistic coefficients*
Age at THA
Years of education
Physical demands of work
Other household member employed
a. Osteoarthritis
b. Trauma
Failed prior surgeries
Hilateral hip involvement
Pain in other joints
Duration of disease
Duration of limitations
Operated 1975-1977
Presurgery hip rating
Pain medications
Range of motion
Ambulation dysfunction
Postsurgery hip rating
I year postsurgery
4 years postsurgery
Pain medications
Range of motion
Ambulation dysfunction
Surgical and sequelae
Mueller prosthesis
Bilateral THA
Biomechanical complications
Wound complications
Systemic complications
Revision of THA
Other health problems
Working 1 month presurgery
Model chi-square
Goodness of fit
Prediction accuracy
D = 0.36
D 0.31
not in model after backward elimination (P > 0.1). N E - group of variables has no effect
(P > 0.1) on work status.
t 1' < 0.05.
.$ P < 0.01.
0 P < 0.001.
Table 9. Predicted probability of working after surgery for representative patients with different
preoperative characteristics, based on logistic regression coefficients
Probability of working
1 year postsurgery
Presurgery characteristics*
Good hip functiont
With unilateral hip involvement,
no prior hip surgery, and
no other joints involved
With bilateral hip involvement,
prior hip surgery, and
2 other joints involved
Poor hip function$
With unilateral hip involvement,
no prior hip surgery, and
no other joints involved
With bilateral hip involvement,
prior hip surgery, and
2 other joints involved
4 years postsurgery
* Variables whose values are not specified are held constant at representative levels.
t Intermittent pain requiring aspirin. full range of motion, slight limp. cane part-timc, can walk
mile, needs handrail on stairs.
$ Constant pain requiring narcotics, Harris range of motion
walk < I block, stairs one at a time.
Several benefit-cost evaluations of THA have
been based largely on the assumption that the “typical
person having a total hip replacement in his or her
middle or late fifties would be able to work for five
more years than otherwise” (14,15). This figure is
double the most liberal estimate of per person workyears gained after 8 years in the present study. Our
data suggest that the benefits accruing from reductions
in severe work disability among younger recipients
may not be as great as previously assumed. Further
economic evaluations of THA will need to quantify
benefits derived from reduction of partial work disability and of limitations -in work activities, as well as
ascertain any costs of prosthesis failure attributable to
The multiple logistic regression analysis of factors affecting work outcome produced evidence of a
powerful underlying relationship between work status
before THA and work outcome. This is consistent
with other studies which have found that persons not
working for more than a few months due to poor health
were unlikely to return to work (24,28). The tendency
for multiple joint complaints to limit the work gains
due to THA is also an important finding, since a large
number of THA recipients with arthritis can be expected to already have, or to develop, generalized or
polyarticular disease manifestations. We caution that
when many variables are tested in an exploratory
40. severe limp, 2 supports full-time, can
fashion in multivariate analyses, the chance of incorrectly stating significance is increased. Our results
need to be cross-validated on additional groups of
Our inability to detect a significant relationship
between variables measuring the surgical result and
the probability of working after surgery may be explained by factors such as multiple joint problems and
labor market liabilities that limit many THA patients’
vocational rehabilitation potential despite hip function
improvement. Although the data are not reported
here, nearly all patients showed improvement according to the hip rating instrument. It is possible that the
instrument was insensitive to the sort of variation in
surgical result that affected work status. Another
plausible interpretation is that successful THA has a
selective impact on work status that is conditioned by
social, labor market, psychological, or medical factors
not directly altered by surgery. Other studies suggest
that this may be the case for chronic disease rehabilitation efforts in general (28-3 1).
It is possible that return to work rates could be
somewhat greater in groups of patients who have
undergone THA more recently than those in our
sample. Many patients from early years of this series
were individuals who had been seriously impaired for
several years. Similar patients would now be operated
on at an earlier stage of disease, perhaps before severe
work disability has developed, and may now be less
likely to undergo other reconstructive procedures prior to THA. On the other hand, current poor economic
conditions increase the risk of total work disability for
a given level of impairment (32). Moreover, a “loosening” of the indications for THA over time would also
result in more patients having disabling coexisting
medical conditions.
In any event, if current cohorts of THA patients
are less severely disabled before surgery than those
constituting our sample, this further emphasizes the
importance of surgical intervention in preventing work
loss. THA offers most patients with severe degenerative or anatomic hip disorders the surest means of
functional improvement. Yet, we found that the work
prognosis was better for patients who remained on the
job until surgery, independent of the short-term functional benefits of the procedure. Maintaining the lower
extremity joint replacement patient as a productive
member of the workforce requires rehabilitation strategies to address the many factors that can precipitate
the initial break with work. At the same time, additional effort should be devoted to determining whether
certain types of work elevate the risk for prosthesis
component loosening and failure (33).
1. Rogers D: Medical care and health status: do they relate
to one another? President’s statement, Robert Wood
Johnson Foundation Annual Report 1981. Princeton,
New Jersey, 1982, pp 10-19
2. McDermott W: Absence of indicators of the influence of
its physicians on a society’s health. Am J Med 70:833843, 1981
3. Wood P, Badley E: An epidemiological appraisal of
disablement. Recent Adv Commun Med 149-173, 1978
4. Colvez A, Blanchet M: Disability trends in the U S
population, 1966-1976. Am J Public Health 5:464-470,
5 . Liang M, Cullen K , Larson M: In search of a more
perfect mousetrap (health status or quality of life instrument). J Rheumatol 9:775-779, 1982
6. National Institutes of Health: Consensus Development
Conference: Total Hip Replacement. Conference Summary 4(4), 1982
7. Salvati E, Wilson P, Jolley M, Vikali F, Aglietti P,
Brown G: A ten year follow-up study of our first one
hundred consecutive Charnley total hip replacements. J
Bone Joint Surg 63A:753-767, 1981
8. Beckenbaugh R, Ilstrup D: Total hip arthroplasty: a
review of three hundred and thirty three cases with long
follow-up. J Bone Joint Surg 60A:306-313, 1978
9. Harris WH: Total joint replacement. N Engl J Med
297:650-65 1, 1977
10. National Center for Health Statistics: Detailed Diagnoses and Surgical Procedures for Patients Discharged
from Short-Stay Hospitals, United States, 1979.
USDHHS Pub. NO. (PHS) 82-1274-1, 1982
11. Sledge C: Indications for total hip joint replacement,
NIH Consensus Development Conference: Total Hip
Joint Replacement, Program Abstracts. National Institutes of Health, 1982, pp 3-12
12. Gerson EM: Quality of life. Am Sociol Rev 41:793-806,
13. Najman J, Levine S: Evaluating the impact of medical
care and technology on the quality of life: a review and
critique. SOCSci Med 15(F):107-1 IS, 1982
14. Taylor D: The costs of arthritis and the benefits of joint
replacement surgery. Proc K SOCLond [Hioll 192:145155, 1976
15. Kelsey J: Total hip replacement: epidemiology and
impact, NIH Consensus Development Conference: Total Hip Joint Replacement, Program Abstracts. National
Institutes of Health, 1982, pp 23-30
16. Harris WH: Traumatic arthritis of the hip after dislocation and acetabular fractures, treatment by mold arthroplasty: an end-result study using a new method of result
evaluation. J Bone Joint Surg 51A:737-755, 1969
17. Cannell C, Oksenberg L , Converse J: Experiments in
Interviewing Techniques: Field Experiments in Health
Reporting 1971-1977. Survey Research Center, Institute
for Social Research, University of Michigan, 1979
18. Coventry M: Surgery for arthritis of the hip. Bull Kheum
Dis 22:674-678, 1972
19. Fleiss J: Statistical Methods for Rates and Proportions.
New York, John Wiley and Sons, 1981
20. Berkowitz M, Johnson W, Murphy E: Public Policy
Toward Disability. New York, Praeger, 1976
21. Kaplan E, Meier P: Nonparametric estimation from
incomplete observations. J Am Stat Assoc 53:457-481,
22. Schecter E: Commitment to work and the self perception of disability. SOCSecur Bull 44:26-30, 1981
23. Pindych R, Rubinfeld D: Econometric Models and
Econometric Forecasts. New York, McGraw-Hill, 1976
24. Levitan S, ‘I’aggart R: Jobs for the Disabled. Baltimore,
Johns Hopkins Press, 1977
25. Harrell F: The logist procedure, Statistical Analysis
System: Supplemental Library Users Guide. Cary, NC,
SAS Institute Incorporated, 1980, pp 83-102
26. Sutherland C, Wilde A, Borden L, Marks K: A ten year
follow-up of one hundred consecutive Mueller curvedstem total hip replacement atthroplasties. J Bone Joint
Surg 64A:971-982, 1982
27. Murray WR: Results in patients with total hip replacement arthroplasty. Clin Orthop 95:80-90, 1973
28. Hammermeister KE, DeRoucn TA, English iMT, Dodge
HT: Effect of surgical versus medical therapy on return
to work in patients with coronary artery disease. Am J
Cardiol 44: 105-1 11, 1979
29. Yelin E, Nevitt M, Epstein W: Toward an epidemiology
of work disability. Milbank Mem Fund Q 58:384-415,
30. Berkowitz M: The search for a health variable, Policy
Analysis with Social Security Research Files. Office of
Research and Statistics, Social Security Administration,
USDHEW Pub. NO. (SSA) 79-1 1808, 1978, pp 41 1-432
31. Levitan S, Taggart R: Rehabilitation, employment and
42 1
the disabled, Alternatives in Rehabilitating the Handicapped: A Policy Analysis. Edited by J Kubin. New
York, Human Sciences Press, 1982, pp 89-150
32. Howards I, Brehm H, Nagi S: Disability: From Social
Problem to Government Program. New York, Praeger,
33. Chandler H, Reineck T, Wixson R, McCarthy J: Total
hip replacement in patients younger than thirty years
old: a five year follow-up study. J Bone Joint Surg
63Az753-767, 1981
Без категории
Размер файла
998 Кб
hip, tota, work, disability, arthroplasty
Пожаловаться на содержимое документа